Inverter for single phase 230vac induction motor (A/C)

[jyanof]

For the most part, everything has been working pretty well. The hottest it's been so far is about 105F (which means its 125F in the cabin when you first get in) and the a/c has performed comparably with a 'normal' vehicle's a/c.

Typically, the vents are blowing 50F air within a few minutes and down to 40F and lower once the inside of the vehicle cools down.

I put a thermistor on the motor field windings (as best I could), another on the inverter heatsink, and another just in the air near the heatsink inlet fan in an attempt to measure the ambient air temp near the inverter.

So far, it appears that the ambient air temp rises about 5-10C when the a/c is running (it's pumping a bunch of heat into the engine bay). This is especially true when the vehicle is still, but motion will get some airflow going and purge the engine bay.

The inverter heatsink, as measured by this thermistor at least, will measure just a few degrees C above the ambient temp measured.

The motor temp will level off about 10C above the ambient temp when the vehicle is in constant motion (like on the freeway) and there's extra airflow. When the vehicle is stationary, the motor temp continues to rise. I'm never stationary too long, but the worst it's been has been in severe rush hour traffic in 38C temps (like 100F) and the motor temp I was measuring was about 62C. The measured temp will also rise a few degrees when the a/c is shutdown from soak-back. The windings are supposedly rated to 130C, so I think there's quite a bit of margin - at the current conditions, at least.

It'll be interesting to see how things change when temps are higher in a few weeks. Not only will the ambient temperature be greater (approaching 50C) but the compressor, motor, and inverter will all have to work harder.

[jyanof]

There's been a couple issues so far, one still unresolved.

The belt was a little loose at first and would squeal when the clutch closed, then squeal hard again about 2 seconds later - not sure why, maybe the refrigerant takes a while to really get flowing? Anyway, tightening the belt solved that right away.

This other issue is a bigger deal. After the a/c runs for a while, if it is shut down, the motor won't restart due to the code sensing an internal fault. Well, it'll begin to restart, but once it gets up to full speed and the clutch is about to close (usually it'll close for a split second, but sometimes not), it'll shut down due to a 'dead voltage sensor' fault.

I actually had the diagnostic code incorrect in the software. At first, I thought it was an over-current condition, but correcting the code so that the LED blink count correctly indicates the fault yielded the true culprit. I'll upload that code at some point.

The sensor outputs a voltage in the range of 2.5V to 4.8V or so. If the ADC reading indicates that the voltage is less than about 2V, I'm interpreting that as a failed voltage sensor and the software shuts the motor down. But why is this happening? Not sure. I did scope the sensor output and sure enough, the signal goes dead occasionally as the motor nears full speed.

Also of interest, waiting a while will allow the motor and a/c to start successfully, so I think it's heat related. Maybe something to do with the motor saturating at higher temps while unloaded and the high currents causing some problems. Maybe there's a ground loop problem too.

It'd be nice to figure out the root cause of this problem, but my first plan of attack is to try a different voltage sensor. Avago now makes an isolated linear opamp specifically for voltage measurement. Feed it 0-2V on one side and you get 0-2v on the other (kinda, needs another opamp for scaling and referencing). I'm planning to set this up on it's own little pcb and tap into the power/control boards appropriately.

I'm hoping that either the new chip or new layout configuration will solve the problem. It'd be nice to be able to cycle the a/c once the car cools down. As of now, I have to wait too long before it'll start up again.