Hi frequency = radical voltage fluctuations?

**Oklahoma Wolf** · 07-25-2006, 11:21 AM

Re: Hi frequency = radical voltage fluctuations?

The 600W is doing it too? Very interesting... I wonder if the secondary caps aren't just a bit undersized. I was somewhat concerned about the multitude of Capxon 1000uf in there. This isn't helping me think of these units as FSP's top of the line next to the independantly regulated older generation server units.

**jonnyGURU** · 07-25-2006, 12:55 PM

Re: Hi frequency = radical voltage fluctuations?

I don't even think that's it. I've seen a lot cheaper PSU's with a lot worse worse quality components look better on the Oscope. It's got to be something else. Less obvious.

**Oklahoma Wolf** · 07-25-2006, 01:08 PM

Re: Hi frequency = radical voltage fluctuations?

At any rate, it's probably some side effect from the higher switching frequency. Just looked at Xbit's review again... he said he got ripple as high as 78mv on the 12v.

**jonnyGURU** · 07-25-2006, 01:10 PM

Re: Hi frequency = radical voltage fluctuations?

If this has an effect on an o-scope, certainly this can have an effect on computer components. You think?

**Oklahoma Wolf** · 07-25-2006, 01:14 PM

Re: Hi frequency = radical voltage fluctuations?

Unless the components in question are filtered enough to be resistant to such short pulses...

These seem to be getting real popular real fast though... haven't heard any major complaints about instability, even when overclocking to the moon. I guess when they designed these, they went for power first and compromised a wee bit on stable regulation.

Certainly they're no replacement for the regulation in Zippy or other high end units.

**Super Nade** · 07-25-2006, 01:49 PM

Re: Hi frequency = radical voltage fluctuations?

I was reading an article of EDN about how ground bounce is a common complaint these days. Maybe it has something to do with this case?

http://www.edn.com/index.asp?layout=...edesc=features

For n00b's like me, I find the articles in them very useful. I subscribe to the print version because its free.

Edit*

Upon thinking on this a bit more, I think the problem could be one of resonance/capacitive ringing. I'm not sure about high voltage components, but I've seen severe ringing in simple amplifier ckts, where the bypass cap (limits bandwidth) was the culprit. To my utter surprise, the issue was because the cap had LOW ESR! I added a 1 ohm in series to kill the ringing!

**PeteS in CA** · 07-25-2006, 02:02 PM

Re: Hi frequency = radical voltage fluctuations?

Are the fluctuations periodic? If so, does the period match the switch frequency. If it is periodic but doesn't follow the switch frequency, but is several hundred Hertz to several KHz, you may be observing oscillation. OTOH, if the fluctuations aren't exactly periodic, is the +12V O/P powering a DC fan? If it is, DC fans draw current in large pulses, and that might be the source of the "problem"; also, if there is a separate fan O/P, but the mfr. goofed the PCB layout, fan noise may be getting into the feedback loop.

Just some ideas. One way to see how much radiated noise is affecting your oscilloscope is to clip the ground lead and the probe tip to the same point (usually circuit ground). Whatever spikes you then see are due to picking up radiated noise.

**gonzo0815** · 07-25-2006, 02:21 PM

Re: Hi frequency = radical voltage fluctuations?

I think this is due to the active PFC circuitry they are using in that psu`S. I am not realy familiar nor i have understand the full benefit from this new circuit desing, but i have seen some differences, an APFC circuit can deliver. Firsyt, the APFC controler is normaly an stepup regulator, e.g it is desinged to bring the voltage from any moment of the sinwave up to the nominal 300 or 400v, wich is then stored in the i/p cap. As the voltage is very high, the capacitance of the I/P cap could be degreased, but voltage musst be higher.

Then this b

**Super Nade** · 07-25-2006, 02:34 PM

Re: Hi frequency = radical voltage fluctuations?

Guten Tag Gonzo. Was ist das? Ich kann nicht verstaehen Sie. Es tut mir Leid, mein Deutsch ist nict so gut.

What do you mean by "capacitance should be degreased" ?

Pete, how would fan noise couple back into the system? As I understand it, there are seperate analog and digital ground's which are tied together. So there should be no ground loops per se. However, RF pickup can occour via any pointy surface, so at times the clips themselves may be the noise source.

**gonzo0815** · 07-25-2006, 02:41 PM

Re: Hi frequency = radical voltage fluctuations?

I think this is due to the active PFC circuitry they are using in that psu`S. I am not realy familiar nor i have understand the full benefit from this new circuit desing, but i have seen some differences, an APFC circuit can deliver. First, the APFC controler is normaly an stepup regulator, e.g it is desinged to bring the voltage from any moment of the sinwave up to the nominal 300 or 400v, wich is then stored in the i/p cap. As the voltage is very high, the capacitance of the I/P cap could be degreased, but voltage musst be higher 8in normal psu`s the v is maxed to 300 or some 320v, but unregulated.)

Then this preregulation of the i/p voltage enables the desingner to use a somhow sepciall desing, wich will act like a dual phase vrm (in my noobish understanding ;-)).
So, even if the nominal fruequency is the same, there are two independent pulses with a 180° shift.Therefore the o/p capacitance can be dramatically degreased, and the regulation will be very fast, like they will have choosen the doubled switching frequency (may be this is as trivial as an full bridge smps, but i don`t know why this needs the APFC circuitry).

So in the end, it is obviousely, that you can`t reach that perfect stable power like an simpler unit with lots of bulk capacitance, as a good capacitor with very low esr will close to an perfect power source. But adding to mutch capacitance will slow down the regulation loop as far as i have understand that,but this means to sacrifying the fast regulation capability of this circuit. But of course, as caps will cost some money, the PSu maker wants probably only reach the specifikation, wich will tolerat IMHO about 120mV ripple on the 12v rail. And those fluktuations are no subjekt in those specifikations, as long as they are within the voltage tolerance (e.g +-5%)
I have seen similar behaivior in all APFC units i have owned (e.g Seasonic ss400SFS, 300FS, Amacrox Frei Erde 500w, FSP Bluestrom).
Some of those APFC units e.g Seasonic S12 will itroduce some 10MHZ noise into the power lane. As those frequencies aren`t in any harmonic regulation....they probably skimped out on the filtering.

This theory is based on the 80+org sample shematic, wich is IMHO from ST micor devices.May be some other sources i have read too.

**arneson** · 07-25-2006, 03:01 PM

Re: Hi frequency = radical voltage fluctuations?

Originally posted by Super Nade

What do you mean by "capacitance should be degreased" ?

Simple Green ?

**gonzo0815** · 07-25-2006, 04:14 PM

Re: Hi frequency = radical voltage fluctuations?

ups some ugly typo, i mean reduced, e.g you can use some lower values for the I/P and the o/p capazitors.

WTF i have posted this two times

Admin, pls delet the cripled message, if possible.

**jonnyGURU** · 07-25-2006, 04:20 PM

Re: Hi frequency = radical voltage fluctuations?

These are rail readings. I'm not oscoping the main transformer.

I'm sure I'm wrong, but how is the APFC circuit going to affect the DC output so dramatically?

Here's the 12V2 on the OCZ 700W during test 5:

X is 2ms, Y is .05V. As you can see, I'm seeing about a .12V wave there.

By comparison, here's the 12V2 on an Ultra X-Finity 600W during a similar test:

Here's the 12V2 on an Antec TP2 550W:

BFG 650W:

Understand the concern?

**Oklahoma Wolf** · 07-25-2006, 04:42 PM

Re: Hi frequency = radical voltage fluctuations?

According to Xbit, the switching frequency is around 55kHz for the FSP... would I be correct in thinking the sampling frequency of the scope may be set a bit too low at 25kHz? Been so long since I last used a scope, I'm not sure I could use one now if my very life depended on it

**PeteS in CA** · 07-25-2006, 04:55 PM

Re: Hi frequency = radical voltage fluctuations?

johnny, try changing the trigger source to "Line". Looks to me like you're observing line frequency ripple. If it syncs up, then it's line frequency ripple.

**jonnyGURU** · 07-25-2006, 05:43 PM

Re: Hi frequency = radical voltage fluctuations?

Guys....

The O-scope is not set any differently than when I tested any of the other PSU's. The OCZ was done prior to my ST75ZF test. The ST75ZF's test results didn't warrant concern.

I put another PSU (a Mushkin) back on the tester after I did the FSP 600W and the results of that do not warrant any concern.

Please read: I can take any of these other PSU's and plug them in and not touch any settings on the O-scope program and get predictable results (see photos.) I can unplug any of those PSU's and plug in either an FSP or OCZ and get the > .12V fluctuation.

The only variable here is the PSU. Two different PSU's from the same manufacturer.

Not the o-scope, the load or the o-scope settings. The o-scope settings are the same in all of my tests. I don't change them so I can make side by side comparisons like this.

I also checked to make sure there was no interference with the o-scope and cables. I've seen where sometimes if the o-scope shielding touches the chassis or the o-scope is too close to the PSU or the load tester, etc that I get interference. O-scope is free from any of this. It sits on the bench. Nothing touches it. Same place it sits during all the tests.

EDIT: Looks like Blogger isn't liking my link. I've uploaded my O-scope screenshots for the OCZ now, so here's a local JPG of it:

**jonnyGURU** · 07-25-2006, 06:39 PM

Re: Hi frequency = radical voltage fluctuations?

Originally posted by Oklahoma Wolf

According to Xbit, the switching frequency is around 55kHz for the FSP...

Really? That would be very disappointing. I can see using a higher frequency in an effort to use smaller components, but if the frequency is the same as any other PSU on the market, and the components are smaller, and the heatsinks are smaller... then I'm not in the mood to conclude this is a decent power supply. Would you?

**Oklahoma Wolf** · 07-25-2006, 06:50 PM

Re: Hi frequency = radical voltage fluctuations?

According to Oleg in the review, most PSU's are below 30kHz... that's why I was thinking the sampling frequency on the scope may be perhaps too low for this particular PSU. Then again PeteS knows a lot more about the engineering side of these than I do.

Here's the quote from Xbit:

Originally posted by Oleg Artamonov, Xbit Labs

Another interesting feature of this power supply is the increased frequency of the PWM regulator. It is usually near 30kHz (this looks like fluctuations at 60kHz in oscillograms of a PSU's output voltages since the push-pull inverter doubles the frequency). In the Epsilon this frequency is 55kHz or almost two times higher than usual.

http://www.xbitlabs.com/articles/oth...roundup_7.html

**Super Nade** · 07-25-2006, 06:54 PM

Re: Hi frequency = radical voltage fluctuations?

John, that looks like a nice AM signal to me. You could discern the frequency components by looking at the FFT of the signal. I'm not sure if your scope has an option to do this or not.

If you noticed, all your PSU's display the same AM envelope, but the OCZ has a higher frequency carrier. The problem of higher harmonics is usually due to small valued filter caps. My wild guess would be that the envelope is due to the input caps, while the higer harmonics are due to the output caps.

Hi frequency = radical voltage fluctuations?

Hi frequency = radical voltage fluctuations?

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Comment

Related Topics