Can't Kill A Toyota Inverter

[Jack Bauer]

Ok. So I pose the question why did the smallest cap with the highest voltage rating blow? Why do we have a 0.5uf 900v cap in parallel with the big 888uf cap? Well, its a snubber. Designed to soak up excess spikes and ringing on the main DC bus. So there must have been excess spikes and ringing on the dc bus. Why? BECAUSE SOME MORON INJECTED THE HVDC INTO THE WRONG PART OF THE INVERTER THUS BYPASSING THE DC LINK!

I committed the cardinal sin of power stage design. By injecting the high voltage into the terminals I used we effectively created a star connection of battery, igbts and caps instead of a parallel connection. Its a miracle the car made it out of the driveway!

[johu]

How is that possible? Will you enlighten us?

[Jack Bauer]

Video shortly. Bit too much to type:)

Edit : Short version V=L di/dt

[arber333]

Video shortly. Bit too much to type:)

Edit : Short version V=L di/dt

OK... what was supplying the current then? I thought you bypassed inductor not put it in parallel?
Hm... i am very interested in your results since a lot of my assuptions depend on your results.

A

[Jack Bauer]

Never assume anything Arber. I did and it blew up in my face:) Video later today / early tomorrow.

[Jack Bauer]

[TFirenza]

not entirely familiar with the internals if the gs450h inverter are there similar considerations needed , does the side port connection that bypasses the buckboost converter create a similar star connection?

[clanger9]

So, basically:

[johu]

Hmm, thats exactly how my Polo inverter is wired up. The battery cables connect right to the same terminals as the first snubber on the right. No blow for 7 years. Maybe the low power (500V, ~100A DC) and 10 Ohm gate resistors are beneficial there?

[clanger9]

Would be interesting to stick a 'scope across the snubber terminals and see what sort of voltage spikes are on there...

[Ryp]

Hmm, thats exactly how my Polo inverter is wired up. The battery cables connect right to the same terminals as the first snubber on the right. No blow for 7 years. Maybe the low power (500V, ~100A DC) and 10 Ohm gate resistors are beneficial there?

Hi!
I think the key in your configuration is that your large caps are closer to the igbt:s. Yo have also 3 snubber caps. Damien connected battery bethween igbt:s and almost all caps.

[clanger9]

Ah, so more like:

[arber333]

I am more interested in how to use inverter to drive MG2 power stage AND use MG1 to charge HV battery through inductor.
The way you describe wiring your inverter inductor will be shorted and you cant use it.
But if you connect to the 220Vdc terminals and have the buck/boost transistor closed when diving MG2 you could only get like 30kW out of it no? Am i missing something?

Therefore i propose 500A contactor across inductor which would be switched off when charging. It could be wired with one AC relay so it would be off when you get AC on MG1 terminals...

[Dilbert]

not entirely familiar with the internals if the gs450h inverter are there similar considerations needed , does the side port connection that bypasses the buckboost converter create a similar star connection?

Yes definitely worth considering, we will have to review the internal layout of the caps. The original e46 conversion using the GS450h inverter from a few years back used this battery connection point and reported blowing up an inverter, but I’m not sure if this was a control issue.

[arber333]

Hmm, thats exactly how my Polo inverter is wired up. The battery cables connect right to the same terminals as the first snubber on the right. No blow for 7 years. Maybe the low power (500V, ~100A DC) and 10 Ohm gate resistors are beneficial there?

Probably that and also with addition of you using large film caps which serve as a quick reservoir AND a snubber at the same time.

[celeron55]

not entirely familiar with the internals if the gs450h inverter are there similar considerations needed , does the side port connection that bypasses the buckboost converter create a similar star connection?

Yes definitely worth considering, we will have to review the internal layout of the caps. The original e46 conversion using the GS450h inverter from a few years back used this battery connection point and reported blowing up an inverter, but I’m not sure if this was a control issue.

In the gs450h that is the most direct connection point to the main cap, can't get closer to it and there's no other "side" to it.

Can you short the boost converter and find large enough contacts for feeding enough current there? It's after all where Toyota makes the power go through, there's going to be more capacitors there to offset the cable inductance...

[Jack Bauer]

Please folks lets not draw a parallel with the GS450H inverter. It's a different beast and as celeron55 said we are using the correct terminals in any event. I was not aware of the original E46 guy having a failed inverter. No surprise I suppose as we can't be sharing info. Also if I'm remembering correctly he ran two volt packs in series as of course you need 650v....thus leaving very little headroom on the cap voltages for transients.

[Jack Bauer]

Also no need for any contactors. When in drive mode just turn the high side of the buck boost on. I'm going to try and write some muppet software for the atmega 328 to do just that.

[xp677]

The original e46 conversion using the GS450h inverter from a few years back used this battery connection point and reported blowing up an inverter, but I’m not sure if this was a control issue.

As far as I'm aware, the inverter blew due to issues with the control hardware/code (this was before he switched to using the Toyota serial protocol and keeping the original control PCB), and not related to the power connections to the inverter. Also, see below:

not entirely familiar with the internals if the gs450h inverter are there similar considerations needed , does the side port connection that bypasses the buckboost converter create a similar star connection?

However, as said, this is outside of the scope of the discussion. There is no other place to inject power to the GS inverters.

Also no need for any contactors. When in drive mode just turn the high side of the buck boost on. I'm going to try and write some muppet software for the atmega 328 to do just that.

This amuses me, its a clever idea. Sounds rather lossy at higher currents, though!

[Jack Bauer]

Sorry I meant just to run the dcdc from the HV side without having to resort to a contactor to bypass the buck/boost stage. DCDC only draws about 5 or 6 amps max at 360v so no big losses there. Muppet code :

Code: Select all

/*
Just turns on the boost hi side always. Low side off always. Runs on prius and yaris/auris boards.
*/


void setup() {

  pinMode(9, OUTPUT); //boost low side
  digitalWrite(9, LOW); //low side always off
  pinMode(10, OUTPUT); //boost Hi side
  digitalWrite(10, HIGH); //hi side always on
}

// the loop function runs over and over again forever
void loop() {
                     // wait for a second
}

Tested in a yaris inverter and seems to work on the bench.

[arber333]

Also no need for any contactors. When in drive mode just turn the high side of the buck boost on. I'm going to try and write some muppet software for the atmega 328 to do just that.

Ok then. High switch will be ON. But what about inductor? Everything will go through poor inductor and at 300A i gather it will show what it is capable of. Or not! I think i will just use the Auris as a charger. We will benefit best from that. But i am eager to know if it would work good as motor drive with some 60kW or so...

[Jack Bauer]

Gahhhhh! I am not talking about running the inverter through the buck boost stage. Just running the dcdc from the main hv via the top side of the buck boost.

[Bigpie]

If I understand correctly, you're not going to bridge the high voltage side of the buck boost physically with a new wire but by just switching on the high side of the half bridge to enable the DCDC?

[Jack Bauer]

Exactly:)

[Jack Bauer]

Tested on my Yaris inverter and model 3 drive unit. Muppet code works and lets hv through to the dcdc.