Hijack of Evans stepper driver thread

Under $20. I would love to see the design.

Do you have a link?

This is a link to the design provided by Mariss F. on cnczone.
Its a complete tutorial on the coolrunner XC2c256..the ide..and a blow by blow explanation of the design..
I believe it to be in essence a 0.75 Gecko (G203?)design...it does not incorporate mid band resonance control..

The complete verilog code has been tweaked and posted on cnczone by pinmo.

http://rapidshare.com/files/370379490/5_microstep_tutorial.pdf

On my version..I have sacrificed the high voltage capability of the discrete H bridges and used MC33887 H bridges on 24v. I made a Fxxx up though..the MC33887 is only rated for 10kHz..but I'd already built the boards ..but they work fine..I hind sight I may have been better off using the Nat Semi H bridges.

Rob

[Paul Alciatore]

The L297/298 combination produces a horrible racket of squeals and squelches..like having a bag of angry cats next to you router.
The increasing current slope is always shallower than the decreasing slope.which leads to unstable waveforms and the noise.

You should research the free 5 microstep TinyCPLD coolrunner based design Mariss F. The increasing current slope is steeper than the decreasing current slope leading to a stable waveform and virtually no noise under 20khz.

I used the Mariss front end combined with dual motorola MC33887 5A 50v bridges with internal current sensing.Much better imho.
Mariss's 5 microstep tutorial and pinmo's code are available on cnczone.



Using the Marris disign ..I can get in at under $20...excluding time.

Rob

Under $20. I would love to see the design.

Do you have a link?

[EVguru]

The L297/298 combination produces a horrible racket of squeals and squelches..

They also seem to leak smoke without much provocation and are a pretty dated design these days. They must be what, 30 years old?

[John Stevenson]

I don't have time to argue Paul, I'm going to town for a blood test.

Ask for nurse Dracula

I have recently designed a stepper driver board using a different chip, an L298 by SDMicroelectronics.

The L297/298 combination produces a horrible racket of squeals and squelches..like having a bag of angry cats next to you router.
The increasing current slope is always shallower than the decreasing slope.which leads to unstable waveforms and the noise.

You should research the free 5 microstep TinyCPLD coolrunner based design Mariss F. The increasing current slope is steeper than the decreasing current slope leading to a stable waveform and virtually no noise under 20khz.

I used the Mariss front end combined with dual motorola MC33887 5A 50v bridges with internal current sensing.Much better imho.
Mariss's 5 microstep tutorial and pinmo's code are available on cnczone.

As for cost, it would be really difficult to get any kind of decent driver board for less than $30

Using the Marris disign ..I can get in at under $20...excluding time.

Rob

I made a stepper driver for bipolar motors using a bunch of half H bridges with 60 amp fets. Was a lot of work and high PWM frequencies would FRY my mosfets... I can do full and half stepping. I am too lazy to try micro-stepping.
Used AVR atmel mcu for it.

[Evan]

I don't have time to argue Paul, I'm going to town for a blood test.

[Paul Alciatore]

I have recently designed a stepper driver board using a different chip, an L298 by SDMicroelectronics. It is good for 2 Amps and about 46 Volts so that comes to about 90 Watts. But you can combine two of these chips in parallel to get almost double the current and power. A second chip, the L297 is used for all the control circuitry. But I don't know if 175 Watts qualifies for "higher power" in your view. The board has provision for either one or two chips, as needed. I think there is a lot you could do with that and although my first project is for a Unimat, I hope to use them on my full sized lathe and mill. These chips also have a bunch of bells and whistles built in to them, including current sensing and chopper to control it. When using them, you basically worry only about the current capabilities of the motor, and not the Voltage rating. When properly set, the chopper cuts the output before the current is too high to damage it. So a 12 Volt motor can be run on a 24 or 36 Volt supply for faster response.

My board is 2.5" X 3.8" and costs about $20 to have commercially made in small quantities. Of course, the parts are on top of that but it can be assembled for about $45 or so depending on weather you use just one or two of the driver chips.

I haven't tested the design on a machine yet as I need to make some additional control circuitry and also get a computer hooked up and running some software. The control circuitry is for manual control on an immediate basis, until I can afford the software. But the board, with just a single chip assembled on it, has run some small motors and the torque was very good. I think it would be sufficient to run any machine that I have, perhaps with a 2:1 step down with gears or a belt drive. One disadvantage if this chip/board is that it is limited to half steps, no microstepping. But I don't really trust microsteps anyway so I don't consider it a very big consideration. When you use microstepping, the holding torque for these increments is only a small percentage of the rated torque of the motor and even a small amount of resistance may produce an amount of lag that is many microsteps in magnitude. In short, the positional error for a microstep can easily be several humdred percent of the step size. To me, the oinly real reason for using microstepping is smoother acceleration and motion. However, by avoiding microstepping I can use smaller motors and smaller drivers. The other tradeoffis speed. A motor operated in half step mode and geared down for fine movements will have a slower top speed and spend more time getting from one side of the work to another in rapid slew mode. You can't get around this as the motor will just reach a point where it simply stalls.

These problems with microstepping may be the real reason why such large motors and drivers are often used. If you are only getting 10% or even 5% of the motor's rated holding torque, then you need to use a much larger motor, 10 or 20 times larger.

I am sure there are cheaper ways to go. A few standard logic chips ($0.50 to $1.50 each) with some high current transistors or FETs would work to produce the pulses needed, but would not provide any of the extras that can improve the performance so much. You would have to use a power supply that is the exact Voltage required by the motor or you would either loose torque or risk burning the motor out. This is because there would be no current sensing or limiting. Ths same company (STMicroelectronics) has a simpler H Bridge chip that could be used for just increasing the current from a logic chip's output. I forget the number of this device.

I am documenting and photographing the project as it unfolds and plan to both publish an article with full details and offer the boards for sale when all is proven to work. I will also ask that the files for ordering the boards be posted on the magazine board for download by anyone who would wish to have them made that way. They are for PCBExpress and can even be modified with their free design software. But it may take a bit of time for this to come to pass as my project time is limited and I have other, more pressing things at present.

As for cost, it would be really difficult to get any kind of decent driver board for less than $30, even if you do all the work yourself and have some of the parts already in stock. Just making your own bare board would cost at lease $10 if all costs were taken into account. (I am sure Evan will argue with this, but what is the deprecation factor on his CAM setup? And the cost of elecricity to run it? Etc.)

[Evan]

That is one of the reasons for using an all on one chip design. It has all the inbuilt protections and all you do is hook up the wires.

[BobWarfield]

I was going to over simplify and view that H-Bridge as the "amplifier" that John covets. Taking an existing drive and adding more circuitry to make a bigger drive is ultimately going to be more expensive than designing a more potent drive to start.

Cheap is I think a matter of time and perception.

Things have gotten measurably cheaper just in the short time I've been playing with CNC. The cost of a Gecko G540 versus discrete Geckodrives is down quite a lot, for example.

And, there are quite a lot of "cheaper" (note I don't say cheap) high capacity servo drive designs on CNCZone. Compare their cost to commercial drives (particularly new, but even many surplus) and they're pretty cheap.

However, as is being alluded to, electronics are fragile and CNC is a somewhat hostile environment for them. The chances even a pretty good EE gets their first time drive design right are not great. You have to make a lot of mistakes and learn from them. Even Mariss F. does that at Gecko. So far as I can see, the best approach is a group of talented engineers building enough of these drives and interating the design that they get better. That seems like what happened with the drives like the Granite drive.

Otherwise I think what you thought of as a cheap drive can wind up being easily blown up or not performing as well as you'd hoped. Of course the Chinese are copying everything they can get their hands on and driving prices down too!

Cheers,

BW

[Bruce Griffing]

To add to what has already been said - The higher voltage is one of the real cost issues. Most integrated circuit lines today produce devices that run on voltages less than 5 volts. This is because high levels of integration - lots of devices in a small area - requires low voltage. If you want to use an 80 volt supply for the stepper, then you need devices that can handle more than that voltage. At that point you have a couple of choices. You can use a line that makes HVIC's (high voltage integrated circuits), but this process is much more expensive. Your second choice is to integrate most of the functionality in a low voltage IC and add external components. If the H bridge were the only external component, that would be almost reasonable, but you also need what are called high side drivers - devices to drive the high side of the H bridge as well as some sensing circuitry. With that combination you can do it, but I think that comes down to a problem of demand. The vast majority of stepper motors can be driven with the fully integrated devices. There are not many commercial uses for steppers as big as John wants to use.

[EVguru]

I've played with DC motor controllers up to 600,000watt and tried to build my own 144 volt (nominal) 400 Amp controller. Lots of people think they can do it becuase they've built a 48 volt 200 amp version and will just scale it up. The trouble is that when a high voltage power stage fails, it tends to do so rather spectacularly and wipes out the evidence of why it failed. Trying to get clean traces on an oscilloscope so you can 'see' what's going on is an art in itself when you're dealing with high dI/dT (500amp per microsecond and up).

There's also the danger the developer is exposed to!

[Circlip]

Lot to be said for damn great glass bottles wi fires inside 'em

Regards Ian.

[J Tiers]

All the control stuff essentially costs the same regardless of power. There may be more control and monitoring applied to a "high value" drive and motor, but they are not 'essential" to functional operation.

The key is in the integrated circuit. They tend to be priced according to silicon area..... more = higher. You have to charge more because with larger area, you could have made something else of higher value, and also your yield is reduced, netting out to fewer devices per wafer processed.

A high power driver is also higher voltage, typically.

High voltage circuitry in an IC takes up more room, more die area. And higher power, higher current devices take up more room.

Higher current devices also need heat removed, and that becomes less practical in an integrated circuit when you start having to remove larger heat loads.

So then you have to go to separate power devices, which means EACH has a cost of the die, packaging, etc. then there is the cost of heatsinking arrange,ments, larger copper clugs as a base, etc.

The labor cost of putting them together is higher, and you must add circuitry, because some sorts of techniques work very well in an IC, but not at all in separate devices. In an IC, you can accurately proportion parts for proper function, allowing shortcuts in design. Not so with separate devices, you must add parts to perform functions which are almost inherent with an IC.

As soon as you cannot put it all on an IC, the trouble starts.

However, certain parts are made to do a full "H" bridge driver up to say 70V or so. I have used them in designs. They may now be obsolete, but they were esssentially a "just add mosfets" device.

The HIP4080 was one such device, there were others in the same family with more general applications.

[Evan]

That is a pretty good question John. I can buy retail a 1000 watt PWM brushed motor controller for $54 USD. That isn't a lot different from a stepper driver in terms of parts count and the type of parts used.