25 amp motor controller help

Thread Starter

RobbJohnson

Joined Dec 15, 2009
76
Ok, here is a question? Should we monitor the heatsink or the motors? My original design was to monitor the motor temp and shut down if a track got stuck and stalled the caused the motor to overheat. Then I went to current, and then temp of the heatsink.
 

SgtWookie

Joined Jul 17, 2007
22,230
Bad news - my computer took a dive on me. I think the drive controller went; it recognizes the HDD, but isn't reading from it.

I'm on the wife's computer for a few minutes.

Until I get that back up and running, won't be able to make much progress.

The motor current will give the most rapid indication of overload. Monitoring the heat sink temp isn't a bad idea.

Have you seen the VNH2SP30 and VNH3SP30 from ST Microelectronics? They incorporate a full H-bridge and PWM. The VNH2SP30 has a current sense feature and lower Rds(on) specs than the VNH3SP30 does. I'd forgotten about them - I have a couple of these IC's I'd ordered several years ago, but never put them to use.

Here's a company that offers them on a carrier board:
http://www.pololu.com/catalog/product/706
... but their implementation would not be useable for your situation; as they omitted the heat sink area from the board. Too bad.

Now that we've gone along for a good while, I'm kicking myself for not asking this earlier; is the 22A rating for your motors at stall, at rated HP, or running at no load?

Stall current (or current during operation at low-to-moderate PWM%) could get quite high.
 

Thread Starter

RobbJohnson

Joined Dec 15, 2009
76
I was not able to find the specs on the motor. I am playing with several motors and used the 22 amp spec because that was listed as load current. Most motors so far have been less than that but if we stick with some thing the mosfet will handel I believe we will be in good shape not matter what motor we end up with. Lots of motors out there but no specs.

Sorry about your computer. I lost my hard drive this week on my Lab computer so im rebuilding also.

No hurry, This is a project with no schedule, It gives me a chance to do more research.
 

SgtWookie

Joined Jul 17, 2007
22,230
Gee, I made a reply last night and it posted, but it's gone now. :confused:

Got my computer fixed; replaced a flaky drive cable and had to reinitialize the bios to default settings; ran a diagnostic scan with SpinRite; that took care of everything.

Darn sorry about your HDD. Hope you didn't lose anything important.
 
Last edited:

Thread Starter

RobbJohnson

Joined Dec 15, 2009
76
Well, we have mixed results. seems to work well except,
For a pulse in of about 12uS (18% duty cycle) we are getting 51uS gate on time (77% duty cycle) so we dont have much control.

Can we use a zener or schotky in the base of Q1 to cut it off faster?
 

Attachments

SgtWookie

Joined Jul 17, 2007
22,230
Somehow, I thought you were using a much slower PWM frequency - don't know where I got that idea? I guess you're somewhere around 10kHz - 15kHz. If you drop the frequency down to 1kHz - 1.5kHz, the error will be quite a bit less.

At least what you are seeing is pretty much in agreement with what I was seeing in the simulation.
[eta]
Are you really seeing just around 6v p-p on the gate? :confused:

I've tried a few different ways to speed up the opto's response - it's darn slow. :(

I've tried the diode thing - not good. A comparator that triggers just when the opto output is near +dvr could work, but now we're talking adding more complexity....

Did you happen to acquire any of those HPCL-3120's? If so, perhaps the thing to do is to use a DC-DC boost converter to get it's Vcc up to 20v; then it'll have the characteristics you're looking for.
 
Last edited:

Thread Starter

RobbJohnson

Joined Dec 15, 2009
76
I'll get back to you I am seeing something funny between the proto and the breadborad circuits.
I am at 15KHz. everything I read said 10 to 20 KHZ. I slow er down and see what happens.
 

Thread Starter

RobbJohnson

Joined Dec 15, 2009
76
Just for fun I tried a single rail comparitor. Don't know what the problems might be, but with just 3 components (probally there is more to it) this is what it looked like. My first attempt at a simulator.

I try to slow down the PWM in the morning.
 

Attachments

SgtWookie

Joined Jul 17, 2007
22,230
This is one of those situations where a simulator's results won't work with real components.

Have a look at a datasheet for an LM193/LM393/LM2903. You'll find that while the input common mode voltage range includes ground, it can't "see" within 1.5v of Vcc, or 2v over the full temperature range. Since heat in Nevada in July inside a tank running full power may be a problem (ya think?) you'll have to plan for the worst-case scenario.

Plan on Vcc being down to 10v under full-power operation. Instead of that 1k pot and 10k fixed resistor, you'll need more like 2k on the top (to Vcc) and 7.5k on the bottom.

You will also need hysteresis, or you will probably wind up lots of switching near the trip point.
 

SgtWookie

Joined Jul 17, 2007
22,230
I say old chap, things are looking a bit better!

I am most definitely curious as to why you are only reading 6v on the gate?

From the noise on the green trace (assumed to be the gate), I can only assume that you are not using the MOSFET source terminal as a ground reference? As far as the MOSFET is concerned, that is what determines when it is turned on, and when it is turned off.

Is the driver board +V dropping down to 6v when the MOSFET is turned on?

Maybe there is a voltage drop on either the +V or ground between the battery and the driver board?

Try grounding your probe at the battery negative terminal, and see what kind of P-P voltage you get at the source terminal of the MOSFET on the driver board. If it's under a volt, try using the battery positive terminal as your probe's ground reference, and see what the P-P voltage is at the driver board's +V.

If your MOSFET's gate relative to it's source terminal is not going from nearly 0v to around 10v, you will have heating problems.
 

SgtWookie

Joined Jul 17, 2007
22,230
Recalibrated the scope. A new version of software messed it up.
Here is the new pattern.
I am not to grab the ground at the mosfet so that may be the noise your seeing.
Ahh, ok... but did you mean to say that you did not measure this using the source terminal of the MOSFET as a ground reference? Because that's really what we need to see.

It LOOKS like during turn-on, the gate goes right from 0v to 12v, and then slowly increases to about 14v, which is good ... but then during gate discharge, it slows down quite a bit at 3v. That might be OK.
 

Thread Starter

RobbJohnson

Joined Dec 15, 2009
76
I let the Motor run for about 20 Minutes and no elevation in temp. This is a smaller motor the call it a 100watt motor. It might bhave enough torq, but if not i have some 250 watt motors.
 

SgtWookie

Joined Jul 17, 2007
22,230
Is that 8.3A at basically no-load? Or is that when it's loaded down somewhat?

You know, the lower your PWM frequency is, the less % of the time your MOSFETs will be in that linear (resistive) stage. Would it make any difference to you if you cut back the PWM to around 400-500Hz? Your motors shouldn't care; they'll sing at a lower frequency, but that's about it. Basically, it's just a quick change in the programming of the uC, right?

That will also further reduce the error due to the slow opto turn-off time.
 

SgtWookie

Joined Jul 17, 2007
22,230
The inductance of the wiring will have less of an effect at a lower PWM. The wiring inductance is relatively trivial compared to that of the motor, but it was plenty to mess up your drive signal. It would still cause lots of grief if you had the ground loop in there; but now that the optoisolator disconnected the ground loop, that problem no longer exists.

You should get pretty much of a full range of PWM now - have you tried it?
 
Top