CPU Vcore VRM - capacitance or ripple?

[trodas]

Guys, I have a question that beg some deeper explanation. For Vcore input caps replacement (used Chemi-con KZG (!) with 1000uF 16V d8) should I use polymer with low capacitance but very high ripple current... or elyte cap with same capacitance, only higher ripple current?

3 phase Vcore VRM, ASRock 775i65G mobo, input 4x Chemi-con KZG 1000uF 16V d8, output 5x 1500uF Samxon GE 1500uF 6.3V + 1x 680uF 4V polymer cap.

Replacement for the input either:

elyte Nichicon HZ 1000uF 16V d8, ripple current 2.88A (UHZ1C102MPM6)
http://www.digikey.com/product-detai...PM6-ND/1628340

polymer Nichicon RNU 330uF 16V d8, ripple current 6.1A (RNU1C331MDN1)
http://www.digikey.com/product-detai...DN1-ND/4991607

...

I guess that the VRM output is w/o question where maximum ripple current matter most, so a 2200uF Nichicon LG (d8x12) with 6.7A ripple current (PLG0E222MDO1TD) is the way to go.
They used one polymer there already ( http://postimg.org/image/i2e14qssf/ ) and I did not remember that Samxon made any GE series ( http://www.manyue.com/?pg=products&b...ity=&LoadLife= ) but they also did not show any GC/GA caps there too, so I guess I rather replace these with known good caps with best specs.

But the question is - do the 330uF on the Vcore input matter more that the ripple current or not?
Also polymers produce slight noise in the signal, so that could be why Intel did have plenty of designs, where Vcore input caps are elytes and output are polymers:
Intel board D955XCS or overclockes board Intel D975XBX "Bad Axe" - or Intel D925XCV - there phases with elytes on the input:

So I'm torn between a elytes and polymers. Who will side with which and why?

[ben7]

Re: CPU Vcore VRM - capacitance or ripple?

https://cdn.badcaps-static.com/pdfs/...3565ddcda4.pdf

I like that explanation. I'd go with ceramics as well. They will also help get rid of high frequency noise/emi (Note: they won't get rid of ALL of it though, you need a ferrite to do that. "Choke" it out. Ha. Ha.).

If not using ceramics, then I'd go with polys.

-Ben

[trodas]

Re: CPU Vcore VRM - capacitance or ripple?

Looks like the explaination is fairly good. So the high ripple current wins, go all polymers, even for the Vcore inputs?
And add a 10uF caps on the legs of these polymers to get additional filtering?

The output caps make me almost hit the "greater that 2mOhms" limit for 20 - 200kHz switching frequency. That is because there is 6 pieces of caps, so using 8mOhm caps get me to 2mOhms. Luckily, there will be contacts resistances, atc. So it should be fine

Maybe I can even add there tantal polymers on the legs of the coils... or add ceramics too on the legs of the polymer caps on the output? Will not that be an overkill?

How to determine the switching frequency? If over 200kHz, then I could get away with even lower the 2mOhms output capacitance ESR...

[ben7]

Re: CPU Vcore VRM - capacitance or ripple?

Maybe I can even add there tantal polymers on the legs of the coils... or add ceramics too on the legs of the polymer caps on the output? Will not that be an overkill?

Might be overkill, but I personally don't see anything wrong with overkill

How to determine the switching frequency? If over 200kHz, then I could get away with even lower the 2mOhms output capacitance ESR...

Measure the period of the switching node. (The connection to the upper and lower FETs, and one pin of the inductor)
As a side note, I know that some advanced converters/controllers/regulators also might employ some sort of variable frequency mode, but I think this is usually done for saving energy at very low load.

Also, when replacing caps, I do see how it could possibly cause trouble if you don't match the capacitance. The change in the capacitance would change the feedback loop, which could move it into an unstable region. I dunno if different amounts of capacitor ESR can cause instability as well, however I suspect it can. Though, don't quote me on it, I'm only a hobbyist at the moment :P
Actually, I do recall someone at a seminar showing how adding an electrolytic cap (with "high" ESR, compared to ceramic caps) in parallel with the already present ceramic output caps would stop the circuit from having sub-harmonic oscillations (instability). Interesing, eh? It makes it look a little more like this, circuit wise:

The resistor would be the much higher ESR of the e-cap. (C16)

-Ben

[c_hegge]

Re: CPU Vcore VRM - capacitance or ripple?

Don't use PostIMG here. All images are to be attached locally to your post.

Also, one note on your ASRock 775i65G: There are no polies on that board - that one near the CPU is actually a Panny FL electrolytic.

[ben7]

Re: CPU Vcore VRM - capacitance or ripple?

Don't use PostIMG here. All images are to be attached locally to your post.

I just fixed that.

Sorry, been used to doing that in the funny stuff thread.

[c_hegge]

Re: CPU Vcore VRM - capacitance or ripple?

Haha. I didn't even notice it on your post. I fixed it on trodas's post.

[trodas]

Re: CPU Vcore VRM - capacitance or ripple?

...well, I wanted to update the thread(s), but I tried & failed... There is no way to place the thumbnails of the local attachments to the thread.

Therefore the local use of the images is not possible as long, as it is not possible to place them where they are need in the post.

On 100+ forums PostImg thumbnails links works 100%. This is the only one forum, where they are not working. I have then post full screen pictures from PostImg (again, attachemnts previews cannot be placed into the post as need) with the unfortunate possibility that when PostImg go bad, then all the images will have to be restored from my backup...
(and I could done that myself, but the forum does not allow me to edit older threads, so... again fail not on my side)

Another way will be to host the images localy and use their links as IMG links... that way they show up where they are need as full fize images (I will miss thumbnails...) and once again at the bottom of the post.

Witch choice I can take?

[ben7]

Re: CPU Vcore VRM - capacitance or ripple?

...well, I wanted to update the thread(s), but I tried & failed... There is no way to place the thumbnails of the local attachments to the thread.

Therefore the local use of the images is not possible as long, as it is not possible to place them where they are need in the post.

On 100+ forums PostImg thumbnails links works 100%. This is the only one forum, where they are not working. I have then post full screen pictures from PostImg (again, attachemnts previews cannot be placed into the post as need) with the unfortunate possibility that when PostImg go bad, then all the images will have to be restored from my backup...
(and I could done that myself, but the forum does not allow me to edit older threads, so... again fail not on my side)

Another way will be to host the images localy and use their links as IMG links... that way they show up where they are need as full fize images (I will miss thumbnails...) and once again at the bottom of the post.

Witch choice I can take?

What do you mean by locally? If you upload them to BCN, then just right click on the attachment, click "copy link address", and using the insert image button on the post editor, insert the image.

Is this what you are trying to do???:


If you do mean locally as in on badcaps.net, then yes, do it that way. It's preferred over having them externally hosted, because they will never be taken down or removed. (Unless highly offensive, etc... but even then, I seem to recall that TopCat doesn't permanently delete any posts, just removes them from the forum)

-Ben

[c_hegge]

Re: CPU Vcore VRM - capacitance or ripple?

We don't usually make that big a deal of it, but 3rd party image hosting is not allowed here, and Topcat has actually asked us mods to edit any posts that use it, and upload the attachments locally.

[trodas]

Re: CPU Vcore VRM - capacitance or ripple?

Well, that means that I cannot post there, untill a method, how to place the image thumbnails into desired places in the post are found...

Again:

There is no way to place the thumbnails of the local attachments to the thread.

Therefore the local use of the images is not possible as long, as it is not possible to place them where they are need in the post.

On 100+ forums PostImg thumbnails links works 100%. This is the only one forum, where they are not working.

So, the only way to post it with full sized images... and hope that they get where I want them...

[stj]

Re: CPU Vcore VRM - capacitance or ripple?

they get placed after the txt.

use multiple posts if you need to break them apart.

[trodas]

Re: CPU Vcore VRM - capacitance or ripple?

I found a way: https://www.badcaps.net/forum/showthread.php?t=46159

...

So in the end - it is the maximum ripple that matter most and I will get best results when replacing all the caps in the Vcore regulator to polymer ones, even the input ones, despite the best ripple caps for 16V have only 330uF?

(Nichicon RNU 330uF 16V (RNU1C331MDN1) )

[stj]

Re: CPU Vcore VRM - capacitance or ripple?

i wouldnt use 16v caps on a 12v line, 25v would be safer.
i know people will claim otherwise, but standard teaching is that the minimum voltage rating should be 1.5xmaximum input.
so 12 + (12/2 (6)) = 18v
that's for electrolytics, for caps that are more sensitive you want to be even more over-rated.
if your using tantalum i would use double.

[trodas]

Re: CPU Vcore VRM - capacitance or ripple?

There are only very crappy polymer caps for 25V, so I believe in my PSU, that 16V is fine.

IMHO a special 12V voltage caps should be created...