What's new
What's new

PSA: Phase Perfect lose capacitor connection, system burn out, WARNING

I've got an American Rotary converter on my Brother. I've been a bit jealous of those with a PP because mine makes an annoying buzzing sound. Its louder than the brother at 10k (non cutting) I was under the impression the PP were quiet.

They are not real loud. But, the noise they do make is annoying. Seriously sounds just like an air-leak.
Also, base anything I say off the dual unit (PT3160). Which is basically two of the biggest single chassis units (PT380).
Including cooling the fans. And, it is when the fans are running that mine is obnoxious.
Its not bad for a few minutes. But, 12hrs in the same room with it, day in, day out? Yea, that gets real old!

My buddy Phil is running his home shop on a 380. It isn't nearly as loud as mine.
 
The sound of the white units is nothing compared to the pitch the old blue units had.
That said my 10hp blue unit has well over 30k/hrs on it and still runs everyday.
Took a good look in both yesterday and everything still look like day 1. Last week I was running a 1 min cycle on the cnc lathe that takes all the power the 355 can give, 6 tool changes and a few ramp up/down in that 1 min and it did it all day long no problem. So the failures like OP had aren't supposed to happen, but loose connections will wreck anything and cause real bad voltage spikes.

Sure would be nice to have extra protection/capacitor failure detection(and easier access to check/replace them) and more line feedback filtering in there though, for that $$$$ surely something is doable these days.
 
We just had a fire in the electrical cabinet in one of our CNC mills a cpl months ago doo to the same thing - a loose warr - at least near as we can tell.

I will post pics in my shop thread soon.

This is not a P/P issue, but any electrical connection anywhere, and maybe even especially known when using alum cable that seems to loosen up more after a number of heat cycles.


--------------------

Think Snow Eh!
Ox
 
I skewl chum built a new building 30 yrs ago and [as the story goes] the electrical inspector didn't like the fact that he used alum wire (at least of that size) as his incoming line, and needed to tear it out and replace.

So he git's to thinkin'.....

He gets a can of that copper spray and paints the end of the wire and shoves it back in the lug and calls the inspector back...

Get's a clean bill of health and onward they go! :cloud9:


Now, you hafta understand that only 50% of what comes outta this guys mouth is true, so how much of that story is true or not IDK, but I remember when it was current, so ... ???? But I don't recall ever looking in the box to see if I could tell the diff or not at the time. ???

While at the same time, this is VERY MUCH in his wheelhouse, and over the years we have come to realize that generally the more unbelievable the story, the higher the truth likelyhood.


-----------------------

Think Snow Eh!
Ox
 
Last edited:
My PT-330 makes a noise like frying bacon or light welding. I looked inside to try to see where the loose connections similar to the OP’s might be. I could not. They are buried.
My PT-355 (20hp) unit, which is about three years old, makes no noise at all....except the cooling fans of course...but when fans are off I hear absolutely no sound from the box... maybe if I put my ear right on the sheet metal... (and yes my hearing is normal... )
 
Thanks for sharing your experience. I will be very interested to hear what any response you get from PP, one was in the plan for this year.

I would second the question, how could you better isolate the Brother? I would have thought something would have tripped?
 
Thanks for sharing your experience. I will be very interested to hear what any response you get from PP, one was in the plan for this year.

I would second the question, how could you better isolate the Brother? I would have thought something would have tripped?

Yeah, I don't know what could have been done. There is a 30A breaker before Speedio and it did not trip. I don't know that putting surge protector would have changed anything since they trip high too, lowest perhaps at 600V. If anyone has suggestions I'd love to hear it.
 
Given the issues with the PP feeding back noise, it seems that there would be a good case for putting a filter on the input to the PP.

Since I do not have a PP, I have no good data on the switching frequency they use. .That would make a huge difference in the proper filter to block the dangerous amounts of switching noise, as well as nuisance effects that have been mentioned,

Usually, the higher power the switching device, the lower the switching frequency is. But as parts have gotten better, the usable switching frequencies have gone up, and where in previous generation equipment they might have been as low as 3 to 5 kHz, they may be above 15 kHz at good efficiency now.

The filter needs to filter well at the basic switching frequency, and above, but have little effect at 60 Hz. .A filter that does that would drastically cut the risks associated with having a lot of fairly high frequency "hash" in the power line, and would tend to cut the radio interference as well.

Many motion detectors, and some battery chargers, some washing machine controls, and other items, have capacitive impedance used to drop the 120V line to create a low voltage non-isolated power supply. Any substantial power at a significantly higher frequency than 60 Hz will "see" the capacitive voltage dropper as pretty much a short circuit. If it is not cut off by a fuse, parts inside the unit may be destroyed. When that type of item is damaged, it pretty much nails the problem as "hash" on the line, unless it is known that there was a large over-voltage..
 
Given the issues with the PP feeding back noise, it seems that there would be a good case for putting a filter on the input to the PP.

Since I do not have a PP, I have no good data on the switching frequency they use. .That would make a huge difference in the proper filter to block the dangerous amounts of switching noise, as well as nuisance effects that have been mentioned,

Usually, the higher power the switching device, the lower the switching frequency is. But as parts have gotten better, the usable switching frequencies have gone up, and where in previous generation equipment they might have been as low as 3 to 5 kHz, they may be above 15 kHz at good efficiency now.

The filter needs to filter well at the basic switching frequency, and above, but have little effect at 60 Hz. .A filter that does that would drastically cut the risks associated with having a lot of fairly high frequency "hash" in the power line, and would tend to cut the radio interference as well.

I've said this before but i forget where it is: those two capacitors in the photo in post #1 are the filter. Adding an additional capacitor across the input of the PP is electrically redundant. However, the fact that they are using motor run capacitors as the main filter element after the inductor concerns me, as i've said before.


Aluminum wire is pretty cheap, 100$ of aluminum wire coiled on a spool will make an effective air core inductor even for a 200 amp circuit, no need to spend hundreds of dollars on a "line/load 3 or 5% reactor" and pay shipping and handling on top of it.

It would be more electrically efficient to use less copper or less aluminum and use a magnetic core but it is not necessary in this case. you will want to size the inductor to drop maybe 2 volts at 60hz at full load, that should be enough inductance to effectively block the 9KHZ switching frequency of the phase perfect (i only have one source for that 9 khz value, it may be different for different models.)

Anyhow, a capacitor of say, 50uf goes on both sides of the inductor, both connected line to neutral on both sides. adding this capacitor to the input of the phase perfect will reduce some of the ripple current that passes through those output capacitors, and the inductor should block 98% of it, with the capacitor on the electrical panel side of the inductor upstream, blocking 98% of the 2% that remains.


The fact that they rely on 2 capacitors in series to handle the bulk of the ripple current demanded of the input boost converter astonishes me, but maybe they have some additional capacitors hidden somewhere. but, if that were the case then a failure of those output capacitors would only affect downstream power and it would only be the third leg that would have the pwm on it. but more than one person has reported the failure of those motor run capacitors has caused other failures upstream on the single phase side. anyhow, the boost converter on the front end of the phase perfect, in theory if they ran the frequency high enough you wouldn't need much of a capacitor to block it all because the ripple current it demands is rather little, the frequency it runs at may be significantly different than the 9khz reported by one person on this site (and that 9khz could have been the input boost converter or the output half bridge since it was a friend of theirs that had the oscope and reported "strong spikes at 9 khz, or perhaps both run at the same frequency)

anyhow, everything i've said concerns differential mode noise at the pwm frequency.

common mode noise is a problem and of a much higher frequency, usually its associated with harmonics of the frequencies of the rising and falling edges of the square waves.. so if it takes 1us for the igbt to turn on, that voltage swing of 0 to +/-350vdc in just 1 micro second, we'll call that 1Mhz.. that noise gets outside the box in a different way than the 9Khz ripple frequency i mentioned in the prior paragraphs does.

if you do have a problem with rfi in the 50khz to 10mhz range from the phase perfect.. there are ways to deal with that but i would need to open one up to deal with that.. you may need to verify that the steel case of the whole thing is sealed up, and put the rfi/emi filters inside the pp.
 
The old blue units had 3 transformers in the back. I just noticed the other day the white units(2010 PT-355) has what looks like 2 at the back, different configuration than the ones in the old blues. I hadn't realize until now while going under it with a flashlight and looking through the grill that the white units still used any, I just assumed they didn't and that was where the big weight reduction had somehow come from... though these new/different looking ones may be much lighter in design?
I don't know what these transformers in there are for though.

And now I'm back to thinking of adding line reactors on the input. I think I'll get one and see what happens before putting one on each unit.
I messed around with adding few capacitor and it did eff all so far, a neighbor said film capacitors like these do eff all for high frequencies anyway.
 
...
common mode noise is a problem and of a much higher frequency, usually its associated with harmonics of the frequencies of the rising and falling edges of the square waves.. so if it takes 1us for the igbt to turn on, that voltage swing of 0 to +/-350vdc in just 1 micro second, we'll call that 1Mhz.. that noise gets outside the box in a different way than the 9Khz ripple frequency i mentioned in the prior paragraphs does.
....

A one microsecond rise time waveform like that has substantial power at the higher fourier components. You'll see substantial signal at even 100 Mhz. You have to
be well versed in RFI containment to keep that in the bottle.
 
I've said this before but i forget where it is: those two capacitors in the photo in post #1 are the filter. Adding an additional capacitor across the input of the PP is electrically redundant. However, the fact that they are using motor run capacitors as the main filter element after the inductor concerns me, as i've said before.


Aluminum wire is pretty cheap, 100$ of aluminum wire coiled on a spool will make an effective air core inductor even for a 200 amp circuit, no need to spend hundreds of dollars on a "line/load 3 or 5% reactor" and pay shipping and handling on top of it.

It would be more electrically efficient to use less copper or less aluminum and use a magnetic core but it is not necessary in this case. you will want to size the inductor to drop maybe 2 volts at 60hz at full load, that should be enough inductance to effectively block the 9KHZ switching frequency of the phase perfect (i only have one source for that 9 khz value, it may be different for different models.)

Not being too familiar with this. When building an inductor with wire as you say. Would this be single wire where 2 individual coils would be made separately one for each leg of the 1Ph input, or just use a 100' coil of say #6/2 to feed a 10HP PP ?

50-60amp 5% line reactors don't seem to be much $ though so maybe I should just get one now and try it on the 10hp unit.
 
I've said this before but i forget where it is: those two capacitors in the photo in post #1 are the filter. Adding an additional capacitor across the input of the PP is electrically redundant. However, the fact that they are using motor run capacitors as the main filter element after the inductor concerns me, as i've said before.


Aluminum wire is pretty cheap, 100$ of aluminum wire coiled on a spool will make an effective air core inductor even for a 200 amp circuit, no need to spend hundreds of dollars on a "line/load 3 or 5% reactor" and pay shipping and handling on top of it.

It would be more electrically efficient to use less copper or less aluminum and use a magnetic core but it is not necessary in this case. you will want to size the inductor to drop maybe 2 volts at 60hz at full load, that should be enough inductance to effectively block the 9KHZ switching frequency of the phase perfect (i only have one source for that 9 khz value, it may be different for different models.)

Anyhow, a capacitor of say, 50uf goes on both sides of the inductor, both connected line to neutral on both sides. adding this capacitor to the input of the phase perfect will reduce some of the ripple current that passes through those output capacitors, and the inductor should block 98% of it, with the capacitor on the electrical panel side of the inductor upstream, blocking 98% of the 2% that remains.


The fact that they rely on 2 capacitors in series to handle the bulk of the ripple current demanded of the input boost converter astonishes me, but maybe they have some additional capacitors hidden somewhere. but, if that were the case then a failure of those output capacitors would only affect downstream power and it would only be the third leg that would have the pwm on it. but more than one person has reported the failure of those motor run capacitors has caused other failures upstream on the single phase side. anyhow, the boost converter on the front end of the phase perfect, in theory if they ran the frequency high enough you wouldn't need much of a capacitor to block it all because the ripple current it demands is rather little, the frequency it runs at may be significantly different than the 9khz reported by one person on this site (and that 9khz could have been the input boost converter or the output half bridge since it was a friend of theirs that had the oscope and reported "strong spikes at 9 khz, or perhaps both run at the same frequency)

anyhow, everything i've said concerns differential mode noise at the pwm frequency.

common mode noise is a problem and of a much higher frequency, usually its associated with harmonics of the frequencies of the rising and falling edges of the square waves.. so if it takes 1us for the igbt to turn on, that voltage swing of 0 to +/-350vdc in just 1 micro second, we'll call that 1Mhz.. that noise gets outside the box in a different way than the 9Khz ripple frequency i mentioned in the prior paragraphs does.

if you do have a problem with rfi in the 50khz to 10mhz range from the phase perfect.. there are ways to deal with that but i would need to open one up to deal with that.. you may need to verify that the steel case of the whole thing is sealed up, and put the rfi/emi filters inside the pp.

A "filter" is composed of inductors and capacitors, we are NOT talking here about another capacitor just glommed across the line, although that would not necessarily be "redundant" nor a bad idea.

A proper filter can and would take care of the common mode as well. They are made and spec'd for their performance in both modes. I spent a number of years doing EMI solutions and testing, and have designed and used a number of them.

The PFC input to the PP is no doubt the problem. The PFC DOES inherently have an inductor in series with the input, and theoretically should have much less input side noise than is being described. that is one of the selling points of the PFC. Even so, any credible design has an input filter in addition to the PFC inductor. And it may need to be fairly etensive. However, designs differ, and it may be that common mode noise is the main issue with 2 phase units. My largest was 2.5 HP single phase 120/240V..

It really does not take much to destroy the simple "capacitive dropping" type power supplies, they are just bound to fail if much HF current is put through them. Given the prevalence of noise on the lines, it would seem they should be obsolete, but they are cheap, so they continue to exist.

PP should include a real line filter, I am amazed they are not falling afoul of the FCC. Maybe they need to be reported.
 
I got a bit more info, the frequency is apparently 5 Khz, (maybe blues were 9? mine definitely a much higher pitch by ear than the white units anyway)
Do you guys think a 5% Line Reactor would help with either a 5Khz or 9 khz harmonic frequency? for $500 I'm thinking of getting on to try it.

Anyhow, it turns out induction cooktops are real picky at picking up on that frequency, someone else reported it too, so I'll just keep flipping that breaker off when I run the 20hp unit. It doesn't seem to pick up on the blue 10hp.
I think otherwise so long at the unit works as its supposed to and caps are in good shape it should all be fine, it has worked for over 10yrs almost 7 days a week for me so...

OP's failure is certainly not a normal thing. What happened to wheelie's was pretty extreme too but probably super rare as well.(running 3-4 cnc's on 1 unit in very hot climate is probably a lot harder for it and the older capacitors to handle, vs running 1 or 2 in 68° temp) The new bigger 480v's for the 60uf units will surely help with that. Not yet available for the 355's though as these are 40uf.

I'm back to feeling less worried now, I'll just check them a little more regularly. My blue unit had spent almost 10yrs under a welding/grinding bench and never an issue, was still super clean inside when I moved it and checked it out.
 
The % relates to power level to get inductance. 10 HP is around 10 kVA, so at 230V is around 40-45A. Probably less if the PFC does its job of making the input look like a resistor.

So then, 45A and 12V drop (approx 5%) gets you 12/45 = 0.266 ohm reactance. about 0.7 mH inductance

At 5 kHz, that has about 22 ohms inductive impedance, so it is not much help by itself, but would work well with a capacitor of around 50 uF. That would resonate around 850 Hz to 1100 Hz, which should be reasonable far from any likely problem frequencies., and nicely below the 5 kHZ. Attenuation should be decent at 15 to 20 dB or so.

If done as 2 inductors one per leg, and a 50 uF to neutral from each on the supply side of the inductor, it would take care of both differential and common mode reasonably effectively. Same 850 hz resonance on each side, possibly a little higher line-to line.

One possible issue is that the parts are not perfect. Inductors pass high frequencies, and capacitors have inductance and resistance, because they are made of actual, not theoretical parts, so performance may not be quite as good as the basic equations suggest. But, I suspect it would no longer cause fires in other equipment. No guarantees, it depends on things we cannot determine from here.

It might be as well to use the filter PP recommends, if there is one. They should have tested it and determined it is suitable. Check if they say it is good to bring interference to "industrial" levels, or if it will bring interference down to "residential" levels. I am betting they never considered residential.
 
yes you can just add another film cap on the single phase input line and that may take care of quite a bit of the problem, it should be inside the pp chasis as close as possible to the inductors used in the boost section of the converter.

two motor run capacitors in series as they are using will have significant parasitic inductance and so yes, adding another one is electrically redundant, but it cuts the inductance down by 66% or something so instead of say, 1 volt rms of ripple at 9khz at the input terminals, you get .25 volts. (is it that high? i suspect it could be)



regarding SND's question. if you simply measure the inductance of a spool of 100 feet of 10/2 romex as its normally wired up in an ac electrical system, you've got nearly zero inductance because the current is flowing in opposite directions. you've got to reverse the connections on that second wire so the current is flowing in the same direction, but yes its more efficient to wire both conductors together than to use two separate coils, more efficient to the tune of about 30-40% more inductance for the same amount of wire.


magnetically inductors are most efficient when they are about crispy creme donut shaped, about a 1:1 ratio for inner diameter and height and width/depth of the spool. anyhow..

common mode inductors need a ferrite core of high permeability designed for that task. they would need to be added inside the steel shell of the pp or the wires between the filter and the pp will still radiate the 100mhz that Jim Rozen thinks you can get from a 1us rise and fall time (though where that noise actually comes from is the ringing at the top and bottom of the square wave, but i think he knows that)



regarding motor run capacitors: i would be surprised if their self resonant frequency is higher than 50KHZ. so they aren't effective above that. you need film capacitors, not motor run caps.
 
Last edited:
Probably, below FIVE Khz. Has anyone ever put one on a network analyzer?

Most capacitors are really inductors when you look close. And most inductors, are
really capacitors.

Resistors now, are actually both!
 
The degree of "being what they seem to be" depends on design. The motor run types, or any wound type where the turns are not shorted by end metallization, or a shorting tab connection, have an interesting property.... they vary in effective capacitance with frequency, as the inductance of the foil wraps "blocks off" part of the windings at any particular higher frequency. Yet they remain reasonably effective, since the same net impedance at higher frequencies corresponds to a smaller capacitance than at lower frequencies.

However, they often remain reasonably effective over a wider range than yu might expect. Even when the capacitor has passed its self-resonance point, it may still be an effective low impedance. Unless you are interested in phase, tou do not need to worry about that. until the net attenuation drops off to an unacceptable level.

Film types often do have something, either end metallization, or just a wire, shorting the turns. But even then, they have inductance simply due to their length, and their leads, Only quite small parts can retain their ideal characteristics to very high frequencies. That would be parts in a 603 or 402 surface mount size.

Obviously, the PP folks are using parts that are effective as an input filter at and above the switching frequency, so long as they are still good.
 
So to update the thread, I have gotten my Speedio back up and running thanks to quick help from folks from Yamazen. We diagnosed DC power supply as a first failed component and Yamazen over-nighted me new one. I got it replaced and luckily that's the only thing that blew. Speedio is making chips again.
 








 
Back
Top