Capacitors on transformer secondary

[dancase]

We've got a dual-conversion split-phase(120/208 transformerless) Powerware 9170+ 18 kVA UPS that is feeding an isolation transformer (208 to 120/240)configured as an SDS. The transformer was added in an attempt to eliminate common-mode noise from the UPS (may not be an issue in a server farm, but a VERY BIG issue in our application, with includes 15-20 radio broadcast and production studios). While the common-mode noise is dealt with quite nicely in this arrangement, there's still objectionable high-frequency noise showing up on the secondary of the transformer, noise that is quite evident in our low-level, high-gain audio circuitry.

Ideally, PowerWare should clean up their inverters and filter them better... but of course, we have to get them past the first step (admitting they have a problem) before we can get them to work the program . Based on the discussions to date, I don't see that happening in the near future.

I know the best way to filter this noise is with a good EMI filter between the UPS and transformer. I also know that I've run out of budget, so I'm experimenting with a few alternatives. I've found that a couple of 30-60 mfd capacitors from the secondary legs to ground do a very credible job of shunting the objectionable noise to ground and reducing it to an acceptable level. I've thought this through and can't think of any problems that would be created by these caps, but then again I don't deal with capacitors on power lines all that frequently, so I may be forgetting something.

Any reason why these two capacitors on the secondary could create a problem?

Thanks!

D.

 

[rattus]

Re: Capacitors on transformer secondary

30 mfd sounds a bit large to me. They must be non polar or they would have exploded, right?

Is this noise in the audio range?

You might try ferrite cores slipped over each conductor. This adds a bit of inductance in series. If Radio Shack doesn't have the cores, you can find some in a defunct light dimmer. Or, you might try an iron or steel nut to see if that helps. Such an inductor upstream from the caps ought to help.

Be forewarned though that the current level may be high enough to saturate the cores in which case they become useless.

[ April 29, 2005, 10:40 AM: Message edited by: rattus ]

 

[dancase]

Re: Capacitors on transformer secondary

Originally posted by rattus:
30 mfd sounds a bit large to me. They must be non polar or they would have exploded, right?

Is this noise in the audio range?

You might try ferrite cores slipped over each conductor. This adds a bit of inductance in series. If Radio Shack doesn't have the cores, you can find some in a defunct light dimmer. Or, you might try an iron or steel nut to see if that helps. An inductor upstream from the caps ought to help.

The caps I'm using are oil-filled caps designed for AC applications (I'm smart enough to not try putting an electrolytic across a power line!)I actually started with much smaller capacitor values, and when they had no effect I stepped out back and yanked a couple of 30 mfd @ 660VAC caps from a defunct Ferrups.

I haven't determined the exact frequency range of the noise -- my "good" oscilloscope is out at one of my sites and I haven't gotten there yet. The 'scope I'm using still lets me see it, but isn't really well calibrated. By speeding up the trace and increasing the gain, you can magnify the slope of the sine wave and see the noise quite clearly. It's not in the RF range... if it were, smaller caps would have more effect than they do. There's a noticable difference between 15 mfd, 30 mfd, and 60 mfd (more cap = less noise), which tells me it's probably in the same range as the audible noise made by the inverters. Of course, it's hard to listen to the noise when it's riding on a 60 hz sine wave! I'll be looking at it with the "good" 'scope today and make a better determination.

One way or the other I don't think it's high enough in frequency to benefit from a ferrite -- a 60 hz transformer would be pretty amazing if it could pass RF unattenuated!

D.

 

[physis]

Re: Capacitors on transformer secondary

Essentially you're building your own filter.

Add an inductor in series to your capacitor in shunt and the response curve will increase from 6 db per octave to 12 db. The cut off frequency has to be calculated though.

I tried to find a good web site for you (and there are a few good ones) but it's getting harder all the time to find things. I found a lot of coffee filter sites.

Try AC filters, or second order response and stuff like that.

Edit: If you're primarily suffering from high frequencies your filter circuit wont need to be as massive as it would if you were after lower frequencies.

Edit: And I can't spell.

[ April 29, 2005, 02:14 PM: Message edited by: physis ]

 

[steve66]

Re: Capacitors on transformer secondary

Physis:

Stop worrying about your spelling. Your making me self conscious (I had to correct two words already. What would I do without spell check?)

I would have thought your high end audio equipment would have eliminated that noise in the DC power supply. After all, it eliminates the 60 Hz noise (at least it should!! Maybe some of the 60 Hz is getting rolled off by the broadcast amp??) So why won't it eliminate the higher freq?

Is your transfromer sheilded?

The only problem I can think of (and I'm not sure it is a problem) is what is the resonant frequency of the transformer and capacitor, and what would happen if this resonant frequency were to somehow appear in the circuit?

Steve

 

[physis]

Re: Capacitors on transformer secondary

Steve, you have to look at low pass, band pass and high pass filtering.

It's pretty common to have some level of low frequency descrimination for 60 Hz. Having an issue with higher frequencies will be somewhat outside of the typical concern.

Ever run a guitar amp at max gain from a generator? Blueckh.

I know, I'm sick of writing "I can't spell" too.

Maybe I'll replace it with "error A", it would be the same though.

Edit: Error A

[ April 29, 2005, 03:39 PM: Message edited by: physis ]

 

[physis]

Re: Capacitors on transformer secondary

And inverters and stuff? I'm sure they're not making sine waves.

 

[dancase]

Re: Capacitors on transformer secondary

Actually, this one makes pretty good sine waves. It oughta for the price. Of course, it also oughta make CLEAN sine waves at this price...

D.

 

[physis]

Re: Capacitors on transformer secondary

I'm sure the load is large enough to make your filter pretty "honkin'" But if you have a sufficient understanding of filter design you can save some "honkin'" dollars too.

 

[dancase]

Re: Capacitors on transformer secondary

Oops. I thought it looked different today -- and as it turns out it did. Somebody turned on the breaker for the fluorescent lights -- there's one fixture in each studio on the UPS panels, going back to the days before we had a generator and were running longer-term on battery. These days, we only need anough battery to get the genset running and switched over.

Of course, those fluorescent fixtures add their own little collection of noise. With those off, I can see more clearly what the UPS is generating. The noise from the UPS is in the 10-20 kHz neighborhood, and definitely responds well to the 30 mfd caps. Actually, so does the noise from the fluorescents.

D.

 

[physis]

Re: Capacitors on transformer secondary

Scratch that. I thought you had 10k and 20k, an octave apart.

Edit: Error B: Didn't pay enough attention to the post I responded to.

[ April 29, 2005, 06:51 PM: Message edited by: physis ]

 

[dancase]

Re: Capacitors on transformer secondary

Originally posted by physis:
I'm sure the load is large enough to make your filter pretty "honkin'" But if you have a sufficient understanding of filter design you can save some "honkin'" dollars too.

Oh, yeah. The transformer is rated at 25 kVA, but with the UPS only able to do 18 kVA the max load is around 80-90 amps and it's fused at 125 amps on the primary.

I have an LEA TVSS/Noise Filter that's got really interesting air-core inductors... several turns of 1/0 copper with a diameter around 3 or 4". When I first saw it, I said to myself "Hmpfh. I can wind coils like that myself. Why pay for them?" Of course, the answer was that I had more budget than time for that project. But' I still coulda built it.

The real unknown to me at this point is how it will all behave under 75% of full load. Could be a whole different game then. Might even be quieter under heavier load. Or, might be worse. Next week I'm gonna properly install a couple of caps, and transition all then loads back to Tech Power (the UPS) and see what happens. If it's not a happy result then I'll get down to some serious filter design.

D.

 

[physis]

Re: Capacitors on transformer secondary

I can't be of too much help to you because I don't have the library I used to have. But with most designs the load is going to shift the response. You'll have to take that into consideration.

Inductors aren't so fancy. They are just coils. But I don't trust sloppy ones either.

Edit: Error C: left out a word.

[ April 29, 2005, 07:15 PM: Message edited by: physis ]

 

[physis]

Re: Capacitors on transformer secondary

even the best EMI filters I've seen won't cut it here

I would wager "the best" or even "the median" would indicate exactly what they do and that is what they would do.


"The Junk" would likely do niether.

 

[physis]

Re: Capacitors on transformer secondary

I've thought this through and can't think of any problems that would be created by these caps, but then again I don't deal with capacitors on power lines all that frequently, so I may be forgetting something.

I'm not familiar with big power stuff myself. I know ideal components don't blow up in ideal circuits that often.

There is one concern I would have with your capacitors. I don't know how durable they are but at 660v. it might not take too much of a noise spike to poke a hole in their dielectrics and damage them. They could go kaboom.

Maybe someone who knows more about bigger power things could offer an opinion about their voltage rating.

 

[dancase]

Re: Capacitors on transformer secondary

Originally posted by physis:

There is one concern I would have with your capacitors. I don't know how durable they are but at 660v. it might not take too much of a noise spike to poke a hole in their dielectrics and damage them. They could go kaboom.

Well, I understand your concern. Actually, I'm not going to use the 660v caps -- they're too big physically. Will probable use 370VAC or 440 VAC, as they're a little more compact.

Consider these points (some of which may not have been mentioned yet). There is VERY, VERY GOOD series TVSS just before the UPS. There's also a manual (MBB) bypass. In the worst of conditions, with the UPS bypassed, there's an excellent series type TVSS feeding the transformer primary from commercial power. The worst load (in terms of spike generation) on the secondary side is a few fluorescent lights (and I may end up taking them off the UPS before we're all through). Under normal operating conditions, the UPS is in double-conversion mode and is regenerating power, which should be transient free (that's the whole idea of a double-conversion UPS).

Now, consider that the capacitors are located at the point where the center tap of the transformer is bonded to ground to create the neutral of a "separately derived system." One capacitor is connected from each leg of the secondary to ground, so there is 120 VAC across each capacitor. With the traditional "good engineering practice" voltage derating of 50%, capacitors rated at 180VAC are required, and 250VAC caps are more than sufficient. I'll use 377 or 440 volt caps just because I'm like that. I'll also use caps rated for higher temperatures, since the only place I can put them is inside the transformer cabinet (being careful not to block airflow).

I don't think they'll be at risk for spike-induced boom-booms. Even if they short, however, it won't be a "boom-boom" situation because the caps are of the sealed "wet film" (oil filled) variety with internal protection. In the event that the internal pressure becomes excessive, an internal disconnect occurs to prevent "boom-boom." It's never good when something electrical explodes.

Thanks for your insights, all. I think this is gonna work.

D.