I have spent over a week on this problem, pulling my hair out and finally narrowed it down to the relay itself....

ISSUE: trying to power a small air conditioner unit from inverter, but when using the factory relay that kicks on the compressor, it trips the inverter protection! IF I BYPASS THE RELAY AND HARD WIRE THE COMPRESSOR, THE INVERTER HAS NO PROBLEM RUNNING IT. (specific relay datasheet below)


Main Question: is there a simple solution to mitigate the spike of the relays contacts closing? Or should I remove the relay and try to use a mosfet instead?

(from searching online I found a couple possible solutions. Using either a Metal-oxide Varistor (Vos) or a TVS Diode. Your thoughts, would these work?)

(I came across this post here on the forum from a while back talking about reducing relay voltage spike on the coil. Not sure if applicable to my situation...) https://forum.allaboutcircuits.com/threads/reduce-voltage-spike-on-relay-coil.184312/

More Info: the air conditioner main board has a time delay to engage the compressor. Originally every time the compressor kicked on it was tripping my inverter. So I went down a week's long rabbit hole trying to implement a couple different soft starter methods. I've tried using NTC inrush current limiters, as well as making a resistive inrush current limiter with a on time delay circuit and I had some varying success...

So I then decided to just completely bypass control board and to my surprise with no inrush current limiting whatsoever, the inverter has no problem powering the air conditioner with the fans and compressor kicking on at the same time.

So I now narrowed it down to the relay for the compressor is what is tripping the inverter! It would be ideal to be able to still use the factory main board in the air conditioner since it provides all the float switch sensors and the delay for the compressor etc.. If I can't figure out the relay problem I'm just going to hardwire the whole damn thing but I'd rather not

Any ideas on how to avoid this? Is there a way to smooth the spike of the relay contacts or should I maybe try to remove the relay and install a mosfet instead?

 

So I now narrowed it down to the relay for the compressor is what is tripping the inverter!

If that's true, the inverter should trip out even when the fan and compressor motors are not connected. Have you tested that to see if it's true?

The compressor and fan motors likely both have start capacitors. The coil, though much smaller current, is a purely inductive load and that's apparently not compatible with your inverter.

 

With the controls bypassed, the inverter starts under load, which that start is not "instant", although it is fast. With the relay closing the circuit multiple times as the contact bounces, it is quite different. Probably using a higher power capability inverter would be one solution, but not the one you seek.
Using a "zero voltage crossing" switch would be one choice that could work, and that switch could be bypassed by the relay once things were started and running.

 

I don't think the relay (or any possible spikes from that), per se, is the problem, it's the sudden application of the low impedance motor load, and changing the relay to a different type won't solve that.

The problem may go away when the inverter is directly connected since, (I presume), you are applying power to the inverter to start the AC unit, which likely allows the inverter to sort of slow-start.
If so, then you could directly connect the converter and connect a relay to start the inverter (unless you use the inverter for other AC power).

I've tried using NTC inrush current limiters, as well as making a resistive inrush current limiter with a on time delay circuit and I had some varying success...

If you need the inverter for other power, then I would suggest continuing to work on such a start-up current limiter.
What was the "varying success..." you had?

 

Using a "zero voltage crossing" switch would be one choice that could work,

Interestingly, for an inductive load (and possibly a motor), starting at the peak of the sinewave generates a smaller surge current.
Starting at zero volts gives a full half-cycle of voltage for the inductive current to build up before the voltage polarity reverses.
Starting at the peak only gives a quarter cycle, and since the steady-state AC inductor current is zero at the voltage peak, it will immediately settle into its normal AC plus and minus current value.

Thus starting a transformer at the zero crossing can cause momentary core saturation with a large input surge current, since the theoretical initial magnetizing current could grow to twice the normal peak.

 

I am wondering if contact bounce is part of the problem. Unfortunately that is neither simple to prevent nor simple to measure.

 

I am wondering if contact bounce is part of the problem. Unfortunately that is neither simple to prevent nor simple to measure.

Possibly, but it would seem that would tend to reduce the surge current, not make it worse.
Of course a solid-state AC relay would eliminate any bounce.

 

I didn't know they were available, but a search revealed they do make peak-switching SSRs (example) which are designed for powering transformers or other inductive loads.
That could significantly reduce the first-cycle peak starting current of the AC unit, (assuming the peak motor starting current is mostly determined by the motor inductance), so that could be another option to try.

 

What has not been mentioned is the connection between the inverter and the AC device. If the trial has always tripped with a close connection, try it with a suitable 100 foot extension cord, coiled, not unwound. This suggestion is based on a SWAG, but easy enough to try. The long cord certainly affected the starting of an induction motor driven table saw many years ago.