Lexus GS450H VCU Support Thread

[Dilbert]

I worked out that 47k should do the trick for mg1 and mg2, I’ve fitted them and updated the firmware.

[xp677]

Great, how did they read? What is the firmware for them?

[Dilbert]

This is the firmware I have written, I haven't had a chance to test it with the gearbox, so have just used fixed resistors plugged into the harness.

float vcc = 5.0;
float adc_step = 3.3/1023.0;
float Rtop = 47000.0;
float Ro = 47000;
float To = 25+273;
float B = 3500;

float readThermistor(int adc){


float raw = adc;
float voltage = raw*adc_step;

float Rt = (voltage * Rtop)/(vcc-voltage);

float temp = (1/(1.0/To + (1.0/B)*log(Rt/Ro)))-273;

return temp;
}

[xp677]

That's great, many thanks.

How did you come up with the 47K value? I'm struggling with 8x thermistors in my project, all I have is their resistance at approx 15C. By looking at datasheets for various thermistors, and trying to work backwards from there, I came up with much different resistor values (249k for MG1 and 2 for example). Not sure where I went so wrong!

Are you able to do the same for the oil and trans sensors?

[Dilbert]

I believe 47K is the nominal value of the thermistor for MG1 and MG2, based on what i observed @ 5C.

So i typically use a 47K pull up, but as we are pulling to 5V, when we have 25C we will get 2.5V into the ADC. As the temp rises we expect this voltage to fall.

I haven't looked at the oil temperature sensors yet.

[Jack Bauer]

[Dilbert]

Looks and sounds great!

[Dilbert]

Is there an stl file or cad model available for this cover?

Not yet, still being prototyped. I've gone through 3 versions now, next one printing tomorrow, should be the final version (fingers crossed)

How did the print work out?

[slow67]

Great job again Damien!

Question, I noticed you still have it stuck in high gear. Have you confirmed the solenoid shift pattern for low gear? Is it both solenoids high instead of low? Also you may want to play with the pressure solenoid PWM (TransSP in the code). This should drop the line pressure during cruising, for less heat and more efficiency. From my experience with GM transmissions, the lower the duty cycle the higher the pressure (most likely the case here, as it isn't slipping). It also may wear out the pump being run at full pressure all the time. I know full pressure in GM Powerglide and TH400 transmissions will wear out a pump after about 10,000 miles of max line pressure (cruising miles, not racing).

Edit: It looks like pin PC18 isn't PWM capable, so I will have to use software PWM, but shouldn't be a big deal. FYI using the TransSP solenoid will only be needed for those who will be hooking a motor to the input shaft instead of locking it (and thus the mechanical pump will be spinning).

[Dilbert]

Very impressive, I was wondering if there is any way to get a feel for how much voltage overhead the super yacht actually has?

For example, could similar performance be achieved at 320V etc.. I'm not sure if it is easy to measure, but maybe the dyno run would give an insight.

[xp677]

Is there an stl file or cad model available for this cover?

Not yet, still being prototyped. I've gone through 3 versions now, next one printing tomorrow, should be the final version (fingers crossed)

How did the print work out?

Not bad, still a few areas to clearance on one part, I got it to fit with some grinding but went through the wall so need to rectify that.

The other part needs the o-ring cord recesses sorting - the supports are almost impossible to clean out. TBC.

[Kevin Sharpe]

Very impressive, I was wondering if there is any way to get a feel for how much voltage overhead the super yacht actually has?

Even with the test battery dropping to 260V under load we experienced no drop off in performance when accelerating beyond 70mph. Damien has also undertaken some driving with a 230V battery (~180V under load?) and even with the torque limited to 33% the car was very spritely (see video below).

Our focus is now shifting to the long range pack installation (nominally 400V) and think it's time for other people to step up with GS450H builds that test the higher and lower battery voltage limits if they're interested

[Dilbert]

That’s very interesting, I’ve been debating the whole Toyota use of the higher voltages for the hybrids. There must be some efficiency gain to be had at higher speeds in the hybrid configuration, higher voltages and lower currents. Also I know from the gen 3 Prius when power is required the engine really spins up.

[xp677]

In hybrid use, MG1 especially has a high rpm. That's likely part of the reason for the higher voltage.

As much of the electrical power in the hybrid system is going between MG1 and 2, keeping currents to a minimum will minimize losses.

I failed my EMD module at uni, but it probably also affects wire size for the motor windings, allowing for a more compact motor.

[Kevin Sharpe]

People love to speculate but lets bring this thread back on topic and focus on Lexus GS450H VCU Support

[xp677]

I'm currently revisiting my original code, not much to add, but one useful thing to add is to have the code recheck the inverter state (MTH byte 1). This has a number of states, the important ones are:

0b00000000 which is "not ready"
0b00001101 which is "boot"
0b00001100 which is "ready"

There are also various error codes available, bits 1 - 4 indicate the presence of the 4 motor current sensors.

Anyway, the importance of rechecking the state byte is to allow us a chance to re-send the setup data should the inverter go back to it's "not ready" state (in event of a power cycle, etc). I noticed this while testing - if I had my controller connected to a laptop for serial communication, then power cycling the car would not reinitialize the inverter, as the controller was kept awake by the USB port power.

For the VCU code which Damien has since made his own, the appropriate change is to change the line:

Code: Select all

if(!htm_sent&&inv_status==0){for(int i=0;i<80;i++)Serial1.write(htm_data[i]);}

to:

Code: Select all

if(!htm_sent&&inv_status==0){for(int i=0;i<80;i++)Serial1.write(htm_data[i]);if(mth_data[1]==0)inv_status=1;}

This way, when the inverter comes back online after power loss, the presence of a 0 in mth_stat[1] will reset the status flag, forcing another round of setup attempts.

My own code for this has forked a bit further, it now includes error code parsing for the various states which can then be sent over CAN, serial, whatever. But this is the main change which I think may benefit others. Implementing my other changes would need a more major code-rewrite, which may break other parts of Damiens code.

[slow67]

Question........ I can't get MG1 to spin (forwards or backwards) independantly . MG2 spins just fine. Any Ideas? Position feedback works just fine. I hooked up MG1 to MG2's resolver and phase connections, it spun just fine.

I guess more specifically, is there any other diag information we can get from the mth/htm data?

[xp677]

You can get individual rpm and rotor position from the MTH data, I believe I had that integrated in the code that Damien used.

MG1 Speed (mth_data[6] | mth_data[7] << 8)
MG1 Speed (mth_data[31] | mth_data[32] << 8)

MG1 Position (mth_data[12] | mth_data[13] << 8)
MG2 Position (mth_data[37] | mth_data[38] << 8)

I was able to spin any combination of motors independently without issue. Maybe check your resolver wiring? Remember that MG1 spins in reverse so needs negative torque for "forwards".

[slow67]

You can get individual rpm and rotor position from the MTH data, I believe I had that integrated in the code that Damien used.

MG1 Speed (mth_data[6] | mth_data[7] << 8)
MG1 Speed (mth_data[31] | mth_data[32] << 8)

MG1 Position (mth_data[12] | mth_data[13] << 8)
MG2 Position (mth_data[37] | mth_data[38] << 8)

I was able to spin any combination of motors independently without issue. Maybe check your resolver wiring? Remember that MG1 spins in reverse so needs negative torque for "forwards".

I was doubting resolver wiring, as position feedback reports perfectly. Later tonight or tomorrow Ill debug to make sure my variables are making it to the array. I zeroed out mg2_torque, and the ThrottleVal and mg1_torque was not making it spin either direction. Being junkyard parts I’m thinking that it may be hardware, but I want to troubleshoot further before returning the inverter.

[xp677]

If position feedback is working correctly, then the code should spin the motor just fine. I'd double check you don't have something silly like sine and cosine swapped somewhere, this would give accurate feedback in the opposite direction.

If you're confident that the wiring is correct then I believe you. Unfortunately I can't offer any more info on the matter, but feel free to ask questions.

[slow67]

If position feedback is working correctly, then the code should spin the motor just fine. I'd double check you don't have something silly like sine and cosine swapped somewhere, this would give accurate feedback in the opposite direction.

If you're confident that the wiring is correct then I believe you. Unfortunately I can't offer any more info on the matter, but feel free to ask questions.

Thanks. Position feedback is working perfectly. MG1 never even tries to spin. I had the resolver plugs switched once, and MG2 just jumped around momentarily. I confirmed I'm writing ThrottleVal to htm 5&6, 11&12, 26&27, 32&33. And 4x ThrottleVal to htm 75&76. Just for giggles I wrote 0 to 75&76 with no change. I'm guessing this inverter is junk

[xp677]

You're writing to [0] as well? I guess so since MG2 works:

htm_data[0] = (byte)((mth_data[31] | mth_data[32] << 8) + (mth_data[6] | mth_data[7] << 8)) /113;

Also note that you have to write negative torque to 5,6,11 and 12 for forward rotation. No idea if this motor works in reverse, I've never tried it, it may be locked out by the inverters logic.

Have you tried MG1's power and resolver into MG2's outlet? Have you metered the windings of MG1 and it's resolver?

The power stage of this unit is very well protected!

Verify that your mth data byte 1 is "12".

[slow67]

You're writing to [0] as well? I guess so since MG2 works:

htm_data[0] = (byte)((mth_data[31] | mth_data[32] << 8) + (mth_data[6] | mth_data[7] << 8)) /113;

Also note that you have to write negative torque to 5,6,11 and 12 for forward rotation. No idea if this motor works in reverse, I've never tried it, it may be locked out by the inverters logic.

Have you tried MG1's power and resolver into MG2's outlet? Have you metered the windings of MG1 and it's resolver?

The power stage of this unit is very well protected!

Verify that your mth data byte 1 is "12".

Ill check reverse again, but currently I have mg1/2 swapped so I can watch the output shaft. Ill check htm[0] to make sure Im writing the speedsum.

[xp677]

Spinning MG1 only will spin the input shaft if neither are locked. If you lock the input shaft then spin only MG1 then I have no idea how that will react.

[slow67]

Ok yes I'm sending htm_data[0] as well. It's sending (mg2_speed + mg1_speed)/113 (then convert to LSB).