openinverter forum

Hello all,
we have this 911 conversion with a SDU and a pack of i3 batteries.
It works quite well but the Inverter Temp is raising VERY fast when a lot of torque is required, for instance when starting from 0km/h and/or driving uphill.

In our parameter set, we have a derating at 85°C and wonder about this limit. Several users seems to have this value but no one seems to know where it came from. We've tried 90°C for a few runs and it helps getting faster before a derating cutoff.
Anyone has good or bad experience with higher values ?

I had similar issues with quick overheats. It seems to be a simple ballpark number; raising it to 100 caused me no issues in the short term (as in, made the change last week) and is helping performance. Make sure you don't have any air in the system either!

Measure the temperature of the water and compare that with the values from the system a too big delta might tell you something.

We don't have a water temp sensor but it likely follows the inverter's temp with a delay and an offset due to it's mass, but the inverter temp goes down as fast as it went up, which tells us the radiator works well.

We don't have air in the system as two features capture the remaining air : an expansion tank and a hose that is the highest part of the water network.
We can hear when air bubbles go through the water pump, and we have none.

Question is : how much temp can the inverter electronics handle and for how much time ?

The SDU is made tobe helped in a setup with his sister in the front. This devides the pain between them two. I was thinking about exactly this the other day. Some Model S D version is rated as 270kw peak that's 135kwper engine. Not 220kw what it is capable off.

I think that normally it starts degrading at 50 with a limit to 80. Electronics in general don't like heat. The fact that it cools of fast means IMHO we are pushing the thermal limits of the device.
Can't we control it differently so less heat is generated?
Lower switching frequency?

It seemed that running at 8.8 kHz was too slow for the low inductance motor. Not quite scientifically proven though

johu wrote: ↑Wed Sep 08, 2021 7:55 pm It seemed that running at 8.8 kHz was too slow for the low inductance motor. Not quite scientifically proven though

It would be nice if someone could get a scope on the analog current sensor and observe the waveform at 8.8kHz and 17.6kHz to see if there's a substantial difference. A quick search suggests that Tesla run it at 10kHz, and running it at 8.8 would likely allow us to push a lot more current, but I wouldn't be comfortable making the change without carefully observing the current waveforms on the bench, which I'm not in a position to do

Edit: just realised this information comes from Damien's readme, so I guess he discovered somehow that 8.8kHz is problematic. Would be interested to know what the problem was.

Back then a racing team had blown like 5 SDU inverters and nobody really knew what the cause was. So Damien experimented with the 2 frequencies and found nasty transient currents at 8.8 kHz. I don't remember the details, maybe he took some notes.

At the same time he also tuned the other usual parameters for them and we introduced iaclim back then. Which measure contributed how much to success is hard to say.

EV_Builder wrote: ↑Wed Sep 08, 2021 9:55 am The SDU is made tobe helped in a setup with his sister in the front. This devides the pain between them two. I was thinking about exactly this the other day. Some Model S D version is rated as 270kw peak that's 135kwper engine. Not 220kw what it is capable off.

The specs I could find show the peak power as 193kW per motor for the 60D model. I don't know where the often quoted 220kW figure comes from, but you're right that Tesla don't always run it at that much power. From what I've seen, with the right settings, the SDU seems comfortable up to around 500A (180kW) for short periods using openinverter. The Model S has 2 motors, but it's also twice as heavy as many of our DIY vehicles

EV_Builder wrote: ↑Wed Sep 08, 2021 9:55 am Can't we control it differently so less heat is generated?
Lower switching frequency?

I just wanted to mention on this thread that in my tests, the inverter seemed to produce dramatically less heat at lower slip values. When I reduce fsipmax from 3.5 to 3.0, the peak temperature reduces from 82C to 64C (for the same current flow). So if temperature is the limiting factor, definitely try reducing slip (fslipmax).

I would strongly recommend that anyone using the sdu or considering working with it review these videos. The lessons learned were painful and expensive caused primarily because I made arrogant assumptions thinking I knew everything. I didn't and don't so use my knowledge to build your own.