PDA

View Full Version : OT-SCR's for power control



Forestgnome
12-13-2011, 11:09 PM
I was going to try building a current control circuit for my welder, and I was going to base my SCR spec on the maximum circuit current which is 50 amps. So I looked for a 50 amp SCR, and came across the BTW69-1200. I think I'm missing some critical info, because I can't get my hear around how this TOP3 cased SCR can handle 50 amps. The leads are 1.20mm x 0.055mm. Seems to me it would be more of a fast acting fuse. Anyone experienced with designing SCR circuits?

Black_Moons
12-13-2011, 11:18 PM
Nah its simple, Your wire acts as a heat sink for the lead to keep it from melting.

At least, thats the only way I can explain my 120A rated TO-220 mosfets (Even smaller then TOP3)

Some day I wanna try it and see if it works!

J Tiers
12-13-2011, 11:21 PM
They DO mention the average current at 32A.... I'm not sure how they square those two.....

The biggest problem is not the leads, but the solder connections..... at that current many parts use multiple leads, or a different sort of connection.

it is a 'gray area" of the specs..... not all the limit specs apply at once.... and the "real" limits may be actually outside the part..... getting the current away from the lead without too much resistance and heating

However, the innards WILL handle the specified currents.... assuming you can hold the case to the temps they quote right with the current.... for the TO-3P case (the one you complain about) you must hold the CASE temp to 75C or less.

Dissipation is basically current*voltage drop when on, and there can be relatively small drops in these parts, resulting in lower power dissipation.

Of all semiconductors, SCRs and rectifiers are among the most rugged and hard to kill.

macona
12-13-2011, 11:50 PM
Your not going to chop the power before the transformer, are you? Thats going to be a mess with the inductive load. I would switch the output of the transformer. Replace two of the 4 diodes in the bridge with scrs. Then you will need a gate drive circuit. There are commercially available ones out there. This is pretty much how it is done in commercial machines,

Evan
12-14-2011, 12:27 AM
Switching that kind of load at power line frequency is going to produce an enormous amount of radio noise. It will wipe out everything on the radio and TV bands for a long distance from your shop. There is a name for a similar circuit that was the first transmitter to send a signal across the Atlantic. It's called a spark gap transmitter.

EVguru
12-14-2011, 04:29 AM
Nah its simple, Your wire acts as a heat sink for the lead to keep it from melting.

At least, thats the only way I can explain my 120A rated TO-220 mosfets (Even smaller then TOP3)
Check the data sheet again. There will probably be an asterix against the current rating and a foot note stating that it exceeds the capability of the package. You still get the advantage of the low Rds-on and lower power dissipation.

It's quite common to use the same die in several different packages.

EVguru
12-14-2011, 05:51 AM
Your not going to chop the power before the transformer, are you? Thats going to be a mess with the inductive load. I would switch the output of the transformer. Replace two of the 4 diodes in the bridge with scrs. Then you will need a gate drive circuit. There are commercially available ones out there. This is pretty much how it is done in commercial machines,
I was planning to do just that to build a TIG. I downloaded the Miller Syncrowave patent which gives you square wave current and the ability to vary the clean to weld current ratio. I bought a Chinese TIG (Jasic WSE200P) before getting very far and still have a 250 amp full bridge and current sensor all mounted on a heatsink sitting on the bench unused.

J Tiers
12-14-2011, 07:57 AM
Before, after, all same-same, really, so long as you are not using the SCR as a rectifier. (Transformers need AC)

The Monarch 10EE "Wiad" version has what may as well be SCRs (thyratrons) directly off a transformer.... for DC motor speed control.

Both SCRs and thyratrons stay on until the current drops to a low level (hold current). That means they are not quite as nasty with inductive loads as a device which can turn off while current is still flowing. SCRs are still used in some DC motor controls.....

A decent amount of filtering will take care of radio noise..... it may not take care of "line harmonics" of lower order, which are controlled by law in some places (europe, for instance).

TheAndroid
12-14-2011, 01:18 PM
Transformers need a change in the current. That doesn't necessarily mean AC. Pulsing DC works also, you just don't get the polarity switch. Even though it's rectified, it will reach 0 and shut the SCR off.

MaxHeadRoom
12-14-2011, 02:11 PM
For an SCR controller full wave bridge 2 quadrant control, you may be able to recruit a DC motor SCR controller, just replace the bridge with higher power out-board components and feed the board itself from the normal 120vac supply.
They use a 5k pot for control.
I have used these in the same manner with higher rated bridge components for bridgeport DC table motors.
You may want to place a large inductor (choke) in the bridge output.
Max.

Black_Moons
12-14-2011, 02:13 PM
Check the data sheet again. There will probably be an asterix against the current rating and a foot note stating that it exceeds the capability of the package. You still get the advantage of the low Rds-on and lower power dissipation.

It's quite common to use the same die in several different packages.

Not that I saw. There was a operating limits chart, that actualy.. EXCEEDED 120A at I think below 50C case tempature, but that area was shaded out with a note "Exceeds die bonding limits".. hahah. I assume that means the mosfet vaporises from the inside out.

Swamp Donkey
12-14-2011, 02:19 PM
Current ratings are "based", (mis)using Ohm's law. Its calcualted on the resistance through the device when it is in full saturation, ie fully on, and the maximum rated voltage, where Current = Voltage/Resistance.

These ratings do not mean the the device will survive very long at these loads without catastrophic failure due to overheating, its just the maximum possible current that it can conduct...unless you run some massive heat sing with cryogenic cooling, LOL.

Kind of like a 220 MPH speedometer on a car...just because it goes 220, doesn't mean that you are a good enough driver to keep it under control!

Black_Moons
12-14-2011, 02:24 PM
These ratings do not mean the the device will survive very long at these loads without catastrophic failure due to overheating, its just the maximum possible current that it can conduct...unless you run some massive heat sing with cryogenic cooling, LOL.

Kind of like a 220 MPH speedometer on a car...just because it goes 220, doesn't mean that you are a good enough driver to keep it under control!

Oh yea, thats another thing to look out for, Lots of devices current ratings are speced assuming.. 20C case tempature! Ie, basicly watercooling with a water chiller! Or maybe a heatsink in -40c weather.

When you consider a realistic case tempature of say 70C you could maintain with a decent sized heatsink and a normal 20~30c ambiant room tempature, with a 125C max junction tempature, you end up with typicaly half the 'absolute max' current rating. (And thats WITH the junction tempature at the 'max')

J Tiers
12-14-2011, 09:32 PM
The SCR in teh datasheet shows a 70C CASE temp max...... actually fairly reasonable compared to 25C case temp on other data sheets.... but in practice may not be easy to hold.... it is 45C total rise from case to ambient....


Transformers need a change in the current. That doesn't necessarily mean AC. Pulsing DC works also, you just don't get the polarity switch. Even though it's rectified, it will reach 0 and shut the SCR off.

You have to take the volt-seconds out of the core somehow. Primary side, or secondary side, once you load up the core you need to take out the energy before the next pulse. Otherwise you effectively have net DC, and it will saturate.

A transformer cannot have more than a small net DC in any winding unless specifically designed for it.

EVguru
12-15-2011, 03:41 AM
Problems with DC offsets causing core saturation is why old sinewave TIG setups had a huge capacitor bank in series with the torch.

I used to have one. I nearly threw it away a couple of times, but it was too heavy to put in the car. I left it out in the back lane and even the local scavengers gave up on it after a dozen yards or so. I managed to give it away in the end.

TheAndroid
12-15-2011, 09:09 AM
You have to take the volt-seconds out of the core somehow. Primary side, or secondary side, once you load up the core you need to take out the energy before the next pulse. Otherwise you effectively have net DC, and it will saturate.

A transformer cannot have more than a small net DC in any winding unless specifically designed for it.

Agreed, that's why you just can't use any old transformer or pulse frequency. The combination has to allow the transformer to reach zero, loaded or not. This is why some boost type switched mode power supplies will modulate the pulse stream to the transformer in DCM mode. They'd never work otherwise as the catch diode would always be conducting thus making it a fancy input voltage matching power supply.

Nonetheless, the assertion that transformers require AC is still false.

macona
12-15-2011, 10:10 AM
Problems with DC offsets causing core saturation is why old sinewave TIG setups had a huge capacitor bank in series with the torch.

I used to have one. I nearly threw it away a couple of times, but it was too heavy to put in the car. I left it out in the back lane and even the local scavengers gave up on it after a dozen yards or so. I managed to give it away in the end.

Never see a cap bank on a tig machine before. MIG machines, yes. Never on a tig machine.

macona
12-15-2011, 10:14 AM
You could copy this guy's design. He uses IGBTs which are much more suited to this project.

http://www3.telus.net/public/a5a26316/TIG_Welder.html

lakeside53
12-15-2011, 10:57 AM
Nice link... good to see guys tackling such projects.

J Tiers
12-15-2011, 11:46 PM
Nonetheless, the assertion that transformers require AC is still false.

Or, alternately, "oversimplified"........... Trying NOT to explain how to make a watch.......

SMPS modulate the pulse stream? Who'd a thunk it? ;)

You can reset the core many ways..... AC works well at that.

TheAndroid
12-16-2011, 03:40 PM
We can settle on oversimplified.

I just happen to be into this stuff right now as I'm trying to figure out if a modulated boost power supply would be better for an EDM rather than the usual linear type charge/dump style like Langolais used.

Although he modulates the charge, at its heart, it is still a linear supply. I simply don't want to mess with all those transformers and capacitors!

darryl
12-16-2011, 09:45 PM
This reminds me of the electronic ac voltage regulators made for those with line voltage fluctuations that could risk expensive gear. These regulators basically measure the input voltage and switch in an appropriate transformer winding using triacs. There would be a triac for each tap on the winding, with the best tap selected to maintain the output voltage within nominal limits. It's done with zero voltage switching, with the transition from one triac to another done without skipping any ac cycles. Only one triac is on at any time, and they are either fully on or fully off, so there is only a small amount of switching noise, same as a rectifier would give.

If the primary of the welder transformer had a few taps, one could be called 'full power', and one or two others would give levels of reduced output. You might in fact be able to get away with only one other tap, so you'd need two triacs (plus associated circuitry). This would go a long way towards helping you control the welding current. You would be able to start a weld at the edge of the metal, for instance, switch to full power for the bulk of the weld, then switch down again as you reached another edge.

Since you would normally select the current level on the welder depending on what you're welding, this might be the only other control you would really need.

It does of course require that you either have a suitably wound primary, or that you can or would be willing to add a few turns of wire to the core.

Another way I've seen used is to pass the welders primary current through a few turns on the secondary of a smaller transformer. The control consists of either shorting or not shorting the primary of this second transformer. With it shorted, the secondary winding has low impedance, and with it open the secondary has a higher impedance. This primary has no connection to ac power- in this case it might be more apt to call it a control winding instead of a primary winding. The current that would flow in the primary of this second transformer is much lower than the current the welder is drawing, so a typical 15 amp microswitch would probably suffice. The beauty in this method is that no electronics are required.

J Tiers
12-16-2011, 11:33 PM
The standard system for a welder depends on whether it is constant current (GTAW, SMAW) or constant voltage (most GMAW).

An "inverter" welder will likely have an output section which is an SMPS set up for whichever type is required.... a buck-derived type, likely a "full bridge" with transformer to the lower output voltage.

The smps will modulate the "bridge" circuit so as to control the output current (or voltage). Fairly standard stuff but rugged and higher power/current than most common SMPS.

In any high power switching situation where applicable, SCRs will typically be used in place of triacs, since their characteristics include much better ruggedness, and less trouble with false turn-on etc. SCRs , and before them, Ignitron tubes, have been used for much the same purpose as the full bridge SMPS, and they work well.