2nd LDU blown in a month - EDIT: False alarm. [SOLVED]

[mane2]

Now checking your parameters.

boost: is only 1400 vs your 2100. I thought it's only related to startup? Does it make difference on higher speeds, more power?

fweak: yours is 238 mine 220. Not that big difference. I'll try that value.

fweakstrt: You have 320 vs my 290. Pretty close

throtramp: I'm not sure which way this parameter goes. I have 55 and you have 5.. Does yours react to throttle faster or slower?

ampmin: you have 1, I have 18. I think this is related to voltage or boost.. maybe both. Your's move easier maybe because higher voltage and boost. I tried to tune this so, that there is no slack in the throttle pedal. I might need to adjust this when upping the boost.

I'll have a go tomorrow and report back how it affects. Thanks again.

Tested those values, and also added udcnom to 260V. Seems to work good, bit more power, I feel. Though it's quite wet and cold outside already and didn't dare to floor it properly.

throtramp 55 that I had, reacts way faster than 5 that you have. Dialed it back after testing. Maybe 5 is good to use if doing acceleration tests etc, to avoid wheel spin.

setting ampmin to 1 in my setup seems to have more dead zone in pedal before car moves. Putting it back to 18, makes the car get going earlier when pressing the pedal.

Boost, fweak, fconst, fslipmin, etc I copied from you. Those seem to work well. Thanks!

[Boxster EV]

My parameters have evolved through from a combination of circumstances, but importantly, I have not had a desat or overcurrent event for ages. It’s a fine balance between finding tune that suits all conditions and doesn’t risk damaging the inverter.

Rather than worrying about performance too much at this stage I’d try to overcome those overcurrent events that you’re experiencing, through softening some of the settings.

[mane2]

My parameters have evolved through from a combination of circumstances, but importantly, I have not had a desat or overcurrent event for ages. It’s a fine balance between finding tune that suits all conditions and doesn’t risk damaging the inverter.

Rather than worrying about performance too much at this stage I’d try to overcome those overcurrent events that you’re experiencing, through softening some of the settings.

Yeah, I would soften my settings if the error would actually be overcurrent. In my case, motor did not get enough power due main contactor OI firmware posted OVERCURRENT. I guess there is no error for that scenario.

Also that DESAT error that I got earlier, is not even possible on Tesla OI firmware without having a short in board. But that was probably the water that shorted it to show that.

I'm now having overcurrent showing in Lasterr when turning into reverse. Car state is normal, moves front and back. din_ocur = ok, status ok. There must be something wrong with the board.

One other thing I started wondering about, is the revision differences between these Sport TDUs. This, my third LDU revision is the same that Tesla is putting in cars when they fix motors, this year. I saw 2019 year model P100D motor on sale, that was already REMAN motor and fresh stickers with same revision. Changed this year. So this is the very latest, but not sure if there are any differences on the Tesla boards or if anyone have reverse engineered differences between those. But then again, this is probably issue with the board.

[P.S.Mangelsdorf]

fweak: yours is 238 mine 220. Not that big difference. I'll try that value.

fweakstrt: You have 320 vs my 290. Pretty close

These are big differences for these values. To quote from JonVolk's LDU tuning guide:

238hz will stick my slicks ... Bringing this to 233hz will just burn the slicks. 258hz worked well for me on street radials.

[muehlpower]

doesn't the battery voltage play a role in fweak? Boxter EV has 360V, I think you have considerably less.

[mane2]

Yeah, that’s my thinking too but I’m not sure. Maybe Johu knows?

Higher fweak feels slower with my lower voltage.

[johu]

Yes it plays a role and can be compensated by using the udcnom parameter. Then fweak is scaled up and down according to the actual battery voltage.

[mane2]

Finally got the new V6 board in. That old V4 card which got wet, had a lot of moisture in one of it's connectors (see picture). I'm sure that caused the random overcurrent errors which eventually caused the car not to move at all. Now new board is in, and wheel turn .. but now there is another kind of problem. Sometimes wheels don't turn at all, and sometimes pressing pedal like 40-45% they start to spin fast. After it gets spinning, I can regulate the speed .. but for some reason, I really need to push the pedal and it agressively gets going. Sometimes it doesn't at all. No errors posted on the firmware. Attached plot .. in the plot, it shows pedal %, potnom % and speed. First pressed the pedal to 100%, and no movement .. then lifted the pedal, pressed brake and pushed the pedal again .. now around potnom 40% wheels suddenly start rotating fast. But as you can see from graph, it requires lot of throttle before that happens.

I copied my previous params manually. Battery voltage is 240-300V. Maybe I missed something?

Any ideas?

Params:

Code: Select all

{
  "boost": 2100,
  "fweak": 238,
  "fweakstrt": 320,
  "fconst": 450,
  "udcnom": 270,
  "fslipmin": 0.84,
  "fslipmax": 4,
  "fslipconstmax": 9,
  "sincosofs": 2048,
  "fmax": 540,
  "dirchrpm": 100,
  "dirmode": 1,
  "ocurlim": 2000,
  "udcgain": 7.43,
  "potmin": 1010.25,
  "potmax": 4095,
  "pot2min": 4095,
  "pot2max": 4095,
  "potmode": 0,
  "throtramp": 50,
  "throtramprpm": 20000,
  "ampmin": 18,
  "slipstart": 50,
  "sinecurve": 0,
  "throtfilter": 4,
  "brakeregen": -25,
  "regenramp": 1,
  "regentravel": 35,
  "offthrotregen": -25,
  "cruiseregen": -30,
  "regenrampstr": 18,
  "maxregentravelhz": 150,
  "brklightout": -10,
  "bmslimhigh": 100,
  "bmslimlow": -20,
  "udcmin": 200,
  "udcmax": 350,
  "idcmax": 600,
  "idcmin": -5000,
  "idckp": 2,
  "idcflt": 9,
  "tmphsmax": 85,
  "tmpmmax": 300,
  "throtmax": 100,
  "throtmin": -100,
  "accelmax": 1000,
  "accelflt": 3,
  "iacmax": 5000,
  "ifltrise": 10,
  "ifltfall": 0,
  "chargemode": 0,
  "chargecur": 0,
  "chargekp": 80,
  "chargeki": 10,
  "chargeflt": 8,
  "chargepwmin": 0,
  "chargepwmax": 90,
  "idlespeed": 250,
  "idlethrotlim": 50,
  "idlemode": 3,
  "holdkp": -0.25,
  "speedkp": 0.09,
  "speedflt": 5,
  "cruisemode": 0,
  "cruisethrotlim": 50,
  "udcsw": 240,
  "udcswbuck": 450,
  "tripmode": 1,
  "bootprec": 0,
  "pwmfunc": 0,
  "pwmgain": 100,
  "pwmofs": 0,
  "canspeed": 2,
  "canperiod": 0,
  "nodeid": 1,
  "controlid": 63,
  "controlcheck": 1,
  "manualstart": 0,
  "fslipspnt": 0.84,
  "ampnom": 18
}

[johu]

Have you tried the "stock" firmware without the new throttle code? To be honest I didn't quite understand the new code

And of course check for moisture damage in the other connectors/plugs as well.

[mane2]

I used this custom firmware with V4 board, and it worked like it should. Didn't test stock firmware yet, need to try that next. Only thing the new throttle code controls is the potnom, and in the graph that is ok. Motor is just not spinning.

I sprayed brake cleaner to that female connectors and lots of compressed air to push it out. Hard to see inside of those female connectors though. New board side connectors were of course dry.

Comparing the parameters with the old, I'm wondering why I don't see polepairs, respolepairs or encmode in my params. Are they hidden? Reading the code, I didn't get how some params get hidden and some don't. Are they made hidden manually after flashing firmware?

Throttle code itself is just pretty much the same code as your original, just increasing the regentravel related to set speed (maxregentravelhz). And set to 0 regentravel when stopped.

[johu]

Ok, then maybe still moisture then...

Parameters can have the hidden flag. I set this on some parameters that mustn't be changed.

flag <name> !hidden

Shows it again.

[mane2]

I think the moisture was on the board connector side, not on the cable side. So they should be dry..

Is there any other value I could plot, to see why motor is not spinning even though potnom goes up?

[mane2]

Flashed the standard firmware, did not help.

I noticed that once motor gets spinning, you can adjust speed and everything seems fine. Problem is that it just won’t get going .. it’s like every 5th time or so, when it decides to start spinning and starts violently.

It’s feels like speed / position sensor issue. Could this kind of issue be caused by loose wire on encoder cable? Like second channel cable disconnected?

[P.S.Mangelsdorf]

It's difficult to trouble shoot the parameters because your udcnom is so much lower than most of the tested LDU parameter sets out there, and everything gets scaled based on that.

Nothing is jumping out to me as wrong. Your boost value is a bit high and fweaksrt is a bit low, but that should be doing the opposite of what you are seeing.

[jrbe]

It’s feels like speed / position sensor issue. Could this kind of issue be caused by loose wire on encoder cable? Like second channel cable disconnected?

It sounds like a positioning issue or an issue with a phase (power stage or winding.) If you have any suspect wiring fix that first.

[mane2]

How could I test the position sensor? I guess I can measure from the sensor end, that it’s getting ground and power. Not sure how to test the encoder channels wires. I guess web interface does not show the readings from encoder, or does it?

[Boxster EV]

How could I test the position sensor? I guess I can measure from the sensor end, that it’s getting ground and power. Not sure how to test the encoder channels wires. I guess web interface does not show the readings from encoder, or does it?

Test continuity from the wiring encoder pins to the 23 pin ampseal and that all pins are pushed home.

Should your ocurlim be 2000 or -2000?

[mane2]

Need to test that sensor today. Thanks for all the help.

Though Yesterday, I tested to disconnect the sensor, as I remember that motor tries to spin but shakes and rattles, but that gave me same result .. few times it rattled but most of time nothing happened. So I’m not very confident that it’s the speed sensor either.. but need to test that sensor still.

One additional thing that I didn’t mention before, is the ticking sound when pressing accelerator. So when motor won’t spin, I can hear this ticking coming from the motor. Any ideas on that?

[Boxster EV]

Need to test that sensor today. Thanks for all the help.

Though Yesterday, I tested to disconnect the sensor, as I remember that motor tries to spin but shakes and rattles, but that gave me same result .. few times it rattled but most of time nothing happened. So I’m not very confident that it’s the speed sensor either.. but need to test that sensor still.

Have you tried switching around the A and B signals?

[mane2]

Checked the speed/position sensor connector voltage, it shows constant 5.28V. Took it out, and gave it 5V from power source and measured the channels (tested separately both inner pins + ground), but it just showed about 5V, no change when bringing any metal closer / touching to sensor end. Then noticed that I had opposite polarity for power .. argh. Took a picture from connector side, so got confused. Well, swapped the polarity but still the same, channel cables give around 5V and no change when any metal closer to it. That's at least the way in youtube they show how to test crankshaft sensors with 3 wires.

Now, I don't know if I broke it or was it broken, or am I measuring it wrong :/

[mane2]

Got readings! I tried to tap metal against the flat end, but the magnet is actually on the side of the sensor. Other channel shows 0V when metal close to sensor, other channel shows 5V when metal close to sensor. So speed sensor is working!

[mane2]

What is the angle parameter in spot values? Should that show the motor position?

[jrbe]

Checked the speed/position sensor connector voltage, it shows constant 5.28V. Took it out, and gave it 5V from power source and measured the channels (tested separately both inner pins + ground), but it just showed about 5V, no change when bringing any metal closer / touching to sensor end. Then noticed that I had opposite polarity for power .. argh.

A hall sensor will go high or low when an Iron target is in range, it can be motionless and read properly.

There are VR sensors with electronics in them to output a square wave. These require the iron target to be moving to output a signal. I'd be surprised if this style is used here. But it's worth trying to swing something with iron in it past the sensor. Waving the shaft of a screwdriver past it should do it.

Could also be a plain VR sensor that gives a sine wave.

If you have a way to watch both channels (2 computer safe test lights, oscilloscope, etc.) you should see one channel go high, then the other as the metal (with iron) swings by.

[mane2]

Yeah, like I said earlier, other channel goes to 5V when iron comes close, and other goes 0V, and vise versa. I think it works same way as most throttle pedals with two signal wires.

I pulled the motor out once again, and tested if this happens with old V4 board. To my surprise, V4 status is just "Waiting for start" - even though start signal wire should be hot. I'm starting to think that there is moisture inside that one large connector as I'm seeing different weird issues. Connector itself seems to have all the pins pushed through and didn't see any problem there. I'm wondering how to dry that. Or if it the inverter internal wiring harness could be bought somewhere..

[mane2]

Got it working! Yes!! It was one or two loose connector legs on IGBT connector! Some flux and solder, acetone to clean flux and compressed air. Put it back and it starts to spin smoothly!

What lead me to discovery was when I started thinking that it must be one phase not working, so it randomly starts to rotate depending on the rotor position. And when it’s rotating, one missing phase is hard to notice. So I checked phases with diode mode, all were healthy 0,38. Then checked the wiring harness connectors and they seemed to be okay. No red lights on phases when powered. Then pulled out the board and zoomed in with iPhone macro on hard light. Went through each leg with multimeter to see if they are shorted, and when touching the last leg, it moved. I was like what.. this is not good. Then realized this must be a problem..

Funny thing here, that I didn’t solder these myself just for this reason. This is my 3rd board and I wanted to be sure that it’s not my soldering that causes problems. Would be great if these connectors could be done at pcb factory. Probably I’m not the only one having these problems..

Just strike of bad luck here, as the first board got water and broke, second board had chip solder issue and third one has connector solder issue. But I’m so happy that it works now!

Thanks for all the tips and suggestions, they led me to the source of the problem.