90VDC PWM motor control advice.

[SgtB]

I have a treadmill I've been trying to get running for quite some time. It originally had a bad motor controller and I have replaced it with a couple of industrial drives over the course of about a year. Both have met their fate from a simple error on my part while prototyping.

Originally, I used the treadmills original control panel and an 08m to translate its speed pulse signal to a servo angle, and linked the drives pot to the servo. It worked well, but is a little clunky, and honestly I'd like to scrap everything but the motor and start over.

I have a IR P102W SCR pack from my last drive that is good, but unless anyone has a straight forward idea how to drive it with a micro I'm probably going to abandon it all together.

My thought is that a FET and FET driver setup with PWM from the Picaxe would be the most straight forward approach. The motor is 90VDC with a FLA of 10A. I'm having trouble finding a good circuit to base off of, and honestly need some help with selecting the FETs and driver. If anyone has some suggestions,stories, and application notes I would greatly appreciate your advice. Thanks.

 

[fernando_g]

The reason for 90 VDC is that if you half-wave rectify a 120 VAC you get an average of 90 VDC.

Why half-wave rectification? Because it was done with a single SCR. Phase-control (PWM'ing a sinewave) would adjust the voltage.

As such, this motor may be optimized for 60 Hz chopping.

 

[SgtB]

I understand the motor theory and whatnot. I'm just curious on how to vary the speed with low level logic. I'm trying to not use a pot, but the drives I've used all use 24v reference, and I really am trying to get away from a digital pot. The drive I used uses a pair of scr's and a freewheeling diode in the pack. From what I see though, you need a resistor to control the angle of an scr and my tiny mammalian brain can't figure a simple solution to control that with a picaxe. If there is a simple way to drive the scr pack and have good speed control I'm interested, but from what I read FETs would be easier for me right? Thanks again!

 

[fernando_g]

The software aspect of phase control via a microcontroller entails a couple of things:

1) Detecting the zero-voltage crossing.
2) Apply a delay before the firing pulse for the SCR.
I would use an SCR...they are far more rugged and turn off by themselves every zero crossing. As such, the firing pulse width is not very critical, as long as you meet the SCR's minimum latching time (from the datasheet)

Hardware wise, one must be careful to isolate the low voltage logic from the mains-derived voltage. The best way to fire a SCR is with a photo-triac coupler (MOC3xxx series)

 

[SgtB]

Ah, that's painfully obvious now that you mention it. I was trying to figure out how to deal with the scr's latching characteristic. Thanks for the advice on the driver. I was a little worried about switching the gate with so much potential on it. Now I have to do my homework on zero crossing detection.

 

[fernando_g]

Forgot to tell you about the photo-triac coupler. It must be of the random-triggering (non-zero crossing) device.
If you use a MOC3011, it also only requires 10 mA to fire, and as such it can be driven directly from a picaxe output. (with a limiting resistor, of course)

 

[SgtB]

Thanks! My other tab is a search for which moc to get. That was exactly what I was trying to figure out.

 

[gengis]

I'm confused.... Phase control is what you use for SCR and Triacs and PWM for mosfets. You mention you want to try a mosfet, and don't know how to drive it... The first problem I see is that your 90 VDC motor is running on phase control now - and there's an excellent chance (almost certain) that the AC line is not isolated from ground. That means that you'd need a degree of competency to make it safe - and not wind up with something like the bike frame at a lethal voltage. The second (minor) problem is that PWM control requires steady DC voltage and that means rectified and filtered. Rectify and filter 120 VAC and you get ~180 VDC, way more than the motor wants to see. You could take care of that in software providing there's some fail safe method of dealing with a shorted mosfet in the event that happens - or any number of problems you might inadvertently design in (like a connector coming out and full 180 on the motor as a result) If it were me I'd stay with phase control. There's info on the web dealing with adapting a lamp dimmer to a basic stamp. You use PWM and convert that to a steady variable voltage via an "integrator" (fancy name for resistor and cap) then use that level of voltage to control a linear optical coupler which controls the lamp dimmer. That deals with the problem of isolating the 120 volt line from the low voltage control. Instead you'd be using an SCR that fires on half wave and only sends 1/2 the sine wave to the motor so the most it gets is (effectively) 60 VDC. (which may be how they do it - with no schematic or knowledge of treadmills... I can't say) PWM and mosfet would work too - but you'd really want a big (70 VAC / 14 amps, what you need to get 90 rectified and filtered, is physically large, heavy, and costly) isolation transformer, to play it safe for isolation and to make damn sure the input motor voltage could never exceed 90. But it might help to see some schematics of what you have that works or worked, to get a better idea of what voltage the motor really wants and how it is isolated from the power line (or not). Further out in left field you could use a mosfet for phase control also - but that's another whole new set of problems to deal with - or the same problems but handled differently. Is that what your were thinking of doing?

 

[fernando_g]

I can't find the mathematical proof right now, but when one 1/2 wave rectifies 120 VAC, the resulting is an average 90 VDC, not 60 VDC.

And indeed these half pulses will cause torque ripple, but the motor design is such that it accomodates that.

 

[gengis]

fernando_g I can't find the mathematical proof right now, but when one 1/2 wave rectifies 120 VAC, the resulting is an average 90 VDC, not 60 VDC.

That is the peak voltage, and I was using the modifier "effectively 60 volts" for that reason. Rectified and filtered, it is ~90. That is why I'm interested in seeing a schematic... 10 amps (if that's what it actually needs ~ 1+ hp) would require a filter cap of ~20,000 uf for ~ 10% ripple at full load - I've seen a few industrial 90 VDC controllers and they didn't have any filter caps in them.... The formula, if my memory is working, is 1.4142 times the RMS to get peak, or .7071 for the reciprocal.

 

[SgtB]

Good points on all accounts. I'm an industrial electrician/ automation tech. It's nice to get some math and theory behind all of this.

FWIW, I will use the SCR power pack I mentioned as my power control. It's a International Rectifier P102W. (PDF datasheet)It has the diodes, dual SCRs and the flyback all in one pack. I was unsure about the different triggering mechanisms. I will use zero detection and delayed fire to get the speed control I want like Fernando suggested. The build will be assembled with all the proper safetys and current protection. All of that is natural to me. I play with low level at home, and work with medium voltage as a profession, I just have never bridged the gap between the two. I absolutely understand the danger, but also understand how to control it. Isolation, etc. Most of my projects have been isolated at a relay, so I am trying to be as sure as possible as I go forward.

Most of the drives I've worked with either have a power pack like the one of above, which is out of an old AB 1365, or discrete SCRs . When I set them up for maximum speed, I usually set the speed pot to %100, and then turn the MAX pot up until I get 90vdc. In my experience, I am able to approach 120v line voltage with the MAX pot. I don't know if that's a function of reactance, or just freewheel voltage between cycles, but it's there. It's definitely higher than 90v too because the brushes start shedding heavily at those voltages. I'm not an EE though, so who knows. Feel free to enlighten me, this is all very interesting. Hopefully once my wife finishes her masters next year, I can start working on my EE now that I'm old enough to actually stick with it.

Thanks again guys.

 

[SgtB]

Wait, no. It's full wave rectified so that's why I can hit line voltage right?

 

[gengis]

Wait, no. It's full wave rectified so that's why I can hit line voltage right?

That is a common enough way to do it. Full Wave Bridge with two diodes replaced by SCR's. A slightly less efficient way is with an ordinary FWB rectifier feeding a single SCR on the output.

You mention the P102W version - looks like that is the "doubler" version? without the diode across the output? No idea how that is a "doubler" though... I can see how it could control full wave power, just don't understand why they are calling it a doubler. And a voltage doubler circuit capable of 10 amps running at 60 cycles would need some large impressive capacitors.

I'd still employ some means of control signal isolation from the power mains no matter how safe I can make it - even if it is all closed up, sealed, bullet proof, water proof, ground fault protected, and "people safe." The simple expedient of signal isolation can often save a lot of heartache with things like noise and ground loops causing problems in the control circuitry.

 

[SgtB]

I'm going to use opto-fets with the main drive components in their own metal enclosure for noise suppression. Most of the drives do not employ a smoothing capacitor, but I will be using one I have handy. I don't believe it's the doubler one. That diagram confused me at first too. It's a p1x2, the one on the left with the freewheel. The AB schematic would suggest so.

 

[SgtB]

So I’ve got the Z.C. circuit figured out, but now I'm worried. I didn't have enough bench time to just "try and see" last night, but I believe my resolution will be a problem. (1000/60)/2 = 8.33 which isn't a whole lot of resolution. I've considered skipping cycles since it's a motor with a flywheel, but I'm not sure about that just yet. I'm considering an x1 or x2 part just for the clock speed too. What can I expect here, and do you have any best practice advice? I've looked at all sorts of command and hardware setups and am starting to confuse myself. I've also considered a raw picmicro just for a speed driver. Thanks again.

 

[fernando_g]

Indeed, the Picaxes, at least the lowlier Picaxes, don't have the computational speed to provide enough resolution within the 8.33 msecs window that a half-sinewave allows you.

You most certainly will require an X2 part running full blast, or go to a raw PIC approach.

Time to experimet a little with some actual hardware.

HINT: after you've completed your prototype, test it without the motor being actually powered up. By this, I mean that with an oscilloscope, in the line trigger mode, is used to verify that the pulses are correct, that they don't have excessive jitter or that they are missing altogether. Sweep you control, and check for jumpines or any othe anomalous behavior. Only after you are satisfied with the operation, then and only then power up the motor.
This simple step will prevent a lot of magic smoke from being released.