3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

[Behemot]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

I was not refering to 3300uF caps this time, more like other sizes and voltages. I have run out of higher-voltage caps with 20mm height for displays and from what I've been quoted, I can get e. g. KYA 1000 uF/25 V in D10x20 for lower price than KY with just 680 uF at DigiKey.

[trodas]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

In the 10mm caps is there also a good candidate - Nichicon HN, 1800uF 16V d10 - yet it have 3190mA rippe current...!
http://www.digikey.com/product-detai...MPD-ND/2428108

WOW!

Now this is a good time to ask one simple question. What does matter more in the PSU? The capacity or the maximal ripple current?

Time to take out a scope and check the voltages results on the PSU 12V rails with 3300uF Samxon GT and then replace it with 1800uF Nichicon HN Any takers?

[mariushm]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

HM and HN are discontinued

[Pentium4]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

I think ripple current and ESR are way more important. The ESR might be too low for in a PSU, but yes post the results please

[Behemot]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

…

If I remember your previous experiments correctly, only thing you have achieved is making the PSU oscilate and have even worse results. I even have this passive Fortron which has such bad ripple that it kills new cap within hours.

So, you are still not cured from that?

[trodas]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

Behemont, if you remember correctly, it was the Eurocase (Eurocrap) PSU that was causing problems and not with the replacement of the output fitering caps, BUT with the replacement of the 1uF elyte caps with SMD ceramic caps...

Reminder: http://www.svethardware.cz/forum/sho...ocase-350W-PSU
...and YOU was the guy, who reminded KEOSAN, that the whole trouble begin with increasing the capacity of caps in the control phase of the PSU: http://www.svethardware.cz/forum/sho...l=1#post142762
Remember?

That was a huge NO*NO, not the caps at the filtering stage

IMHO is also the ripple current AND the ESR way more important that capacity. And since this is a d10 cap and it punch neat 3190mA - I did not see why for example Enermax Liberty should not like it much!

[Behemot]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

- HoNY - himself told me you two experimented with Samxon GC in PSUs and switched back to RS because of too low ESR. Did anyone measured the results or everybody has been just blind-trying that?? I even replaced some GCs with RS after him few years later…

I see why, Liberty use CEC TUR which probably has less than half the ripple. You are way far from what the cirtcuit was probably designed for.

[mockingbird]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

c_hegge has an excellent writeup about this very topic on his site:

http://hardwareinsights.com/wp/poly-...actually-work/

The article isn't about poly-modding per se, but rather about the effect of using caps which characteristically have a much lower ESR than was called for in the original circuit design which is the topic of the argument.

Ultimately, what your argument is getting down to is how low can you actually go beyond a 'KZE class' cap with an ESR of around 0.020 before you start experiencing failure?

[Behemot]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

So ripple-wise, it was more or less the same two times from three. However, two times from three it does not work at all.

If you are interested, I have some numbers of my own here, first hard data I have in hand. Recently, I have also recapped some older Seasonic S12 (OEM model) and the ripple after recap was just excelent (~13 mV@100 % labeled power on +12 V), but I haven't taken any readongs before so I cannot compare.

This is the Evolve Pulse 80+ 500 W, I think I posted it here before. OEM is Guangzhou AOIJE Technologies Limited. As follows (links only as it would be too many images to load), all measurements are at +- the same (100 % nominal) load:

default
http://diit.cz/sites/default/files/33_v_27.png
http://diit.cz/sites/default/files/5_v_28.png
http://diit.cz/sites/default/files/5_v_sb_28.png
http://diit.cz/sites/default/files/12_v_28.png
http://diit.cz/sites/default/files/-12_v_28.png

added 3 Y caps on primary (ground-ground) + between primary and secondary
http://diit.cz/sites/default/files/33_v_28.png
http://diit.cz/sites/default/files/5_v_29.png
http://diit.cz/sites/default/files/5_v_sb_29.png
http://diit.cz/sites/default/files/12_v_29.png
http://diit.cz/sites/default/files/-12_v_29.png

increased input cap from 220 uF to 330 uF (268 uF measured)
http://diit.cz/sites/default/files/33_v_29.png
http://diit.cz/sites/default/files/5_v_30.png
http://diit.cz/sites/default/files/5_v_sb_30.png
http://diit.cz/sites/default/files/12_v_30.png
http://diit.cz/sites/default/files/-12_v_30.png

complete output recap (Samxons RS 3300uF/16V + 680uF D10 Chemi-Cons KY; however, I have somewhat disbalanced +12 V rails with this)
http://diit.cz/sites/default/files/33_v_30.png
http://diit.cz/sites/default/files/5_v_31.png
http://diit.cz/sites/default/files/5_v_sb_31.png
http://diit.cz/sites/default/files/12_v_31.png
http://diit.cz/sites/default/files/-12_v_31.png

So at least in this case, with somewhat disbalanced PI filters, increasing total output-side capacity made the result worse Using all previously-vacant Y cap positions and increasing input capacitance decreased output ripple.

However, compare this:



First is 220 uF@575 W, second is 286 uF at ~500 W. Different scope, yet there is no clear improvement. So I think that seeing power-grid's ripple on output is not always connected with inefficient input capacitance rather than with PFC's design.

Anyway, I am getting strong feeling that you can somewhat improve/worsen the ripple by used caps, but absolutely highest impact has the overall PSUs design quality.

[Pentium4]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

Is it possible to make motherboard grade caps more usable in PSU's such as nichicon HN by leaving more of the capacitor leads, by having it sitting taller thus increasing the ESR a little bit? Because they have amazing ripple rating for their size, or does increasing ESR just lower the ripple current?

[Wester547]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

^ I think it's a bit more complicated than that. Older series of capacitors from decades ago tended to have high ripple ratings for being so high ESR. It might take more than a bit of an ESR increase with longer leads to compensate for altering what was originally tuned to a certain "frequency" or "amount" of noise. PI filters (most of the time seen as one capacitor, a PI filter coil, and another capacitor to form a PI filter, but there are many other alterations) are optimized and configured in a very meticulous manner, so deviating from what was configured ideally for a certain feedback loop (or rather controller) could yield undesired results. Switch mode power supplies don't really benefit from ultra low ESR capacitors (giving exception to the +5VSB output and linear regulated rails) the way the Vcore circuit of a motherboard does (as an example).

Heating up a capacitor does increase ESR and ESR does increase as said capacitor ages. I'm not sure if increasing ESR lowers the ripple rating - by increasing ESR, you are decreasing the amount of ripple current "shunted to ground" by the capacitor (increasing the end ripple voltage going to the load, decreasing the ripple current handled by the capacitor). That being said, I also noticed that lower ESR capacitors do have higher ripple rating than higher ESR capacitors, which means that they dissipate less power still when passing the same amount of ripple so it doesn't "escape" to the load (though that too has limited to no benefit in SMPS applications when you go low enough, I believe as noted by Pete).

Lower ESR capacitors (especially ultra low ESR lytics, this isn't counting polymers) do tend to be on the more aqueous side of electrolyte, and such electrolyte expands considerably with heat by comparison to the less aqueous capacitors, so heat related failures can be a problem. Obviously not to the outrageous extent that Taiwanese and Chinese capacitors (or KZG) fail from that, but of note nevertheless. To answer your question without a large rant on my part, however, HN isn't really suited for SMPS use, neither is HM or anything equivalent. I wouldn't really go lower than Nichicon HD as far as grade of capacitors go for SMPS applications.

Ultimately, what your argument is getting down to is how low can you actually go beyond a 'KZE class' cap with an ESR of around 0.020 before you start experiencing failure?

You could go a bit lower (thus HD grade), but anything lower and the increased ripple voltage may be problematic. Personally, I think Nippon Chemi-con KY grade is the best choice for SMPS, because capacitors of that ESR range still do away with lots of ripple without causing problems.

Also, if HN impresses anyone here in terms of ripple current, well, polymers go much higher than that at similar or lower ESR ratings, but of course they aren't really necessary for SMPS either (though it's nice to see them being used more in modern PSUs).

[Pentium4]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

You have some good points. No wonder GP caps have higher ESR and poor ripple ratings (Rubycon PX) not a rant, more like good info. KY series are definitely one of my favorites for SMPS, they're fairly priced, good specs and very reliable. Once I get my scope I will experiment with some HN series caps in some PSU's and see what the results are

[c_hegge]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

Is it possible to make motherboard grade caps more usable in PSU's such as nichicon HN by leaving more of the capacitor leads, by having it sitting taller thus increasing the ESR a little bit? Because they have amazing ripple rating for their size, or does increasing ESR just lower the ripple current?

Leaving the legs longer only increases the ESR very slightly (less than 0.1 ohm), so it's not a meaningful enough increase to make the cap workable for an SMPS. The only way you're gonna make a HN/HZ workable for a PSU is if you modify the PSU and re-design the feedback loop around them, like what you would have to do in order to make it work with polies.

[momaka]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

Sorry for this sort of epic thread revival, but I am just beginning to clean my summer mail and going through some of the more interesting threads.

The only way you're gonna make a HN/HZ workable for a PSU is if you modify the PSU and re-design the feedback loop around them, like what you would have to do in order to make it work with polies.

Not necessarily. Some PSUs will work just fine with Nichicon HN grade and equivalent caps. It all depends on the PSU really. I fully recapped a Bestec ATX-300 12Z and partially an ATX-250 12Z. Both have been working fine. The only thing that I can't confirm 100% to be fine is the ripple output because I do not have a scope to test them with. But the computers they were in seemed to be perfectly stable.

[Pentium4]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

^ Interesting. Considering the stock caps were probably GP Jamicons? How long have they been running? And I am under the impression that if a PSU is outputting lots of ripple, the hard drive will run much warmer. Is this true or only with severely high output ripple?

And Behemot, thank you for sharing your testing results. I was kind of surprised that increasing the input filtering capacitance lowered the output ripple. Is this because with the ability to filter more power, the wave is more smooth for the switchers to chop? I don't understand how that lowers the output noise but that is very cool

[momaka]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

^ Interesting. Considering the stock caps were probably GP Jamicons? How long have they been running?

Yes GP Jamicons in the ATX-300 12Z and CapXon KM in the ATX-250 12Z.
Not sure about the running time. I guess I should have checked the hours on the HDD in which the ATX-300 12Z was in since that PSU was original to that computer. I would guesstimate under 10k hours for sure. Quite possibly even under 5k since there wasn't that much dust in there.

And I am under the impression that if a PSU is outputting lots of ripple, the hard drive will run much warmer. Is this true or only with severely high output ripple?

In my experience, this has been true. Had it happen with multiple HDDs on multiple computers. This was back before I knew about ripple and bad PSUs. But I did take note.

[Behemot]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

And Behemot, thank you for sharing your testing results. I was kind of surprised that increasing the input filtering capacitance lowered the output ripple. Is this because with the ability to filter more power, the wave is more smooth for the switchers to chop? I don't understand how that lowers the output noise but that is very cool

Well, for me it is expected restult, why else increase the capacity in the first place. But not always as if you breake the Pi filter too much, than oscillation outweight the capacitor effect.

[Pentium4]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

Well I knew the capacitance of the input capacitor can affect the total power capability but I didn't know it could affect the output noise. I've been told that the entire input has nothing to do with the noise on the output side

[Behemot]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

Oh sorry, missed the "input" cap. Well, yeah, there was some strange low-frequency waving on high power. Well, on theory, if some high-frequency ripple comes through input filtration, than there is the bulk capacitor to filter it out. Here the ripple is in MHz range where PSU works only in kHz range so basically when the transistor is opened, these things may pass through on secondary. That's my understanding anyway.

[Behemot]

Re: 3300uF low-ESR caps suitable to PSUs in D10mm - may be possible!!

Nippon Chemi-Con KYA caps arrived, seems legit, will update website in a while.