Paul and Sabrina's Cheap 3 Phase Inverter (AC Controller) with Field Oriented Control

[thingstodo]

Originally Posted by MPaulHolmes

(canhi_1a, canlo_1a, gnd)
(canhi_2a, canlo_2a, gnd)
for one of the interfaces, and for the other CAN interface, just change the a to b.

Hmm. I thought CANbus communications was just the CanHi_1 and CanLo_1 differential pair. The 12V and Gnd were for supplying power to the end devices? Maybe I got that wrong.

Two CANbus channels would be nice if you need to talk two different speeds. For example, if you have a BMS on CANbus it may be a different rate than the ABS system. Not that I have that situation (yet) but I'm hoping!

[MPaulHolmes]

Yes it's a differential pair, but you have to have 2 connections on the board for the same differential pair. I worded that really confusingly. You need a "into the board" can1H and can1L and an "out of the board" can1h and can1L since they are daisy chained together.

Well, I guess you could have a cable that juts out a "canh & canL" every now and then, but they don't usually make cables like that.

[e*clipse]

Thanks for the CAN info.

I'm fine with doing the work to make this happen; and it doesn't necessarily apply directly to the inverter physical design. They could share a common schematic.

Before I get too far on this, I have a few questions.

- Moving to the 6010A 80 pin pic - it's the 6015's twin with more I/O. Thoughts?

- Moving to surface mount components on a "do it yourself" project. Thoughts?

- I'd like the result to be as pin/code compatible with the work you've already done. However w/ 40 extra pins to work with there's a lot of room for future expansion. Could you put together a necessary I/O list - with a "gee that would be nice" list?

- If the 6010A seems like a good idea, should we set aside hardware for 3 or 4 synchronized current sensors?

- Driving multiple TO-247 switches instead of big IGBT modules: Can a driver design ( with component changes ) drive IGBT's, FET, and SiC FET's??

-I'm currently working off of "ACControlAndDriverBoard1" files. Is there a newer one you'd prefer?

- The design I'm working from uses an external 24V supply. Do you want to stick with that?

I'm sure there are other questions, but that will help me get moving on this. If you already have a design to do this great! If not, I'm more than happy to help make one to your specs.

[thingstodo]

Originally Posted by e*clipse

They have a board to case measurement of 15mm. If the LEM current sensor is replaced with the Tamura version, it will fit. The Tamura sensor is only 11mm thick.
http://www.tamuracorp.com/clientuplo...L08PXXXD15.pdf

The only issue I see is the temperature specs. -10F for operation and -20F for storage. I regularly see -20F (30+ days per winter) in the mornings, with bad days at -30F (maybe 4 or 5 days per winter) and worse days at -40F(1 or 2 days per winter)

Specs for temperature in vehicles, I'm told - I'm not a mechanical guy - run -40F to -45F operating with storage to -50F or -60F

Below -30F a lot of LCD displays update so slowly they are not usable. But they still work after things warm up.

[e*clipse]

Originally Posted by thingstodo

The only issue I see is the temperature specs. -10F for operation and -20F for storage. I regularly see -20F (30+ days per winter) in the mornings, with bad days at -30F (maybe 4 or 5 days per winter) and worse days at -40F(1 or 2 days per winter)

Specs for temperature in vehicles, I'm told - I'm not a mechanical guy - run -40F to -45F operating with storage to -50F or -60F

Below -30F a lot of LCD displays update so slowly they are not usable. But they still work after things warm up.

Huh. Boy, that raises a good point - I've usually been worried about the high side of things. On that end of the spectrum, we see ambient temps over 110F. I'm told cars in a parking lot can see something like 150F in those conditions.

I wonder what would happen with the cooling fluid, seals and all that stuff when it gets to -40C ? I've heard they just run the trucks 24/7 when things get that cold up north.

Another issue - for EV's would be can the batteries handle those extremes? As with many compromises, it really wouldn't pay to match an extreme rating in one critical part if another critical part couldn't handle the same conditions.

- E*clipse

[thingstodo]

Originally Posted by e*clipse

I wonder what would happen with the cooling fluid, seals and all that stuff when it gets to -40C ? I've heard they just run the trucks 24/7 when things get that cold up north.

Diesels, I'm told, sometimes have trouble starting even when they have had engine block heaters used. I have heard stories from truckers that they leave their diesel rigs running all winter, but that is further north.

50/50 glycol/water mix is good to -50F. Thankfully I have not seen that mixture start to gel. Gasoline DOES start to freeze up around -40F and our winter gas has an antifreeze additive included.

On my gas vehicles, I normally plug in the engine block heater, at work and at home, plus a small interior warmer on a timer - about 4 hours before I go to work and 4 hours before I leave. They vinyl seats are much uncomfortable when they are frozen. If your battery is a couple of years old, a trickle charger on the battery when parked helps as well.

My 92 mazda MX6 has CV boots that do NOT like that cold weather (early Japanese design). If I run it at -40F the life of the boots is measured in weeks, not years. Al of the rest of it - seals, gaskets, etc - appear to have paid attention to the full operating temperature range

Another issue - for EV's would be can the batteries handle those extremes? As with many compromises, it really wouldn't pay to match an extreme rating in one critical part if another critical part couldn't handle the same conditions.

The performance of the batteries is expected to be poor at the extreme temperatures. The energy is not lost, but the batteries need to heat up in order to have the chemical reaction release it's stored energy.

I have some ideas on battery box insulation, a battery warming system that I can plug in and circulate warm glycol around the battery box, etc. Of course, I have to get the MX6 finished (to a point) and on the road .. then when it's running I can experiment with that sort of stuff. I have it from EVTV that charging lithium cells below 32F does permanent damage, so unless I get some sort of warming going on, it will be a 3 season car.

The controller could very well be in the engine compartment or near the motor. No warming available until the electric motor needs cooling. The current sensor is one of those things that needs to work, within temperature spec, to make the controller function properly, right? It's not optional, like the LCD. If the accuracy drops to 5% instead of 1% - no big deal. If it stops working at all, or is damaged ... well, that's another story!

I do not mean to make a big deal of this. This is one design decision in a long list of design decisions. If the design decision is to have -10C as the lower limit, it's simply a limitation that must be listed. Anyone that has to deal with lower temperatures will need to deal with the limitation somehow or spec a part with a wider temperature range.

[e*clipse]

Wow - what a PIA. Thanks for reminding me to be thankful for my sweltering heat rather than the mind-numbing cold!

Also - it does bring up a very good point. It's generally considered very bad practice to design something to work with only one part. I would like to know why the Tamura ones are limited to only -10C. The company does make sensors that can take the -40C to +85C range.

So, here are a few options - I haven't included the $200+ options.
1) LEM makes a similar part that could fit. It would require a pigtail wire, but could work with some slight design changes. Some cleverness would allow both the Tamura and LEM options.
http://www.lem.com/docs/products/hais_e_rev11.pdf

2) Design it for these little sensors:
http://www.allegromicro.com/~/media/...-Datasheet.pdf
The problem here is that they would need to be soldered to the busbars, they would be submerged in the coolant and the I/O wires/pins would have to be sealed. Or a different design would split the output pin. This would require some work to ensure the pin's strength.

3) Design it for these little sensors:
http://www.lem.com/docs/products/fhs%2040-p%20sp600.pdf
Paul already made a bunch of little circuit boards to take sensors like these. I suppose a little design tweaking and calibration would make this a good option. Also I don't know what the current limitation **really** is. It would be good to be able to handle 300A.

Frankly, I'd prefer the last option, if it can be accurate enough.

- E*clipse

[P-hack]

fwiw, silabs has integrated their isolation technology into an isolated shunt monitor Si8920, signal delay measured in nanoseconds.

http://www.mouser.com/ds/2/368/Si8920-766188.pdf

-100 to 100mv range

shunts are what tesla uses for fast and accurate.

It is conceivable to run an op amp and an isolator off the high side driver supply as well and blast out a digital current shunt reading back to the controller, or do other distributed gate logic stuff (i.e. hysteresis current control, shoot through protection, saturation detection, "formal" digital protocol between gates and controller, etc). I mean if you can have an attiny just to make a square wave, you can throw a $1 arm 24mhz 12bit adc on the high side gates and slave the low sides to them with another isolator, and even capture leg vs "neutral" voltages between pulses, or?? for pennies (shunts are cheap)

the controller basically then is the clock/position source plus user demands, and reporting back to user.

just my $0.015

[e*clipse]

Originally Posted by P-hack

fwiw, silabs has integrated their isolation technology into an isolated shunt monitor Si8920, signal delay measured in nanoseconds.

http://www.mouser.com/ds/2/368/Si8920-766188.pdf

-100 to 100mv range

shunts are what tesla uses for fast and accurate.

It is conceivable to run an op amp and an isolator off the high side driver supply as well and blast out a digital current shunt reading back to the controller, or do other distributed gate logic stuff (i.e. hysteresis current control, shoot through protection, saturation detection, "formal" digital protocol between gates and controller, etc). I mean if you can have an attiny just to make a square wave, you can throw a $1 arm 24mhz 12bit adc on the high side gates and slave the low sides to them with another isolator, and even capture leg vs "neutral" voltages between pulses, or?? for pennies (shunts are cheap)

the controller basically then is the clock/position source plus user demands, and reporting back to user.

just my $0.015

I like that - if I understand the data sheet correctly, this part isolates small analog signals. This could be a really good solution for a lot of isolation challenges. In fact it may be wise to put these on all analog signals to protect the ECU... hmmm...

I have messed around with digital signal isolation a bit. I was trying to come up with a very robust rotary position sensor setup for the throttle. There are really good sensors out there for position and current sensing that put out pwm and/or pulse frequency modulated signals. They have the advantages of easy isolation and an easy failsafe checks, and are less susceptable to noise. I was able to get them to work using the PIC's input change notification feature with interrupts. The main thing that bugged me about it in the end was the large amount if interrupts I ended up using. Perhaps a better programmer could find a better solution?

- E*clipse

[thingstodo]

Oct 28 - testing AC controller (but not quite)

Connect controller to 12V - contactor pulls in correctly - no hi voltage pack

Connect hyperterminal. Powerup OK. config shows no throttle signal still.
I forgot to change that pot.

Change encoder-ticks to 64 ppr

It will not show pulses ... Oh yeah ... that's a different version of firmware.

So - burn AC Controller Project Encoder.hex from June 1 - connect PICkit 3, power up controller, burn firmware.

Success. Turn off controller, remove PICkit 3, start a new capture,Siemens encoder test 2. Power up controller

config - oh yeah - no commands on this one, just encoder counts
run-encoder-test
00000x

Turn the motor shaft by hand - no result

Turn off the controller and stop the capture.

Ring out the cable - which I should have done before now!

The 5V and GND to the encoder were in the wrong pins.

ctl board - cable - cable - siemens encoder
gnd       - blu   - blu   - pin 5
5v         - orange  - or    - pin 4
cha       - brown   - brn    - pin 2
chb       - white - wh  - pin 1

No joy. Power down, then Remove the siemens encoder connector
Connect female pins on cable direct to the siemens motor pins

Same result. Check for 5V on orange versus blu + 4.95V
check white versus blu. 0.00V
check brown versus blu. 0.00V.
check white versus orange. 0.00V
check brown versus orange. 0.00V

Open collector outputs? I wonder if they need to be tied high or low? High, silly. They are low now!

To the EVTV web site! Nope - no mention of pullup resistors. Maybe the DMOC645 has them built in? the DMOC 645 is the AC controller that is sold for the Siemens motor.

I expect that there should be pull-up resistors ... I'll start with maybe 10K .. tomorrow!