VFD Input Power; Negative Power Factor on One Phase?

[WWWJD]

Hello all,
I have something going on here that's really got me vexed. I have a 2HP VFD running on a 208/120 3Φ service, and I'm trying to measure input power with a power quality meter. One phase is reporting a negative power factor value, and I really have no idea why, unless this meter is having trouble dealing with the input current pulse train or something.

This is a regular 6-pulse VFD, nothing fancy. I've looked at the input current waveforms on a 4ch scope, they're all equal amplitude, and 120 degrees apart. Input voltages are all nominal.

Situation #2: A 1HP 480V VFD fed from a pretty well balanced 480/277 secondary is reporting a negative power factor on all three legs, and a negative total power factor of -.42. This drive includes an EMC filter. Is the EMC filter turning this drive into a capacitive load? Looking at the waveforms for this guy, two of the input phases show a large leading current pulse followed by a smaller lagging current pulse for each positive and negative half cycle. The third input phase shows a smaller leading current pulse followed by a larger lagging current pulse. The AC input current values read like this: A) 1.4, B) 0.5, C) 1.0. This has got to be throwing the math off in the meter.

The power meter I've got to play with is a HIOKI 3169-20. 3 voltage probes, 3 CTs, no neutral measurement (cause the VFD doesn't use it). I'm just wanting to get some honest to god power numbers out of this equipment, but these goofy readings are making me second guess everything I'm seeing.

Add: The metering IS set up correctly; verified with a 3Φ motor fed straight off the mains. Nothing out of the ordinary.

Thoughts?

 

[ScottyUK]

Power factor of a distorted waveform has at least two definitions. Which are you using? And more to the point, which is your meter displaying?

 

[WWWJD]

Scotty,
You're talking about PF displacement vs. distortion? There's no selection for that on this meter. I seem to remember a Fluke 434 I used years ago let you select between the two. It will let me toggle between THD fundamental and THD rms, but makes no difference in the readings, (I thought maybe this setting would factor into how it was calculating the measurements).

As an experiment, I tore the 480V drive apart and removed the EMC filter then retested. No change. I then put a 5% reactor between the drive and the metering on the input side, and noted that instead of displaying negative PF on all three legs thus a total negative power factor, it behaved more like the 208V drive, with one leg's PF reading flopping back and forth from +.42 to -.42.

This is making me feel really inadequate.

 

[waross]

I would not be concerned with what the meter indicates as much
"What what am I paying for?"
I would attempt to borrow or buy at the model of KWHr meter that the local utility is using and see if it indicates that you are about to be charged PF penalties.
The range of uneven currents may indicate supply power issues. Is the phase with the highest current showing the negative PF?

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

 

[WWWJD]

Waross,
This is a lab environment, where I'm trying to accurately measure the input power to a VFD that for lack of a better description, is on a dyno type load. I've looked into these meters a bit; the Hioki meter that I'm using is referred to as a 'demand' meter. I'm thinking that it's DSP isn't fast enough to process the switchmode current waveform on the input of the drive. I suspect that in a real world situation, where this waveform is ganged up with a bunch of other stuff on a branch circuit, this meter would be fine because the current stays above zero through the complete half cycle.

I've rented a Fluke Norma 4000 power analyzer for comparison. It should be here tomorrow; I'll update the post whenever I get some numbers out of it.

 

[LionelHutz]

What are the specifications of the current probe you're using?

Does the meter display current waveforms and current levels and phasors?

I'm thinking the current is too low for the probe you are using.

 

[ScottyUK]

If you can find one in your part of the world, Voltech's PM3000 is a pretty good analyser. I believe there's a newer PM6000 too but I have never had the chance to play with one.

The PM3000 made sense of a measurement job which totally confused a Yokogawa WT2030. As I recall we had a 3-wire star-connected VT system with an internal floating neutral point and grounded 'yellow' phase (B phase) plus regular CTs. Measurements relying upon phase displacement were all over the place. All analysers are not created equal.

 

[Marke]

The meter may well be looking at the displacement power factor rather than the true power factor, but the real problem will be that the VFD does not have an internal reactor, AC or DC and so any very small imbalance in the supply voltage or phase angle will result in a major imbalance in the input current.
The conduction angle of the current on each phase will be very narrow and this will confuse any displacement angle measurement as well.

Adding 3 - 5% reactance and operating close to rated load will cause the currents to balance and be stable, and will lengthen the conduction angle of the rectifiers and this will give you more consistent displacement readings.
Adding reactance, but still operating at very light load will not create much improvement.
The true power factor, (KW/KVA) could be quite low, but the displacement PF could be quite high.

Mark Empson
Advanced Motor Control Ltd

 

[ScottyUK]

Mark - third post down noted a change in behaviour after introducing a line reactor so you're possibly on the right track. I suspect the instrument is struggling with the current waveform.