[david85]

Long Post, you have been warned.....

Some of you might remember this thread and for some back round it might be helpful to look over it briefly:

http://www.diyelectriccar.com/forums/showthread.php/diy-ac-motor-controller-super-size-30075.html

Others no doubt are aware of my own personal issues with the BLDC motor controller for my car which has died a second time roughly 2000 km after I brought it back to life from a mild internal fire last winter. See my build thread for details on that saga......

So far I can't quite figure out what is wrong with it this time and I am beginning to wonder if its worth bothering to fix it a second time even if I could. Options I am exploring right now are to try and replace the logic/driver board with an evaluation board but I am not quite sure which one to use.

The biggest stumbling block is finding one that I am sure can control an external driver since most appear to be intended for powering fractional horsepower inverters and are fully integrated with their own built in power stage - which is naturally way too small to push a 3000lb car up to highway speed. I'm wondering if they could be modified to run an external power stage but I'm not sure if it's worth doing this. Opinions wanted!

Of some of the things I found in the process, this was fairly interesting:

http://www.eefocus.com/设计我最赞/blog/08-09/155607_39e2c.html

The ideas in that link could solve my problem for how to drive my existing IGBTs since I *think* the powerex BG2B-5015 units can run the IGBTs I already have. Some reorganizing inside the controller would be needed to physically fit them though. At roughly $100 each, they aren't cheap but it would get me running again and digikey does carry them.

Problem is I would still need a logic board to control the inverter. They show that too in the link, but they also had to modify an existing MCU developer board a few times before they could settle on something......I'm not sure if I could pull something like that off yet.

This board seems to have most of what I'm after since they claim it can run an external power stage, but finding a source proved challenging:
http://www.fujitsu.com/ca/en/servic...microcontrollers/datalib/devtools/91f267.html

And again, I'm still not sure if other boards that come with low power integrated power stage can be modified to drive an external power stage. Gaps in my knowledge are rather annoying in that area.



I stumbled over these during my research when a link showed up on an ecomodder discussion that google found for me:

http://www.ehitex.de/p_info.php?products_id=544&xID=574668e34207918b45cdf663e9e9cd2b

It's extremely unlikely that I am going to dish out that kind of money at this point but I did find the price to be fairly reasonable considering the 80kw power rating. I have looked over the PDF file here:
http://www.infineon.com/dgdl/AN+Hyb...47094&fileId=db3a30432239cccd01225a7cfb166c8a

and its not exactly plug & play, but all the major hardware is there. Most of the work appears to be in setting up the software to run the motor safely. I have to admire the compact and clean package layout (water cooled).

A smaller 30kw version is available here:
http://www.ehitex.de/p_info.php?products_id=543&xID=574668e34207918b45cdf663e9e9cd2b

The HybridKIT1 and 2 are built on the idea of an integrated single piece 3 phase IGBT power stage, and if you looked more closely in the AC motor thread I posted at the top, Technologic found a similar setup offered by Fuji. It would be nice, but I'm not particularly picky at this point.




Applied Power Systems in the USA does offer an impressive IGBT half and full bridge driver board that is rated for IGBTs up to 1700 amps. Not that I need that much but there seems to be a wide gap in the products I have been able to find with most being under rated for what I need.

http://www.appliedps.com/drivercircuits.htm


No, I haven't called and asked for a price yet, but I probably don't want to know anyway. Besides, that would still be only half the battle and I would still need a motor control logic board.

The general idea I am working on at the moment is to get a basic control setup together that could get the car running again but that could also be reused in the future if I decide to upgrade the power stage.

I haven't quite given up 100% on repairing the current setup, but after the second failure I felt compelled to try putting together a backup.

 

[yarross]

Improve power stage design, use planar DC bus, put closer modules, ddd snubbers and it shuld live much longer

 

[Harold in CR]

Have you checked on the Ecomodders forum ?? mpaulholmes was working on a 3 phase controller experimental build, quite a while back ?? He was also trying to build 2 separate boards, one to control the power board with, so it didn't need upgrades to power a more powerful power board, if the owner wanted to upgrade to higher power, later on ?

 

[tomofreno]

The kit looks promising if you are looking for a lengthy project where you get deeply into the software/firmware, and have the necessary background to take it on. Any chance you could purchase the needed components for your present controller from the company you bought it from? Probably easiest and cheapest, but maybe you are concerned you will just have the same problems over again, and yeah, I know you are not enamored of them, but always keep the end goal in mind. Another possibility might be to purchase another controller that will learn your motor parameters and "autotune" to it. Should have a detailed description of that procedure from the controller manufacturer before you purchase it. Seems the tradeoff will be money or time invested. Hope you find a good solution!

 

[Weisheimer]

I think that you would be well advised to consider IPM modules today.
They will save a huge number of drive "head-aches" and they are not that much more expensive.
Eric Tischer used one in his build/mod of an existing 3 phase controller and has achieved nothing short of stunning results.
I think his thread is titled something like "Homemade 90 HP..."

BLDC has it's own issues, and you may be able to find an existing drive to modify in a similar fashion.

What controller were you using that failed?
What kind of failure did it experience?
Did your existing controller suit you with the exception of this new failure?
Perhaps it is a candidate for a make-over!

 

[david85]

Thanks for the replies. At this point I am open to any and all suggestions.

The setup I had was a little underpowered but did get the job done with enough power to sustain 55 MPH even up gentle hills. The failure happened when cresting a rather steep hill at ~25 MPH. First the current limiter automatically cut out (which isn't unusual in itself since it happens all the time), but when I tried to reapply the pedal it didn't quite feel right. After turning onto a side street it became apparent that one phase was not working right. As it is, I can get the motor to spin but if it stops in the wrong position, it will not restart - this is how I was able to limp the car home.

I had a failure before, which is well documented in my build thread (originated at a loose terminal screw on one of the IGBTs) and I was able to fix it that time but considering the marginal performance and reliability of the over all design, I am leaning toward trying to make something better even if I do salvage most of what I have.

The controller also had other drawbacks as it seemed to operate in a constant torque mode and did not have a very useful regenerative braking. I could still drive the car and was able to get used to it but it wasn't very drivable for some one not experienced with it. With all these issues I wasn't confident letting other people drive the car.

Even if I was able to get a new controller/driver logic board from the OEM, the tech support would be horrible if something should go wrong again. At least with a developers kit, there is some documentation to go on and even the possibility of getting some one on the phone that speaks english. Not to mention having a larger knowledge base to call upon if I use an off the shelf PLC. The controller was a purchase direct from mainland china. I would consider buying IGBTs from china but am reluctant to get finished assemblies like a controller board.

What is meant by "planar DC bus"?

 

[Weisheimer]

I'll look at your build thread later tonight, as I am pressed for time at the moment.
I think I remember your thread about the controller issues, the screw was loose, and you had arcing that damaged a cap and such?
Planar DC bus...otherwise known as "laminated bus".
You have bus bars in your controller now, and so does the one shown in your pix above.
A laminated bus is plate material separated by an insulated (mylar) layer, one plate above the other.
It has a lot of advantages, but is more difficult for the DIY'er to implement.
Not impossible, but more difficult. Probably best done if you have CAD and waterjet services available.

Cheers
Mark

 

[david85]

Ok, I have read about laminated bus bars. APS mentions that in some of their product descriptions.

CAD is part of my line of work and our shop does have a CNC machine that could probably do the cutting out of a copper plate if it came to that. Not as good as waterjet cutting but we have been able to cut aluminum in this manner. It would make for a pretty high end product if we could pull that off...

An arc happened right at the terminal for the IGBT. A snubber was reduced to molten goo, the outside case of the IGBT was burned and some parts on the driver board also had to be replaced before it worked again for me.

I wouldn't worry too much about my build thread since its very long. If you have specific questions I can just answer it here if that would be easier.

Some one suggested this motor developer board:

http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1406&dDocName=en537020

I'm not sure if that can run an external power stage but I still need to do some more reading. Price is fairly decent though.

*sight* and on that note, I have to head back out to build another job for tomorrow.

 

[bga]

Hi David,
This sounds like my project.

The battery is 160 x 40AH TS cells (512V)
Charger is 2 x Elcon 288V
Motor is a 18.5KW ABB 4Pole, rewounbd for 220V delta operation

I'm using a DSPIC30F6010 controller on a custom board based on the Microchip MCDM? board.
I have sourced the following high power components:
IGBTs 300A/1200V (CM300DY-24A) from Powerex.
Gate drivers also from Powerex.
Film capactors from EACO capacitor, China (6 x 390uf/900volt/50A RMS)
Better then electrolytics because of exremelylow ESR, the above is about 500 microohms. I expect the overal ripple to be similar owing to the lower capacitance.

IGBT snbbers 2uF/1000V per IGBT, also from EACO.

The pics in this thread and its links from the Circuit Cellar project described about 2007. it is also known as the Camosun College - EV Drive team

I analysed their source code and it looks like almost all of the variation is in the PC interface module, the main body is virtually identical to the Microchip reference application. There is no EV logic or setting management stuff in the code.

The DSPIC30F6010A has a good amount of code space and speed beyond that needed for the 3PH PWM generator, so there shouldn't be any big problems in getting an EV application to work nicely. About 1/2? of the code space and 1/3 of the clock cycles are used.

 

[Weisheimer]

I finally understood your present controller after reading your posts in the AVEA forum.
I almost didn't even follow that link because it seemed to be about formula data!

The primary control is an MC33035 which is fine, but it is not a programmable controller, per se.
You might be able to increase performance by tuning the feedback circuits.
I suspect that solution will not bring you the results and control that you really want to have.

The Microchip route has a BLDC library, and that would appear to be a preferred solution.
I still suggest that you consider the IPM module route, with built in drivers and temp sensing.
Perhaps:
http://www.pwrx.com/Result.aspx?q=PM300CL1A060
It can be had for appx US$400.00

BGA has already suggested using Metalized Poly caps (on the DC bus) that are well suited to handling the ripple current.
I didn't see a price for the Eacon, but consider this from Digi-key:
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=478-5823-ND
A bit spendy, but not compared to your already sizable investment in batteries!
You would only need one!

BGA, how far along are you on your design at this point? It looks interesting.

Mark

 

[peggus]

You might be interested in the tumanako open source inverter project:

http://sourceforge.net/projects/tumanako/

It is based on an stm32 arm cortex processor, the development board is only $200 and there are open source tool-chains for it.

 

[david85]

I already have data sheets printed off for the motorolla MC33035 IC and did some reading on that a while back. The main board in my current setup does have what I believe are trim pots on each of the 3 main resistor adjusted feedbacks to the microchip, but I never got the chance to investigate further since I wanted to have some proper meters in the car before I took the risk of adjusting anything.

One thing I was never able to verify is if the chip is capable of offering an effective regenerative braking. I think the three adjustments were for the acceleration, deceleration, and peak current limit. In theory the deceleration constant could affect regen, but the datasheet also mentioned the chip was designed to "coast" the motor if rotor speed exceeded the commanded speed, and the actual brake switch could be dangerous since it's basically uncontrolled and amounts to shorting out the motor.

What's strange is I did observe a fairly powerful electric braking effect when the motor was spun above the normal operating range from driving downhill in a low gear. Although again, I had no way to know if power was really going back into the battery in that instance. I guess I'll never know It felt like hitting a speed governor.

I've seen that line of powerex IPMs and they are very nice. Integrating the gate drive into the IGBT sounds like a great way to simplify everything and improve reliability. Not to mention space savings. 300 amps is a bit low for what I need (currently have 400amp modules) but $400 is a decent price for what it is. Its already getting close to what I would spend on external gate drivers that would be needed for rebuilding my existing setup.....

 

[DawidvC]

David, if you wish to use a PIC development board, I would suggest using the dsPICDEM MC1, and using their BLDC library with that board. This should interface nicely with the driver board you indicated, or with an IPM setup.

I hope you get your ev up and running quickly. Let us know how it goes.

Regards
Dawid

 

[yarross]

One thing I was never able to verify is if the chip is capable of offering an effective regenerative braking. I think the three adjustments were for the acceleration, deceleration, and peak current limit. In theory the deceleration constant could affect regen, but the datasheet also mentioned the chip was designed to "coast" the motor if rotor speed exceeded the commanded speed, and the actual brake switch could be dangerous since it's basically uncontrolled and amounts to shorting out the motor.

DS conclusion is correct, and this is not a regenerative braking. After turning on lower transistors, the current circulates through the motor and transistors, dissipating power in their resistances. Fortunately, most controllers allow external PWM control (as a last resort, you can apply APWM directly to brake input pin), and this should allow regeneration: when transistors are on, circulating current rises and stores energy in motor phases leakage inductances, then after switching off, that energy discharges through aniparallel diodes to DC bus.

What's strange is I did observe a fairly powerful electric braking effect when the motor was spun above the normal operating range from driving downhill in a low gear.

This is perfectly normal for speeds beyond base speed. BEMF has exceded DC bus voltage and regenerative current has flowed through antiparallel diodes even in the case the transistors have been completely switched off.
See http://en.wikipedia.org/wiki/Brushless_DC_electric_motor#Kv_rating

Although again, I had no way to know if power was really going back into the battery in that instance. I guess I'll never know It felt like hitting a speed governor.

It was a real regenerative current. Uncontrolled, so it might be dangerous for your controller.

 

[Weisheimer]

...I've seen that line of powerex IPMs and they are very nice. Integrating the gate drive into the IGBT sounds like a great way to simplify everything and improve reliability. Not to mention space savings. 300 amps is a bit low for what I need (currently have 400amp modules) but $400 is a decent price for what it is. Its already getting close to what I would spend on external gate drivers that would be needed for rebuilding my existing setup.....

David,

There are other choices for amperage. The sister module is 450 amps and runs about US$700 as I recall.
You would probably want to consider using liquid cooling with a chill plate, pump, and radiator regardless of which module you used.

You do have 400 amp rated modules in your existing controller.
Do you REALLY get anywhere near that current through them though?
I suspect not.
And you do have a BLDC motor with magnets inside too, don't you?
Be careful not to over-amp the motor and risk losing some of the magnetism.
You probably can't totally demagnatize them, but you can permanently reduce it and lose some level of performance.

Do you have, or have access to, test gear such as an o'scope and peak reading current probe?

All of that considered, it could be a great project to explore.

Mark

 

[david85]

David,

There are other choices for amperage. The sister module is 450 amps and runs about US$700 as I recall.

Hmm........

You would probably want to consider using liquid cooling with a chill plate, pump, and radiator regardless of which module you used.

I was already thinking to do that since I can easily make an aluminum liquid cooling setup in our shop.

You do have 400 amp rated modules in your existing controller.
Do you REALLY get anywhere near that current through them though?
I suspect not.

You are very likely correct. The car struggled to reach 65 MPH and I highly doubt it really needs 50Hp to sustain that speed. I wonder if I was getting around on as little as 20 kw of power? Eventually I should be able to do some comparisons once its running again and I have working meters. Its one of a few questions that kept me up at night.

Although I heard from one source that IGBTs should not be run more than about half their rated capacity so 200 amps would be the safe limit on a 400 amp module. Does that sound right to you?? The infineon spec sheet for the FF400R06KE3 IGBTs states a peak current of 800amps for 1ms. I'm not sure if that means all of the 400 amp rating is safe to use even with adequate snubber protection.

http://www.cntongling.com/guowai/pdf/ff400r06ke3.pdf

And you do have a BLDC motor with magnets inside too, don't you?
Be careful not to over-amp the motor and risk losing some of the magnetism.
You probably can't totally demagnatize them, but you can permanently reduce it and lose some level of performance.

How do I know if it gets over amped? It never overheated, but I realize thats not much to go on.

Do you have, or have access to, test gear such as an o'scope and peak reading current probe?

haha, no - and no. But if I need to I will get it. I'm in too deep to back out now so I might as well start looking for the tools.

All of that considered, it could be a great project to explore.

Mark

I agree, but after all thats happened, there doesn't seem to be much choice.
Thanks for your input, it helps a lot.

 

[david85]

Its been a few days of reading bits and pieces of various datasheets in my spare time and I'm now 90% sold on getting the dsPICDEM MC1 board:

http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1406&dDocName=en023540

Seems to have everything I could ever want and then some including a brake chopper.

I'm still mulling over what to do about the power stage though. The integrated IPM route is tempting but I would feel bad not to use the IGBTs I have considering the cost of an upgrade.

Of course starting a controller from scratch would allow me to make something much cleaner and more compact. Hopefully more reliable too.

Thinking more about what Weisheimer for power rating of my current controller also has me mulling over if I really need more than 400amp rated units. As it is the car was underpowered but drivable and my best guess is it wasn't putting more than 20kw to the motor. Comparing to other conversions, that power seems about right for a car of this size and weight to sustain reasonable highway speeds. Realistically, my battery pack shouldn't be asked to deliver more than 30kw constant anyway due to the 1C continuous rating (200A cells x 156 system volts = 31.2kw). Peak is 5C for 15 seconds but I'll probably never see that considering the size of controller and motor needed for that. I wonder if 40kw really would be enough.....

Twice the power of what I had would probably come close to breaking traction and make for adequate power. Maybe I can take it snow boarding someday after all..... (a few 12% hills to face on the mountain)

So what do you guys think about the IGBT rating verses the actual controller output? What sort of safety net should there be?

 

[Tesseract]

So what do you guys think about the IGBT rating verses the actual controller output? What sort of safety net should there be?

Relying on rules like "derate current by 50%" will end up costing you a lot of money with IGBT modules. I generally assume that whatever current rating is given at a realistic case temp (like 70-80C) is an attainable maximum with good heatsinking (e.g. - liquid cooling).

In that eupec module's datasheet they claim a rating of 400A at a case temp of 70C and a junction temp of 175C. Now, folks, you CAN run these chips up to 175C, but you really SHOULDN'T. They won't withstand too many thermal cycles that extreme (causes the die-attach solder to crack). Anyway, for a first pass approximation just multiply Ic by Vce[sat] by Rtheta[j-c] to get the rise in case temp at that current... E.g. - 400 x 1.6 x .012 = 76.8C rise, so if ambient is 25C then the case will be at 101.8C. That's a tad higher than the 70C spec but this is just a first pass approximation. The actual conduction losses are split between the IGBT and FWD according to "modulation depth" while the FWD is extra lucky because all 6 diodes in a 3ph. inverter share the freewheeling current (not evenly, though!).

If you bother to calculate the switching and conduction losses more accurately, what you typically find with IGBT modules is that you can run at somewhere between 65% and 80% of the current rating given for the case temp of 70-80C. This assumes a reasonable sized heatsink and derating current as temp rises.

Here's a good paper by Infineon on calculating the losses more accurately: http://www.google.com/url?sa=t&sour...iQQXT3vOQ&sig2=JyyBlmepH3aelDawSSTJyg&cad=rja

 

[david85]

You are telling me what I want to hear, thanks!

I never measured temperature direct from the source on the controller but I found that no matter how hard I was able to run the car, the cooling fans never came on unless it was a warm or hot summer day. During a heat wave, they came on just from being switched on and parked in the sun......

I think the fan cycling switch was set for 30C but I'd have to double check that.

Regarding temperature, the hybrid kits by infineon that I linked to earlier claim to be able to run at ICE coolant temperature in order to share the same cooling package. It made me wonder if an inverter like this can be used for heating the interior of the car but I wonder how wise it would be to have the controller that warm all the time. It's tempting because a separate heater does limit an EV's range and it would be a way to increase overall efficiency without sacrificing comfort. I'm getting ahead of myself though.

Either way, I am looking at a water cooled heat sink and a cooler up front in the car.

 

[david85]

Well, I found that digikey does list a decent price for PM450CLA060 and PM600CLA060 IPMs but the availability is 6 months!

http://parts.digikey.com/1/parts/1637416-mod-ipm-l-ser-6pac-600v-450a-pm450cla060.html
http://search.digikey.com/scripts/D...k=hp_go_button&KeyWords=PM600CLA060&x=28&y=21

Found another source that has ~50 in stock but they want close to one grand for it - heck no....

Going to keep looking because if I can get a 400 or 600 amp setup for those prices, I would be very tempted to go that route.