Loud compressor

[rih5342]

Hello Motor Experts,

I'm new to motors.

I'm trying to make an air conditioning scroll compressor QUIET.

The motor has a capacitor wired in.

A signal analyser spectra of the current has really strong harmonic content every 60 Hz, as does an accelerometer mounted on the compressor, and as does a microphone mounted near by.

The "nasty" current (spectra looks like a comb!) is causing the noise.

My question: is there a simple way to get rid of the harmonic content of the current?

I thank you in advance.

Rob

 

[MikeHalloran]

Scroll compressors are alleged to be relatively quiet, and capable of near-perfect balance.
Find out what's binding or clashing or out of balance in the mechanical assembly, and fix it.


Mike Halloran
Pembroke Pines, FL, USA

 

[MiketheEngineer]

Mike is right!! Supposed to be the "quietest" around!! Something is wrong

 

[rih5342]

Measurements were made in an anechoic environment with a 19 dBa ambient, using a microphone array, carefully tested before every use.

There is high correlation between motor current and the acoustic spectra.

This is a new unit, working just fine.

Looking at the spectra, you realise there is an opportunity to make it quieter.

If the unit had a mechanical problem, there would be maybe a once, twice or thrice shaft rate noise, but not at every single multple of 60 Hz way beyond 3000 Hz.

I've read on this forum, that there could be a "startup" capacitor which is turned off via a centrifugal switch.

I'm in the process of verifying that the centrifugal switch is functional.

I'm convinced the current spectra is causing the acoustic spectra, so one solution is to eliminate the current harmonics.

How do you do that?

Thanks,

Rob

 

[MikeHalloran]

19 dbA is typical of what? A church in the middle of nowhere on a country road on a weeknight in winter?

Getting the output below ordinary office noise or home noise or night hospital noise might be an interesting exercise, but whoever is paying for it may not be fully appreciative.



Mike Halloran
Pembroke Pines, FL, USA

 

[waross]

Hi Mike, I understood that the 19 dBA was the ambient noise and the compressor was much louder.
Have you tried more than one identical unit?
Is this driven by a VFD?
How big is the motor?
Single phase or three phase?
If this is a single phase motor, it may be failing.
A spectra looking like a comb sounds like a VFD. Not compatible with a capacitor. If this is a VFD application it is a wonder that the capacitor hasn't damaged the VFD.
Please give us the specs on the motor.
Thanks.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[electricpete]

Good questions above.

Can you provide more detail about current spectrum and acoustic spectrum. Ideally a screen dump. Or at least tell us:

What is the current THD? Or what do these harmonics look like as fraction of fundamental?
Which harmonics of line frequency are present (odd only?).
What frequency range are we talking about?
Are you sure you are not seeing 2*LF sidebands around RBPF?
Are there specific harmonics in the acoustic that go far above the others (suggesting resonance... which is always likely when you have rich harmonics)

* Does the noise go away when load is decreased (by uncoupling motor or somehow reducing compressor load) ? If so, this is typical of slot-related noise.

* Time waveform view (vs spectral view) may also provide clues as to source of current harmonics.

=====================================
(2B)+(2B)' ?

 

[rih5342]

Answering several of the above posts all together.

The air conditioner makes noise at about 60 dBa, with the background noise being about 19 dBa.

This is the only unit I have, I might get another unit soon, either bigger or smaller.

My understanding is that the motor is single phase, with a start capacitor. If you disconnect the capacitor while it's running, it will still run.

The unit is not using a variable frequency drive.

At this time, I have no way to unload the motor.

The motor is integrated into the compressor, completely inaccessable, will try to find information.

My frequency range of interest is 50 to 2000 Hz.

I've uploaded a jpg with the acoustic spectra, current spectra,
and the current time trace.

Thanks for your input.

 

[waross]

Is there a shipping brace still installed?

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[electricpete]

That is good data. And a head-scratcher as well.

The audio spectrum looks quite a bit different than the current spectrum. Current has odd harmonics, with 5x highest among them. Audio has both odd and even. Prominent harmonics at even harmonics like 10*LF, 12*LF, 14*LF (600hz, 720hz, 840hz) in audio spectrum have no corresponding peaks in current spectrum.

I don't have any ideas, but some more thoughts on things to check:

Is it possible these current harmonics arise from power system rather than motor. Perhaps compare voltage THD and current THD. Or better yet check voltage THD with motor off as Bill suggested before in another thread.

Compare performance as voltage changes. If you don't have ability to manipulate voltage, check during daily highs and lows of supplied voltage. If current harmonics decrease at lower voltage, could be a sign of saturation.

Now another less productive shot in the dark. Consider running motor without starting cap to see if that has any effect. Perhaps you can disconnect starting cap while deenergized, then use mechanical means to get motor spinning IF you can do it safely (perhaps spin and then apply power, or apply power than spin). I am not sure why cap would have an effect... if can easily be done would be interesting but perhaps this one is a wild goose chase.


=====================================
(2B)+(2B)' ?

 

[itsmoked]

If you have the recording equipment you could also disconnect the running compressor and see if the noise goes away instantly or not. Even 100ms would allow you to exclude the motor as the source.

If this is a fully hermetic, perhaps the motor is banging on the inside of the pot? A chance bad luck thing.

Is this a new compressor, or a used, or re-manufactured one?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[rhatcher]

Theoretically the even harmonics cancel out in a purely sinusoidal waveform. That is why you do not normally see even-order harmonics being an issue on normal power delivery problems. It is the odd-order harmonics that are always mentioned.

This brings my point to the front. When I look at rih5342's waveforms, I do see small even-order harmonic content from the 4th order on out along with a lot of odd order content well beyond the normally expected range. This strikes me as unusual.

In addition, when looking at his current waveforms, I am struck by the idea that this is something that I have seen before.

My thinking is that the anomolies in the current and frequency spectrum waveforms are not a product of the air conditioning compressor or the motor from either a mechanical or electrical standpoint.

I am thinking that they are coming from the power supply or from some other load that is attached to the power supply.

If rih5342 can post some spectrums showing current versus voltage then this may be helpful. More information about the power supply would also be helpful.

Perhaps I am reading too much between the lines but it almost sounds like these are readings from a sound test chamber by someone with specialized equipment to measure sound, vibration, electrical frequencies, etc.

If this is the case, then it may be also be the case that the power supply is electronically regulated via an inverter or some other means.

It also may be that the source power supply is purely sinusiodal but there are other loads on the system that have fully controlled rectifiers or other similar devices on the front end that are causing such high harmonic content.

However, my guess is that the harmonic source is some sort of regulated and/or inverted power supply. See the attachments.

 

[rhatcher]

and the other attachment...

 

[rih5342]

WOW!, good job rhatcher!

I would never have suspected that the power coming out of my wall had "noise".

I wired up 2 500W construction halogen bulbs, 2 voltage splitters
(to accomodate the data-acq voltage range) and 2 current probes, and repeated the measurements.

For a pure ohmic load, see attachment.

Plot 1 = V1 and V2, the 2 phases
Plot 2 = (V1 minus V2)and Current
Plot 3 = FFT of (V1 minus V2)

Next step try 2 isolating transformers.

 

[rhatcher]

rih5342, I appreciate the compliment but I am not that good. I would be more satisfied if your second waveform looked more like the first (the notches).

The test that you performed does prove that there are harmonics/power supply waveform distortions that are not related to the load.

But now, the questions that remain are:

Your new waveforms do appear to be non-sinusiodal. What is the power source that you are measuring? Is there any electronic voltage regulation in the source or is it a pure rotating (local or utility) generator supply?

Also, there was a distinct 'notching' in the first current waveform that you presented that is not present in the second. I would guess that this is because the load of the lamps that you tested is much less than the motor and that the increased load of the motor when compare to the output impedance of the source makes the harmonic distortion, ie. notches, more apparent.

Or, there may be something else entirely going on. Again, I will say that I would be more satisfied if the new waveform looked more like the first.

Can you show the same type of voltage and current readings for the a/c compressor? I am very curious to know if you can create that 'notched' current waveform from that smooth (distorted and harmonic rich but no distinct notches) voltage source.

Also, if the lamp test was at lower current than the motor test, can you provide a lamps test (pure resistive load) at the same amperage value as the motor test to see if the waveform becomes notched under increased load?