PDA

View Full Version : OT: Solid state winch control



jacampb2
07-14-2007, 07:47 AM
I am interested in building a solid state control for my Warn 8274 self recovery winch. The current control scheme uses 4 constant duty 200 amp solenoids. 2 switched in parallel to connect the field 1 terminal on the motor to the armature for power in, and the other two in parallel to connect field 2 to the armature for power out. The solenoids are quite expensive, and during prolonged use, will occasionally weld themselves closed, and require replacement. Recently several winch manufacturers have began offering solid state m.o.s.f.e.t. switching controls to run the winch, instead of solenoids. This looks to me to be the perfect solution, with no moving parts to get screwed up by the harsh conditions.

The catch is, the winch will draw upwards of 500 amps at full load. Designing a power supply to handle this kind of current is beyond my abilities. What I am wondering, is if any of you know how I would go about doing this, or if there are any online sites that may have similar project instructions that would get me heading in the right direction?

Bellow is the wiring diagram for the solenoid control:

http://www.rollmeover.com/bronco_fab/odds_n_ends/8274_wiring.jpg

Any thoughts or information is of course appreciated.

Thanks,
Jason

Evan
07-14-2007, 08:39 AM
Designing a power control circuit for that sort of amperage isn't for the casual pretend engineer, like myself. It could become very expensive very quickly as the devices blow up for no apparent reason. I doubt that you will find plans for such a controller.

jacampb2
07-14-2007, 09:06 AM
Ah, but where is the fun in life if something doesn't blow up at least occasionaly :D

Evan
07-14-2007, 09:12 AM
Ah, but where is the fun in life if something doesn't blow up at least occasionaly :D

Agreed, but I prefer it to be intentional.

jacampb2
07-14-2007, 09:42 PM
So, Evan, what do you think I am going to end up blowing up, the control circuitry, or do you think I will do damage to the winch itself.

I think what I need to design is a standard H-Bridge drive, with appropriately sized mosfets. I found some power mosfets on digikey, rated for 160A @ 30 VDC that might work out nicely. They still are a bit small for what I need, can mosfets be wired in parallel for more current capacity?

I am not so much worried about my controls going out in a spectacular puff of smoke. The components themselves are not going to amount to that much money. A replacement winch motor is, however, quite costly. I just dont see how the control circuit could fubar the motor.

If any of you can help me along, I am not going to hold anyone responsible if I burn it to the ground. I just need to know if I am heading in the right direction. Do you think the H-bridge will work if I can source the components appropriately? Here is a link to the power mosfet I found on digikey: digikey link (http://www.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=497-3252-1-ND)

If you all definitely think this should not me a amateur project, for whatever reason, are their any commercial controllers that you can think of that I should investigate. The control pack from Warn goes for about $500, which seems mildly excessive for maybe $100 worth of electronics. Gotta pay for that R&D though I suppose.

Thanks,
Jason

Dawai
07-14-2007, 10:42 PM
My cheap winch, draws 450 amps.

Under full load, most winch loads are only minor tension wheeling in and out cable.

my thoughts would be to find a electric golf cart or hyster control..

BadDog
07-14-2007, 10:51 PM
I don't know about that, my 9000hs routinely has (had) over 5k fully dangling from it, taking my buggy up vertical walls from 15-30+ feet, with the tires just barely against the wall only because of the geometry of the cable going over the lip. And I have also taken my buggy to the top, anchored mine at the top, and pulled several other rigs over, one after another. I would love to have solid state, but it would have to be able to sustain the high amp loads for 5+ minutes continuous at a minumum.

jacampb2
07-14-2007, 11:03 PM
I'm with you baddog, my winch routinely sees hard pulls.
That is why my solenoids fail. They weld shut due to carrying high current for too long. I typically get 1 year out of the constant duty solenoids. Obviously they are mechanical devices, and subjected to some pretty severe environments, but the root cause of failure is, I believe, prolonged high current use.

The more I am looking into this, the more I think it is doable. It is going to require a few power mosfets in parallel for each portion of the h-bridge. If overheating is still a problem, even with a good heat sink, I can always build an enclosure in the passenger compartment of my buggy, and fan cool them. If I figure something out, I will be sure to detail my project on here.

Hopefully someone will chime in and tell me if I am on the right track so far.

Later,
Jason

J Tiers
07-15-2007, 12:17 AM
The winch I assume is always "on" unless it is "off", in other words it is not speed adjusted in any way.

Then, your major specs are only a few.

1) the max current rating at upper temp limit.

2) the "Rds(on)", which is the effective resistance when fully "enhanced" (turned on) at max temp. The high temp Rds(on) may be 2.5 times the 25C rating, sometimes more.

3) the gate-source voltage to fully turn on the device at maximum current and worst case temp etc.

4) the thermal resistance from junction to case.

the application of (1) is obvious

(2) affects the power dissipation, and hence the heating

(3) affects your drive circuit. You have two "high-side" switches, which may be a problem to keep turned on. You may need a boost power supply to run them.

(4) affects the heating and limits the amount you can skimp on your heatsink.

There are lots of very high current Mosfets around for 100V and below. I have used 45 amp units at higher voltage, larger currents are available for lower voltages. For the low voltages, you are better off with Mosfets.

You will need to handle the currents without trouble, but the dissipation can be 'cheated" depending on duty cycle.

I don't know how long the winch is to be "on" at a time, nor the percent "on" time overall. With a short "on" time, and low "duty percentage" the heatsink need only handle the "average" dissipation, and have enough mass to absorb the higher dissipation for the shorter "on" time.

A suitable anti-turn-off-spike diode is already in most Mosfets. Just be sure it has sufficient current rating. Some are NOT rated for as much as the Mosfet. That may or may not be an issue, the time of conduction is dependent on the stored energy in the rotor mass and inductance.

Also, a resistor-capacitor damper may reduce spiking. The power supply must have a local bypass to allow the Mosfet internal diodes to conduct spikes without a large voltage rise.

One other issue is the voltage drop thru the Mosfets. This must allow the motor sufficient voltage remaining. The solenoids presumably are essentially short circuits when closed. The Mosfet drive may also have a re-designed motor, capable of operating on slightly lower voltage.

sch
07-15-2007, 01:05 AM
One other concern is that normally, transient snubber
diodes across the coils assume power is applied in the
same direction, but the winch reverses polarity to reverse
direction, so the circuit would have to somehow flip the
polarity of the snubber diodes across the coils, electrically
by preference. Your suggested STM 160A FET has a
nominal R(ds) of about 3 milliohms, so device drop would
approach 0.5V, probably not a problem for the power supply
or motor. Four of these 160A FETs in parallel at $5 or so
each for each leg begins to add up... Some sort of debounce
and optical isolation would
be a good idea to separate power ckt from the control ckt.

jacampb2
07-15-2007, 01:48 AM
upon looking at my own wiring diagram some more, it appears that an H-bridge will not work. The motor in question has 3 wiring connections, field 1, field 2, and armature. The motor is grounded via it's case and the winches case to ground. I was mistaken in my initial description of the solenoid control scheme. It actually uses one solenoid to connect 12vdc to F1, and the opposing solenoid to connect F2 to the armature for one direction, and in the opposite direction, the solenoids connect 12vdc to F2 and F1 to the armature.

It looks like I need two separate unidirectional high side drivers, but, the problem I see with this is that current will be applied backwards to one of the drivers, while the other is running. Perhaps I can isolate them with diodes, I don't know. I will continue to research.

The other option, would be to hard wire the motor so it is unidirectional, and then proceed with a H-bridge design. There is a plethora of schematics and information for them online since it seems to be the driver of choice for the robot builders. The biggest problem with rewiring the motor, is that it will have to be electrically isolated from the case of the winch itself. Maybe I can disassemble the motor and remove the ground from the motors case and put in a fourth lug for ground. I will have to think about it more.

Keep the info coming though, I appreciate all of the help. I found some 490amp mosfets, about $18 each, however, it say that the case style is limited to 160 amps. It is in the same package/case style as the link I posted above. I also found some 120 amp darlingtons, which may be an option, and they were a bit cheaper.

Whatever I do, I will be trying to find used/surplus components if it is looking to get very expensive.

As for the question on how long the winch will have to run at any given time, I would say, worst case scenario is 10 minutes. My battery reserve will not go much longer than that at full current draw, and if I haven't rescued myself in 10 minutes, it is time to rethink what I am trying to do, and go at it from a different direction.

Later,
Jason

darryl
07-15-2007, 02:02 AM
When I was still repairing electronic stuff, I'd get car stereo amps in with blown mosfets. It became obvious to me that in many cases the problems were due to dragging down the power supply voltage (12 to 14 volts in vehicles) to the point where there wasn't enough voltage to fully enhance, or turn on, the mosfets. Under this condition, the rds rises, which means that more power is dissipated within the mos devices themselves. At the same time, less power is delivered to the amp circuit, and it then asks for more from the power supply, and current draw tries to go up.

If you're driving a motor through a controller, it's much the same. As less voltage becomes available to it, it slows down, develops less back emf, and behaves more like a short. I'm talking about a loaded down condition here, and a winch certainly is a heavy load.

My feeling is that if you're going to be loading the battery to the point where its voltage is dropping to 12 v or less (which is easy to do), then you should have a separately regulated voltage supply for the control circuitry. This voltage can be derived from the vehicles battery, but it would deliver a constant voltage of not less than maybe 14 volts, even if the battery voltage drops to 9 or 10 volts. Your high current mosfet arrays will live longer.

Dawai
07-15-2007, 02:27 AM
Years ago, I had a winch with a chainsaw motor on the side of it.
It'd snake a log up a mountain. No pictures sorry. I had large drum one laying in the shop with a hydraulic motor on it. I don't know where it has gotten to now thou. The new Ramsey I bought? it is chinese made, no doubt, it looks exactly the same as the ones HF sells. Plastic planetary gears in one end. I got to take one like it apart recently. After seeing that I hope I never really need it.

I got a old hand winch with a hitch-ball hole in a mount plate, it has dragged more cars out of bad places, I almost had to slam my buddy Doc to get it back from him after loaning it to him for a year or so.
Everytime I use it I seem to bleed. The cable is all fish-hooked up and with the mount the way it is it seems to pinch blood out every use. Them winch cable fish hooks are mean, kinda like a barb on a real hook, once they get into the flesh, you tear them out. It is laying on the porch right now after moving the big truck out of the shop.

Swarf&Sparks
07-15-2007, 02:28 AM
Have a look at
http://www.siliconchip.com.au/cms/A_105741/article.html
Hope this is some help
Lin

jacampb2
07-15-2007, 06:21 AM
Lin, thanks for the link. Those have been commercially available for a while too, I really have no need for a remote control though. I have my winch's controls hardwired into the cab of the buggy, as well as the plug in remote still. It is an idea I have toyed with though, buying a cheap RC car and using the bits to run relays to run the winch, but it is just something I have thought about. No real need for it IMO.

Anyhow, I have one last question for the night. I am now looking at this HEXfet as my number one choice: IRF13245-7PPBF-ND link (http://www.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=IRF1324S-7PPBF-ND). It is rated for 24VDC, 429 Amps, costs $5, and capable of operating at temps of up to 175* C. Anyhow, my question is this. The data sheet says that the packaging limitation for current handeling is 160 Amps. Does this mean, that this is the maximum current that it can handle while relying on the FET alone to dissipate the heat, i.e. if I have a robust heat sink, and can keep it within normal operating temperatures, can I use it to switch the full 429 amps which it is rated for?

Thanks again guys.

Jason

sch
07-15-2007, 09:39 AM
Looking at the spec sheet I would say it is a combo
of thermal and electrical. The chip die is large enough
for 400+ amps but even with the 5 source leads, the
package size is such you are trying to push 30+ amps
out each of those leads. Then on the Drain side you
have a plate with a slot to mount the device to the
heat sink, not a very big connector for 160A is it,
much less 400+. Mention is made of 3mm mount screw
but it is not clear where this is. At 160A, device will dissipate in
neighborhood of 40 watts, for a nominal (ideal case)
junction temp rise of 20D C, not too bad.
24Vds is pushing it a bit and means transient control
will need to be even more rigorous.

J Tiers
07-15-2007, 10:26 AM
1) I don't see why the H-bridge will not work.......that seems to be how the solenoids are set up. H-bridge is a very standard setup.

2) The motor seems (as is normal for hoists, etc) to be a series motor. Therefore it is not possible nor is is desirable to separate the armature and field.

3) 24V vs 12V for the IR part is a possible problem, but NOT due to local transients. LOCAL transients can be adequately handled by the addition of one large diode across the armature.
The main problem is that 24V is low for automotive "load dump" transients in the general electrical system. A mosfet of the size required will "probably" have an avalanche rating sufficient to handle it, and the H-bridge doubles the rating to 48V inherently, which is better. 100V is best.

4) The IR data sheet indicates a "real" rating of 160A almost certainly due to the leads and solder connections. No way can you carry full table rating on those leads without trouble.

5) The issue of drive power I alluded to in an earlier post. You need a way to provide constant drive power to the UPPER gates, and a "bootstrap" means will not reliably do it, unless you KNOW EXACTLY how long you will EVER need to have the winch "on".
You will need a power supply giving about +24V. A simple "charge pump" type will do that nicely, since there is very little current involved. You are not switching at a high rate, so the gate charging current will not be an issue. Looking at fig 6, it appears that 10V min will ensure full enhancement on that particular part.

6) When looking at the Rds(on) DO NOT BELIEVE THE SPEC SHEET RAW FIGURE. IT IS USUALLY "FALSE" (at least it does not reflect your reality).
The real figure for Rds(on) is found when you correct for temperature. observe figure 4, which shows the variation with temperature at 160A. In the case of the IR part, it isn't actually too bad, only 2:1 over the full range shown. The range may be larger for some devices.

7) snubber diodes.... They will always be in the correct direction with the H-bridge, that should be no concern. They will also clamp positive transients of the armature. But they will NOT clamp the expected negative transients on the armature terminal, which will need an added high current fast diode across it to limit negative transients.

8) The original circuit has a problem if you have 200A solenoid relays and 500A draw.
They will CARRY 500A, but they may not BREAK 500A regularly without welding shut or arcing badly. There are no snubbers shown to reduce arcing, which suggests that the inductive load may be also contributing towards killing the contacts.
Possibly the designers assumed that you would not shut off or turn on into max load..... if so they were very "optimistic".......

9) The transients you need to consider for Mosfet control are significant. I do not know the inductance of the field or armature. However, the energy involved will be related to the inductance x the current squared. That energy will have to be handled somehow, probably simply by the diodes. But it means that the armature diode will need to be rated similarly to the mosfets.
Since this is a DC motor, there will be some commutation transients, probably low, but maybe not so low at high current. So don't skimp on the rating of that diode.
If not given a "place to go" the transients will rise to whatever voltage is required to get rid of the "volt-seconds" in the inductance. So the power supply will need local bypassing with a capacitor sufficient to absorb the charge without an excessive voltage rise.

10) Your figure of 10 min "on" time continuous is essentially "forever" as far as any thermal or upper gate drive issues are concerned. You will have to figure on 'constant" duty for heatsinking,, power supply, etc.

fasto
07-15-2007, 01:45 PM
Don't negelect the turn-on current of the DC motor. A stopped DC motor looks like a short circuit, until it spins up enough to generate some back EMF. You will be shorting out the battery through the mosfet with the only impedance coming from the resistance of the circuit for a few hundred microseconds. If, indeed, you are pulling 500A at full load that implies the circuit resistance is really small (like 24 milliohms - neglecting the motor's back emf contribution). It may be necessary to add a sizeable inductor to keep the turn-on transient from far exceeding the mosfet's current rating.

BTW, I suspect that this turn-on transient is responsible for the welding of the existing contactors.
--
Aaron

jacampb2
07-15-2007, 09:24 PM
Thank you guys, I am getting further in understanding what I need to do. I intend to draw up my proposed drive in Photodraw tonight, and would appreciate it if you guys will take a look and tell me what my design flaws are.

I am thinking that 10 of the IR HexFet's will do the trick 2 each for power out side of the bridge and 3 each for power in. My thinking is that under power out, there is basically no load on the motor. The current draw of the motor w/ no load is 55 amps. The price of the FETs will indeed add up quickly, but 10 of them is still less than the price of 1 continuous duty solenoid.

I am still having a bit of trouble figuring out how current control scheme works. The motor is indeed a series wound motor, HP rating for 4.6HP. The way I understood it is that when the control is engaged in one direction, current flows from bus->solenoid->field 1->field 2->solenoid->armature->ground, and for the opposite direction current flows from Bus->Solenoid->Field 2->Field 1->Solenoid->Armature->ground. So basically the polarity of the field is reversed to reverse direction. Maybe I am very confused, I am going to look at it some more.

As for the current control design, I think it is more a case of this is the way it has been done for 50 years or so, why re-engineer. Full load through the controls is a worst case scenario, and I would imagine that if they are rated for 200 amps continuous, then they probably have a much higher intermittent rating. I think the major problem is breaking the connection while under any serious load, being mechanical switches, every time the contacts arc is one step closer to welding them closed.

The other issue, is that the more my sport advances, the more capable the vehicles become, and the easier it is to put them in a more perilous position. Then we ask more from the winch, then was asked when the solenoid controls first became the norm. My winch was built in the early '70s, at the time, I think you would be hard pressed to find a vehicle electrical system capable of supporting the winches current draw for a long enough period of time to put the controls in danger. The vehicle it is in now has two 950 CA deep cycle batteries and a 140a alternator. I believe I can now pull harder for a longer period of time than could have been done when the winch was designed. The problem is, even as the vehicles and winches advanced, the controls did not. It is very recent, w/ in the last 2 years, that the mosfet controls have been available for our application.

Thanks for all of the info so far, stay tuned for more questions, and hopefully a preliminary circuit design.

Thanks,
Jason

J Tiers
07-15-2007, 10:45 PM
I am thinking that 10 of the IR HexFet's will do the trick 2 each for power out side of the bridge and 3 each for power in. My thinking is that under power out, there is basically no load on the motor. The current draw of the motor w/ no load is 55 amps. The price of the FETs will indeed add up quickly, but 10 of them is still less than the price of 1 continuous duty solenoid.



The 160A ones will give you 480A capability at 3 in a set. I tend to agree with you on the directions, you can't 'push a rope". You could maybe even go to 1 and 4.

As with the 12V battery, the terminal voltage is likely to be 11 V or so max, even with 950A CCA, the resistance needs to be VERY low overall to get there. Assuming 10V, for ease of mental calculation, and 500A draw, that is a total resistance of no more than 20 milliohms. Ain't much, and that includes the battery, all cables in series, all FETS in series, the motor, including back EMF (if any).

I'd need to go look at some wire tables, but I'm dubious about that. DC current is hard to measure. The "lay it by the wire" meters are junk. I DO have a set of Columbia Electric clamp-on AC/DC meters that go up to 500A, but they are the only ones I ever saw (old analog ones). So meter readings are dubious at best.

My guess is that the 500A is probably nearly the highest "stalled winch" draw likely.





I am still having a bit of trouble figuring out how current control scheme works. The motor is indeed a series wound motor, HP rating for 4.6HP. The way I understood it is that when the control is engaged in one direction, current flows from bus->solenoid->field 1->field 2->solenoid->armature->ground, and for the opposite direction current flows from Bus->Solenoid->Field 2->Field 1->Solenoid->Armature->ground. So basically the polarity of the field is reversed to reverse direction. Maybe I am very confused, I am going to look at it some more.


Nope, that is how you reverse a series motor, reverse either the field or armature, but not both.

The H-bridge will be fine.

The only caveat is that the voltage across the armature will SUBTRACT from the gate voltage of the low-side Mosfets if they are driven from a ground referenced source. You need either to compensate for that, OR drive from a "floating" source directly from gate to source of the low-side Mosfets.




As for the current control design, I think it is more a case of this is the way it has been done for 50 years or so, why re-engineer. Full load through the controls is a worst case scenario, and I would imagine that if they are rated for 200 amps continuous, then they probably have a much higher intermittent rating. I think the major problem is breaking the connection while under any serious load, being mechanical switches, every time the contacts arc is one step closer to welding them closed.

That would be correct. The arc is what kills them, eventually it either welds them directly, or increases the closed resistance until they weld from conduction heating.

As for teh design, ONLY solenoids were capable of 500A switching until reasonably recently. At least at reasonable commercial cost.




The other issue, is that the more my sport advances, the more capable the vehicles become, and the easier it is to put them in a more perilous position. Then we ask more from the winch, then was asked when the solenoid controls first became the norm. My winch was built in the early '70s, at the time, I think you would be hard pressed to find a vehicle electrical system capable of supporting the winches current draw for a long enough period of time to put the controls in danger.

Not for Mosfets. The time to overheat may be as low as microseconds (internal heating) or low numbers of seconds for poor heatsinking or poor conduction away from the case.... i.e. what is called "effective thermal resistance from case to ambient".

Then also, if it is 120F in the desert, AND controls are under the hood, the temp may be so hot already that you are close to 100 deg C before you even turn it on. Under those conditions, you need very considerable heatsink mass to avoid overheating. You almost cannot get rid of the actual heat as fast as it is input, unless you go to ridiculous extremes.

See below



The vehicle it is in now has two 950 CA deep cycle batteries and a 140a alternator. I believe I can now pull harder for a longer period of time than could have been done when the winch was designed. The problem is, even as the vehicles and winches advanced, the controls did not. It is very recent, w/ in the last 2 years, that the mosfet controls have been available for our application.

Thanks for all of the info so far, stay tuned for more questions, and hopefully a preliminary circuit design.

Thanks,
Jason

The OTHER value of MORE parallelled mosfets is lower net Rds(on). THAT reduces the heat load, and is one of the things that makes a Mil-spec electronic device so much more expensive than commercial.

If the AMBIENT is close to 100 deg C, and the MAX JUNCTION temperature is 150C, you must have very low dissipation per device to avoid instant overheating.

Schlumberger well loggers operate in the heat. They at least used to assume 220 C junction for their Mosfets (rated 150C max). They also only wanted a few hours life, they toss or refurbish tech logger when done. But, they could put down another, at some cost, of course.

When you get "off data sheet", there is NO guarantee. There may be voids in the solder holding the chip, or any other sort of issue that makes the local temp higher or the failure point lower than you think. The guaranteed specs usually allow for those things.

YOU need guaranteed working, and no nonsense, or your vehicle may be lost. So over-design the heatsink and number of devices to lower your max junction temp.

A failure may keep the winch running until it runs down. That may mean it rips the hook off the wire, and becomes useless for further pulls, as well as possibly running away, or dropping the load (you and the vehicle).

The solenoids at least usually fail one at a time. Something that kills a Mosfet may kill more than one, and they fail shorted nearly every time, until they burn open.

That reminds me..... don't rely on PWB traces to carry that current. You need wires. Older Trace inverters used individual 10GA wires to each 25A mosfet in parallel. That is the sort of design feature you need.

CCWKen
07-15-2007, 11:32 PM
David had the best idea--Use a golf cart controller. 24-48v but it's ready made and can handle 1000 amps.

http://cgi.ebay.com/Navitas-Golf-Cart-Motor-Controller-DSE1000-48-USED_W0QQitemZ260138530752QQihZ016QQcategoryZ75208 QQcmdZViewItem

jacampb2
07-16-2007, 12:12 AM
Thanks for that link, if it stays cheap enough, I may buy it just to take it apart, but I looked up the datasheet for the controller on the navitas web site, and it still uses solenoids to switch motor direction. I did look into golf cart controls before deciding to try to do this, but 99% of them use a solenoid control to reverse, and the controller is basically a PWM controller for speed. Golf cart solenoids are an option for control if this doesn't work out. They make some very large solenoids for some of those critters, and from what I have found, they are more reasonably priced as well.

I have the H-bridge almost done, downloaded deign works lite, pretty slick and intuative software.

Later,
Jason

J Tiers
07-16-2007, 01:15 AM
That Navitas controller has problems for your application......

1) it is NOT 1000A, it is 325A continuous, read the specs. For these purposes, teh continuous rating is almost certainly the ruling spec, given the 10 min "on" time suggested by the original poster. Even 5 min might easily be 'continuous" for their purposes.

2) As it is designed for 24 to 48 V, the internal voltage drop may be considerably higher than a 12V controller, and will reduce motor power.

3) Since it is 24 to 48 V, the power supply may not operate the control correctly at your 12V. I would be very surprised if it did, in fact. Very substantial modifications might be required to make it operate at 12V input.


your idea of golf cart solenoids is quite direct and sensible. The only difficulty I can see, and it's a big one, is if they are set up for 24V to 48V coil voltage. Then they probably won't pull in on 12V.

But you could provide another 12V "boost" battery for the lower coil current, and be quite confident it would last your 10 min.

jacampb2
07-16-2007, 01:20 AM
Okay, here is what I am thinking for the H-bridge. It seems to me like it should work, I may be way off base, because this is my first endeavor short of fooling around with things.

http://www.rollmeover.com/bronco_fab/odds_n_ends/h_drive_winch.jpg

I am a bit unsure of the diode inline w/ the armature. It is a Schottky
barrier rectifier rated at 400Amps, 100 VDC. Data sheet here: link (http://rocky.digikey.com/WebLib/On-Semi/Web%20Data/MBRP400100CTL-D.pdf)

Also, I think I am on the right track for wiring from the bridge and through the rectifier->armature->ground but if not, let me know, this just made sense to me.

I am unsure of how to drive the Fet's as of yet, but if I am understanding correctly, you are telling me that I can design a charge pump to fully enhance the gate and I can basically turn it on and off via a SPST switch. From what I was reading online, it was looking like I needed a seperate mosfet driver to drive the power mosfets, however, I don't need it to do anything fancy, or be logic controlled or anything. Can I just power the gate with said charge pump and turn the power FETs on?

Thanks for all of the help so far. This is really looking doable. I am sure there will be a few expensive, and spectacular puffs of smoke, but I am willing to keep trying until I get it. Even if it costs me the same as buying the commercially produced controller from Warn in the end, at least I will have learned how to build it.

Later,
Jason

J Tiers
07-16-2007, 01:43 AM
You have a good start......... A few details........

1) the diode D1 should be in PARALLEL with the armature, and must be "poled" so as NOT to conduct unless there is an "inductive spike" that goes negative. That will happen when turning off the winch.

In other words, it should have the "bar" end at the bottom of the h-bridge, as you show, BUT the base of the arrow at ground, pointing reverse from what you show, and NOT in series. Should be rated at over 25V, for safety.

2) The "upper" gate drive really cannot be in parallel with the lower, as you show. The lower will be maybe 15V above ground when 'on", but the upper must be at 12V + 'enhancement" voltage, or about 25V. That would put the lower gate in danger, since they are normally rated at no more than 30V, gate-to-source, and may be 20V.

You need a "floating" gate drive for the upper switches, that goes 10V over the voltage at the SOURCES of the upper Mosfets, which will be at about 12V.

Essentially, you need at least 3 gate drive power supplies.

#1 handles both lower mosfet "banks", supplying the +15V or so that is required to keep them "on". You need to know the voltage drop in the armature to know exactly what voltage you need. There will be two separate gate drive circuits to drive the two lower "banks".

If the armature voltage is too high, you will need two separate supplies, and you will need to 'float" the gate drives to avoid possibly exceeding the gate voltage rating in fault conditions, or transients.

#2 and #3 are independent, and each handle a "bank" of upper Mosfets. Each has an associated gate drive, which as you have surmised, are each linked "logically" with the opposite lower bank, although they are at different voltages.

The good news is that all you require is a gate drive that will reasonably quickly raise the gates to fully "on". That requires a bit of charge for each Mosfet, which you can determine from the graph (I believe it is Fig 6) of gate charge vs voltage.

Since 'reasonably fast" may be in milliseconds, not nanoseconds as with PWM setups, all you really need is a fairly low impedance drive that will stay at the high state as long as needed, and shut off to below about 0.5 volts when commanded to.

Some CMOS gates will do that, old style "B" series inverters will do it and draw little current on their own. An "open collector" output would also do it.

It looks like a drive capable of supplying about 1 mA per device at 6V output would do fine. That would be 3mA total for the larger banks, or driving from 15V, about 3000 ohms in an "open collector" circuit. Those Mosfets have rather low gate charge requirements for high current devices, which is natural at their low working voltages.

jacampb2
07-16-2007, 02:13 AM
Sounds good, I will be going back to the drawing board in a little bit. I am working 12hr midnight shifts right now, so this is a good distraction from the grind.

I will start researching power supplies as well, and see what I can come up with. Thanks for all of your help J, I wouldn't be this far along with out you.

As for some of the other comments you made in a previous reply, I have been meaning to mention, I intend to mount the mosfets to solid copper buss, I am thinking either a mechanical connection, sandwiching the source leads between the buss and another copper plate, or soldering and clamping them. Also, as for a run away control, I will probably mount the controller in the cab, where it will be somewhat protected from the elements, and use a large manual battery disconnect switch in the lead from the bridge to the armature to break the connection in case of serious problems. The winch has a built in brake capable of holding the full rated load of the winch, so suddenly loosing power should not cause issues. Until the new controls have been field proven for a while, I will probably carry my solenoid pack as a back up control source.

Thank you for all of your help.

Jason

jacampb2
07-16-2007, 07:05 AM
Well, here is the revise H-bridge.

http://www.rollmeover.com/bronco_fab/odds_n_ends/h_drive_winch_rv1.jpg

I have been looking at power supplies, and am significantly confused. I actually ordered a book tonight that looks promising, geared toward robot builders, who knows, maybe I have found a new hobby... The book is Intermediate Robot Design, by David Cook, and it looks like it has a good section on motor controls and mosfet driver circuits. We will see how it turns out. It looks like it is written on a "robots for dummies" kind of level, so it should be easy to follow. Reading the technical documents from various semiconductor manufacturers has my head spinning.

Is there any reason I can't use a mosfet driver like IR2112 to drive each half of the bridge? It looks like it may be designed for bootstrap applications, but I also found a circuit described somewhere that uses a 555 timer to keep the bootstrap cap charged in constant duty. Feasible?

I am out for the night, got to get home and get some sleep soon.

Later,
Jason

J Tiers
07-16-2007, 10:20 AM
Looks much better.

As for the general setup of the drive and power supply, maybe this added block diagram may be helpful. I only added one, since the other is similar, and I did it in the windows 'paint" program, which is not the nicest schematic program......

http://img.photobucket.com/albums/0803/jstanley/drive.jpg

The drive is "floating", referenced to the sources of the upper Mosfets.

Some form of "level shifting" is needed inside the drive circuit to get the ground referenced "logic" signal (the "off" or "on" signal) to be referenced to the sources, which are 'off ground". An "optocoupler" can be used, but there are simpler methods that may be OK at these low voltages and low speeds.

The entire 'drive" COULD be as simple as a big multi-pole switch, although I do not recommend that.

Regardless of how the drive gets done, you still need a power supply that has an output that is also "floating" so that it can supply voltage to the gates of the upper Mosfets. But it's input is the normal +12 versus ground that you have available in the vehicle.

There are a number of different ways to do that.

Your robot book may give some of them. But, watch out, the robot folks may have variable speed "PWM" circuits. You do not need that, apparently.

because of that, your gate drive can be a lot simpler than the typical PWM gate drive, which must switch the mosfet in nanoseconds or microseconds. That speed takes a lot of current, and you do not need it for a simple 'on-off" drive.

One other detail.

Each mosfet should have a small resistor in series with the gate, located VERY CLOSE to the gate terminal. About 47 ohms should suit your needs. This resistor is to prevent high frequency oscillations which might otherwise occur.

jacampb2
07-17-2007, 01:48 AM
Thanks again, I have done more research, and revised my circuit again. Here is what I came up with.

http://www.rollmeover.com/bronco_fab/odds_n_ends/h_bridge_winch_w_driver.jpg

**Edit for a closeup of the bridge driver**

http://www.rollmeover.com/bronco_fab/odds_n_ends/h_bridge_driver_closeup.jpg

The block diagram for HIP4082, the full bridge driver is here:

http://www.rollmeover.com/bronco_fab/odds_n_ends/HIP4082_block_d.jpg

I took some of my cues from the manufactures spec sheet, the chip has built in logic that protects against a lot of possible bridge failures. It looks like in order to use this chip, I only need to provide it w/ the bootstrap caps and diodes, and a floating 12vdc supply.

Am I still heading in the right direction?

Thanks,
Jason

jacampb2
07-17-2007, 06:58 AM
Here is a picture of the bootstrap charge pump circuit I found online, claims to be able to keep the boot strap cap charged indefinitely, keeping the high side gates enhanced w/ 100% duty cycle.

http://www.rollmeover.com/bronco_fab/odds_n_ends/bootstrap_charge_pump.jpg

I am working on revising my circuit w/ it included, it will be a while though, I just downloaded Pspice and am trying to learn how to use it.

later,
Jason

Bruce Griffing
07-17-2007, 09:31 AM
I would consider paralleling more smaller FETS. This would be cheaper and give you more margin - though it would be more of a packaging challenge.

Here is an example:

http://www.bgmicro.com/index.asp?PageAction=VIEWPROD&ProdID=9588

Ten of these would give you a 0.0022 ohm on resistance and 1500watts max dissipation - a 3x margin at 500 amps. The board and heat sink design would need to be well thought out, but doable.

I would also consider using a separate battery to supply the gate drive. It would draw very little current and give you less problems during battery voltage drops at start up.

Evan
07-17-2007, 10:10 AM
To give another perspective on the problem there is the example of my 1959 Land Rover. The starter solenoid on the Rover cannot fail as it doesn't have one. Instead it has a manually operated push button switch that handles the entire starting current.

What I am getting at is that an appropriately rated switch such as a knife switch could handle that duty and wouldn't wear out or be susceptible to the failures a solid state unit can suffer. Not very sexy but it works.

J Tiers
07-17-2007, 10:14 AM
1)
The HIP4082 part, if still readily available, would be OK. I have used the similar HIP4080 part in commercial products before, although it started to become hard to get, at least in the "thru-hole" package we were using. In SMT it may be available OK.

2)
I think you will need to drive all the logic inputs, though. You show connecting AHI and BHI to +12, i.e. "strapping" them "on".

What you need to do is drive AHI and BLI together, BHI and ALI together. That will turn on and off the opposite "corners" of the bridge together.

3)
For the "floating" drive supplies using that chip, it can get simpler.

You can pretty much IGNORE the "bootstrap" setup. it won't do much for you.

What you can easily do is to set up ONE charge-pump or other boost supply that will give you about +30V.

Then, put a 12V zener diode from AHB to AHS, and one from BHB to BHS.

A resistor from the +30 to AHB and another from +30 to BHB, and you have a constant drive supply. The resistors need only supply the quiescent drive circuit current, which will be shown on the data sheet. The zeners avoid excess drive voltage. (the resistors must supply that current at the minimum voltage across them, which occurs when the upper drive is "on")

You keep the capacitors that you show from AHB/BHB to AHS/BHS

I say +30, because that gives you some margin over the required high drive output voltage and should keep the resistor value reasonable.

4)
The charge pump circuit looks to be a special-purpose one, most of the parts shown are irrelevant to you.

You should be able to use a standard LM555 timer (similar to their 7555) without the other chips they show. National Semiconductor should have some applications data you can use. if not, and you can't find one, I can help you.

5)
One condition to observe is that when NO Mosfet is "on", neither the bootstrap NOR the circuit I suggested will keep a charge on the high side drive, because there is no current path to ground. That path is only established when a low-side Mosfet is "on".

I didn't look up to see if the chip has an undervoltage lock out on the high side drive. If not, or if the UVLO is at a low voltage, there will be a transient high dissipation condition while the high side voltage builds up through the dropping resisitor, AFTER the low side comes "on".

To avoid that condition for the non-bootstrap supply, a "leak" resistor can be put from one side of the field to ground. A value about the same as the resistor for either drive supply should be OK.

J Tiers
07-17-2007, 10:17 AM
To give another perspective on the problem there is the example of my 1959 Land Rover. The starter solenoid on the Rover cannot fail as it doesn't have one. Instead it has a manually operated push button switch that handles the entire starting current.

What I am getting at is that an appropriately rated switch such as a knife switch could handle that duty and wouldn't wear out or be susceptible to the failures a solid state unit can suffer. Not very sexy but it works.

And using the heavy-duty golf cart solenoids is another brute-force approach that may be entirely appropriate.

The problem with switches is that they are no longer easy to find in 500A ratings. The days of mechanical control of large DC current has passed. I would expect the golf carts to rapidly move to SS control as well.

sch
07-17-2007, 11:05 PM
Seems like there ought to be a Schottky diode paralled across the field coil as well to suppress inductive spikes there when directions are reversed. Switching 300+ amps would generate a nice spike.

J Tiers
07-17-2007, 11:16 PM
Seems like there ought to be a Schottky diode paralled across the field coil as well to suppress inductive spikes there when directions are reversed. Switching 300+ amps would generate a nice spike.

Not needed. It's not obvious, but they are actually already there.

Mosfets have an "intrinsic" internal diode in parallel (IGBTs don't). In the specific parts he is using it is exceptionally fast for the current rating. IIRC is is about a 115 nS diode, which is fast enough for several hundred khz switching.

So there are 4 intrinsic diodes which will clamp any field spikes, regardless of polarity. The negative end of the field uses one intrinsic diode plus the one in parallel with the armature to clamp to ground. The positive end will clamp to the supply voltage. And, with the 4, it does not matter which way the field is poled.

That's why I mentioned earlier that the supply needs a good bypass cap to absorb spikes. Those spikes that are "steered" into the supply will need to be absorbed without allowing the transient voltage to rise too far.

The series combo of any two Mosfets will provide a clamp to the supply for any positive spikes from the armature, although the normal spike there should be negative-going.


NOTE:

I forgot to mention for the last schematic that there should be 47 ohm or so resistors in series with EACH gate, close to the mosfet. They are to suppress parasitic oscillations, and should not be forgotten.


NOTE again...

In case anyone is getting bored with what is nominally an OT subject here, recall that aside from the specific currents involved, this exact same circuit is ALSO a workable circuit for a DC drive, either for a spindle, or for a feed. So it is not as OT as it might be.

Else-wise I would have done it back-channel.

jacampb2
07-17-2007, 11:34 PM
Seems like there ought to be a Schottky diode paralled across the field coil as well to suppress inductive spikes there when directions are reversed. Switching 300+ amps would generate a nice spike.

I was kind of under the impression that the built in freewheel (correct term?) diodes on the Mosfets were supposed to take care of this. I am still learning as I go though, so if I am mistaken, let me know.

I am starting to get the hang of OrCad's Capture, the software that I guess is replacing Pspice. A bit more of a learning curve than the first design software I downloaded, but it appears to be much more robust.


1)
The HIP4082 part, if still readily available, would be OK. I have used the similar HIP4080 part in commercial products before, although it started to become hard to get, at least in the "thru-hole" package we were using. In SMT it may be available OK.

2)
I think you will need to drive all the logic inputs, though. You show connecting AHI and BHI to +12, i.e. "strapping" them "on".

What you need to do is drive AHI and BLI together, BHI and ALI together. That will turn on and off the opposite "corners" of the bridge together.

J Tiers, the chip is still readily available from Intersil. On paper, it looked like the perfect solution for me. As for strapping AHI and BHI together, from the truth tables in the tech sheet, it appears that w/o a low input, the high inputs are ignored, and when the BLI is 1, then BHI gets disabled, allowing you to drive the H-bridge in either direction w/ only 2 inputs. I will post a link to the spec sheet later after I finish messing with Capture for the night (I am not working midnights tonight).

I am kind of considering buying a surplus DC-DC laptop power supply to use as the floating power supply for the gate driver. Have not had a lot of luck finding any information on line for a way to "float" an ordinary supply. If I understand the concept correctly, I need a power supply that does not share a physical ground connection with the winch/vehicle. I think the easiest way that I could come up with to actually build it, is to invert the DC to AC and then use a transformer to isolate the supply and rectify the AC back to DC, but this seems to be a very round about way of going at it, and I would think a Laptop supply w/ charge pump for the high side would work well.


To give another perspective on the problem there is the example of my 1959 Land Rover. The starter solenoid on the Rover cannot fail as it doesn't have one. Instead it has a manually operated push button switch that handles the entire starting current.

What I am getting at is that an appropriately rated switch such as a knife switch could handle that duty and wouldn't wear out or be susceptible to the failures a solid state unit can suffer. Not very sexy but it works.

Evan, I know for a fact that I can design a electro-mechanical system that would work fine, and not suffer the frequent failure I currently have. The reason I am trying to do this w/ solid state components is that it presses my limits, and furthers my knowledge. Starting out I had a very basic understanding of FETs and such, but now, thanks to all the help I am getting on here, I am starting to understand how it actually works, and that is just as important to me as building something that will work correctly.

Thanks for all of the help so far. It will probably be tomorrow night before I have my updated designs up. I only slept about 3 hours today, and I am starting to loose my clarity of thought.

Later,
Jason

jacampb2
07-17-2007, 11:39 PM
NOTE:

I forgot to mention for the last schematic that there should be 47 ohm or so resistors in series with EACH gate, close to the mosfet. They are to suppress parasitic oscillations, and should not be forgotten.


NOTE again...

In case anyone is getting bored with what is nominally an OT subject here, recall that aside from the specific currents involved, this exact same circuit is ALSO a workable circuit for a DC drive, either for a spindle, or for a feed. So it is not as OT as it might be.

Else-wise I would have done it back-channel.


Not to worry, the resistors are in the new design, and yes, I did understand each gate needed them. From what I understand this is to prevent "ringing" correct? I have noticed them in most of the designs I looked at, and figured we would get there eventually.

On your Note again, Add PWM and you have bidirectional, and speed control. There is lots of neat information I have found online for PWM applications, feedback for rotational speed and what not. Almost makes me wish I have a need for it :)

You beat me to it on the diodes, I guess I am really starting to understand. Thanks for all of your help.

Jason

Evan
07-18-2007, 12:13 AM
The reason I am trying to do this w/ solid state components is that it presses my limits, and furthers my knowledge.

One of the best reasons there is, too. Buy a few extra MOSFETS though... ;)

J Tiers
07-18-2007, 12:33 AM
J Tiers, the chip is still readily available from Intersil. On paper, it looked like the perfect solution for me. As for strapping AHI and BHI together, from the truth tables in the tech sheet, it appears that w/o a low input, the high inputs are ignored, and when the BLI is 1, then BHI gets disabled, allowing you to drive the H-bridge in either direction w/ only 2 inputs. I will post a link to the spec sheet later after I finish messing with Capture for the night (I am not working midnights tonight).

If it does work like that, defaulting to a 'low/high" choice on a bridge, then you are correct. Then it will act pretty much the same as the "HIP4080A", but with 2 inputs instead of 1. At least one of the other chips in the series does NOT do that, so I questioned it.

However, that choice does not allow an "all off" mode, unless you use the "DIS" pin.



I am kind of considering buying a surplus DC-DC laptop power supply to use as the floating power supply for the gate driver. Have not had a lot of luck finding any information on line for a way to "float" an ordinary supply. If I understand the concept correctly, I need a power supply that does not share a physical ground connection with the winch/vehicle. I think the easiest way that I could come up with to actually build it, is to invert the DC to AC and then use a transformer to isolate the supply and rectify the AC back to DC, but this seems to be a very round about way of going at it, and I would think a Laptop supply w/ charge pump for the high side would work well.


Aside from one question, you maybe don't need to "float" the supply, the bootstrap supplies for the chip will do that.

The question is the voltage drop of the armature.

At this point, before you get excited about a power supply to run the gates, you need to know what the voltage from the armature pin to ground is in the running condition, preferably somewhat loaded.

That voltage determines what your choices are, and whether that chip will actually work well for you at all.

Aside from that, the entire gate drive setup will draw so little power that an actual laptop style power supply would probably be several hundred times overkill.

darryl
07-18-2007, 01:10 AM
Just a note on mounting the mosfets- depending on the type you will be using, it may be best to 'sandwich' them to the heatsink with a solid bar of aluminum across the tops, instead of using the mounting holes . The bar would have bolts through it into the heatsink, between the mos devices. One reason I suggest this is that it gives better contact between the power device and the heatsink, for best cooling ability. Couple more reasons- it eliminates the bolt insulator, which can be a weak link in maintaining tight contact between device and the heatsink- another is that it helps withdraw heat from the device package (the epoxy part) and still another reason is it helps to keep the package together under taxing conditions.

Usually when I see this done, there's a thin silicon rubber pad or similar on top of each device before the bar is bolted down over them. This helps to even out the pressure that's being applied to each device by the bar.

Another sensible reason for using this method is that it presses down on the mosfets right over the spot where the die is, the active part of the device. The die is where the heat is generated, and that is where you'd want the best contact to the heatsink (through the insulator if that is required), so that is where the pressure should be. Usually the mounting hole is off to one end of the metal tab, and the tab can actually lift off the heatsink where the die is if the normal mounting method is used. The heat then has to travel along the tab a bit before it can escape into the heatsink.

You're dealing with hundred of amps flowing on a relatively continuous basis, and you want the circuit to be as reliable as possible. Take a cue from some manufacturers of high power car stereo amps- some of which are using several tens of mos devices. This is how they do it.

Another way is a short bar across the top of each mosfet, with a bolt on either side of it. This way you tork each device down individually, and you know that it's going to be solid.

J Tiers
07-18-2007, 11:30 AM
Just a note on mounting the mosfets- depending on the type you will be using, it may be best to 'sandwich' them to the heatsink with a solid bar of aluminum across the tops, instead of using the mounting holes . The bar would have bolts through it into the heatsink, between the mos devices. One reason I suggest this is that it gives better contact between the power device and the heatsink, for best cooling ability. Couple more reasons- it eliminates the bolt insulator, which can be a weak link in maintaining tight contact between device and the heatsink- another is that it helps withdraw heat from the device package (the epoxy part) and still another reason is it helps to keep the package together under taxing conditions.


Another sensible reason for using this method is that it presses down on the mosfets right over the spot where the die is, the active part of the device.

The devices in question don't seem to have holes (d^2 package) so that's what he will have to do.

But there are good and bad ways of doing it. The simple way tends to put the force on the package edges, not right in the middle. The "end" devices tend to get squeezed like a watermelon seed, all on one side.

The best way is to have a discrete "bump" over each device to make sure pressure is centered.

In the case of the D^2 device, they may really expect you to solder down to a solid heatsink, as with smaller devices.

Wirecutter
07-18-2007, 02:55 PM
David had the best idea--Use a golf cart controller. 24-48v but it's ready made and can handle 1000 amps.


Having done a bit of research on this particular topic, I can tell you that you don't want this controller for what you're doing.

1. It won't do much on 12v other than complain about "low voltage".

2. Most controllers of this type can handle peak current for only about 30 seconds. That's not you.

3. Just about all golf cart motor controllers are potted in epoxy, so you'd get no educational value out of it.

4. I haven't done all the research on this particular model, but I do know that Navitas makes controllers that can do motor reversing without a contactor. (Usually these are for shunt-wound or separately excited motors, not series wound.) Also, most golf cart motor circuits have a contactor in them as a kind of safety, not just for reversing. I can also confirm that the "reversing contactors" are expensive, and are basically 2 single pole double throw relays strapped together.

That is, however, a seriously nice motor controller for what it's designed for. I doubt it will stay that cheap.

I realize that you're close to a circuit, but if you're interested, I can steer you in the direction of a nice 4000 amp disconnect switch if you need a safety backup. I use one on my newest gokart - it's the "Oh Spit!" switch.

-Mark

jacampb2
07-18-2007, 09:53 PM
Okay, I have made some more progress. Not much more, but I got the whole thing so far redesigned in Pspice. Here is the picture as it sits. I am read over the truth table again, and it looks like I was wrong, what I described was correct in that it would have controlled correctly, but it looks like according to the table, that with either ALI or BLI energized it will run one direction, if they are not energized and the AHI and BHI are left energized, then the driver automatically reverses, requiring you to use the Disable pin if you want to shut it down. I will run it the more traditional way.

http://www.rollmeover.com/bronco_fab/odds_n_ends/H_bridge_v3_low.jpg

Click here for a high resolution version of the image: Super Picture (http://www.rollmeover.com/bronco_fab/odds_n_ends/H_bridge_v3.jpg)

More questions for J in the next post.

Thanks,
Jason

jacampb2
07-18-2007, 10:14 PM
5)One condition to observe is that when NO Mosfet is "on", neither the bootstrap NOR the circuit I suggested will keep a charge on the high side drive, because there is no current path to ground. That path is only established when a low-side Mosfet is "on".

I didn't look up to see if the chip has an undervoltage lock out on the high side drive. If not, or if the UVLO is at a low voltage, there will be a transient high dissipation condition while the high side voltage builds up through the dropping resisitor, AFTER the low side comes "on".

To avoid that condition for the non-bootstrap supply, a "leak" resistor can be put from one side of the field to ground. A value about the same as the resistor for either drive supply should be OK.

The chip does have a UV lockout, if I am reading the correct specifications, it looks like typical rating is 6 volts referenced to BHS or AHS, low rating is 5 volts, high is 7 volts.

By a leak resistor on one side of the field, do you mean the motor field? If so, would it burn up as soon as the field was energized? Or, is the idea to get a resistor with a higher resistance than the fields resistance, so the current will travel through the field to ground being the path of less resistance?

Thanks for bearing with me, I am trying to learn as fast as possible; I don't want to just copy stuff down and not understand how and why it works.

Thanks,
Jason

J Tiers
07-18-2007, 10:14 PM
D4 is reversed, "bar" end should be +.

Make circuit for D5 conform to that for D4.

The idea is that the +30 offers a 'higher than needed" supply, and that the zener diodes limit the floating drive voltage to what the driver and gates can take.

It LOOKS (but is not) like the +30 could pull up the voltage to a damaging amount, but in fact that can't happen, unless the +12 is disconnected and +30 is still on.

Otherwise, the Q2 and Q6 intrinsic diodes will clamp AHS and BHS to the supply, and the zeners will limit the added voltage within chip limits (AHS or BHS +Vdd, i.e. AHS +16V max, or BHS +16V max).

We STILL need that armature voltage value...... This is a parameter that affects the usable approaches to the circuit, so we really need it before getting too much more into circuit design.

Another note. The chip shows +16V max for its power (Vdd). That means you may need to put a protective zener and resistor arrangement on the chip power supply (Vdd) to prevent charging voltage or transients from damaging the chip.


For later questions......

Yes, it seems that is the UVLO, but that is too low to enhance the Mosfets well, so the UVLO is not helpful there. The overall chip UVLO is 6.8 to 8.25V, and may be almost good enough. I'd be very tempted to put a voltage detector on teh "DIS" to hold the chip "off" until the +12 AND +30 were above say 10V and 24V at least, respectively.

With the chance that the +12 might drop, I'd also be very tempted to run the chip itself from a voltage derived from the +30 in any case, so that you KNOW the LOW SIDE is also going to be fully enhanced, and no overheating occurs.

Yes, the resistor would be large enough to avoid burning up. But, it may be that the clamping effect (mentioned above) to the +12V will actually keep them charged OK in the absence of a ground connection. I overlooked that effect, and the resistor may not be needed in reality.

One other decision you need to make is the priority.......

Is the thing going to be set up to be "protected", or is it going to be set up to run until it can't, no matter what, so that a safety circuit doesn't get you hung up in a bind. Sounds like the latter, like a ship....... no fuses in the firepump power lines........

And don't forget to get that *&^%$# armature voltage. if it is more than a couple volts, and I bet it is, then some more thinking may be needed.

jacampb2
07-18-2007, 10:29 PM
I should have the armature voltage drop for us tomorrow. I pulled the tarp off my buggy today and got it fired up. I am rebuilding the brake assembly on it, and have to get it back together before I can put a load on it and take some readings. The way I intended to do this, is just to pull my pickup up the driveway by the winch, and measure with my DVOM (it's not a real high quality one, unfortunately HF tig welding scrambled my fluke...) from the armature to ground while the winch is under load. I will also measure the voltage at the winch solenoid bus while under no load, the difference should give me the voltage drop, correct? I can put the winch under a serious load if you think it is necessary, I can actually fully stall it if I need to. I can't take any current ratings however. Will the above be sufficient.

Thanks,
Jason

J Tiers
07-18-2007, 10:33 PM
Should be.

We want the voltage that would be across D1, i.e. armature.

jacampb2
07-19-2007, 03:05 PM
Okay, I have some numbers. These are averages from multiple readings.

Voltage at the winch solenoid buss, no load: 12.53 Vdc

Voltage from Armature to ground, Rolling load, approximately 6000 Lb truck, rolling up a 3* incline: 10.51 VDC

Voltage from Armature to ground, Heavy load, above vehicle w/ parking brake locked, dragging in gravel up a 3* incline: 8.59 Vdc

**Edit** for reference, the above load nearly stalled the winch.

Voltage from Armature to ground with no load, spooling in empty cable: 12.33 Vdc.

All of the above readings were taken with the buggy running and the charging system operational.

I measured voltage at the buss with the buggy off and had 12.23 Vdc. It is rare that I would be doing hard pulls with the vehicle off, should I be concerned with voltage drop while the vehicle is off?

On a side note, while testing, I believe I saw some of the high voltage transients you all told me would be there. Between armature and ground, starting pulling with the above mentioned heavy load already on the cable and braked by the winch, I saw voltage spikes as high as 50 Vdc. They averaged in the range of 24 Vdc. When instantly reversing the winch from load to spool out, I saw transients in the neighborhood of 18-20 Vdc. Not sure if it is anything you needed to know about, but I learned something new again. I was surprised to see spikes that high.

Thanks,
Jason

J Tiers
07-19-2007, 03:23 PM
No time to reply now, but that is good data, and may indeed make some radical changes in the design necessary

Bruce Griffing
07-19-2007, 03:28 PM
I think your data reinforces my earlier suggestion of using a different voltage source (battery) for the gate drivers. It would make the design much simpler also.

hoof
07-19-2007, 03:31 PM
One thing that comes to my mind is the cheap and dirty Ford starter solenoid's. Have you seen the little cans with 4 bolt terminals sticking outthe side's. They are good for 500 A 12vdc. My favorite part is they cost about 7 bucks a piece. Good Luck with you endevour

jacampb2
07-19-2007, 04:55 PM
One thing that comes to my mind is the cheap and dirty Ford starter solenoid's. Have you seen the little cans with 4 bolt terminals sticking outthe side's. They are good for 500 A 12vdc. My favorite part is they cost about 7 bucks a piece. Good Luck with you endevour

Actually, that is the exact same style solenoid that the winch uses, except it uses a continuous duty version. The starter solenoids can only take that kind of current for less than a minute or so. I do typically carry starter solenoids as a backup, just in case I need to do a quick fix to get out of a bad situation. Starter solenoids are much, much cheaper than constant duty solenoids.

Later,
Jason

J Tiers
07-19-2007, 11:19 PM
OK.........

The reason I was harping on the armature voltage is that it is normally greater than the field voltage, and I suspected just that sort of issue. I was rather hoping for a better split, but so be it.

The problem now is that the "ground" or "common" for the driver chip is actually the armature voltage point, and it is WAY "off-ground".

A couple issues come from this.

1) it is off-chassis-ground, and that makes another hassle driving the chip from a ground-referenced source. It would be in that case easier to drive from a completely floating control, although that can be a hassle also, and makes for a problem with power supply.

2) Getting power to it is more difficult, although not a very big problem if the chip is supplied from the +30 supply. There is enough voltage compliance to take care of the issue.
Floating battery is another solution, but it is another battery you need, that might be not-charged when you need it, and that you can't do without. It also does not really make things THAT simple, since the floating battery STILL needs to hold a high side drive "indefinitely" at an "on" state at a voltage greater than its terminal voltage (unless you use 2 or more). That just "translates up" the issues present with using the original 12V vehicle battery.

3) Any transients on the armature voltage (and there will be some) will be "messing with" the chip ground and potentially the whole control circuit ground. That may cause mis-operation of the chip, and/or the controls by coupling in noise.

Bottom line is that using that chip and the winch motor as-is involves a number of potentially needed "work-arounds" that complicate the circuit.

Solutions:

1) Treat ALL FOUR of the mosfet drives as "floating", and use "high side driver" chips for ALL 4 of them. This means not using the chip you selected. It might possibly mean using NO chips, and instead a "roll your own" drive. But it does not really end up needing any more complicated external control system than the suggested chip would.

2) float the chip you selected with its ground at the armature voltage, and deal with whatever transient issues, power supply issues, and signal input/control issues that come up.

3) Check on the possibility of re-wiring the winch to not be hard-wired to ground on armature negative, reversing the relative positions of the armature and field.


Initial evaluation of solutions:

I suspect that #3 is impractical. Normally when the ground is made inside something like this, it is "designed-in" in such a way as to make it pretty obnoxious to change.

#2 is possible, but leads to possible added problems and needed "work-arounds". Unlikely to be simpler overall.

#1 is perfectly possible, puts the least strain on the +30 supply, and directly uses all parts in ways that they are designed to be used. I think I would tend to recommend this approach. It will allow a conventional control system operating from chassis ground, and uses one +30 (approx) supply for all gate drives (which can operate off the vehicle battery).
At that point, I'd also tend to look at a "roll-your-own" drive circuit, since for this application the chip-type high side drivers really only offer better speed and other features that you will not be using at all.

jacampb2
07-20-2007, 12:07 AM
Thanks, I will go back to the drawing board tomorrow. I really liked the idea of the driver chips, if I can find individual high side drivers for inexpensive enough, I may still go that route. But it doesn't look like it will be all that difficult to "roll my own" either.

I would rather not add an additional battery to run this thing. Aside from having to be battery operated, I also need a way to charge said battery, and it can't be coupled to the existing power supply to keep it charged. It could easily end up costing more than the whole project, just putting together some form of redundant charging system.

I am thinking #3 is unlikely as well. I have had this motor apart before, and I am pretty sure there was nothing easy about the removing the armature ground connection. Worse case scenario, I will tare it apart and try, but if avoidable I will avoid it.

Thanks again, I am going back to work for midnights again tomorrow night, so I should have some extra time to devote to this.

Jason

J Tiers
07-20-2007, 12:24 AM
I suppose one additional and potentially helpful feature that the high side driver chips offer is the ability to ignore transients.

After all, they are MADE to deal with transients, sometimes 500V worth of "transient" at the time of switching on or off.

jacampb2
07-21-2007, 12:41 AM
Okay, I have a new updated version of the design. Drivers are all IR2117, floating high side driver. Relatively inexpensive, and I can get small quantities. I think I set things up correctly. I am kind of busy tonight, but I am going to try to educate myself on diode, capacitor, and resistor selection to supply the chips and gate drives.

Here is the picture.

http://www.rollmeover.com/bronco_fab/odds_n_ends/H_bridge_v4_low.jpg

Super Size Picture, Here (http://www.rollmeover.com/bronco_fab/odds_n_ends/H_bridge_v4.jpg)

I will post more as I come up with it.

Thanks,
Jason

J Tiers
07-21-2007, 01:00 AM
Re-arrange the zeners per the top left one.............

http://img.photobucket.com/albums/0803/jstanley/H_bridge_v4.jpg

You are using them as voltage limiters.

jacampb2
07-21-2007, 01:20 AM
I was just reading about this, and thought I had it wrong again. Am I correct in that I want my zener breakdown voltage to be ~10 Vdc, the required voltage above supply to fully enhance the FETs?

Also, as for R11, 12, 13 & 14, I need them to supply the quiescent drive current for the gate power, or for the chip, I think I know how to figure it out, but the spec sheet for the driver lists a specification for Iqbs and Iqcc, Iqcc is quiescent current for the logic side of the chip, and Iqbs is for the gate drive side. I assume I need the current for the gate side, but I am unsure. Do I just use Ohm's law to find the resistance I need?

Thanks,
Jason

jacampb2
07-21-2007, 01:36 AM
Made the changes to the circuit. A refresh should show them in place of the originals.

J Tiers
07-21-2007, 11:09 AM
I was just reading about this, and thought I had it wrong again. Am I correct in that I want my zener breakdown voltage to be ~10 Vdc, the required voltage above supply to fully enhance the FETs?

Also, as for R11, 12, 13 & 14, I need them to supply the quiescent drive current for the gate power, or for the chip, I think I know how to figure it out, but the spec sheet for the driver lists a specification for Iqbs and Iqcc, Iqcc is quiescent current for the logic side of the chip, and Iqbs is for the gate drive side. I assume I need the current for the gate side, but I am unsure. Do I just use Ohm's law to find the resistance I need?

Thanks,
Jason

Zener should be enough larger than 10V that you can supply 10V to the gates for guaranteed full "on". The gate driver has some inherent voltage drop in most cases, and does not allow a swing to the full supply voltage.

And enough smaller than 16V so as never to exceed the chip supply limit.

The quiescent current is correct, supply that with a small excess to make sure.

The relevant current is the worst case gate drive current, and you need to supply it when you are in the "on" condition. That is when the difference between the "+30" and the sum of source voltage + gate drive supply voltage is least.

Then also you need to make sure that resistors and zeners will be OK for dissipation with the MOST voltage they will see on them.

That would be for the lower gate drives, with the drive "off" but energized. In that condition, the lower drives source voltage is ground thru the (un-energized) armature, and the +30 is on.

It does not appear that the upper gate drives have a similar condition. They should clamp to +12 if all Mosfets are off.

jacampb2
07-21-2007, 10:47 PM
I have been doing more research, and I think I have figured out the ratings for the components in the circuits to drive the HO on the chip. Better yet, I found an excellent web site that explains things very well, allaboutcircuits.com, and I finally understand how the zener/resistor arrangement works.

Here are what I have tentatively selected for these components:

R11,12,13,14= 5033ED100K0F12AF5, Metal Film, 100K, .5 Watt, +/- 1%

D2,3,4,5= 1N5352 15Vdc Zener, 5 Watt

C2,3,4,5= P10397, electrolytic, 50V, 100 uF

I figured the resistances using the data sheet for the IR2117, the spec sheet says that the quiescent Vbs supply current is a maximum of 240 uA. I calculated the resistance, based on 300 uA. .000300 = 30/R, yielding a result of 100K. I then calculate power dissipation required by P=.000300*15, where 15 v is the voltage dropped across the zener, assuming the 30 Vdc supply. P= .0045 watts, 4.5 mW. I selected the .5 watt resistor for a bit of a safety margin, and since they are readily available.

The Zeners selected are 15 VDC models, the IR2117 chip selected has a voltage range of 10-20 Vdc, so 15 volts should keep me well with in the limits of the chip. I was unsure of power dissipation for this component, so I went with the highest readily available wattage. I assume that the calculation should be based on the same current as the resistors above, but I was unsure. A higher wattage rating than necessary will not be detrimental, will it?

The capacitors are kind of a seat of the pants guess, the IR data sheets said that a minimum of 47 uF was required, and basically bigger is better to prevent overcharging. I am not 100% sure I understand what the caps are doing there w/o using a bootstrap supply. I understood that for a bootstrap supply, they are there to provide the necessary gate charge, are they serving the same purpose with a non boot strap supply?

If the above looks adequate so far, I am going to look at designing a voltage regulator to run the logic sides of the chip from the same supply as J Tiers suggested. My only question here, is I have read repeatedly that the logic side of the chips should use a separate floating power source from the gate side, mainly I believe because sharing the same ground as the high power components renders the noise immunity of the chip useless.

So, how an I doing so far? Am I on the right track calculating this stuff, or did I go at it all wrong?

**Edit** I just thought about this some more, and should I have calculated the resistance needed based on the voltage dropped by the Zener, and not based on supply voltage? I think I probably should have.

Thanks,
Jason

J Tiers
07-22-2007, 12:16 AM
Here are what I have tentatively selected for these components:

R11,12,13,14= 5033ED100K0F12AF5, Metal Film, 100K, .5 Watt, +/- 1%

D2,3,4,5= 1N5352 15Vdc Zener, 5 Watt

C2,3,4,5= P10397, electrolytic, 50V, 100 uF

I figured the resistances using the data sheet for the IR2117, the spec sheet says that the quiescent Vbs supply current is a maximum of 240 uA. I calculated the resistance, based on 300 uA. .000300 = 30/R, yielding a result of 100K. I then calculate power dissipation required by P=.000300*15, where 15 v is the voltage dropped across the zener, assuming the 30 Vdc supply. P= .0045 watts, 4.5 mW. I selected the .5 watt resistor for a bit of a safety margin, and since they are readily available.


You want the resistance to be based on Iq and (30 - Vzener - Vsource) because that is the worst case, when the Mosfet is "on" and source voltage is highest. I have not looked up the chip, so I am ASSUMING your quiescent current is OK. It may change with temperature and voltage.

That voltage is (30 -15 - 10.5) or 4.5V (A). Therefore the resistor ought to be no more than 15k ohms, call it 10K for margin (for now). Lower is better to assure the zener of regulation current. More later.

High wattage zeners have a problem for you, which is high leakage currents and poor zener voltage regulation at low current. So a 5W is way too high, and not needed because dissipation will be much less than that. A 1 watt should be OK.

If we stay with the 10K ohms, then the worst case zener current for lower drives is at the condition of all "off" with the +30V still ON. Then the source voltage is essentially zero (shorted thru armature) and the current is (30V-15V)/10K or 1.5mA. That gives 22 mW, still fine (B).

Now, we would like to have more zener current, for decent regulation. If we assume 5mA or so at least current in condition (A), then we can get 4.5mA with 1K ohms. Then in condition (B), we will have 15mA (30-15)/1K.

In condition (B), dissipation is worst case, and we have 0.225W dissipation in the zener, or 1/4 of max using a 1W zener. We also have 0.225 W in the resistor, and so we would use a 1W resistor, due to the high temp potential environment.

Condition (B) only occurs when the winch is powered but not in use, and it only occurs for the lower drives. In every other condition, dissipations are lower. But it is a good model for all 4 of them.

As far as the capacitor, it can be substantially smaller than 100 uF. All it ever needs to do is to supply the charge that is transferred to the Mosfet gates, without much voltage sag. In this case, with continuous re-charging, even a 0.1 uF capacitor should be fine. Their cautions (which I have not read) are for different conditions. If you like, a 1uF would also be fine.

Tolerances can be 5% resistor, 5% zener, and standard (20%) for the capacitor on the bootstrap.

looks like 15V 1W, 1K 1W, and a 1uF capacitor will be fine.

Now, there is ANOTHER issue, and that is the REGULAR power supply for the logic part of the chip. That does not need to "float" off ground, but it DOES need to "exist". It could be the battery, but I'd advise running it regulated off the 30V to avoid issues with the UVLO at max winch draw. But, ONE supply will power all of the logic sections, that is the point of them being "high side drivers", the input and output can have a voltage between them. You could use a UA7815 or UA7812 regulator for that, and a substantial capacitor on input and output sufficient to keep the "78XX" happy.

I know it seems silly to raise the voltage and then regulate back down. But the point is to be sure it meets the minimum at all times.



The Zeners selected are 15 VDC models, the IR2117 chip selected has a voltage range of 10-20 Vdc, so 15 volts should keep me well with in the limits of the chip. I was unsure of power dissipation for this component, so I went with the highest readily available wattage. I assume that the calculation should be based on the same current as the resistors above, but I was unsure. A higher wattage rating than necessary will not be detrimental, will it?



See comments on leakage and minimum current for correct zener voltage

Chip dissipation will be minimum, since you have essentially zero dissipation from switching. use the base line of the curves they give.




The capacitors are kind of a seat of the pants guess, the IR data sheets said that a minimum of 47 uF was required, and basically bigger is better to prevent overcharging. I am not 100% sure I understand what the caps are doing there w/o using a bootstrap supply. I understood that for a bootstrap supply, they are there to provide the necessary gate charge, are they serving the same purpose with a non boot strap supply?

If the above looks adequate so far, I am going to look at designing a voltage regulator to run the logic sides of the chip from the same supply as J Tiers suggested. My only question here, is I have read repeatedly that the logic side of the chips should use a separate floating power source from the gate side, mainly I believe because sharing the same ground as the high power components renders the noise immunity of the chip useless.


Capacitor
Most of that should be due to boot-strap. I didn't see the part of the data sheet giving that, it may be on an applications note. The context should make it clear why they specified that, but their test circuits use less capacitance.

power supply floating
Well, the high power components should not share a current path with the logic, but there is no reason the logic cannot be ground-referenced. As mentioned, that is the point of having the high side driver.

Running them regulated off the 30V will avoid a "brown-out" triggering the UVLO in case of a heavy winch current and weakening battery. The thing should run until there isn't any juice left, if possible.

jacampb2
07-22-2007, 12:55 AM
I am pretty sure I understand. You are basically telling me to design with a resistance that will still supply the needed current during worst case voltage scenario, and a power dissipation large enough for a best case power scenario with the same lower resistance.

I read a lot about zeners over the last few nights. I was not aware of the higher leakage and poor regulation characteristics of the higher power devices.

So, J, are you getting sick of me yet :D ? I honestly am not typically such a slow learner, but this is a lot of stuff to try to grasp, and I do appreciate your patience.

Jason

J Tiers
07-22-2007, 01:02 AM
The application is not unique to you, and I like unusual problems.

You may want to be sure you read all of it, I have been editing and posting a bit at a time.

jacampb2
07-22-2007, 01:19 AM
I will defenitely re-read it. Most likely about 50x before this is done. As for where I got the 47 uF value, and also the yammering about sharing grounds was fron AN-978, a publication I found on IRF's website, about using their gate driver IC's. I knew that the recomendations they were making were specific to a boot strap supply, but I did not understand the function of the cap in the current design. It makes sense now, that in bootstrap operation, it has to provide the charge to the gate, and with the constant supply, it need not be nearly as large.

I am going to go back and re-read the thread again, and then start on the power supply section.

Thanks,
Jason

J Tiers
07-22-2007, 02:14 AM
Part of the reason they advocate a large cap is to prevent "pump-up" in the case of repetitive voltage swings below ground at the switching rate. Usually this is some sort of a transient and anomalous condition.

Those are usually quite short, but can cause the boot-strap voltage to get built up over the chip limit, either causing excess dissipation, or breakdown. Larger capacitors charge slower, giving more chance to drain the charge in the course of normal operation.

It is a real concern in some cases, and I have had to put in zener limiters to take care of it before. However, it won't bother you, because you have no repetitive switching, you are not boot-strapping, and in any case you have already got zeners.

jacampb2
07-22-2007, 07:11 AM
I have designed a 30V power supply based on the LM2587-ADJ boost voltage regulator. I calculated the resistance needed to achieve 30V output based on the equations in the spec sheet. L1's inductance is the only thing I could not make the formula work for. According to the spec sheet, L1 is what limits the duty cycle of the circuit. The sample circuit provided for 12V to 24V had an inductance of 33uH. When I tried their formula, w/ various duty cycles, I almost always got a number near 33uH, but never arrived at the same. Oddly, it didn't mater what percentage I used for the duty cycle, the inductance was always nearly the same. It is listed as the minimum rating, so I selected 100uH for now.

The voltage regulator for Vcc on the IR2117 drivers is the suggested uA7812. I basically copied the circuit shown in the spec sheet, I don't know what constitutes "substantial capacitors" so I used the spec sheet values to start with.

I also found a neat 12V DC-DC filter regulator for the 12V supply if needed, but if I understand this LM2587 chip correctly, pre filter the 12VDC should not be necessary. The LM2587 can use a supply voltage of 8-16 VDC and produce a stable 30VDC.

Without further ado, here is the new picture.

http://www.rollmeover.com/bronco_fab/odds_n_ends/H_bridge_v4_ps_low.jpg

And of course, the super size picture: Here (http://www.rollmeover.com/bronco_fab/odds_n_ends/H_bridge_v4_ps.jpg)

Thanks,
Jason

J Tiers
07-22-2007, 11:50 AM
It looks like a reasonable approach.

I haven't checked out the boost chip in detail, but it looks reasonable. Dunno about the "comp" pin components, presumably they are app note values from a closely similar type supply.

Sounds like you are about ready to proceed with checking things out for this portion, presumably starting with your 30V supply.

You still need to have input logic. it will need to have an absolute guarantee that under NO circumstances will it produce a demand for "winch in" and "winch out" simultaneously. That would turn on all mosfets at once through the armature, while the field gets no power.

While if the parts are selected right, that won't damage anything directly, it won't drive the winch and may let it overheat rapidly. It would be about the same as a stalled winch.

Swarf&Sparks
07-22-2007, 12:27 PM
I must be missing something here.
Unless this is a personal challenge you've set yourself, why not just use relays?
By all means, use logic to drive transistors, but it doesn't take a big transistor to drive a relay coil. (or 2)

J Tiers
07-22-2007, 12:58 PM
I must be missing something here.
Unless this is a personal challenge you've set yourself, why not just use relays?
By all means, use logic to drive transistors, but it doesn't take a big transistor to drive a relay coil. (or 2)

You missed the fact that the original DOES use them, but they weld shut/go bad.

So he wants to do it this way...... OK.......

Swarf&Sparks
07-22-2007, 01:26 PM
Fair enough, but it's my experience that decent relays in sockets, are a damn sight easier to replace in the field!

J Tiers
07-22-2007, 03:27 PM
Fair enough, but it's my experience that decent relays in sockets, are a damn sight easier to replace in the field!


BTW, current level is 500A......................... Sockets anyone?

Swarf&Sparks
07-22-2007, 03:32 PM
back in the 70s, telephone exchange
daily occurence
What the hell sort of battery are you running that will allow 500A draw?
(for more than 30 secs)

Bguns
07-22-2007, 04:40 PM
Dual Battery 4x4 winch rigs with 8,000 10,000, 12,000 lb electric winches...The 4 relay power switch system used now is bulky, has cheap rustable can type relays that stick or fail. Many a 4x4'r would love to replace the crap bank of 4 relays with one solid state (well heatsinked and with thermal overload of course) potted device.... I use Braden winch and PTO's myself, No battery needed, and if engine is dead, spin pto shft by hand to get out...slow tho...

Oh forgot to add for the mechanical relay fans...These will be repeatedly dunked under water, are mounted on front of rig for best exposure to road salt and grime, failure in some modes can kill or maim people.....

jacampb2
07-22-2007, 05:24 PM
Here is the spec sheets on the batteries. They are Optimas, dual purpose, starting/deep cycle, and they will supply all the current I need in most cases. 500 Amps is not going to be drawn on a regular basis. The winch is currently powered through 1/0 g cable, with both batteries in parallel. Somewhere in this thread, I made mention that I would carry my solenoid pack as a spare for a while until the Mosfet control is proven in the field. And yes, the reasons are part personal challenge, and part out of necessity. There are probably literally thousands of people in my sport that would love to be able to do this if it works out.

http://www.rollmeover.com/bronco_fab/odds_n_ends/optima_chart.jpg

jacampb2
07-22-2007, 05:37 PM
It looks like a reasonable approach.

I haven't checked out the boost chip in detail, but it looks reasonable. Dunno about the "comp" pin components, presumably they are app note values from a closely similar type supply.

Sounds like you are about ready to proceed with checking things out for this portion, presumably starting with your 30V supply.

Yes, that circuit was found in their data sheet for the chip. It was set up for 12->24 volts, I used the instructions given to modify it to produce 30V. The data sheet specifically said that the components shown on the "Comp" pin would be suitable for most scenarios.


You still need to have input logic. it will need to have an absolute guarantee that under NO circumstances will it produce a demand for "winch in" and "winch out" simultaneously. That would turn on all mosfets at once through the armature, while the field gets no power.

While if the parts are selected right, that won't damage anything directly, it won't drive the winch and may let it overheat rapidly. It would be about the same as a stalled winch.

It will be a few hours before I begin work on this, but if I am not mistaken, this can be as simple as a SPDT switch that supplies 12v to one of the two logic inputs at one time, center off. Or, I can go as crazy as I want. I am going to have to do a lot more reading again before I take a stab at it, but hopefully I will have something to look over this evening.

Unless someone wants to point out something that we have "cocked up" so far, I may put an order together tonight for digikey. At least for the power supply, and possibly for the whole shebang as it sits.

Thanks,
Jason

Swarf&Sparks
07-22-2007, 05:46 PM
errm, so the universal motor (brushes, etc) is mud/water/rock bulletproof?

BadDog
07-22-2007, 07:15 PM
I can attest that NOTHING is "rock proof".

I mounted my Warn 9500 with a home-made "multi-mount" systems. That included a nice "roll cage" to protect the winch from ANYTHING that I could imagine coming at it. First trip out, the front fell off a shelf and a pointy rock went right between 2 bars to tag the motor housing and break the main spool mount on that side. Cost me $145 for another one, for that price, I shoulda made one myself out of "billet" but had a run coming up and no time. ;) I've broken transmissions in half (not exagerating, you could see the rear drum) and transfer case adapters even with them behind skid plates, caved in an oil pan that you would have a hard time getting your hand to without being in just the right spot, bent a 3/8" wall tie rod, and the list goes on...

jacampb2
07-22-2007, 08:58 PM
errm, so the universal motor (brushes, etc) is mud/water/rock bulletproof?

Not even close too it, as BigDog points out. The winch motor will continue to run in adverse conditions though. I take pretty good care of my winches, the 8274 on my buggy has the original motor still, and is approximately 35 years old. Any time it gets submerged, I tear it down and clean it out. The motor Warn chose for this winch has never failed to run for me.

I have a couple of other winches around, a 2 Ramsey 12K worm gear winches, and one of them, when I dragged it into my shop would not run at all. Took the motor apart and found it packed with dirt. Cleaned it all up, polished the commutator, reassembled and it works fine. Obviously continuing to run them with them packed full of crap will destroy it, but I am fairly diligent about cleaning the stuff up that takes a beating while off road.

Later,
Jason

jacampb2
07-23-2007, 12:49 AM
Well, here is my attempt at the logic section. The IC is MC14070B, CMOS Quad xOR gate. I am not 100% sure this idea will work, as it seems the xOR gate requires both inputs to be at the same level, either high or low to keep the gate off. I am going to think about it some more, but here is what I have so far.

http://www.rollmeover.com/bronco_fab/odds_n_ends/H_bridge_v4_ps_r_low.jpg

Here is a larger view: Super Size Picture (http://www.rollmeover.com/bronco_fab/odds_n_ends/H_bridge_v4_ps_r.jpg)

**EDIT** I added inverters to the XOR gate circuit, I think this should work much better than before.

Thanks,
Jason

jacampb2
07-23-2007, 09:34 AM
Ordered parts last night. $120, including a bread board to build the PS's on for trial. I was going to just order 4 of the FET's for now to do testing, but their price break is at 10 pieces, and the drop was from $5.80 to $3.44, so I bought all of them. If for some reason they are not going to work out, I will eBay them.

Later,
Jason

Evan
07-23-2007, 10:33 AM
so I bought all of them

Good plan. ;)

Where's the remote? :D

J Tiers
07-23-2007, 10:59 AM
Did not check through your logic.......But......

Exclusive "OR" is indeed "a" or "b" but not both.


Switches "bounce" * , and can assume any state when bouncing, no matter what they "nominally" are.

Switches come in various types

1) "break before make", where the second contact (whether part of the same switch section or another one) makes contact only after the first has opened.

2) "make before break", where the second contact is made BEFORE the first opens

3) "unknown/unspecified", where you don't know.

But due to bounce, you cannot *guarantee* that any particular contact type will act like that at "logic speeds" even though it may act correctly when , for instance, turning on a light.

* contact bounce is a rapid series of makes and breaks that occur while a switch is changing from a closed to open, or open to closed state. May last milliseconds or tens of milliseconds, depending.
It is normally required to take some care to make sure that bounce will not affect the circuit and results of the logic.

Evan
07-23-2007, 11:49 AM
Re switch bounce, as long as the circuit is level sensitive and not edge sensitive a simple RC across the switch contacts will usually take care of any bounce issues.

jacampb2
07-23-2007, 06:06 PM
Had to do a lot of re-thinking on the logic, I did a truth table for it, and it was not going to work, it would allow both outputs on simultaneously. I have redesigned it using the two inverters, and two AND gates. Here is the updated logic section. I neglected to save my design to my flash drive last night, so I don't have a working copy here right now.

http://www.rollmeover.com/bronco_fab/odds_n_ends/H_BRIDGE_LOGIC.jpg

Here is to hoping this one will work :cool:

Thanks,
Jason

jacampb2
07-23-2007, 06:11 PM
Where's the remote? :D

What do you mean? Physical location, or where in the schematic? I plan to use the current remote which is just a hard wired DPDT center off momentary switch. The current switch is DPDT, but it is wired as a SPDT, I assume for more current carrying capacity because it is the actual connection to 12V to pull in the solenoids.

Bottom center of the schematic is a very basic representation of the remote. I believe in the new design, it will have to get it's 12V from the regulated supply, instead of the common 12V supply to the vehicle.

Later,
Jason

jacampb2
07-26-2007, 02:26 AM
Well, I got my big box-o-goodies in the mail today. I built the 30V PS on a breadboard, and it provided exactly 30VDC according to my DVOM. I found a power supply designer application on the manfacturers website for the LM2587, after using it, I had to adjust some of my values I had planned, but the thing looks like it will work. I ran it off a 9.5V wall adapter, and it happily put out 30 volts. Obviously, I have no idea yet what it will do under load, but supposedly according to their designer software it will do .5 amps with no trouble, starts getting out of spec for junction temp if I actually try to use it to supply it's rated 1A, but I shouldn't need that much from this supply.

On a side note, I cannot believe how small those power FETs are, I kind of have my doubts about their current carrying capacity. I guess if I smoke them, it's back to the drawing board.

Later,
Jason

J Tiers
07-26-2007, 03:10 AM
They have a very small, 10 x 7mm heatsink contact area. On the other hand, they also have a need to dissipate perhaps 42W max at 160A per device.

before someone corrects my math, I am assuming the maximum 25C Rds(on) of 1 milliohm, and assuming the value is 1.6x at max junction temp, as the graphs indicate.

In that instance, the DIRECT dissipation will result in an INTERNAL temp rise of only 21 deg or so.

There may be added dissipation or heat load from lead resistance, but it will be from CONTACTS since the Rds(on) should include the leads etc.

YOUR challenge is to get rid of the heat without adding too much more thermal resistance.

About the only way to do that is to solder the drain tabs to a chunk of copper bus bar, and then heatsink THAT to a finned thing with whatever insulation is required (hopefully not much). That is pretty much what is expected, I believe.

I can't think of much in the way of individual insulators that would work at all for you, without a large increase in thermal resistance.

My thought is to de-rate further and use at least 1.6 x the number you absolutely have to have for the maximum sustained (over 5 seconds) current you will have. Say 5 devices instead of the 3. That treats them each as 100A devices, and reduces heat load from 42W to about 17 watts per device.

That will also reduce the added heat load and problems from having only 3 leads carrying 53A per each. Now they carry 33A each.

Of course, you STILL have to get rid of 17 x 5 x 2 watts, or 170W total, through some heatsink.

But starting from an assumed 60C ambient, you now have a possible junction temp rise of 115C max. Of that, you lose 9 deg in the device, and probably at least another 9 deg in the soldering or other attachment of drain to bus, leaving 95 deg approx.

Then also you lose more in the insulated thermal contact to heatsink and heatsink to ambient. I don't know what to estimate for them, but if you do no worse than 10 deg in the insulator etc you will be doing well. In that case, you would have 85 deg to get rid of 170W, or a need for a 0.5 deg C per watt heatsink. Almost surely forced air, i.e. fan.

If you did NOT over-design, I think you would have a much less realistic heatsink requirement.

jacampb2
07-31-2007, 03:32 PM
More progress... I learned all I could about etching my own PCB's and gave it a shot for the 30V_12V power supply. It went pretty good, I tried to iron on the traces from a laser printer print out, I have read using paper works, but transparency film works better. The paper transfered most of the pattern but it was sketchy in places. I filled in the bad spots w/ a sharpie, and let it dry over night. This was my first attempt, and I don't think it came out too badly. I bought some transparencies today, and will give that a shot for the logic/gate drive board since most of that stuff is SMT components. I don't have a steady enough hand w/ the sharpie to draw pads for SOIC16's... Anyhow, I tested the supply, and the 30V side fluctuates around 30.5V, the 12V supply is rock steady at 11.9 volts. Here is the picture.

http://www.rollmeover.com/bronco_fab/odds_n_ends/30v_12V_ps.jpg

I have a plan for the FET's. Did more digging on the IR website, and found mounting and heat sinking suggestions. I ordered a 8x10 sheet of double sided 2oz PCB, I am thinking of making a board for them that uses short wide traces for the drain and source conductors using both layers of the board, and terminating in a solid copper buss for the high current connections. The FET's can be heat sinked on the case side, as well as the copper trace for the drain providing some heat removal. I will use an aluminum heat sink on the case side and fan cool it. I also have a few peltier (sp?) piezo cooling sinks around that may help, if fan cooling it is not enough, I will come up with something :D

Thanks,
Jason

Bguns
07-31-2007, 05:33 PM
If you get it up and running ...you might try adding a thermal sensor to cut off power, and light up an overheat/cool cycle LED on cover, with cooling fan/Peltier applied in overheat mode. Peltier sounds better (but power hungry) than a fan in 4x4 use :)

J Tiers
07-31-2007, 08:59 PM
I think you may find that the dissipation is way too much for a copper-trace-only heatsink scheme.

Then also, remember that what works at 25C at home may not work at a higher temp in the field.

I made some worst-case assumptions in the analysis I made above. But the 3 device situation is much worse than the 5 device suggestion, heat-wise. And the 5 device is bad enough.

A little analysis saves time and money........ I think you need more heatsink.

jacampb2
07-31-2007, 09:22 PM
I would much rather solder the drain directly to a solid copper buss, and then heat sink the buss like we talked about originally, but I don't see any way that the FETs are going to take the heat necessary to solder them directly to say a 1" sold square of copper. Off the top of my head, I think the data sheet said 300* for 10 seconds was the max soldering temp. I think it will take far more than that to get the solder to "take" to solid copper. The only other option is a mechanical, sandwich type connection. Any thoughts? My new plan here was not to rely solely on the copper traces for heat sinking, but rather to use it to carry the current and some heat for a very short length, and to supplement it with a robust heat sink on the non-conductive side of the FET's case. I am not against using more FETs either, oddly enough, they quickly became the cheapest part of this project. I spent more on one power diode to do the neg. transient clamp, than on all 10 FETs.

One possibility I was thinking of for drain buss, is solid copper bar and mill a recess for each FET, then a top buss to clamp them in place.

Let me know what you think, I haven't committed to anything yet. Just trying to figure it out.

Thanks,
Jason

J Tiers
07-31-2007, 09:32 PM
There are some "low temperature solder" formulations, which I have not used.

The deal would be to use the thinnest possible copper, probably in the 1/8 to 3/16 area, and the lowest soldering temp. heat the bar, tin it, and slap all the devices for one "bank" down, then cool it as quickly as practical without too much thermal shock....

Those cases IIRC are MADE to solder to.... there is a way. I am not an expert on that, as we never did that fancy of a construction.

The block-type diodes and IGBTs are made with multiple die all soldered to a common header.........

jacampb2
07-31-2007, 09:43 PM
Yeah, that sounds like a defenite possibility. I somehow had it in my head that I was going to use copper large enough to drill the end for the 4 gauge wire used for the motor, and slap a set screw in it and kill 2 birds w/ one stone. A buss for the FETs and a connection to the large gauge wire. I could however easily solder a block to the buss bar before the FETs are mounted, or mill them out of a solid 1" or so bar. It is going to be a kind of convoluted way of doing things, since the source leads will need a buss of their own, but I think it will work. Thanks again for getting my head out of my arse :)

Later,
Jason

darryl
07-31-2007, 11:41 PM
I also have not had experience soldering devices to a copper bar or plate, but that surely would be the way to do it. I would probably want to make an array out of the mos devices, maybe by soldering the leads to a piece of copper wire- just something to keep them spaced properly. To facilitate soldering them to the copper, put a drop of resin on the solder side of each one. Heat the copper, tin it, then lay the mos array onto it and quickly apply some pressure to each device. You might not want to use your fingers for this :) The resin will make the solder take to the devices quickly, and you'll see the solder form a meniscus around each device. I'm sure there would be some advice on the IR site about types of solder, and techniques to use.

Once the arrays are made, they can be insulated from the heatsink using a piece of sil pad or similar. That's basically just a piece of thin fiberglass cloth that's been impregnated with silicone rubber. I've made that myself and it's not hard. You want to fill the gaps between strands of cloth but not build up the thickness. You might even consider making it in place- essentially you'd be using silicone rubber to glue the copper plate to the heatsink, with a piece of fiberglass cloth between them, and having worked the air bubbles out.

I would still be wanting to apply pressure evenly to the tops of each device to hold the array to the heatsink, and I'd have that prepared before doing the silicone thing. It would be difficult to improve on this method of heatsinking.

As far as the size to make the copper plates- figure out the surface area of the mosfets combined ( the solderable area) then multiply that by about 3 to 5 to get the surface area for the copper piece. That should spread the heat out enough to compensate for the insulator without the plate becoming overly large.

By the way, you do want the copper plate to sit flat on the heatsink before you silicone it down. You might have to file the contact area to remove high spots. I'd do this after the devices have been soldered to it.

jacampb2
08-02-2007, 03:40 PM
Well, I finished the gate driver and logic board. I used a transparency to transfer the toner this time, the results were much better. I did however get a bit of adhesive from the tape used to hold the transparencies in place ironed on to the board as well. It didn't etch very well around the corners. I had to clean it up with a dremel, and it still isn't perfect, but it should be electrically sound. Those tiny SOIC chips are a PITA to hand solder. I will probably spend a bit of time today figuring out the FET mounting, but I don't have much time left to devote to this this week, I am about to head out on a wheeling trip. Here is the picture of my... Um... Masterpiece :D

http://www.rollmeover.com/bronco_fab/odds_n_ends/gate_drivers_and_logic.jpg

Anyone know where I can buy small quantities of conformal coating? Ironically, we make it at Dow Corning, where I work, but they don't tend to do any small quantity sales, and although I can get a pass out for a gallon or so of it, it is a ton of red tape, and I would just as soon buy some. I am thinking I will dip the PS and Gate driver once they are tested.

Later,
Jason

jacampb2
09-05-2007, 08:49 PM
Well, I made a bit more progress, I got all of the FETs mounted and the armature clamp diode. I have room for more FETs if needed. My fan cooled peltier is mounted, sorry no pics, and that thing is awesome. ~6 amp current draw and it gets down to 0* F on the cooling side. It is mounted in an ammo box, ducted to cool the hot side of the peltier, and hopefully keep the fan relatively clean.

Here is the pic of the FETs, hopefully I didn't kill them, the copper bar is 3/4"x.125", I had to solder them to it with a torch... I went as fast as I could and cooled them with compressed air as I went. We will see how it came out.

http://www.rollmeover.com/bronco_fab/odds_n_ends/odds_n_ends017.jpg

Later,
Jason

J Tiers
09-05-2007, 11:09 PM
Hopefully you remembered that some of those little leads are gate leads!

Almost looks like you soldered all of them down.............. :eek:


I was just wondering how you were doing. And hoping I remembered to mention that static precautions are very much in order when working with Mosfets and IGBTs.......

uute
09-05-2007, 11:15 PM
Hi Jason,

Missed this thread, and didn't read it all - so maybe this has been covered, BUT:

Looking into elec vehicle controllers in the past, one thing I was always told was to put a small resistance (say 5 ohms) right at the FET input (gate) to keep them balanced. I guess this helps ensure all (3 in your case) FETs turn on at the same time, avoiding cascade failures. Put the resistor right on the FET if you can, or as close as possible. One restor on EACH mosfet.

Electric vehicle controllers were a place to study.

Hope I'm not just "killin bandwidth"! :D

Good luck
uute

jacampb2
09-06-2007, 01:15 AM
Gate leads are not soldered down, they are the far left lead on each FET, and I bent them each up before I soldered the rest down. The center truncated leads are drain leads, and obviously left open due to useing the tab as the drain. I have bolt on blocks from McMaster carr that accept up to 2/0 g wire and are bolt mountable. I will use one for each connection. I also found some high dielectric strength kapton tape that is a minimal thermal barrier, but good electrical insulator. It will be going on the back of the copper bus, between it and the peltier.

We did cover the gate resistors earlier in the thread. They are in the schematics, and I will probably solder and heat shrink them at each gate.

As for static sensitive, I hope they survived. There was no good way to do this really, seeing as I had to pound a thumb tack with a 20 lb sledge, but the buss and myself spent the whole ordeal bonded to my grounded welding table, so time will tell. With any luck, I will get it all mounted and installed tomorrow and give it a trial run...

Later,
Jason

jacampb2
09-07-2007, 09:20 PM
Well, I got it all together today, and I can't get it to work. I when I power it up, I have 12V on every buss. W/o power there is no connection between them, so it is not a simple short. I finally took my gate drive board right out, and still the same thing, I had 12V at each buss, plus oddly enough I have 12V at the gate of each FET w/o anything but the 47 ohm resistors connected. I have checked and rechecked for a short, and cannot find anything. I finally found that if I hooked the armature buss to a resistor and then ground, that the voltage on the other buss' went to near zero, If I then grounded the FET gates, the transistors totally switched off. I thought that may be I had found the problem, since the armature buss was supposed to go to ground, through the armature, then perhaps the fact that none of the FETs were referenced to ground was allowing the gates to slowly charge and turn on. I reconnected my driver board and tested it with the Armature terminal still to resistor/ground, and still cannot get it to work. It acts like my gate drivers and/or logic section are doing nothing.

I think I may have an issue with my logic design. I thought I had it, but I am not so sure now. I have to do some thinking on it again. I used a quad and gate, and an inverter. The controller inputs are pulled to ground via 1k resistors, the controller inputs, say channels "a/d" go to one and gate, and one side of the inverter, the inverter's output goes to one of the inputs for the opposite channel, "c/b", so when "a/d" is 1, "c/b" gets 0 on one of it's and inputs, the other "c/b" channel is pulled to ground via the pull down resistor, and then inverter should invert this to 1 and feed the second half of "a/d" 's and gate, causing it to also output 1. The switch debounce for both channels is a simple .1 uF cap that has to be charged between the input and ground. Make sense? I will try to dig a schematic of the logic up to post later.

I am stuck at work again, so it will be a while before I get back to it, but I am open to any input.

BTW, the peltier may work too well, after running for about 5 minutes today, while troubleshooting, I had ice all over the FET's, that probably isn't going to help matters...

Later,
Jason

J Tiers
09-07-2007, 11:49 PM
One thing is that if the gates had a positive charge before you unhooked them, they will retain the charge until it is discharged. So the parts will be "on" for a long time if the gate charge is not allowed to leak off somehow.

If the gates control conduction the mosfets are probably good. Keep them that way by avoiding static until you have them in-circuit!

The drivers, if they are well-behaved, should be OFF until their "UVLO" (undervoltage lockout) is satisfied that they have enough voltage to drive OK. THEN they will turn on, and accept inputs.

The supply voltage (30V) to the gate drivers will charge things and 'clamp" to the 12V, if the other voltages and a ground path thru armature are not present.

I didn't look closely at the logic.

jacampb2
09-08-2007, 12:13 AM
The logic in my original schematic somewhere in this thread is not correct, and not what I did. I looked on my flash drive, but it appears the design I came up with is not on there, must be on my on my desktop at home. I will post a pic when I get home tomorrow.

One thing that had me looking all over for a short, and which I do not think is correct, is that with the 12V buss powered and connected to the mosfets, and nothing else connected, I have ~12 Vdc on all of the gate leads. There is very little current there, somewhere in the order of .6 mA when measured with my multimeter. Not enough current to light an LED, but I am not sure at all how 12V is getting there unless I screwed up some of the mosfets.

To test my gate driver board, can I try to switch a channel on with it disconnected and measure for the gate charge from the driver chip output (Ho) to ground? I been all over the thing looking for a trace or solder point I screwed up, but I haven't found anything yet. I will get out my loupe in the next few days and examine it closely.

Thanks,
Jason

J Tiers
09-08-2007, 11:08 AM
unless I screwed up some of the mosfets.

Possible. A gate punched through by static may short gate to drain. But it usually results in an overall dead short if you short gates to sources, so if THAT wasn't a net short, you may be OK.

jacampb2
09-08-2007, 09:12 PM
Okay, here is a pic of the final schematic of the gate driver ICs and the logic and switch debounce board.

http://www.rollmeover.com/bronco_fab/odds_n_ends/odds_n_ends018_low.jpg

Super Size Picture (http://www.rollmeover.com/bronco_fab/odds_n_ends/odds_n_ends018.jpg)

I am going to go over it again tonight against my PCB traces, make sure I didn't seriously screw up some path or another.

I tried doing of a meter check of the FETs before I knew I had a problem. After I had them all soldered to their respective buss, I thought it would be a good idea to try to check them out according to instructions I found online. What I read said that with the source and gate tied together, the FET resistance from drain to source should be very low. It wasn't, the FETs read as having a very high resistance from drain to source, somewhere in the order of 30-100 Mohms. It depended on what bank I was checking, and the resistance would change the longer I had the meter connected. I kind of assumed that the differences were due to the FETs being in their parallel circuits. I assumed that the changing resistance was due to their gates slowly accumulating a charge (I did not leave them shorted for the test). I eventually decided that I didn't know what the heck I was doing, and that as long as none of them showed a low resistance path, I was probably okay. If their any good way to test them with them still soldered in place? I really don't want to subject them to more heat to remove them if it isn't needed. If I did total the FETs due to heat or static, I need a much better plan for the next batch.

I still kind of think the FETs are okay, and that my problem is elsewhere. I will continue to research, and see what I can come up with. If I have to do the FET section again, I am probably going to look for something other than SMDs, I don't like having to have the gate leads all bent up, and the unsupported resistors directly attached, it seems that vibration is eventually going to make them fail.

Thanks for all of the help, let me know any testing or trouble shooting suggestions you may have.

Later,
Jason

J Tiers
09-08-2007, 10:41 PM
I tried doing of a meter check of the FETs before I knew I had a problem. After I had them all soldered to their respective buss, I thought it would be a good idea to try to check them out according to instructions I found online. What I read said that with the source and gate tied together, the FET resistance from drain to source should be very low. It wasn't, the FETs read as having a very high resistance from drain to source, somewhere in the order of 30-100 Mohms. It depended on what bank I was checking, and the resistance would change the longer I had the meter connected.

With source and gate tied, mosfet should be OPEN and no conduction. They must be talking about "depletion mode" Fets, typically "Jfets" not Mosfets.

Mosfets are "enhancement mode" and need a voltage on the gate to establish conduction.



I kind of assumed that the differences were due to the FETs being in their parallel circuits. I assumed that the changing resistance was due to their gates slowly accumulating a charge (I did not leave them shorted for the test). I eventually decided that I didn't know what the heck I was doing, and that as long as none of them showed a low resistance path, I was probably okay. If their any good way to test them with them still soldered in place? I really don't want to subject them to more heat to remove them if it isn't needed. If I did total the FETs due to heat or static, I need a much better plan for the next batch.

I still kind of think the FETs are okay, and that my problem is elsewhere. I will continue to research, and see what I can come up with. If I have to do the FET section again, I am probably going to look for something other than SMDs, I don't like having to have the gate leads all bent up, and the unsupported resistors directly attached, it seems that vibration is eventually going to make them fail.

Thanks for all of the help, let me know any testing or trouble shooting suggestions you may have.

Later,
Jason

The best way to check N-channel Mosfets is to get a 9 V battery with a series resistor (just for safety). Remember to take static precautions when doing this.

With meter across source and drain and meter's "positive" lead* on drain, connect battery thru resistor to make gate +9V. Meter should read very low resistance.

Now short gate to source. Meter should read a very high resistance.


if that happens, the mosfet (or bank of them) is probably OK.

* Positive lead is the one that has a positive voltage on it, and may not be the red lead for some meters.

jacampb2
09-12-2007, 10:32 AM
Okay, I checked my FETs. I used a similar procedure to what you described, but used my meter in diode test mode which I read is safer for the FETs. It appears that the 3 transistor banks are both okay, when I charge the gate with a 9 volt battery, they allow the full 1.8 volts or so of my meters diode check circuit to pass with the meters + lead to drain, and - lead to source. W/ gate shorted to source, they pinch off and the meter reads 0V.

The two banks of two do not appear to be functioning correctly. After charging the gate, they read .6 volts or so drain to source. Gates shorted to source, they continue to read the same. So, evidently they are partially enhanced, and stuck that way. I have read a lot about checking FETs and how they typically fail, and I have not seen one mention of this... Most places I have read say when they fail, they typically fail with drain shorted to source. Maybe they failed due to excess external heat though from soldering them. I am going to pull the assembly back apart today, get the buss' off the peltier, and see if they behave the same way on the bench. If so, I will de-solder them and test each of the pair individually.

I also found some issues with my gate driver board. It appears the logic is working correctly. I didn't have a good solder connection to the +30v side of the board though. I fixed that, and now with a meter check from HO on 3 of my 4 IR2117 drivers I have 30Vdc regardless of if they are receiving the signal to switch on or not. The 4th one seems to stay off whether or not it is receiving input to turn on.

I really have to clean up my work area before I go much further, things are getting far to cluttered to do any kind of work...

Later,
Jason

jacampb2
09-12-2007, 06:47 PM
Okay, I have good news. The FETs are all fine. I pulled them out of the enclosure, left them on the peltier though, just in case that was causing problems, I wanted to isolate one thing at a time. Anyhow, using the method J Tiers outlined above I have 0 Ω from drain to source on every bank. With the gate shorted to source, they all switch to somewhere in the neighborhood of 10 MΩ. I am unsure why they weren't switching in the enclosure, but I did find some corrosion or tailings of metal between the truncated drain lead and the source terminals on one of the 2 transistor banks that were not switching. I went and broke all of the truncated drain leads off to hopefully avoid this happening again. My guess is the condensate from the peltier allowed it to bridge the gap and short them out. I still may have to go to some form of passive, or just forced air cooling. My hope is that the when it is up and running the FETs are going to generate enough heat to keep the peltier from condensing moisture and freezing.

I do still have trouble with the gate driver. I found I screwed up a few things. One, I had the 15v zeners in backwards. I went back to my original schematic, and realized I put them in backwards when I drew it up for PCB express. I de-soldered them and put them in the correct orientation, cathode toward BS, however, maybe having them backwards already damaged the gate drivers. I found an open trace going to the driver that had no output regardless, and now it behaves exactly like the other 3, I have 30 V at the output regardless of if there is 12V at the input or not. I tested them all with a 1.5K resistor and LED to ground, with my DVM measuring current in the circuit. They all put out 5 milliamps, so I assuming they are all on. I am not sure what is going on, the logic is working correctly, and pulls the inputs down to 0 volts when neither main switched input is hot, it pulls the correct outputs high, to 12Vdc when switched. Is it possible that the high side drivers need a pull down resistor on their input, they work similar to the FETs they are controlling, don't they?

Thanks,
Jason

J Tiers
09-12-2007, 10:36 PM
I didn't check the schematic before.......

Check the data sheet for the drivers.

VD is the Plus line, and VS the negative for the drivers.

VS should go to the sources of the driven fets, which is not a connection shown on your schematic.

(back from dinner now)

The connections in the schematic above are all sorts of backwards. If that is truly the connection that was made, the drivers might have gotten reverse voltage through the VS pins. I don't know what that might have done, since the whole driver portion of the IC is in an "isolation well" that should have kept current out of the main portion. But I don't know if the driver part can stand reverse voltages. Could be OK, particularly if VD was not connected to the part of the mosfet circuit that should have had the VS connection.

And, due to the way power is supplied, it is assumed that there will be a 12V bus supplied and that the lower fets will be grounded through the motor. Otherwise the voltages will be goofy.

Power leaves the drivers through the VS connection, through motor and the "on" mosfets. If all mosfets are off, the upper drivers will be "grounded" to +12V through the "intrinsic diodes" in the upper mosfets. The lower ones remain grounded through the motor as always.

jacampb2
09-12-2007, 11:40 PM
Okay, first the source leads were connected to Vs, I forgot to draw them in that schematic, but I did not forget them when I wired it. In the first picture bellow is how it is currently setup, I mentioned above that I put the zeners in backwards from what was in the original schematic, the one that we did go over. I redid it in a different software to link with the PCB trace design I made. I highlighted the change in the circuit bellow, this is how it is currently setup.

http://www.rollmeover.com/bronco_fab/odds_n_ends/gate_drivers_current_setup.jpg

I see the problem now. I just went back and looked at the original schematic we did. The bellow should be correct, right?

http://www.rollmeover.com/bronco_fab/odds_n_ends/gate_drivers_current_prop.jpg

I am a freaking moron. The I will try again tomorrow, w/ any luck I didn't totally fubar the drivers. About the most expensive piece of the puzzle if I did, but that will teach me not to redraw thing after a 12 hour midnight shift and 4 hours sleep...

I will keep you all posted, I think I am getting very close.

Thanks,
Jason

jacampb2
09-13-2007, 08:59 PM
Drum Roll Please..........

I am pleased to announce that the Mosfet control is now functional. I have not yet used to to drive a winch, but preliminary tests are complete with a 12V starter motor on the bench and it performed admirably, with absolutely no puffs of magic smoke. Once I rewired the gate drivers power supply to comply with the original schematics, it functioned perfectly.

I put the whole thing back together and found that I still had 12V on each rail. I was starting to pull my hair out, and I finally said, well lets hook up a motor and see what it does. If it caught fire at that point, so be it. The motor showed no signs of current flow, even though voltage was there by previous meter check. My guess is that putting the meter in the circuit is like a parallel resistance with the FETs, the meter being a lower resistance, and allows current to flow. I will not have time to make up cables and hook it to a winch for a few days now, but that will be my next step.

I did screw a few more things up. Specifically the mosfet configuration on the buss'. I somehow managed to solder it up with a 3 mosfet bank going to 2 mosfet bank, and visa versa. It will be easy enough to fix, as I left myself room for additional mosfets.

Many thanks to J Tiers and the others who helped to guide me along so far. And a big thanks to International Rectifier as well since they evidently make some idiot proof components.

Here are pictures of the almost finished project. I still need to add a master switch and the remote control jack. I am thinking that I will wire up the controller jack so that the power to the PS is disconnected with it unplugged.

http://www.rollmeover.com/bronco_fab/odds_n_ends/odds_n_ends019.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/odds_n_ends020.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/odds_n_ends021.jpg

(Cont. Next post for Picture limit)

jacampb2
09-13-2007, 08:59 PM
http://www.rollmeover.com/bronco_fab/odds_n_ends/odds_n_ends022.jpg

http://www.rollmeover.com/bronco_fab/odds_n_ends/odds_n_ends023.jpg

I know it is not the prettiest creation, but I didn't want to spend the $$$ on a nice project box, so the ammo box was a good fit and I had it in the pile...

Thanks again,
Jason

BadDog
09-13-2007, 09:49 PM
Man, I love those amo boxes. Got 2x50, 4x5.56, 8 or so of the big 20s, one big boxy who knows what, and a LAW (I think, long and thin, just right for spare D60 shafts!) box. Bullet proof storage, only problem is they are so heavy EMPTY, you need a fork lift even for the moderately sized boxes once much stuff is inside...