Poly and 'lytic mix questions

[Gariarto]

Hi guys, I have a Dell Precision desktop in the shop that has burst Nichicon HNs in the output side of the VRM. All other caps appear to be visually OK. All I've done at this stage is open the case and look around.

As you'll see in the picture they are four 6.3V - 820uF mixed with five 4V - 560uF polymers either side.

Now, I've seen this before on motherboards but with one electrolytic at each end of a row of polymers. It raises a couple of questions if I may...

1- Why ? What would be the purpose of the board manufacturer doing this ? Is is to effect timing? (though I don't think how this would make sense) or is it saving them a couple of dollars ?

2- Given the answer from 1- would it work to just put 4 polys in there or does it have to be the same electrolytics?

Thanks in advance for your input

[delaware74b]

Re: Poly and 'lytic mix questions

Judging by the date code on the power supply (0412), your HN's are of the bad production run by Nichicon. You can replace all of the caps with the same series and ratings from this site, Mouser, or Digikey. Nichicon has fixed the problem with the HM and HN series capacitors (overfilled with electrolyte, IIRC).

All of the HN's on your board should be replaced, regardless of being burst. If not burst, it won't take long for the remaining to fail.

[Gariarto]

Re: Poly and 'lytic mix questions

Thanks for your input, I appreciate it. Yes its an older Dell. The caps in the PSU are visibly fine and the output voltages are within tolerance and stable so thats OK right now.
I know I have to replace the failed caps, I have replacement HN's ordered and on the way, I just wanted to know what the answers to my first 2 questions might be.
Why the mix ? Anyone ?

[momaka]

Re: Poly and 'lytic mix questions

Why the mix ? Anyone ?

Electrolytic caps are cheaper than polymers, so probably just cost saving for the manufacturer indeed. Also, notice the 2 empty cap spots in your picture - I bet you $50 that this board was designed to take all polymers and have all of the spots filled. But the manufacturer likely decided that replacing some of the polys with electrolytics gives good enough noise suppression for the CPU so they went that route to save some $$. That also answers your other question - no the timing won't be affected by using all-polymer caps.

[Pentium4]

Re: Poly and 'lytic mix questions

I wonder if the ripple is still in spec with those poly's having to work overtime. But I'm sure those lytics being dead throws everything off balance

[tom66]

Re: Poly and 'lytic mix questions

The polymers are for high frequency ripple (200kHz~500kHz), along with ceramics. They handle the vast majority of the switching ripple.

The electrolytics are for the lower frequency ripple (1kHz~200kHz), and for bulk decoupling (used during CPU power saving switches.) However, they would also take some ripple current from the VRM. Switching off a load from a VRM quickly can lead to sharp spikes.

It is not necessarily cost saving, but very common engineering practice to use different types of capacitors to handle different transients/frequencies.

[Gariarto]

Re: Poly and 'lytic mix questions

That about covers everything I could want to know guys and I appreciate it greatly. Thankyou very much ! I'm now better armed to decide what to do and why in future. Incidentally, the board is now repaired with the same make and series (though now half the height) and is back with a happy customer.

Tom66 I'm intrigued by your response and have another question from your answer that myself and I'm sure many here would make use of in choosing capacitors for "poly-modding" an older board... Being by original design. If polymers are better at handling higher frequency ripple current and electrolytics at lower freq', how do board makers handle that lower freq' with the newer all- polymer boards ? Do you think they rely on newer PSUs handling that or outputting less ripple ?

It almost says poly modding an older motherboard NEEDS some electrolytics leaving in to run as originally designed ? It would certainly explain why most boards (in a kind of cross-over era) were made with electrolytics on the VRM input side and ploys on the output. Sorry if I'm opening a can of worms here !

[tom66]

Re: Poly and 'lytic mix questions

Tom66 I'm intrigued by your response and have another question from your answer that myself and I'm sure many here would make use of in choosing capacitors for "poly-modding" an older board... Being by original design. If polymers are better at handling higher frequency ripple current and electrolytics at lower freq', how do board makers handle that lower freq' with the newer all- polymer boards ?

You generally see them on higher performance motherboards, where energy saving isn't a big concern. Also, I suppose you could use polymers for low frequencies -- there's nothing stopping you. It just doesn't make sense from a cost perspective -- it costs a lot more than is necessary. Maybe marketing?

[Pentium4]

Re: Poly and 'lytic mix questions

Interesting. So if the CPU and VRM filtering caps are all poly's then how does the lower frequency get filtered?

[tom66]

Re: Poly and 'lytic mix questions

Poly caps will work from low frequency (not DC) to 200~500kHz. They're just expensive so not always used. And normal electrolytics are fine in this application. The lower esr caps will take the majority of the ripple.

The failed HN/HMs were really bad. Not just slightly shorter lifespan, significantly shorter.

[Gariarto]

Re: Poly and 'lytic mix questions

Companies like Asus must be trying hard to re-build trust along with others like MSI in making entry level stuff with the poly treatment like the Asus M5A78L-M LX Plus at $60. Their so-called "Military Grade" higher end boards at 3x the cost with all polymer (plus the extra gadgets) seem to have set a precedent. But a lot of this just has to be more towards marketing besides cost (unless we're going to find in a couple of years they were using cheapo sub standard polymers?) Hopefully it won't be true and remain a benefit to consumers.

Anyhow.. Thanks again for your time and knowledge Tom66, that filled in a couple of gaps for us electronically challenged, looks like the overall picture could be just as much a mix as the caps. Design driven by cost and the need to maintain market share. Wait a minute.. Isn't everything ?

[tom66]

Re: Poly and 'lytic mix questions

You could make stuff like motherboards last for 20 years easily, it wouldn't cost much more, but why bother? Most users upgrade every 2-3 years nowadays. At least, that's what the companies think. I like to keep holding onto my hardware.

[Gariarto]

Re: Poly and 'lytic mix questions

You could make stuff like motherboards last for 20 years easily, it wouldn't cost much more, but why bother? Most users upgrade every 2-3 years nowadays. At least, that's what the companies think. I like to keep holding onto my hardware.

Thats obviously true for perpetuating sales but for many cases such as some of my business customers, they have dated software on equally dated OS's which they don't want to let go of so its easier to fix a board and put it right back in place. My worst case (if you can call it that) is a local radio station who runs 3 networked pcs in broadcast production that are based on Asus P5A's. The OS is Windows 98 running DOS level programs !! It does what they want and they know it inside out so suggesting new stuff meets with cries of anguish !

[Gariarto]

Re: Poly and 'lytic mix questions

Forgot to mention, the radio stations main issue is that their software is only written to work with a particular ISA sound card that they have. So that's the sticking point.

[LLLlllou]

Re: Poly and 'lytic mix questions

Forgot to mention, the radio stations main issue is that their software is only written to work with a particular ISA sound card that they have. So that's the sticking point.

There are a few manufacturers that still make motherboards with ISA slots. Google industrial motherboard and ISA. Expensive as all hell though.