Toyota Prius Gen 2 Inverter Controller

[MattsAwesomeStuff]

I had to go back to the Prius manual to work out what you were referring to.

Sorry, I meant that for Johannes, to be clearer about exactly what terminal I was referring to.

Connect your inidcator bulb between one of the phase terminals (U, V, or W) and HV+ on the bus supply.

Okay, so I think that's what I put in my diagram above, except you split the system 12v from the bus 12v.

Good so far, I have two power supplies.

By default, the lower transistor in each half bridge is on when the IPM is enabled, so the bulb should light on all three phases.

Surveyyyyy SAYS?:

... nothing.

Well, 0.26v on between 12+bus and any phase. -0.26v between Gndbus and any phase.

Connecting or disconnecting MSDN to 12+system has no effect on this voltage.

I even got paranoid and swapped my gator wires just to reconfirm.

Also swapped out a new Dupont wire on MSDN, just in case. No change.

Find the PWM signal for each phase, these are;

MUU (Blue/Black) pin 18 on the IPM, pin 9 on I-10
MVU (Blue/Yellow) pin 19 on the IPM, pin 10 on I-10
MWU (Yellow/Black) pin 20 on the IPM, pin 11 on I-10

The voltage on these connections should be similar (or I recall a little higher) than the system supply. This is the internal pullup.

Yes. 14.6v across any phase to ground, with a 12v system supply. So, a little higher, yeah.

...

Oh, and, one other thing no one's mentioned, but just to rule it out, since the lid has been off this whole time (months), I've got a jumper jammed into the enclosure switch to jumper it closed (have for months, nothing's changed). Gave it jiggle, seems solid. However, removing it has no effect on the voltage I was measuring either. Still 0.26v.

There's also a little white connector with a single wire, that connects to the case switch, that I do nothing with.

Here's my setup just in case I'm communicating this wrong.

[MattsAwesomeStuff]

Minor update, no progress.

- Switched over to testing MG1 instead of MG2, wondering if perhaps MG2 was blown. This means connecting pin 20 to System 12+, and testing the other set of terminals.

- ... measurements as MG2. No change.

Are *both* inverters burnt out?

...

So, I'm stumped.

What next, start disassembling the inverter itself and looking for exploded transistors?

I have a feeling that the problem is something so basic, and so stupid, than everyone has presumed I would have followed it already, and thus not mentioned it to me.

And if I knew what that was, I would un-stupid it.

Time to just abandon ship and buy something else?

[johu]

Unlikely both are dead.
The jumper connects the "??????" connector to GND when the cover is on. Not relevant for now.
Have you connected GINV?

[SciroccoEV]

Have you connected GINV?

I wondered about GINV. I have my PWM pulldown transistors referenced to GINV, but there is continuity between GND and GINV, so it shouldn't make any difference.

I haven't really touched the inverter since I got it to spin a motor and I have a 1990's Brusa/Solectria inverter blocking the workbench at the moment, but I'll try and replicate a test setup.

[MattsAwesomeStuff]

Unlikely both are dead.

Good... if it leads to a solution :p

The jumper connects the "??????" connector to GND when the cover is on. Not relevant for now.

Hmm, maybe, but not directly.

On the lid of the case is a jumper that lines up with where I put the wire jumper, labelled "Jumpered case switch" in white. Nothing plugs into the ???? connector that I see or that I can recall. There's no wire or other connector on the lid. What's it there for? What's the puffy electrical-tape-wrapped green foam thing wrapped around it?

I never got wires with my inverter, so whatever was plugged into there, I don't think I ever saw.

I went through Dr. Kelly's Weber Automotive disassembly walkthrough, and he never mentioned what it's for:

Damien's does have it, but it doesn't look like he does anything to it:

Have you connected GINV?

Yes. Pin 16 on i10 to the same system ground. I think it's just common to the main ground on i9 anyway.

...

Something worked, once upon a time, when I tapped on the control board. So, whatever's "broken" it's gotta be something to do with a wire, somewhere, right? The control board is completely disconnected right now, and the inverter still isn't working as it should as just a basic H-bridge or however it's configured. To me, it sounds like either some very basic isn't getting power, or, some kind of safety switch is still disabled.

[johu]

You can choose to connect something to the "????" connector as a switched ground. I actually route inverter power through it to make sure the contactors drop when the lid is opened. Not sure what's inside the foam.

How much current draw on "System 12V"?

[SciroccoEV]

Not sure what's inside the foam.

A ferrite suppression core as I recall.

[MattsAwesomeStuff]

How much current draw on "System 12V"?

Hmm, it's through a 12v power brick, not sure.

My knockoff Kill-a-Watt was showing basically zero or 0.5 watts without being connected, and 14.7 watts when powering the inverter. It might not be accurate at low ranges, but... that'd be just north of an amp.

Lemme intercept that with a multimeter to check... 970mA.

Something in that ballpark. This is without the control board connected, just powering up to the i9 connector.

[johu]

That is also normal. Hmm

[MattsAwesomeStuff]

That is also normal. Hmm

I wish I could just have a working example, and then start unplugging things and making a list of what kills it.

I guess I could try unbolting the capacitor and, I'm not sure all what else is under the black plastic there. See a bolt, remove a bolt, until I'm looking at a circuit?

[MattsAwesomeStuff]

Well, taking a stab at things for lack of anything better to try.

I took off the capacitor and the i10 connector and used (new) Duponts to tap right into the inverter's ... IPM? Pin 1 was the brown wire, as Scirocco confirmed. I tested the wire while I was at it, had continuity, but no matter. Tied Pin 1 (MSDN) to 12+ system. And, whatever the last pin is there is ground, so I tied that to ground just in case.



... no difference. 0.22v on the phase outputs. If I crank the bus voltage up to 24v from 12v, I get 0.28v, and other such minor increases the higher the bus voltage goes. Same as it always has.

And, maybe I misunderstood what this lid connector does, I see no connections to the rest of the inverter from it, just the negative side is grounded and the other side is that weird terminal after the foam and ferrite. So, doesn't matter I guess, if I had it jumpered or not. It doesn't seem to go anywhere.



...

I guess I'm ready for more advanced diagnostics if need be.

[MattsAwesomeStuff]

...

Whelp. I know this is my problem only, no one else's, but today has now been the end of my second 3 month stint where I had almost unlimited time to tinker on this inverter before the summer. It's been sitting on my desk and had me avoiding doing my taxes on said desk because it's perpetually "almost working, just one more thing" for 3 months now, and... I just don't have any skills to troubleshoot this other than to be handheld the whole way, and I'm frustrated being a nag, I like contributing and getting things done, not making others do things for me.

I had a lovely thought today, that... the whole reason I started getting into EVs is because I was a new driver, and I had no intention of learning about engines in my life, where motors and electricity were things I thought I had a good base for. But, that, I dunno that that's even true anymore, it would be a lot easier to get my car up and running if I just put an engine in it and not go yet another year without it. Even as a complete novice, it'd be faster and easier to rebuild an entire engine myself, and less frustrating, because at least the answers to every problem I might have are out there. And, even for a 50 year old and unpopular car, there's still hundreds of people who've done what I'm doing, where there wouldn't be surprises to them if I had questions.

So... I dunno. I also have a Gen 3 inverter I pulled a few weeks ago which was my backup plan if my Gen 2 hit a diagnostic dead end, but, I think I'm more interested in throwing in the EV towel than going that route.

And yet, if I'd spent a bunch of time helping someone diagnose their project, only for them to abandon it, I'd feel less like helping the next person that comes along, so, that's no good. The only reward people get for helping others in the community is the joy of seeing them succeed, and seeing progress updates. So, it's not just my own efforts I'm tossing away, it's also the long list of those that've pitched in where they could over the last few years.

Can anyone give me a realistic guess as to whether I am "just around the corner" from this working, or, whether this slog is likely to continue on like this for quite sometime still? Naturally, "could be the next thing you try" might be the answer, but, when not, a better guess than I'd be able to make? I don't have a lot of perspective of where I am in the troubleshooting process.

[johu]

It seems like you've hit the point where you'd need someone stand next to you with a scope in their hand and checking exactly what's going on. With this short feedback loop it'd probably take 10 minutes to get it running or at least diagnose why it isn't. I am out of ideas but like yourself I think it's something really silly, impossible to see from the pictures. The feedback loop over the forum is just too slow.

If you have a fast and stable internet connection and a good camera I could offer an engineering hour, provided you want to throw that kind of money at it.

[mjc506]

Where abouts in the world are you?

[MattsAwesomeStuff]

It seems like you've hit the point where you'd need someone stand next to you with a scope in their hand and checking exactly what's going on. With this short feedback loop it'd probably take 10 minutes to get it running or at least diagnose why it isn't.

Yep, I have the same feeling.

The feedback loop over the forum is just too slow.

It's not the speed, it's more just how limited it is. And, many times having been in the opposite position, it's far more frustrating being the helper than it is the one being helped. To the point of... I'd rather get in my car and drive an hour over to someone's house, than try to diagnose why their network isn't working, or, why they can't find their excel file, or, whatever vague technical problem someone doesn't even really know how to accurately describe to me. So much of it is just... the person doing the work *noticing* what seems out of place, not necessarily specific measurements.

If you have a fast and stable internet connection and a good camera I could offer an engineering hour, provided you want to throw that kind of money at it.

I'm closer to just buying a finished Gen 2 board from you, or buying another inverter, than doing that. I've paid for help twice on my project, and each time really felt like cheating. Like, why am I even doing this if I'm just paying someone else to do it? But I know I was given a great rate and it was for small technical things I had no interest in learning, where I would have missed things (brake caliper rebuild and headlight rotating disassembly/zinc plating/reassembly, involved specialized tools and setups that dwarfed what I was charged).

I've had that in the back of my head for a while though. I'm more or less freeloading off the community here, you should be writing new software, not debugging someone else's deprecated designs. If push comes to shove, I might just get the scope out and sign you a cheque.

...

I had the damned this working... ish... depending on how the control board was tipping. How can I not have even basic continuity when it's stripped to nothing?

... I think my last effort will be to try to bully the scrapper who sold me the inverter to give me another, and if symptoms are the same, I'll know it's not the hardware, it's me.

[MattsAwesomeStuff]

Where abouts in the world are you?

Western Canada.

[mjc506]

Ah, a bit of a trek then from here

[MattsAwesomeStuff]

A glimmer of hope...

One of the local junkyards now has a Gen 2 Prius, which, they're not supposed to have. Same ones that sold me an inverter for $50 and the hybrid battery for $33.

Hopefully I can confirm or deny whether it's the inverter.

[MattsAwesomeStuff]

And... I've shot myself in the foot.

After the last time I was there and found a battery with the service plug still live, I told them about how anything with orange cables was the high voltage system and dangerous.

... so this time they cut the battery out, and, the inverter was also removed. I doubt anyone else salvaged it, it's only been there a few days.

i did manage to get all my missing wiring harness, the MG1 and MG2 wires, the 32-pin connector, etc. But, no inverter. There's a chance it's on a pallet in the processing area, but, manager is gone for the long weekend. Oh well.

[Isaac96]

Here's a baseline test of inverter functionality that I ran today. I've used my inverter to drive small DC motors and it seems to be very functional.

Wiring as follows.

I9 gets 12v and ground.

HV into inverter (this being 12v at the moment).

On the white 32-pin: (I10 I think?)

Ground to pin 16 (black)
12V+ to pin 25 (red/white, MSDN) --- NOT GROUND!
And signal to pin 9 (pink/blue, MUU)


Default behavior. No signal, MUU floating.
Bulb does not light when between phase U and ground or phase U and HV+. So phase U is floating.

Then we ground MUU.
Bulb lights when between phase U and ground. Phase U is supplying HV+.

Now we let MUU float again.
Bulb lights when between phase U and HV+. So phase U is grounded.


So it seems that until you send a signal to a phase that phase is 'asleep' and will neither source nor sink current. Once it's received a single ground pulse it will then respond to signals.

[MattsAwesomeStuff]

And signal to pin 9 (pink/blue, MUU)

Hmm, what are we calling "signal"?

Default behavior. No signal, MUU floating.
Bulb does not light when between phase U and ground or phase U and HV+. So phase U is floating.

Okay, this is the behavior I've been having that it's not supposed to have. No power.

Then we ground MUU.
Bulb lights when between phase U and ground. Phase U is supplying HV+.

... Well I'll be damned. Me too.

Now we let MUU float again.
Bulb lights when between phase U and HV+. So phase U is grounded.

... indeed.

So it seems that until you send a signal to a phase that phase is 'asleep' and will neither source nor sink current. Once it's received a single ground pulse it will then respond to signals.

Thank you thank you thank you. That's an interesting tidbit.

And hurray, I got the same behavior!

I'm wondering what makes it reset other than powering down the inverter, because I'm sure I've done this before and never got a bulb to light no matter what. I guess, if I was expecting a bulb to go out, and it never lit in the first place, I might have missed it when swapping to the other side of the rail? But, that doesn't make sense, I definitely left it connected when I grounded the PWM phase, it should have come on the second I let go.

Perhaps, I know I was shutting down the inverter between every change to be as methodical as possible, and so that I didn't short anything out when clipping and unclipping leads. So when I tested each phase expecting it to light before grounding the PWM, none lit. And then powering down, clipping MUU/MVU/MWU to ground, changing my test points, and starting back up, none were working then either.

I guess at no point did I ever make changes after a wakeup signal that would've told me it was going to work.

So...

This is great news. I have now verified that the power transistors are not destroyed and, to some very limited degree the inverter is working.

Each phase works once woken up, conducting to Bus 12+ with the phase signal disconnected, and conducting to Bus Neg with the phase signal connected.

Also, just for note, *each* phase had to be woken up individually, even if the inverter stayed powered up the whole time. When I switched the bulb to the V or W phases, they were not illuminated until woken up.

...

.

.

Backing up to what Scirocco and Johannes had me testing... what's the next step?

This quirk shouldn't have mattered, if the signal lines were being given PWM, they'd have their "wakeup" signal in a small fraction of a second from any PWM amount at all.

[RetroZero]

Great to see progress

[johu]

Oh wow, that wakeup is new to me as well. This Prius board has quite a life of its own.

[MattsAwesomeStuff]

Oh wow, that wakeup is new to me as well.

I'm now wondering what else might make it reset, that might be the culprit in why it was intermittent before, or, why it doesn't work at all for me now.

[MattsAwesomeStuff]

Does anyone have any suggestions of what to next pursue for diagnosing my problem?

Troubleshooting the problem from both ends (keyboard on one side, HVDC / systemDC on the other side) towards the middle...

---
From the OI software: Connection to wifi module seems to work. Interface loads. Settings save. Starts and stops.

From the Blue Pill/signal: PWM is being generated. Signal appears to be correct waveform (soundcard oscilloscope). Voltages appear correct before transistors and after transistors to the terminals.

From the connection to the inverter: Wires between control board and inverter seemed to connect correctly, PWM signal was getting to the inverter harness as judged by backstabbing the wires. Even removing harness and connecting directly to the inverter IPM resulted in no change.

Then: ??? one or several gaps here ???

From the inverter side: HVDC is getting into the inverter. System 12v is getting into the inverter. Power transistors all seem to be working, and conduct HVDC+ (after wakeup) to the motor phase terminals, and after powering PWM line, the transistors toggle to conducting to HVDC-.
---

I'm not really sure what could possibly be breaking down at this point, or, what could possibly be the missing link. If I understand correctly, my latest test of powering the PWM line and manually connecting it to 12+ and seeing the transistors invert... is just a really slow mechanical version of PWM. So why does an actual PWM signal not work, when it does the same thing to the same pins in the same way?

The discovery of a "wakeup" requirement is interesting in terms of theory, but, should not have any practical implications since a PWM signal is going to be giving it a wakeup many times a second. This is probably why neither Damien nor Johannes nor anyone else reverse engineering the inverter ever noticed it was "needed".

And, how does this relate to the inverter working when I tapped on the blue pill board to get it to work, but connecting the wires better somehow made it stop working?