VOLT/AMPERA inverter single controler dual IGBT drive

[Isaac96]

Hi Arber - I've been following this hoping to eventually use a Volt inverter in my car. Fortunately (???) I blew up another IGBT in my DMOC power stage, and it's at least $150 to get a new one, so I just ordered a Volt inverter for the same cost.
I've looked at this thread and also your Volt adapter board viewtopic.php?f=2&t=297 and am wondering which to use. I guess the dual IGBT would be nice just for the sheer power, since I've got lots of battery capacity. Gonna need a bigger fuse...
I love how clean the single PCB is on your dual board

Do you have any spare PCBs I could buy? If not I can just OSHPark a few (I'm in the US) and send them to anyone else here who needs one.

Thanks!
-Isaac

[mdrobnak]

Do you have any spare PCBs I could buy? If not I can just OSHPark a few (I'm in the US) and send them to anyone else here who needs one.

Thanks!
-Isaac

If you end up doing that, I'm definitely interested. Didn't know of OSHPark.

-Matt

[arber333]

Actually i have just one of this boards left. It is a rev2.
However i have a set of several new revision 2.3 interface boards here that include resolver interface. I didnt publish it yet because it is not tested.
Board is a little larger and it incorporates both connectors which are used for driver and sensor board on Rev2 controler. That way you dont need to split cables or anything with sensor board. I recommend this board if you decide to go Volt inverter route.
Additionally i included the inverter chip connected to BOTH driver connectors sou you can drive both IGBT bridges if you want to experiment.

Personally i think it would be awsome to test dual sides with ACIM, since you are not burdoned with FOC code requirements. You can set the OClimit low and observe how the current avaraging goes...
I updated my Github now to include the new files...
https://github.com/arber333/Ampera-inve ... -interface

Another way to go forward would be to use one of my new boards that are tested and proven to work. They still run on Olimex chip and are for troubleshooting purposes the same as Rev2 boards. I show them in the beginning of this thread.
Also they are on my Github
https://github.com/arber333/Johannes-co ... IGBT-board

You can still PM me if you want me to send one board to you.

[johu]

IGBT resistance goes down with higher temperature, so one with more load tends to get even more load.

Thats not generally correct. So called punch-through (PT) IGBTs do as you say. But NPT types will increase Vcesat with temperature, so the hotter one takes less load.
Here is some random data sheet https://www.littelfuse.com/~/media/elec ... et.pdf.pdf

You will see Vcesat=1.7V @ 25C and 1.9V @ 125C

[mdrobnak]

Oops. Didn't look at the thread referenced, didn't realize that was not what I want to get. I want the new all-in-one stuff. Sorry for the noise.

[Isaac96]

Actually i have just one of this boards left. It is a rev2.
However i have a set of several new revision 2.3 interface boards here that include resolver interface. I didnt publish it yet because it is not tested.
Board is a little larger and it incorporates both connectors which are used for driver and sensor board on Rev2 controler. That way you dont need to split cables or anything with sensor board. I recommend this board if you decide to go Volt inverter route.
Additionally i included the inverter chip connected to BOTH driver connectors sou you can drive both IGBT bridges if you want to experiment.

Personally i think it would be awsome to test dual sides with ACIM, since you are not burdoned with FOC code requirements. You can set the OClimit low and observe how the current avaraging goes...
I updated my Github now to include the new files...
https://github.com/arber333/Ampera-inve ... -interface

Another way to go forward would be to use one of my new boards that are tested and proven to work. They still run on Olimex chip and are for troubleshooting purposes the same as Rev2 boards. I show them in the beginning of this thread.
Also they are on my Github
https://github.com/arber333/Johannes-co ... IGBT-board

You can still PM me if you want me to send one board to you.

Thanks Arber! I'm looking at the Rev2.3 now. It looks great! I don't need the resolver hooked up so I'll just skip that part. Looks like IC1 is the chip for that?
I'm liking the dual IGBT plan. It might break the transmission at that point, then I'll be forced to upgrade to a LDU
Looking at TomDB's reverse engineering PDF, looks like there are pullups on the gate drivers so that one of the sides can just be disconnected if I don't want to use it.

Could you send a v2.3? I'm happy to be a beta tester
Of course I will assemble myself, and make a BOM from the schematics.

-Isaac

[kiwifiat]

Running parallel phase cables for each IGBT allows better balancing across IGBTs (if they are the same length). It lets them naturally balance much better.

Ok, how much parallel length should i use for 25mm2 phase cables? Do i use full 0.8m from the motor or can i use like 0.3m?
Actually i think i can just rout cables directly inside the motor. I would just need to seal them inside holes with black automotive glass sealant.
You think it is wise? What about EMI if i take out original cables? Should i wrap them in alu foil?

A

The guys in the Ukraine that pioneered the use of both power stages in the Ampera/Volt inverter appear to have ~1m long cables. They haven't shared any specifics but one thing that was mentioned was that they have blown multiple transmissions.

With regard to the cable length think about the Honda IMA used in the Civic's. The inverter is in the Boot/Trunk and the motor is in the front so I can't see that in practice cable length is a concern. I have no first hand knowledge as to whether or not the Honda phase cables are shielded, maybe someone else can comment?

[arber333]

I feel i need to stress this detail for those that will use Volt inverter with original phase connections.

If you drill M6 thread into those contacts you get a good 35mm2 worth of connection. You just need to find out how to cover tha contacts from weather.
However there is a potential incident lurking. We found out there is a steel brace inside plastic of connector which is in contact with inverter casing. If you drill out the copper there is a space underneath where your phase bolt has a chance to hit this brace and cause a short to GND. Instant smoke from IGBT!
So everytime i drill out that i turn the plastic around and mill out that plastic to reveal where metal brace is. Then i mill a piece of it out. At least 2mm around bolt should be enough.

THIS IS CRITICAL FOR LIFE OF INVERTER! Dont ask me how i know .

[slow67]

IGBT resistance goes down with higher temperature, so one with more load tends to get even more load.

Thats not generally correct. So called punch-through (PT) IGBTs do as you say. But NPT types will increase Vcesat with temperature, so the hotter one takes less load.
Here is some random data sheet https://www.littelfuse.com/~/media/elec ... et.pdf.pdf

You will see Vcesat=1.7V @ 25C and 1.9V @ 125C

Thanks for correcting me Johannes. The papers I read on it must have been older!

[mdrobnak]

What do you recommend for cables to connect to the motor? Is there a way to get those covers without the cable (to make your own?) or do you think the stock cable will work for 600A for ~10 seconds? https://www.ebay.com/itm/2013-CHEVROLET ... 4174304816 they're not exactly cheap cables, and that's a good deal compared to new...

[arber333]

What do you recommend for cables to connect to the motor? Is there a way to get those covers without the cable (to make your own?) or do you think the stock cable will work for 600A for ~10 seconds? https://www.ebay.com/itm/2013-CHEVROLET ... 4174304816 they're not exactly cheap cables, and that's a good deal compared to new...

I use a single 50mm2 cable on one IGBT side. Later if i am successfull with both IGBTs i will use 35mm2 cable from each phase that will meet inside the motor cover. Leaf motor has plenty of space for thick cables.

However the cables you show are actually some 16mm2 inside. They look thick, but it is only shielding. I think it will be ok with 350A per phase for a minute or so max which is what Volt motors can handle. If you are ok with 350A per phase and maybe use 700A in all from both sections i think those cables will be good and safe.

I replaced original phase cable cover with insulating plate to which i mounted original phase links and current sensors.
I intend to have Alu metal cover above both sections with side bolts tightening also this. It will be closed on 3 sides and partly closed from one side where i will make sure you cant get to contacts with your fingers.

[Isaac96]

Hi all,
Just got my Volt inverter in the mail. It's a 2013-2015 with p/n 12643810. Looks identical so I guess I have the right inverter.
Here it is with a Rev2 brain for scale:

The brain doesn't fit inside the lid. Spacer time maybe?

I have parts from DigiKey on the way so we'll see how many components I forgot to buy

I'll be attaching lugs to those weird output connectors, just using 1 side of the inverter for now. And probably also attaching lugs to the power input connectors, since I didn't get those wires with it.
Arber, it looks like you only have the upper power connector hooked up?
Ohmmeter says the two connectors are shorted together anyways. Who knows why they put two on then...

It looks like the connector in the upper right corner is for monitoring HV so I'll just hook that to the interface board.
How do we handle the dual current sensor inputs? If running only half the inverter, then wouldn't the output of one sensor feed into the output of the other and get messed up? Maybe just leave the unused sensor disconnected? I'm no EE so please excuse the silly questions

When (if) I get this working I hereby pledge to document it and wiki it (and also any explosions that happen)

Thanks!
-Isaac

EDIT After some more disassembly, it looks like the two connectors are just different ends of the same pieces of metal. I guess one wasn't enough for the current draw? Anyways I'll probably fabricate a cover and just bolt straight to the bus bars.

[mdrobnak]

If you're talking about the side..it's battery input and output to the AC compressor and I think DC/DC...

-Matt

[Isaac96]

If you're talking about the side..it's battery input and output to the AC compressor and I think DC/DC...

-Matt

Hmm... The AC compressor gets run off the third 3 phase block inside this inverter. But the DC/DC is in the trunk and has much smaller wires so maybe it hooks somewhere else. Maybe it's return power to the batteries?
Oh well, not super important. I'm just trying to understand how this works before I start breaking it.

[mdrobnak]

No, arber made use of the 3rd block...that's usually for the transmission oil pump. There is a great overview on the ever auto YouTube channel explaining the components.

[arber333]

Hmm... The AC compressor gets run off the third 3 phase block inside this inverter. But the DC/DC is in the trunk and has much smaller wires so maybe it hooks somewhere else. Maybe it's return power to the batteries?
Oh well, not super important. I'm just trying to understand how this works before I start breaking it.

Please don't break it.

You are talking about rear side connectors yes? The uppermost is direct connection to the HV battery. Other one which has the same form but it uses different connector, there they connect power for AC compressor and the heater. Both of them are directly connected to DC cap inside.
You can ease your connection and buy one battery dc cable and use it to power inverter like I do. I split the cable and I use the other end to cover the second set of contacts. You can also get the other cable and use it to power your heater... It is eberspracher 5kw unit and it works with SW CAN commands. Or you can use leaf heater which is the same but uses pwm commands.

I will describe how the interface board works later with pictures...
And yes I use printed 18mm spacer with O - ring groove to lift the cover a bit. You can mill it off Alu plate. I'll post solid works file.

[Isaac96]

Please don't break it.

You are talking about rear side connectors yes? The uppermost is direct connection to the HV battery. Other one which has the same form but it uses different connector, there they connect power for AC compressor and the heater. Both of them are directly connected to DC cap inside.
You can ease your connection and buy one battery dc cable and use it to power inverter like I do. I split the cable and I use the other end to cover the second set of contacts. You can also get the other cable and use it to power your heater... It is eberspracher 5kw unit and it works with SW CAN commands. Or you can use leaf heater which is the same but uses pwm commands.

I will describe how the interface board works later with pictures...
And yes I use printed 18mm spacer with O - ring groove to lift the cover a bit. You can mill it off Alu plate. I'll post solid works file.

Aww, buying a cable is no fun! Also they cost about $100 on ebay, for that price why not buy another inverter?
Lugs to those connectors should work (yes I will insulate it properly).
I don't know what I will end up using for heater... Probably just a simple PTC heater core.

I guess it's time to hand-mill some aluminum (or maybe HDPE plastic). Sabre saw and drill, the low-tech way.

No, arber made use of the 3rd block...that's usually for the transmission oil pump. There is a great overview on the ever auto YouTube channel explaining the components.

Okay, I stand corrected. I'll have to go watch that video now...

[mdrobnak]

Stupid autocorrect.
This is the video:


Yeah I only have the propritary phase cables on the way, I'm in the same boat WRT not having cables..

[Jack Bauer]

So based on Arber's design I have decided to whip up a board that can be made at JLCPCB. After all I have two of these inverters so they need something to control them.

As Mr.Lebowski has released his source code we can get the dspic placed by jlc and program it in circuit. So as per Arber's design the big L controls the aux inverter and the big J can control either MG1 , MG2 or both. Thinking of putting an option in there to allow the big L access to one or both of the main inverters also.

Based heavily on the Gen 3 prius design it will use the same 34 way idc to ampseal 35 system for external connections.

[arber333]

So based on Arber's design I have decided to whip up a board that can be made at JLCPCB. After all I have two of these inverters so they need something to control them.

As Mr.Lebowski has released his source code we can get the dspic placed by jlc and program it in circuit. So as per Arber's design the big L controls the aux inverter and the big J can control either MG1 , MG2 or both. Thinking of putting an option in there to allow the big L access to one or both of the main inverters also.

Based heavily on the Gen 3 prius design it will use the same 34 way idc to ampseal 35 system for external connections.

Well thank you JB.

On a second note, since you are using L controler in circuit, please hold a bit... i made some improvements. You may want to include them.
1. CAN bus with dsPic will require code rewrite since L has its own CAN which has nothing to do with our CAN bus.... . But i built 2x CAN bus transcievers onto the board. Could be fun running everything through CAN.

2. I have successfully built autothrottle into Lebowski AC compressor control. I apply 12V signal to throttle line and it now spools up to its RPM and holds it there. You can setup throttle travel in setup menu and have it 80% or 75% etc... Compression works very well and when you remove signal it spools down slowly. Idea is to run it with AC clutch signal and retain whole AC system in the car, save compressor.
Basicaly i put 10K/3K3 divider onto throttle line and put 470uF elcap across 3K3 resistor. Signal from 12V now rises from 0V to 3.4V in about 1s and of course it diminishes when stopped. With Lebowski acceleration limiter set broadly you now have auto-throttle

3. You could just build a second Lebowski brain into it. It doesnt require much space. No need to take away AC drive... You need to use precharge system somehow. I recommend to keep Lebowski in reset via relay until DC switch is triggered.

4. I am well on a way to try Wifi gizmo with Lebowski. Please include MAX232 chip onto RX/TX pins of dsPic. We need to translate TTL into serial for our Wifi module. I want to have everything on wireless setup.
This one did not pan out as MAX232 reverses voltage. So i am left with blocked wifi module. I will try different methos...

5. Additionaly series of 6k8 pulldown resistors are important addition because they drain pwm signals and protect driver section from switching on when STM32 resets under power.

6. You can build original contact headers on the board. They are 2.54mm pitch dual row and i tried to remove original low profile connector and replace it with ordinary 2.54mm dual row socket. However i think there is not enough pins to connect everything including brain I/Os, Lebowski I/Os and resolver lines... You need at least 30pins if not 32. I like to have 12V/GND supply separate from other connectors.

I will post my rev 3.2 on github. I just need to finish drawing. Maybe wait untill tomorrow?

EDIT: I uploaded rev3.2 to github now.

A

[mdrobnak]

Speaking of headers - I had a hard time finding the 10 pin header that you could buy in less than 500 quantity. I found the 18 pin no problem..
Are you talking about the header for the external control signals?

[arber333]

Speaking of headers - I had a hard time finding the 10 pin header that you could buy in less than 500 quantity. I found the 18 pin no problem..
Are you talking about the header for the external control signals?

Yes i am speaking about original Volt connectors in front and the way they connect to original control board. I have unsoldered that from flex strips and soldered normal 2.54mm 26pin female sockets. I assumed we could have the male counterpart on our control board. True i can only use 22pins, but that doesnt mean we cant put it on board.
If you look closely on photos you will see there is not enough wires to use everything. Maybe One Johannes board and one Lebowski AC control...

But original connectors are hard to get anyways. Maybe Ampseal would be better way.
Damien you just have to verify if connector can fit into front panel. There is quite small space with all the coolant hoses and all.

BTW coolant hose connector can be taken off and cut short so you can use any 18mm coolant hose with it.

[arber333]

Further answers inline

Hi all,
Just got my Volt inverter in the mail. It's a 2013-2015 with p/n 12643810. Looks identical so I guess I have the right inverter.
Here it is with a Rev2 brain for scale:

The brain doesn't fit inside the lid. Spacer time maybe?
Yes i have printable 18mm spacer with O-ring grove. I put it in my Github design folder for Volt inverter board.

I have parts from DigiKey on the way so we'll see how many components I forgot to buy

I'll be attaching lugs to those weird output connectors, just using 1 side of the inverter for now. And probably also attaching lugs to the power input connectors, since I didn't get those wires with it.
I have demonstrated how to thread holes for phase cables in original conductors here: viewtopic.php?f=2&t=697&start=20#p9999

Arber, it looks like you only have the upper power connector hooked up?
On the outside the upper connector is connecting the HV battery via the cable and it is a sealed connection. See here on my installation
Lower connector is connected to AC compressor and heater.

It looks like the connector in the upper right corner is for monitoring HV so I'll just hook that to the interface board.
RH side upper connector was meant to supply clean HV to main board sensing circuit. I use it in the same way. I you use my translator board for rev2 Huebner board you should just take the connector off the driver board and replace it with good 2 pole cable to supply the X1 connector with sensing HV.
On the upper LH side driver board has its supply connector. I dont recommend even touching driver board.

How do we handle the dual current sensor inputs? If running only half the inverter, then wouldn't the output of one sensor feed into the output of the other and get messed up? Maybe just leave the unused sensor disconnected? I'm no EE so please excuse the silly questions
Both current sensors have their own input pins on connectors. I use 4K7 pulldowns to decouple them in 0A state. When single side is used they perform as normal hall sensors. My method of using both drive units sensors will let me use resistor divider on both sections with further signal paralelling before division. So normally in single side operation i use 6K8 resistor as a pulldown and i feed it 5V level signal through 3K3 inline resistor. This makes for 3V4 signal divider and true i get 1V6 on sensors in 0A state. You will use this divider for single side.

With dual side operation i only experimented on desk. I replace 3K3 signal resistors with two 6K8 and lead them to a single 6K8 pulldown for each channel. This is how i get 1V6 in 0A state. Now how the gain will behave i intend to see. Do i need to change gain in code or do i just keep the current gain so inverter would think it is operating 500A power section, when in thruth it will power 1000A from both. TBD.

When (if) I get this working I hereby pledge to document it and wiki it (and also any explosions that happen)

Thanks!
-Isaac

EDIT After some more disassembly, it looks like the two connectors are just different ends of the same pieces of metal. I guess one wasn't enough for the current draw? Anyways I'll probably fabricate a cover and just bolt straight to the bus bars.

Definitely, if you have option of alu fabrication you can make the covers yourself. They only use simple flat seals.

[Jack Bauer]

Got a chance to place the components on the board. Lots of space:) Now to go read Arber's comments and design em in.

[Jack Bauer]

Now with Arber's latest mods for 12v big L throttle and option for wifi (includes level shifter).