Communicating with the Lexus GS450h Inverter/Converter

[xp677]

I did more testing and I am definitely getting different results.

If I run Damien's code then results are as follows.
- FORWARD is working fine
- In REVERSE mode no spin at all (just slight movement of the shaft back and forth), keep in mind this is with mg1_torque=0

Now, when I comment out line setting MG1 torque to 0, in REVERSE motors spin in the same direction as FORWARD but at about 1/2 rpm per the same torque value, they sound like they were "struggling"

I fear that Camry may use slightly different HTM bits, if so it would be huge disappointment. Without having access to working Camry I would not be able to capture any data to compare.

Anyways, my workaround for reverse works fine however I am still concerned about some other symptoms.
Here is video during which I am re-applying throttle during motors spin-down (when motors slow to around 100-300 rpm).
You can observe transmission rattling really badly. Sometimes motors will stall completely before they pin up again.

I wonder if either one of you observed this behavior or is it specific to my setup.

What was your workaround for reverse?

Twitching of the motors back and forth happened to me when I had a resolver wiring issue. Maybe check these, as Damien said.

Make sure your input shaft is locked - potentially you are spinning the input shaft instead of the output shaft, hence a lower speed, etc, it all depends which has less resistance.

The "rattling" can also be explained by the resolvers. I'd scope the resolver outputs, should be a sine and cosine with varying amplitudes at 10kHZ, google if you're not sure. Easy to test by spinning the transmission by hand.

I'll message you regarding service manual things.

Note that the service manual is not correct for a RHD regarding MG1 and 2 outputs. Neither is the picture you posted.

MG1 is on the left (RHD inverter).
MG2 is on the right (RHD inverter).

As evidenced by the OEM cables, the one labelled "MG2 RHD" has a longer rubber sleeve, it is intended to run past the MG1 connector.

Note that it may all be the other way around on a LHD inverter.

Another way to check is to look inside at the IGBT arrays, each MG2 phase should have 2 pairs of IGBTs, each MG1 phase should have one pair. See here, page 14: https://www.osti.gov/servlets/purl/928684

You can find which is which using a multimeter. Or by connecting the resolvers and spinning the input shaft by hand while holding the output shaft stationary. The motor which reads rpm is MG1.

[arturk]

Damien/xp677, I appreciate all your suggestions and help.

I did test few things and wanted to share my findings.

Indeed, there was a problem with resolver connections but not with noise or signal quality.
Problem was again with documentation, I wired resolvers according to Camry documentation using inverter's external connector (keep in mind Camry inverter has this dedicated 13 pin connector for resolvers unlike GS450h). All of GXX(X) signals for Generator (MG1) and all MXX(X) signals for Motor (MG2). Turns out that does not work, in this configuration MG1 was matched wih MG2 resolver and vice versa. I was able to spin motors but reported positions were not accurate therefore I was getting rattling noise and stalling (great catch Damien!).

Fix was very easy, just swapped resolver cables at the transmission end and I get silky smooth response from motors! They now seem extremely powerful, spin up super fast and without an effort. Amazing!

However, again I have a problem getting DRIVE and REVERSE working properly
I modified code to control each motor independently in an effort to isolate the problem and I found out I can only get good results by applying negative torque for MG1 and positive torque for MG2. When I reverse torque: positive foe MG1 and negative for MG2 they turn very slooooowly and do not produce much torque. I do not judge speed by simply looking at the shafts but I can tell by noise and RPMs reported by inverter based on feedback from resolvers.
Something is definitely not right.

I do not really trust pinouts for the resolvers from documentation I have. I know that I have excitation connected properly but I wonder if there could be an issue with polarity of Sin and Cos signals. I am going to open up inverter again and trance/verify connections back to internal control board based on xp677 diagram.

Also, I have purchased GS450h inverter which I will be getting in a few days. This way we will be working with the same components.

So, thanks again for your help. I will not trouble you for few days until I am done with my homework

[xp677]

Yeah, I was going to suggest using my wiring diagram for the internal connectors to the PCB. Since it's the same part as the GS450h, I can't see it being different, even if the external connectors are the same.

For reverse, have you tried running just MG2 with negative torque?

[arturk]

For reverse, have you tried running just MG2 with negative torque?

I documented results in following videos.
Keep in mind that motor assignments I referred to in my previous post were "logical" (as they are appear in the code). Physical motors are acty\ually switched. For example, when I control MG1 in the code, I am controlling physical MG2 in the transmission. I know it may sound little bit confusing but this is how it works right now with resolvers assigned to their motors. I could swap all the cables around but that's not the point. Point is that I can run each motor smoothly in one direction only. Below videos show how MG2 (physical) behaves, MG1 does the same but with torque values reversed (+/-).

First MG2 controlled by negative torque (works really well):


Second one shows controlling MG2 with positive trorque (problematic):

On this video MG2 is controlled with positive torque (0-500)
1) Starts spining slowly with torque above 0 until around 150 (0:00s - 0:10s)
2) With torque reaching around 150 motor stops (0:10s)
3) As torque is being increased above 150-200 motor starts spining in the other direction! (0:10s - 0:19s)
4) Keeps spining in the other direction until torque reaches around 500 and stalls with high pitched noise. (0:19s)

I know it is crazy and I am not sure what is causing it. As I mentioned I will verify resolver signals back to internal controller connector.
Also I should have my GS450h inverter on Friday, hopfully it will help clarify something.

[arturk]

I have just received GS450h inverter and want to report results as soon as possible.

I simply swapped Camry inverter with GS450h.
All other components and wiring remained unchanged from my previous setup.

Quick testing revealed that entire system (now with GS450h inverter) works as expected and reported by xp677 and Damien.
I can successfully spin up motors in both DRIVE and REVERSE modes using latest code posted by Damien.

This is both good and little bad news.
Good because I was able to get it working.
Bad because it became apparent that controlling of Camry's is limited (both motors spin reliably in one direction only).
Either there is something wrong with my inverter (I doubt that) or because Toyota changed something in serial protocol .
I do not have means to investigate it further.

Anyways, for now I am going to stick to GS450h inverter to avoid distractions.
Hopefully at some point in the future we will have resources to figure it out.

Here is few comparison pictures of both inverters.

[xp677]

Interesting. Many try swapping the control boards between them, if you dare.

Also perhaps have a look at comparing the MTH data.

The issue may be a compatibility issue between the inverter unit and the transmission, I can't think how, perhaps there are slight variations.

There are quite a lot of unpopulated components around the resolver interface ICs on the GS board, what does the Camry one look like?

[Amoor]

Hi guys i was always wanted to convert my lexsus to ev and now i am ready (with your help) now my main problem is to design the opensorceCommunicating for gs450 i dont know how to start and i dont know any basics about pcb boards and how to put parts on it and programing it can some one lighten me perhaps with a video on how to do it form A to Z there are people like me who have limitation in this kind of work
If some one can make a video or show me how its done step by step

I can not afford to buy it thats why i need your help


Thanks alot

[Kevin Sharpe]

If some one can make a video or show me how its done step by step

Can I suggest you take a basic soldering course at a local technical college and try building some mixed technology boards? IMO this is not a beginners project and you will want a robust and reliable controller in your car.

[sfk]

This is all very much at the sharp end of reverse engineering and bleeding edge prototyping right now. The sort of EV conversion for beginners videos you are hoping for might be about 5 or more years away from being produced (if at all) sorry.

It's like standing at the base of Everest and calling up to Hillary and Tenzing to stop, come back down to the bottom and carry you up.

This is not mainstream technology yet, but it's rapidly advancing! Best to go through the entire thread and many of the others on this forum to get a better understanding of what you're asking.

And Damien's YouTube channel is very informative - https://www.youtube.com/user/pooey1911

[xp677]

Hi guys i was always wanted to convert my lexsus to ev and now i am ready (with your help) now my main problem is to design the opensorceCommunicating for gs450 i dont know how to start and i dont know any basics about pcb boards and how to put parts on it and programing it can some one lighten me perhaps with a video on how to do it form A to Z there are people like me who have limitation in this kind of work
If some one can make a video or show me how its done step by step

I can not afford to buy it thats why i need your help


Thanks alot

As said above, this is not a basic DIY project. If you can't understand how to complete this from the information in this thread and Damiens Github, you should aim to gain the necessary skills before attempting it.

The modifications involve high voltage electricity which, if mishandled, can be dangerous. I think that this particular project is best handled by those who are proficient in electrical/electronic work, and I feel that a certain level of experience is necessary to safely undertake this project.

You do not need to "design the opensource inverter", you can just buy one from Damien. If you can't afford it, you can use the freely available files to order your own PCB, you can order the components, and solder them onto the board. Some PCB manufacturers will do all of this for you, for a fee. Try allpcb.com.

Some of the work required, such as soldering the SAM3X8E MCU, are particularly daunting - I'm not looking forward to doing that myself when I come to order my final PCBs!

[Amoor]

Hi there sorry
I know what i am dealing with i am learning about ev conversion from 5 years now i also build my own lithium ion packs and i do soldering the only down side is i didnt do pcb board before (i can hire a professional for pcb)
But i first want try doing it my self

I know its a hard work to do
If you guys did it i know i can do it
I just wanted you guys to tell me where to aim

Thanks alot

[Kevin Sharpe]

I just wanted you guys to tell me where to aim

Damien publishes everything on GitHub (here).

[Amoor]

I just wanted you guys to tell me where to aim

Damien publishes everything on GitHub (here).

Thanks a lot Kevin

I have a question can we connect a 3rd motor to the input shaft and connect wires to the lexsus inverter or maybe to a 2nd controller and put a button in the dash to run the 3d motor when we need

Thanks a lot for damion and all of you for this great work

[arturk]

I have a question can we connect a 3rd motor to the input shaft and connect wires to the lexsus inverter or maybe to a 2nd controller and put a button in the dash to run the 3d motor when we need

Amoor, attaching third motor to the input shaft does not seem like practical solution due to configuration of this gearbox.
If you wanted to use MG1 for driving then you need input shaft locked, with third motor attached it is problematic.
Now, if you wanted to use all three for driving it is technically possible but would would have to perfectly synchronize speeds for three motors.

I what your motivation behind third motor. Here is nominal power rating MG:134 kW, MG2:147 kW which gives you combined 281 kW (376 HP). For most applications it is more than plenty.
If you wanted more I imagine you could push those motors much harder for brief period of time. If I remember correctly Arlo was able to get 300hp out of Nissan LEAF motor rated only at 110 HP.

... i didnt do pcb board before (i can hire a professional for pcb)
But i first want try doing it my self.

If you really want to learn how to solder SMD PCBs here is excellent video from Johannes:


However please keep in mind that some of the components on our GS450 controller are much harder to solder than ones shown on this video.
As other recommended it may make more sense to have some boards soldered professionally but other you could do yourself as you gain skills.

Good luck with your project!

[xp677]

Nice job on getting the Yacht up and running, Damien!

It's funny how our engine bays look almost identical:

[Amoor]

Hi guys
Can i use Haltech ecu or unichip instead of damion's uc (microcontroler) is it possible ??

[Kevin Sharpe]

Can i use Haltech ecu or unichip instead of damion's uc (microcontroler) is it possible ??

Anything is 'possible'

To help you scope the challenge can I suggest you download Damien's controller code and try running it on your chosen hardware? This will tell you a lot about their development environment and support for third party software.

[Amoor]

hi guys i found a Forum that (mauswerkz) who converted a bmw with the gs450 gearbox and inverter he shared some of his work and what hed did
it might help us improve damions uc controler

https://forum.e46fanatics.com/showthrea ... 667&page=4

https://www.diyelectriccar.com/forums/s ... 42946.html

i know its too late for finding this guys work after damion finished the super youth but i think it will be usefull in some point

[Kevin Sharpe]

i know its too late for finding this guys work after damion finished the super youth but i think it will be usefull in some point

Many of the people on this forum are from DIY Electric Car and will be aware of the historical work on the Toyota/Lexus drive trains. Unfortunately, none of the detailed information was ever shared which is what forced Damien, Tom, and others to reinvent the wheel (same story with the Tesla drive units). It's critical we all learn from past mistakes and share knowledge as open source going forward

[arturk]

Hi guys
Can i use Haltech ecu or unichip instead of damion's uc (microcontroler) is it possible ??

Amoor, Lexus/Toyouta inverters communicate in synchronous way using 4 differential signal lines. None of the controllers you mentioned has this kind of interface built. And then there is still software to be developed. It is really not good application for these controllers because in a way these controllers are almost equivalent of the inverter in ICE world. They manage running of combustion engine and inverter manages running on electric motors. Both have to be controlled by something. In out case you need controller just like one Damien designed.

Another option is to get stock Arduino DUE and just build interfaces around it such as serial communication transceivers and drivers. This would be pretty easy to do since required components are easy to solder. Still, it only makes sense if you are not able to use Damien's board.

[xp677]

hi guys i found a Forum that (mauswerkz) who converted a bmw with the gs450 gearbox and inverter he shared some of his work and what hed did
it might help us improve damions uc controler

https://forum.e46fanatics.com/showthrea ... 667&page=4

https://www.diyelectriccar.com/forums/s ... 42946.html

i know its too late for finding this guys work after damion finished the super youth but i think it will be usefull in some point

This project is what alerted me to the potential of these units, and let me know that it was indeed possible. In the end, I was unable to contact him to discuss any of his code, and instead I purchased a GS450h, and received advice from others who have completed similar projects. As was mentioned, it required reinventing the wheel!

Hi guys <br style="null;/*startDM*/background-color: rgba(0, 0, 0, 0) !important;/*endDM*/">Can i use Haltech ecu or unichip instead of damion's uc (microcontroler) is it possible ??

No idea, but I built an earlier version of the controller that Damien has released, using a $15 PCB from a fab company in China, with around £30 of components, so, what's the point?

If anyone is interested, I can cut down the schematics for a my early controller, which accepts a complete Due board, and post that, this would reduce the soldering required to just the CAN transceivers and associated components.

[EVPanda]

I would love it if you would do that. I have having multiple options.

[Amoor]

[/quote][/code]No idea, but I built an earlier version of the controller that Damien has released, using a $15 PCB from a fab company in China, with around £30 of components, so, what's the point?

If anyone is interested, I can cut down the schematics for a my early controller, which accepts a complete Due board, and post that, this would reduce the soldering required to just the CAN transceivers and associated components.
[/quote]

yes please

[xp677]

OK, see attached, this schematic should work fine. I have laid out the board similar to my original controller, it's up to you to finish the layout and route it - that's too much work for me to do right now.

It should autoroute if you play with the component locations a bit.

Created in Eagle 7.6. Rename the files to .sch and .brd

[xp677]

I've already changed a few components since my last post. The FDT457N is much larger than needed for many of the tasks, a smaller FET such as the PMV20ENR would be more suitable for the relay driving, and you can likely use something even smaller for the PWM taks. I used the larger FETs all round due to laziness.