Frequency Fluctuations

Status
Not open for further replies.
B

brian john

Senior Member
Location
Kilmarnock, Va
Occupation
Retired after 52 years in the trade.
Ran into an unusual problem, I have a 100 KVA Liebert DataWave feeding 3 phase power supplies to some blade servers, AT 38% load no real issues (noisier that I would expect but sounds like 60 HZ. Above 38% load and the unit sounds like an air compressor. Performed a resistive load test on the unit and no issues at any load percentages. Run the blades above 38% with load bank on line and no issues shut off load bank and the compressor sound starts up.

38 % Load

VNONDISTORTED.jpg


Above 38% load bank is not on line at this point.

VDISTORTION.jpg


Frequency

HZ2.jpg


Voltage THD Harmonics were primarily 5, 7, 11, 13 assuming 6 pulse rectifier?

VTHD.jpg
 
Interesting problem! So with the resistive load test as the only load on the UPS, no issues at any load percentage. Resistive load and normal load both on at the same time, no issues at 38%. What about above 38% with both loads turned on? Normal load only at 38% loaded results in the UPS roaring.

The harmonic sprectra definately looks like a 6 pulse rectifier (highest is the 5th and the 7th, next highest is 11th and 13th) and the individual voltage harmonic magnitudes are astronomical. IEEE 519 recommends individual voltage THD limits at 3% at that voltage and 5% voltage THD.

I would be interested in hearing what Liebert has to say about harmonic limits on their UPS. I would suspect the distorted waveform is causing some issues with the power conversion. Some harmonic mitigation at the offending loads might be in order.
 
only load on the UPS,
Click to expand...

This is not a UPS, this is a power conditioner, ferroresonant transformers and input output caps. The unit is designed to minimize impact of utility issues on loads. I think it is a cap issue, possibly the 6 pulse signature may be the problem as I think this equipment was designed with single phase loads in mind with 3rd harmonics. Or maybe the caps are going south, we did follow Lieberts instructions for testing the cap banks and the caps meet their standards and they look OK.

0-38% no issues
>38% compressor like noises
>38% plus resistive load no issues, we did not vary the load as once we got the noise we shut down the critical loads.
 
Were you measuring on the input or output of the UPS?
If on the input, is there a harmonic filter or input transformer?
If on output, what does the V and I look like when on bypass?
 
ron said:
Were you measuring on the input or output of the UPS?
If on the input, is there a harmonic filter or input transformer?
If on output, what does the V and I look like when on bypass?
Click to expand...


It is a power conditioner not a UPS

We measured on the input and output simultaneously 2 meters connected to capture similar data, the information provided is output.

There is no bypass. 480-3 wire input 208/120 4-wire distribution.
 
Are the blade server power supplies wye or delta connected? How about the load bank? Same results with various servers? (Bad PS in one of the servers?)

Something is trying to excape my brain about this, but I've been rewriting a coworker's document all day, and it was -really- bad.
 
zbang said:
Are the blade server power supplies Wye or delta connected? How about the load bank? Same results with various servers? (Bad PS in one of the servers?)
Click to expand...

Not sure on the servers trying to get information on them, the problem occurs with different servers. The load banks are delta connected.

Something is trying to escape my brain about this, but I've been rewriting a co worker document all day, and it was -really- bad.
Click to expand...

I understand I have; to rewrite my reports all the time, write, read, correct, write, read, correct. Writing was never my strong point as all can see.
 
I wonder if the filter is tuned to perform at its best with certain criteria, such as load, slew rate of the load, etc. Just a guess.
Is the input to the conditioner distorted too?
Is it possible to connect some type of bypass and see the results? It would seem weird that a critical load would have a gizmo such as a conditioner that couldn't be bypassed to perform service such as replacement of the caps.
 
ron said:
I wonder if the filter is tuned to perform at its best with certain criteria, such as load, slew rate of the load, etc.
Click to expand...

I think this is the situation the caps banks and associated components possible were designed for single phase loads.

Is the input to the conditioner distorted too?
Click to expand...

I did check this but need to re-look Monday.

Is it possible to connect some type of bypass and see the results? It would seem weird that a critical load would have a gizmo such as a conditioner that couldn't be bypassed to perform service such as replacement of the caps.
Click to expand...

This is what I want to do but temporary bypass will be expensive (I have rental transformers and cables) but I am positive there will be a phase shift between the Liebert and my transformer.
 
Brian,
I'm not familiar with the Datawave product, but some manufacturers produce a line interactive or double conversion static UPS without batteries, and call them line conditioners.
So you may be dealing with a UPS inadvertently.
 
This sounds a lot like the old style constant voltage transformers with the ferroreasonant transformer and lots of caps, and my memory tells me (but I cant produce any data to back up my memory) that you couldn't use ferroreasonant CVTs on non-linear loads as they went bananas, which sounds a lot like what you are experiencing.

The problem with this is that your machine comes from Liebert who one assumes know a thing a so about large non-linear loads, being as they build data centre UPSs...
 
I am fairly sure the unit is built to handle single phase non-linear loads as there are a fair amount of DataWaves in datacenters.
 
Status
Not open for further replies.
Top