Volt inverter interface

[arber333]

Here is another interface that i made to use V2.0 main board with Volt inverter directly without additional boards. See zip file for PCB.
EDIT: I think i can make the same for Prius Gen3. Idea is to just use normal Jonannes V2.0 board with any suitable inverter. I think this way we even help Johannes sell the rest of V2 board kits.

[Isaac96]

I just finished assembling my PCB; I haven't yet added the pulldown resistors or the current sensor resistors, since both of those are waiting on some testing first.

But:
When I gave it power, the OPA344 let out the magic smoke.
I checked the PCB, the design files and the datasheet for the OPA344, and it looks like the PCB and design files have the power pins backwards. So that would explain the poof.
I'll just keep using the original voltage monitoring board, or switch to one of those Isabellenhutte shunts that everyone is raving about.

EDIT I've done surgery on the PCB to correct the polarity, and am ordering another OPA344 to use. All good!

If I use only one IGBT side, which current sensor resistors should I populate? Can I just hook them all up, or will that cause issues?

Thanks!
-Isaac

[Isaac96]

A little update - I'm running 42v into the Volt inverter and it's pulling the fault output low. Is this normal behavior? I guess it wouldn't like to work much under 300v or so.

[arber333]

A little update - I'm running 42v into the Volt inverter and it's pulling the fault output low. Is this normal behavior? I guess it wouldn't like to work much under 300v or so.

Volt inverter works from 38Vdc upwards to 420Vdc. I find it quite broad range of driver board supply.
Yes Volt inverter uses FAILSAFE fault output. This means it actively holds the pullup down if everything is good. The moment something happens this ends and Fault pin climbs to 5V and signals pwm to stop.
What i did i use one NPN transistor (or N-MOS) with 4K7 pullup. When everything is good this transistor is sitting idle and letting MPROT have its 3V3 signal. If something happens NPN will close and pull MPROT to GND. I did this because Johannes uses inverted logic from Volt.

I have seen Fault pin in action on the road when i used wrong deadtime and before when i used sine code adapted to BLDC. Of course regen was kicking and biting back there.

[arber333]

I just finished assembling my PCB; I haven't yet added the pulldown resistors or the current sensor resistors, since both of those are waiting on some testing first.

But:
When I gave it power, the OPA344 let out the magic smoke.
I checked the PCB, the design files and the datasheet for the OPA344, and it looks like the PCB and design files have the power pins backwards. So that would explain the poof.
I'll just keep using the original voltage monitoring board, or switch to one of those Isabellenhutte shunts that everyone is raving about.
EDIT I've done surgery on the PCB to correct the polarity, and am ordering another OPA344 to use. All good!

ERR... I am sorry i wasnt able to respont sooner. But OPA344 is another version of op amp that i havent used really. So i cant testify on how it works.
What i used is Microchip MCP6001RT variant which is inverted sensing. I think OPA344 has inverted variant also. Check for pinout on datasheet.
I think you may have a problem in voltage polarity if you use another positive sensing opamp. I dont think cutting the PCB is a good idea.

Also to clarify for isolated op amp i use Si8920 DIP8
To supply isolated voltage i use Murata NME0505SC which can withstand short on opamp indefinitely.
I see i have used a different parts for the same application. The isolated opamp is NOT ATTINY! I have corrected the names on the layer in my Github now. I will try to replace part from design with correct ones in new version. Rest assured everything works on board i have tested it.
I am sorry if i caused confusion. I have just used the drawing for so long i forgot about some points.

[arber333]

If I use only one IGBT side, which current sensor resistors should I populate? Can I just hook them all up, or will that cause issues?

Ok now!
First you will have to decide which side you want to choose. If you dont use one side then simply take out the driver cable/connector from that side and also remove cable from current sensors on that side.
Then on the board you need to populate only the fault transistor on the side you will use.
Next you decide which temp sensor you will use and fit a jumper there.
Then you go and set the current sensors voltage divider. First you trace which sensors affect which row of contacts.
The LH IGBT is reported current by the top row of current sensor pins. What i use is divider from 2.5V to 1.6V. So i use 3K3 inline resistor R35/R36 near the connector on the underside and i pair it with 6K8 pulldowns further on the line near the signals header. Then you should get 1V6 in steady state on the signals connector.
Additionaly you need to populate R27/R28 with 4K7 pulldown resistors which clear the signal of switching noise.
There is 105 or 106 ceramic cap near the current sensor header.
Then you populate 220R inline resistors from the mainboard to SN74LS06 inverter chip. Next you add 470R pullups the the output pins. I first marked those resistors as 220R, but i found 470R pullups sufficient.

Then there are 6K8 pulldowns to fit from the underside which i found best. I advise simply to use small 6K8 0.25W wire resistors. Solder one wire to SN74LS06 input pins and the other wire directly to GND plane. Clean the isolation off copper on ground plane for that purpose. This is how i did it.
This discharge resistors i found out recently and i had to put them on my every board.
Lebowski brain strangely doesnt need this and just uses direct cmos outputs over 330R inline resistors. Must be the bootloader.

[Isaac96]

That's what I get for just building it quickly and not waiting for advice!

What i did i use one NPN transistor (or N-MOS) with 4K7 pullup. When everything is good this transistor is sitting idle and letting MPROT have its 3V3 signal. If something happens NPN will close and pull MPROT to GND. I did this because Johannes uses inverted logic from Volt.

Okay there's my second mistake. I used PNP transistors. Will swap in a 2n2222 which should fix that issue.

ERR... I am sorry i wasnt able to respont sooner. But OPA344 is another version of op amp that i havent used really. So i cant testify on how it works.
What i used is Microchip MCP6001RT variant which is inverted sensing. I think OPA344 has inverted variant also. Check for pinout on datasheet.
I think you may have a problem in voltage polarity if you use another positive sensing opamp. I dont think cutting the PCB is a good idea.

Also to clarify for isolated op amp i use Si8920 DIP8
To supply isolated voltage i use Murata NME0505SC which can withstand short on opamp indefinitely.
I see i have used a different parts for the same application. The isolated opamp is NOT ATTINY! I have corrected the names on the layer in my Github now. I will try to replace part from design with correct ones in new version. Rest assured everything works on board i have tested it.
I am sorry if i caused confusion. I have just used the drawing for so long i forgot about some points.

Yup, I got a Si8920. I'm using an ROE-0505S which is also short circuit protected.
OPA344 has the same pinout otherwise, it is the op-amp which Johannes uses in the inverter kit board. I'll just see how it works, board was already cut and comparing datasheets and example circuits it looks the same besides the power polarity. I have a couple more on the way too.

I'll scope the signals and see about the 6K8 pulldowns, will probably install them.
Okay I have the right current sensor resistors populated. I did some calibrating and the correct gain was 2.02 on both sensors. Does that match your findings?
Also for the temp sensor - looks like i just connect the correct vias together. What sensor curve is needed?

Thank you so much for your help! I'm looking forward to finally having a reliable and powerful inverter!
-Isaac

[Isaac96]

Update - after switching in the 2n2222 fault is no longer firing. I'm getting beautiful sine wave outputs from the IGBTs when running at 42v. Still haven't checked the pulldowns, that's the next thing to do.
-Isaac

[arber333]

Also for the temp sensor - looks like i just connect the correct vias together. What sensor curve is needed?

See the IGBT datasheet i just added to this post. On the last page is the thermistor data. It is the MBB600 option.
viewtopic.php?f=2&t=697&p=9791#p9791

[Isaac96]

Alright, should I have R33 and R34 populated? (they pull the sensor to 2.5v, 220r according to BOM).
I ask because with those resistors in, the reading stays at 100.00.
I set snshs to 2 (which should be correct according to parameters wiki page).
Thanks!
-Isaac

[arber333]

Alright, should I have R33 and R34 populated?

Yes once more i didnt correct the value after i calibrated resistors.
You should use one 10K pullup on thermistor line for Ampera inverter. So populate R34.
See posts in DIY forum here: https://www.diyelectriccar.com/forums/s ... count=2114

[Isaac96]

Excellent, will do. I will start testing with full power and motor soon!
-Isaac

[arber333]

Excellent, will do. I will start testing with full power and motor soon!
-Isaac

Nice! I like your speed!
There are some details maybe...
The line resistor for fault transistor, i dont use it. Just solder a small piece of wire across. You only need 4K7 pullup there.
Also pullup on fault pin is needed to feed 5V to the fault line on Rev1. I think it is redundant on Rev2 mainboard. You can try powering up mainboard and measure this pin without pullup. If it is normally high you are good.

[Isaac96]

After switching to the 10k pullup resistor and fixing my IDC cable, I have temperature sensor readings. For anyone else who needs this, I'm getting 1.8v at 30 degrees C.
Excellent!!


Fault signal seems to be working fine as-is (all populated). When the inverter is unpowered it signals a fault
Looks like the rev2 doesn't need the pullup, but I'll just leave it in since the transistor can sink enough current to deal with it anyways.

I'd say this is all working well. I will try to document as I go, just created a wiki page. (just a placeholder for now).
https://openinverter.org/wiki/Chevrolet
Of course you are the creator, so please feel free to correct all my mistakes.
I've got exams this week too so can't spend a ton of time on the docs yet. It shall be done.

-Isaac

[arber333]

After switching to the 10k pullup resistor and fixing my IDC cable, I have temperature sensor readings. For anyone else who needs this, I'm getting 1.8v at 30 degrees C.
Excellent!!

Good then. Last time i was playing with temp sensor i forgot to power my pump and inverter threw a fault after it got quite hot. No discernible damage...

Do you need me to collect some instructions to assemble the board? To be honest all my Volt boards are pretty much the same from the STM32 down. Differences are only in details. I can describe the latest board version and add a PDF?

[Isaac96]

Good then. Last time i was playing with temp sensor i forgot to power my pump and inverter threw a fault after it got quite hot. No discernible damage...

Do you need me to collect some instructions to assemble the board? To be honest all my Volt boards are pretty much the same from the STM32 down. Differences are only in details. I can describe the latest board version and add a PDF?

That might be a good idea, it seems like this inverter will become quite popular due to low cost and high power. Whatever works for you is good.
-Isaac

[Isaac96]

Motor spinning!
I first tested at 36v with some old lead-acids which sagged too far. 48v worked great though!

I then went on to full pack voltage and results were similarly excellent. I've only run in manual mode so far but it seems like everything is working well. The inverter hit 50 deg C soon enough so I shut it down. (I haven't plumbed the radiator to it yet...)

Do you think 4.4khz or 8.8khz is better? 4.4 would reduce the switching loss in the inverter but it's remarkably loud (at least in manual mode at low RPMs, seems to quiet down nicely). I am borrowing Joromy's Siemens parameters from viewtopic.php?f=16&t=186&start=60
with some modifications for my higher voltage. With pwmfrq at 4.4 the motor started nicely, 8.8 made the start occur later. So it seems like more boost is needed at higher frequencies, maybe due to the deadtime requirements?

Anyways, I'll keep on testing (adding encoder and cooling soon). Documentation will follow I promise...
-Isaac

[Isaac96]

Cooling is all ready, but I seem to have gotten some water near the board somehow - concerning because I don't know where it came from. Tomorrow after it's all dried out I'll see what's going on. Then the final inverter testing can be done. Can't wait!
-Isaac

[Isaac96]

Okay time for another update. Cooling is working nicely, the inverter stays pretty cool.
But....
I'm trying to run in manual mode with full pack voltage and the motor won't do anything but jump. It stays in the same place and shakes back and forth with a loud banging sound.
I've tried changing boost, fweak, ampmin.
When ampnom is ~10 the motor won't move at all, just makes the pwm whining noise. (this seems normal). 15-20 will get it to start shaking. More just makes it shake harder, then it hits OC (400a).

The motor worked perfectly when powered from 48v - this was with the same boost of 3000. I thought boost might be too high, so I tried dropping it to 500 and raising it to 6000. Still no good.
This is the same problem I had with my old power stage - the motor just kept jumping.

I'm running pwm at 4.4khz because Joromy was having success with that frequency with a similar motor.

I'd rather not blow my wonderful inverter quite yet so I will not test pedal control mode until manual works as it should.

Any ideas as to what might be wrong?
Thanks!
-Isaac

[arber333]

Any ideas as to what might be wrong?
Thanks!
-Isaac

1. and most important! Did you set deadtime at 185?
2. Just go and use 8khz this inverter is good with it.
3. Can you power inverter up and just rotate motor with your hand. What happens with encoder? can you scope it? Do you get clear picture of square pulses?
4. You probably have reversed A and B signals in encoder.

What i would do is completely disconnect encoder for manual run. And try to set boost at 7000. Then set Fweak at 200. Did you setup voltage limiter?
Try manual start and command 1Hz at 10% ampnom. Then set 20% and 25% and 30% and so on until you get rotation. Then command 2Hz and 5Hz and then 8Hz and 10Hz etc...
Remember, when you are in manual mode you command V/Hz not Slip.
Somewhere along your inverter will run out of amps and you will not be able to get more rpm despite higher Hz command. So you need to command more ampnom. You can now play with it...

But anyways, when you see your motor rotating i would suggest you stop and connect encoder and pedal. BTW if you use pedal in manual mode you will get as much rotation as you set fslipmin, which is odd at first, but makes perfect sense regarding you use V/hz then.

When you go to encoder mode at first you set OClimit at 200A just in case. Now see if you connected AB encoder channels correctly. If not you need to reverse them or find correct phase configuration. With AB you have 4 possible quadrants.

Can you post your Json file?

[Isaac96]

1. Yup, I set deadtime to 185. Taking good care of the new power stage
2. Alright, changed that back.
3. I'm running without the encoder. I do know that it outputs nice square pulses - I've checked it quite a few times in the past chasing other problems.


Params attached, I'm trying to copy exactly what Joromy is doing with his ranger motor.

I'm not getting any change with your suggested parameters or method.

This is rather driving me up the wall

Oh well, it's probably something stupid simple that I forgot.

-Isaac

[Isaac96]

Color me dumb.
It's a bad PWM channel that's getting me. PWM 3 TOP channel isn't working right, the "high" output voltage seems to be too low. At least that's all I could see with my scope - the bad channel only reaches half the amplitude of the others, so the inverter output just stays high. On the plus side, it failed in a good way - protecting the power stage!

So I guess I'll order another stm32 board. This will be my fourth. Oof..

Arber, I do apologize for wasting your time with my silly issues. I should have done my due diligence with troubleshooting before asking the expert.

-Isaac

[arber333]

So I guess I'll order another stm32 board. This will be my fourth. Oof..

Dont feel bad. I am with my 12th STM32 Olimex. I just bought 2 more to be sure when i am experimenting. Usually i broke some of the inputs, most notably MPROT or Fault input that governs PWM stop...

Just for that purpose i have a small ACIM motor here in my workshop and every new inverter is connected to it in manual mode to see everything is ok before it goes in a car. I am thinking of getting one Kia HSG for PMSM testing purpose.

[Isaac96]

Alright I've got 2 new Olimex boards on the way, arriving Wednesday. I'm trying to see if I can run some Arduino code on the Olimex board I killed to see if the output is actually dead.

In other news I've mounted the Rev2 board and the 2.3 Volt adapter board onto the heatsink plate in the inverter. Tomorrow I will have the op-amp for the voltage sensor and I'll confirm that it works well. No more manual precharge timing!

[arber333]

Alright I've got 2 new Olimex boards on the way, arriving Wednesday. I'm trying to see if I can run some Arduino code on the Olimex board I killed to see if the output is actually dead.

In other news I've mounted the Rev2 board and the 2.3 Volt adapter board onto the heatsink plate in the inverter. Tomorrow I will have the op-amp for the voltage sensor and I'll confirm that it works well. No more manual precharge timing!
IMG_0050[1].JPG

Nice job! Did i send you the 3D printing file for a cover spacer? I think i added it to github anyway.
I like how you solved the connection for voltage sense. I only recommend you twist the wires to make twisted pair connection. There will be 360Vdc in those wires.

Another consideration maybe? You need to connect the board to GND plane of inverter. Since you use a plastic spacer i would add one cable from GND on each board to the alu shelf and fit it with original screws.

Also take care to connect motor and inverter together with a braid or other kind of thick cable so potential is shared. I felt quite a jolt when i touched them unconnected.