openinverter forum

arber333 wrote: ↑Fri May 08, 2020 1:35 pm
BAT input if you can think of some isolated op amp circuit that would take HV and display working area (say 300V to 420V) from 2.5V to 5V in linear fashion you can use it to show HV to inverter. It makes for smoother drive as far as Arlo told me. I pull it to GND for now.

That mapping is incorrect. There was some ambiguous discussion on this over on ES. I have clarified this directly with Bas. Mapping is between 0-5V for 0V to full pack voltage. You would come to the same conclusion by analyzing the original schematic released by Bas.

kiwifiat wrote: ↑Fri May 08, 2020 9:09 pm
That mapping is incorrect. There was some ambiguous discussion on this over on ES. I have clarified this directly with Bas. Mapping is between 0-5V for 0V to full pack voltage. You would come to the same conclusion by analyzing the original schematic released by Bas.

Oh, I didn't know!
Then I could just use another chips pwm output since I use this chip to work precharge and DC switch.
I have in mind of using one simple smd pic12f chip. It has just enough I/Os to work my requirements.

arber333 wrote: ↑Sun May 10, 2020 1:28 pm Oh, I didn't know!
Then I could just use another chips pwm output since I use this chip to work precharge and DC switch.
I have in mind of using one simple smd pic12f chip. It has just enough I/Os to work my requirements.

That would work and would be inexpensive. I have a design in the pipeline, really just a modified version of johu's isolated voltage sense schematic for a Nissan Leaf 96S pack. I'll post it once I have built it up and verified its operation.

Board layout done.

So having a think here. If we put a summing opamp with a gain of 1 into the current sensors such that it sums the MG1 and MG2 sensors then we could run foc with dual powerstages in parallel.

Jack Bauer wrote: ↑Sat May 16, 2020 3:13 pm So having a think here. If we put a summing opamp with a gain of 1 into the current sensors such that it sums the MG1 and MG2 sensors then we could run foc with dual powerstages in parallel.

That is brilliant! I need/want to do this for my Volt inverter. Do you have a schematic in mind?
I was originally intending to just sum them up across resistor divider, but it would be nice to have an error state if both values diverge too much.

I think FOC would need to have correctly calibrated gain. So i would need to use two current sensors and apply a load through 10x loop into each sensor and then adjust gain to get a correct sum on the other end. Or should i just cheat and set it to 1/2?

Jack Bauer wrote: ↑Sat May 16, 2020 3:13 pm So having a think here. If we put a summing opamp with a gain of 1 into the current sensors such that it sums the MG1 and MG2 sensors then we could run foc with dual powerstages in parallel.

If you choose a summing solution with a gain of one you will run out of headroom very quickly. An alternative solution proposed by Lebowski over on ES is to employ a resistor based averaging circuit then halving the V per amp figure in the current sensor menu. That way you can measure up to twice the phase current. Of course an opamp summer with a gain of 0.5 will achieve that same outcome.

kiwifiat wrote: ↑Sat May 16, 2020 7:33 pm If you choose a summing solution with a gain of one you will run out of headroom very quickly. An alternative solution proposed by Lebowski over on ES is to employ a resistor based averaging circuit then halving the V per amp figure in the current sensor menu. That way you can measure up to twice the phase current. Of course an opamp summer with a gain of 0.5 will achieve that same outcome.

That was my idea of a "cheat" with resistors. I see i will have to employ a halfh gain then. Well see how it goes.
Pug drives again! It is evening how, so tomorrow i will make a video.

Arber, on your rev 2.3 board, what are you controlling? I see:
Precharge out
Temp out?
Overtemp out
Error out
Brake out

What are you doing for contactor control with this board?

All done and on order at JLC.

Resources on Github :
https://github.com/damienmaguire/Ampera-Volt-Inverter

I'll send a board to Arber and mdrobnak once the prototypes arrive.

Really nice!
I have a lonely Outlander motor here wanting to have a date with the new main board.

I also can provide AC compressor to test the 3rd stage. However Damien I hope you provided 5 or 6 pin programming connector. I totally forgot this because I use DIP version of dsPic30 i programm the chip directly through adapter. I have Pickit3.0.

Yep it has the pickit programming port:)

Great JB, so this will be a breeze.
I also see you add some details to serial port for wifi link. Can you explain diodes? This is something i didnt use yet. And i tried much to get serial across wifi gterm.

EDIT: OK i would definitely add two small relays for precharge and DCSW lines. I dont trust darlington array to live through everything DC contactor load can give. You can also use one relay side to keep Lebowski in reset until precharge is done.

I also hope your DCDC for voltage sensing can survive shorting? I use this one exclusively. Murata NME0505SC.
I had several opamps shorted when testing and DCDC survives.

DANG! I need to install DS9!

The diodes allow the 5v dspic to drive the 3v3 wifi module without damage. I use the NCV8402 driver for driving contactors. Its a short circuit and thermal overload protected low side driver designed for automotive applications.

Yes the dcdc is short circuit tolerant.

Jack Bauer wrote: ↑Mon May 18, 2020 8:04 am The diodes allow the 5v dspic to drive the 3v3 wifi module without damage. I use the NCV8402 driver for driving contactors. Its a short circuit and thermal overload protected low side driver designed for automotive applications.

Ah i see! This is then open drain device? So connect 12V to contactor and other side to the board. Good.

Exactly. And if you do something silly like put 12v on it it just shuts down.

Jack Bauer wrote: ↑Mon May 18, 2020 10:37 am Exactly. And if you do something silly like put 12v on it it just shuts down.

Heh, wonder how you know that .

This all looks great, thanks Damien! I can't wait to try it out.

I tried looking through the stm32-sine repo but failed miserably - I'd like to see about turning off the precharge relay after the main relay is up and running. Turns out I mixed economized and non-economized relays. Oops! But that seems like a parameter that is good to support. Thoughts?

It already does this:)

Jack Bauer wrote: ↑Mon May 18, 2020 3:42 pm It already does this:)

Awesome. I'll go back into my corner then.

Hi all - I have received an interface PCB from Arber as of today. I'll probably have it all built up in a couple of days, then I can proceed to testing. I'm open to suggestions on single versus dual IGBT drive
Arber, I will probably have all sorts of questions for you regarding components and current sensors, so I'll probably put those in the other thread.
-Isaac

OK, good to hear.
I send you the dual version so you could be using both sides of Volt inverter.
I have here the drawing and you can see the P/N and values for all the parts. Look for rev 2.3.
https://github.com/arber333/Ampera-inve ... -interface

It also includes resolver interface now. If you dont use it you just dont populate that part.

The HV sensing part is best prepared like this... From Volt inverter remove original 2pin HV supply connectoron the right side of the main board. To its output contacts you fit your own HV cable of sufficient length you can reach the HV connector on the board comfortably. I made space for 2P JST HV 3.96mm 90deg connector that is simple to crimp and cant be mistakenly turned. Well at least not if you are sensible...

There is one thing though. Maybe you should add 6 resistors from SN74LS06 chip inputs to GND. Its because of bootloader that Olimex surges positive signal on driver lines at the time of manual reset. I suggest to solder on the underside directly to the legs of the chip and then chip a bit of insulation layer on the back plane to reach copper and solder the other side of resistors. I get good results with 6K8 pulldowns. So now voltage generated instantly would be less than 1V.
At that time chip would open its PWM ports and send 3V out. That could cause all of transistors to open mommentarily. This would be heared as a loud springy sound from the motor. I am not sure if this could damage IGBTs. It doesnt happen in the car though. It only happens if you manually reset Olimex under power. I would rather be safe than sorry. In my new boards i designed this into the output stage.

Also you MUST setup the PWM as positive in software since you use inverter chip that inverts those signals.
Dead time also MUST be 3uS or more. So set it at 180 in interface.

For all you Volt people, I've added information to the wiki about the Volt inverter. Feel free to check it out and add your own experience or knowledge!
https://openinverter.org/wiki/Chevrolet
-Isaac

Isaac96 wrote: ↑Tue May 26, 2020 8:03 pm For all you Volt people, I've added information to the wiki about the Volt inverter. Feel free to check it out and add your own experience or knowledge!
https://openinverter.org/wiki/Chevrolet
-Isaac

Nice. I'll double check my part number later.

A huge thanks to Tom for his ground breaking work on the Ampera/Volt inverter back in 2018:

https://github.com/tomdebree/Volt-Inverter

Current sensors are 2.85mV/A so good for at least 700A.