[Technologic]

Well I started another thread cause we were nearing 30 pages... and my raw schematic is "vaguely" close to stress testing and part sheets.

I'm not particularly happy with it, for lots of reasons. Though the component choices are excellent (if I do say so myself) it's not cheaply priced enough.

http://img406.imageshack.us/img406/7552/part1e.png

Who would have thought the $17 MCK201 DSP would look cheap at this point? well the gate driver and other components all tallied up are close to $80 in ICs.

I want to drop this to about 1/4th that for the parts minus IGBTs... and the only way I can see doing that is going to a position only sensor V/Hz model...

I must say that the MCK201 is amazing though... it allows for simple motor input parameters (ie. max RPM, inductance etc) and compensates the programming... even allowing for different PWM freq for different motors... Full on vector control without worrying about modifications.

The registry can be updated from any computer with a simple excel sheet they give you through the serial port... really nifty stuff.

The torque command is giving me issues... I'm trying to figure out how to do the throttle on it currently (Hall sensors will be coming in soon but those are simple)

The optoisolators on the current sensors are inverting... I'm sure this will be a problem since the 22771 isn't inverting

They don't give you enough parameters for this stuff... I've been contacting IRF about the 22771 and they keep giving me incorrect answers.

At any rate I forsee setting this aside once I get a solid design and starting in on a simplified (very reasonable 100-200 dollar finished) V/Hz AC controller, since this would be much faster.

This design, however, has been a good learning experience... the IGBTs for it really should be like 1800v x 800A ones if you want to use the full 1200v and 600A it's designed around so far.

Output stage will be on a separate board and in a separate case if you value your motor

Cheers... input always welcome
Just be gentle if at all possible

Edit: I dropped the optoisolators from the current sensors... after rereading I noticed the input for PO (IB1 and IB0) don't need a higher voltage.

[etischer]

Couldn't you just put your current transducer in backwards (current flow in opposite direction) to compensate for your inverting circuit?

Why are you worried about $80 in ICs? Most EV inverters are priced around $4k.

[Greenflight]

What did you have in mind for a source for the MCK201? I wasn't coming up with extremely credible sources on my google searches... Sounds like a nice IC though!

I've been playing around a little with the Freescale MC3PHAC. It's pretty much useless for anything but industrial motors running V/Hz, but it's good for educational purposes.

[Technologic]

Originally Posted by Greenflight

What did you have in mind for a source for the MCK201? I wasn't coming up with extremely credible sources on my google searches... Sounds like a nice IC though!

It's beyond nice.

I can grab samples of it from IRF, the problem is people DIYing with it (building these one at a time) is out the question.

I found sources to get them in 100s...which is fine for a DIY controller (ie. sell them with PCBs etc).

Couldn't you just put your current transducer in backwards (current flow in opposite direction) to compensate for your inverting circuit?

Why are you worried about $80 in ICs? Most EV inverters are priced around $4k.

I'm worried about it considering how pricey the components are (IGBTs) and I still need a contactor for the main DC current... it's all just rather expensive when you are shoving 1800v x 800A IGBTs in as well .

This controller would really need to be preassembled to make it viable for DIYers, which is fine, I just don't want to shove capital into such a project

The vector control isn't practical in my opinion, and it's also clear that most commerical AC controllers (ie. curtis and zapi) probably aren't using it.

And no, that won't work for the inverting optocoupler... it's not really a good idea to have it anyway... the PO output is being pulled up to 5v which is too high for most optocoupler inputs in the first place.

If you judge "practical pricing" by 3 times the component costs ... this controller should cost (roughly) twice-three times as much as a DC controller that can handle the same current range/voltage.... which would put it at $8000... totally unacceptable in my opinion.

[etischer]

Originally Posted by Technologic

The vector control isn't practical in my opinion, and it's also clear that most commerical AC controllers (ie. curtis and zapi) probably aren't using it.

Why wouldn't it be practical? With V/hz all you can do is control speed, with vector control you can control torque. Plus you get much more low speed torque.

IF you're trying to save $60, why not buy smaller IGBTs? 800 Amps is insane, what kinda motor are you running?

[Technologic]

Originally Posted by etischer

Why wouldn't it be practical? With V/hz all you can do is control speed, with vector control you can control torque. Plus you get much more low speed torque.

It's not practical cost wise... of course it's better (Torque control for throttle response is also useful which the MCK201 allows).

A V/Hz model would give you graphs like (unless I'm mistaken):
http://www.electricmotorsport.com/st...-induction.php

Likewise you need encoders in the motor (rotor position sensors), hall sensors, etc for the thing to run properly...Also these things have to be set to specific motors via the on board registry interface... and it's not a simple point/click thing. Things like motor inductances etc must be input for max efficiency

I might still use the MCK201 and step down the design to survive 600v instead of 1200v, which would cut the costs by half for the parts and allow $50 or less IGBTs to be used.

The problem is the sensors/encoders... most motors don't have them installed... plus it requires an external throttle IC chip (though not too expensive) certainly increases board layout costs and complexity.

In the end this particular model after the PSU is added will have a 500-800 part count, not including any bus bars or temp cut offs etc.

The design was me building the largest and beefiest motor controller that had parts available... I never really intended for it to be built and sold. 1200v+ rated gate drivers, current sensors, 1500v rated DC filter etc... all VERY expensive.

[etischer]

Originally Posted by Technologic

It's not practical cost wise... of course it's better (Torque control for throttle response is also useful which the MCK201 allows).

A V/Hz model would give you graphs like (unless I'm mistaken):
http://www.electricmotorsport.com/st...-induction.php

I don't think it is possibe to get 100% torque at zero speed without vector mode.

Think of it this way, at 100% speed you are supplying 230vac at 60hz. At half speed 120vac at 30hz. At 1% speed 2.3vac at 0.6hz. you can't expect 100% torque when supplying the motor 2.3 volts. You can put some DC boost in there, but I don't think you can get close to the performace as a true vector setup.

You can do sensor-less vector mode which does not use an encoder, it just needs current feedback.

[Greenflight]

Have you looked at the double, triple, and hex pack IGBT modules? The 10+ price for a Powerex 1200v 600A module us under $300... 3 of those plus your ICs and parts and you're probably looking at keeping it under $1200. If you're dead set on 1800v and 800A you're probably gonna be spending some more $$ though.

And of course, you could always use a bunch of 50-100A IGBTs or MOSFETs. That'd bring your price down a bunch.

As far as not using vector control- I don't know how well that's going to work. V/Hz is, as far as I know, only useful in applications where there is going to be little difference between the theoretical speed of the motor based on voltage and the actual speed. In a car, it's not uncommon to floor it from a dead stop, which would lead to serious timing issues unless you had incredibly slow throttle ramping. Unless there's something I'm not getting?

[Technologic]

Originally Posted by Greenflight

The 10+ price for a Powerex 1200v 600A module us under $300... 3 of those plus your ICs and parts and you're probably looking at keeping it under $1200. If you're dead set on 1800v and 800A you're probably gonna be spending some more $$ though.

And of course, you could always use a bunch of 50-100A IGBTs or MOSFETs. That'd bring your price down a bunch.

As far as not using vector control- I don't know how well that's going to work. V/Hz is, as far as I know, only useful in applications where there is going to be little difference between the theoretical speed of the motor based on voltage and the actual speed. In a car, it's not uncommon to floor it from a dead stop, which would lead to serious timing issues unless you had incredibly slow throttle ramping. Unless there's something I'm not getting?

I always intended to, if designing a V/Hz model, limit the amount of stop current the controller would allow to pass through by controlling the throttle via an IC controlled amplifier (with a significant delay and low faulting overcurrent setting).

V/Hz works just fine in theory for a car, just doesn't fit the "racing" mentality.

However I will attempt another go at a cheapened down version of this once this one is mostly finished... at the moment though it's difficult to explain just how much more expensive that design is than say the Zilla controller.
Just think parts counts in the 4-10 times range (IC prices at the 8-10 times expensive range) and 4 times as expensive (than the 1000 amp version) output stage... at a minimum... 3 separate PCBs will also be needed

I don't intend to source IGBTs from the US... just way too expensive.

If the Zilla's 1500-1800 price was the market value, such a controller would be $5000-8000... clearly I need to shorten things up a bit...

[aeroscott]

from my reading svm lets us use less robust motors (non inverter rated)