low or super low ESR + impedance caps

[davmax]

Re: low or super low ESR + impedance caps

Hi Linuxguru. I am glad that someone else has supported standard electrical theory. We have tried to help Trodas. He apparently wants to argue/fight.

Help can be offered but if it is not welcomed then, well, move on.

You explained well. Cheers.

[yanz]

Re: low or super low ESR + impedance caps

hi trodas, im glad that you're folding. your team has beat ours last moth

on topic, i think the "fight" is well worthed. i learn so much about inductance and esr from this thread. i think we should bring that to one section on FAQ to explain about esr/impedance and what the relation with ripple current.

imo, the problem you have with the oc and fanless system is because the KZG failed. KZG failed because of external heat, it has lower endurance and given ~1 year duty in your mobo with fanless system its understandable.

now this is my opinion, you could blame me later .. your searching for super caps is because you thinking that KZG is not suitable for your system. mind you, the only KZG not meet your requirement is it lower endurance, not its ripple curent and esr/impedance. as its quality deteriorate, the problem rise on your sound output and hampered the overclocking.

so i would still suggest you putting a caps that match the KZG spec or a little bit lower but has super endurance (5000-7000hrs) or bigger capasitance (higger ripple current). that means bigger diameter caps, but i assume it wont be a problem with your mobo (i don't know about dfi nf2 layout, could you post the pic here?). now you could try it and expect the problem on your audio will be gone. if the ocability isn't better than yes you should find a better caps.

i never see the toroid get too hot. the mosfet does. and i bet your mosfet is very hot that i think you need a bigger heatsink. are your psu fanless? putting a 90mm panaflo fan (low version) on exhaust and operate it at 5v will be silent, as it what i've done. or putting the psu outside (ontop) the case will create a bigger room in your system for heat convection. and the psu would operate at cooler temp, and 5v or 7v for the fans would be enough cooling. the modification posibility is unlimited, you know that (hence you visiting the alternative mod section on ocforums)

sharing information and knowledge is what we love here. keep it on (on fighting)...

(because im not more knowledgeable than the others i will just sit and watch )

[davmax]

Re: low or super low ESR + impedance caps

Hi Yanz, I agree that there should be a good exchange info. But it is not good for Trodas to invent his only theories and trash well established laws of physics and electrical theory and also trash those who give him good information. He will have to learn in a very hard way if he cannot accept these absolute principles. Exchange is OK, fighting or put downs due to poor understanding is not the way to go.

There are many on this forum that are willing to help.

[yanz]

Re: low or super low ESR + impedance caps

That's true and i agree also.. However, if english isn't your main language, sometimes what you're talking is too "sharp" and look like attacking the others, although you do not mean it to be..

I see that Trodas have its own theory and assumption, you have explained that that could be wrong. Now it's up to Trodas to make any decision after getting so many inputs from all of us.. He could be right, he could be wrong, but he'd been warned..

[larrymoencurly]

Re: low or super low ESR + impedance caps

Is it possible for capacitor ESR to be too low for some mobo applications?

I was warned that this can be the case for some TV circuitry.

[trodas]

Re: low or super low ESR + impedance caps

First at all - sorry for big delay. I having lot's of problems, however I also get my hand on Nichicon HE 2200uF 16V caps. They are in pretty wrong casing - 17mm in diamater and only 15mm high. So I somewhat managed to "get em' there" and tried the mobo. First it does not boot at all, only all 4 diagnostic led's on and no autotest, and then, after some bios resetings W/O battery (?!) it kicked in and even it was extremly unreliable, it somewhat works (bios, post). When I added other devices (HDD to boot) it stoped working. I repeaded this two times and I don't know what to blame. The HE caps did not match the KZG specs, true. Is this up to blame? Or the 5-6mm long wires on witchare the caps over the mobo are up to blame?
Well, todays I cleaning up the mobo and preparing it to solder caps where there aren't cap before (yet the space is left) and then I - for fun - tried power it on. I did this few times before but no luck, yet after the cleaning (witch should have no effect on the mobo functionality, I basicaly just sucked the tin from the empty holes, so I can now just "lump" the cap there and solder...) it booted right away. It does NOT show any sign of unreability - even I set up pretty tight timings on memory (my super-tight ones) and 200Mhz FSB and 1.850Vcore (200x10) ...
I was amazed. So, I added on primary channel CDROM and Zip drive to boot for some DOS-stress testing. First I switch the devices both to master and the mobo refused to even post then! 4 lead of death - I know that, all DFI users does... Yet unpluging make it work, then I discover my failure and fixed it, attached again and whoa - it worked.
So, after I realized that everything in DOS works, I decided that the best stress-test will be installing winblows. So, I moved the CDROM+ Zip on secondary channel and attached 80G Seagate testing HDD - and mobo again refuse to post - 4 leds of death (4LOD later).
And this time nothing ever helps. It looks like the mobo need another day or so to "cool down" and erase it's flash rom...

So - good news - the mobo can work. Bad news - there is likely also something else (most likely into the SB powering and stuff) wrong.

Sorry, no time to react on the other posts, guys

[yanz]

Re: low or super low ESR + impedance caps

That's very unfortunate, trodas. Hope you'll figure out what was wrong.

About the lead of the caps, i dont think you should "hang" the caps even for 5mm above the PCB. I know because i've tried it & it just make overclocking worse. The caps need to be as close as possible to the circuit/PCB or it will negate the low esr/impedance value.

About the board, this is yours...



That's not many caps arround the VRM, compared to my 8RDA (NF2)


pic is taken from Epox diagram

So I'm now convinced that your caps need to be very high ripple curent and lower ESR ones. Probably you should replace all KZG there with some sanyo WG or MBZ/MCZ for the possibility of other deffect caps.

Btw this man has a DFI board too.. https://www.badcaps.net/forum/showth...0&page=1&pp=15

.

[trodas]

Re: low or super low ESR + impedance caps

linuxguru - well, ideal inductor is loseless. True enought, but we are talking about toroid coils on mobo, not teoretical stuff.
If toroid is getting hot it is because the magnetic resistance (that is where the loseless go to hell) in core material. Normal stuff. It should not need to mean that the density of magnetic field exceeding it's parameters. It could be (and according to what I read it is) normal stuff. The optimal working temperature is close to 100°C, witch is pretty hot, if you ask me
And I dubt that the ferite changes in any way over time. I have similary hot toroids on my Abit ST6R from 2001 - still not a glitch there.

Can your back claim that the ferite could change in time when facing no more that 100°C with some link/info I don't know about any material changes that should take place there, but maybe I just not payed attention enought.

The hysteresis loss you are talking about could be simply described like the power need to flip all the little magnetic "components" into the ferite. There are far worser materials with notably more loss : However I never ever hear anything about time-factor. Can't you be confused about this AND permament and nonpermament magnetic materials? As the ferite fall to the second group and the first one does lose it's field over time...

davmax - I did not want to fight, I just can't agree with some things you and linuxguru say. But as long, as you both did not try to tell me that the toroids on my mobo heating-up because of resistance of the thick cooper wire, witch (as I measured recently) has not measurable resistance (that it is - the quality multimeter of mine measuring 0.1 ohm for it, but the very same 0.1 ohm it says for the measuring wires only, so...) then we should get along fine :

yanz - no wonder we get beated-up, because my folding went to hell with all the mobos problems... hey, you aren't nowhere neat that fast, stop boosting around and obtain more folding machines. I have 6
The KZG caps are simply nowhere near the quality that they should be. I recently talked with more that enought owners of the very same board and all confirmed the very same - over time the max. OC descreasing, as the caps going to hell. The problem is obvious. And some of them use phasecooling, some aircooling and most have well-ventilated cases. Yet having the very similar symptoms and same problems as me...
Could it be coincidence? Sure, but it is very unlikely.
And yep, the Panasonic FM ones are probably the best ones for this job. They come 7th in the world (check my table) and are the only ones with 7000h rated live - all other only 2000h. Since my machine run about 15h per day, it is 5400h per year... And yep, I managed to fit the bigger FM ones there, yet there is some problems with fiting the heatsink. Even as little space-demanding one, as the Thermalright SI-97A one is... Well, with my WC block it is not a problem, so...
And the toroids are just fine
PSU? No, not fanless. The only one fanless PSU I use is the one I made for my DL server withch run 24/7/365 in same room as I sleep, so... Check it out: http://ax2.old-cans.com/g.php?p=silentpc&d=1&v=v2
That is what I would call modding

davmax :

to invent his only theories and trash well established laws of physics and electrical theory and also trash those who give him good information

Ehm, don't push me : My language barrier did not allow me to better explain what I mean and at least I did not come into theoretical world each time the real world refuse your version

larrymoencurly - I dubt it. ESR too low? Hmmm. That would only mean the circuit is designed to use the limited current from normal caps. To put it simple, the normal cap discharge longer (because he can produce only less current that the low ESR one and therefore it took longer to discharge) and one can think about way what this could be usefull, however the ESR in normal caps aren't too precise thing - so when one want discharge cap slower, then one use a resistor in serial - witch negate the ESR cap value. (quietly assuming that the resistor will have several times higher resistance that the cap, so the cap resistance will play neglecting role there)
But since there is only two infinite things in the world - the universe and human stupidity and I'm not sure about the first, then there should exist circuit, where the ESR of a cap could play a role. Could you show us the circuit?

yanz - working better now, yet still far from purrrfect : And yes, I definitively afree with you - long traces aren't good. And never ever ever for OC. This is mentioned into any voltage regulator - place the damn cap as close to the regulator, as possible. Adding tons of caps on the back side on wires (one suggested make them from paper clip - at least they should use cooper, damn...) can't help things much. I tried twice and twice the mobo go to silicon heaven. Never more. Ever.
And yep, Epox have far more caps, witch is the reason why Samxon GA ones would rule the OC there. I simply need the best ones ever. Since I have onyl there of them. Period.

OTOH, more caps = the ESL is increasing too! (when you put caps in parallel, their capacity increase. Their ESR (resistance) descrease as with resistors. However since the damn current pass now thrugh all of them, the ESL (inductance) is INCREASING!)
So, the more caps = the better is not always true thing.
The question is, however, how much does the ESL play a role there. I heard a lucky user with Osciloscope that complain about the DFI boards that generate too much i-frequency noise into the voltage. He fixed it by adding small keramic caps in parallel to these 3 ones, tough. Ripple is no longer measurable.
Might be, that the ESL could also decrease the hi-frequency ripple a little, because hi-frequency have diffucities to pass through inductance, even low, so...
I could get only hand on Panasonic FM ones, sadly. Rubycon MCZ will be better and Samxon GA superior, tough

I check the thread, looked funny at first and the user is having similar problems - you see, the damn Chemi-con KZG caps are going to hell for everyone...
I heard suggestion that they maybe stole just a part of the procedure to make the electrolyte, witch is why the caps are dying...

[willawake]

Re: low or super low ESR + impedance caps

I check the thread, looked funny at first and the user is having similar problems - you see, the damn Chemi-con KZG caps are going to hell for everyone...
I heard suggestion that they maybe stole just a part of the procedure to make the electrolyte, witch is why the caps are dying...

KZG is working fine for most people. stolen electrolyte formula is old news. if KZG was bad electrolyte we would see many reports of fails like nichicon HN HM. where is the many reports of fails? they are one of the most popular caps on boards these days.

[trodas]

Re: low or super low ESR + impedance caps

willawake - given my experience I can't agree with you Given the fact that over 10 users of DFI LP/Infinity boards complain about the very same thing - in time the max. stable speed decreasing - it become pretty obvious that the caps are the culprit. Yes, they have about 1 and something lifetime, perhaps even 2 years when user is not heawy overclocker, however they does "wearing out" pretty badly.
The stolen formula is just what I heard. And my opinion on the KZG line is just my opinion, not universal truth or something like this. I just say my experience and obviously, I never ever use Chemi-con again. Ever.
And why there are no reports? Because overclockers sold their mobos and bough K8 ones and don't care. And the others? They just accepting the secondhand mobo and think that they perhaps miss the there stages phase cooling to get where the overclocker was previously - yes true, but the caps play their role as well.
Like I say, I could deliver you more that 10 persons with the same problem. Would you take things then seriously?
(okay, I probably can't - most of them did not register and post there, jsut because a) their english is even worser that mine (or so they say) and b) they know what went wrong and did not feel the need to share this with the world, so... or c) they are too lazy/bussy to do it - SAE could be there as prime example)

Last Viking suggested this:



He added the caps from back of the mobo in parallel to the existing ones. Sure it could help the mediocre Chemi-con KZG ones, however it is not a perfect (nor a usable) solution. I believe he could gain much better results by using Samxon GA caps, witch are far, far better the the KZG ones (see the table) and also maybe adding caps where they aren't anymore.

So, when I finally (3 weeks!!!) get my Panasonic FM caps from customs (help!) I started my project. Basicaly, I use the Last Viking recommendations and come up with this:



So, as you can see, I added a lot's of caps there. It looked amazing. However the mobo refused to post, always ending at diagnostic LED3 only on, witch is "Testing memory presence". So, since not even long cleaning of CMOS helping there, I pulled out the 470uF cap 11. It passed now the testing memory presence and end up at LED4, witch is "Early program chipset register before POST". In short - VGA is not present/inicialized. So, I pulled out caps 16, 17, 18 and 10 and it finally posted! Hoooray! Everything seems working well, untill I inserted Audigy and FX5600 card. Then 4LOD and nothing more - time for another 30min cleaning procedure
So, I keep pulling out caps that I added there, but nothing improved. I ended up in mobo w/o any extra caps, but I still cannot insert any more power hungry deviced to the PCI / AGP Looks like the mosfets there or something get damaged as well?
The 12V support power connector get a bit yellow from the power that goes there when things failed and since my PNY 6800GT and Audigy2zs working well (I'm using them right now) then I have to suspect something wrong up there into the powering part...
Other than that - the mobo is EXTREMLY stable and working well right now folding for me - with very poor oldie Barton witch has "wall" at 2200Mhz that not even 2.00Vcore can break - so the mobo running right now at 200x10.5 - 11-2-2-2-9-12-0-0-3-3-2-0-3-E-E-F-E ultra DFI timings and entierly stable. I installed WinXP on it and folding 100% CPU load and no crash or problem are there.
So - sucesfully recaped.

However there is still a large problem with powering of the slots, so any suggestions where to look and what to measure, yet, I'm affraid I have first remove the heatsinsks from the mosfets that powering the slots:
http://ax2.old-cans.com/s.php?p=badt...4&c=8&d=1&v=v2
As you can see, I got two of them covered...


PS. as you can notice, I omited replacing the two 1000uF caps bellow the Vdd / Vdimm mosfets as Last Viking suggested with bigger ones. Basicaly because they are in line with AGP GFX card and therefore I can't add big caps there...

PS2. some caps are definitively not need to be pulled out for sure. I measured the connections and cap 10 is directly connected to the ground AND to the power imput connector, so pulling it out aren't necessary. Same goes for the caps 12, 14 and 15, however not the cap 13 - this one is NOT directly connected to any of the pins at power connector close to him, so I'm not sure about it. Same goes for the rest of the caps, even it should be good add them. All added caps was Panasonic FM ones.

[willawake]

Re: low or super low ESR + impedance caps

Given the fact that over 10 users of DFI LP/Infinity boards complain about the very same thing

if you give me 10 other users of various non dfi board with blown KZG i believe you. otherwise suggests a problem with dfi design not the caps. anyway keep us up to date with the progress. if you dont nuke the FM also later i will believe you.

[linuxguru]

Re: low or super low ESR + impedance caps

Trodas, you need to read a book on basic electric circuit theory (or do some courses on the subject) before you come here and spout off your half-baked nonsense. Both davmax and I tried to help you understand the subject, but you're impervious to good science and engineering.

Firstly, there isn't anything called 'magnetic resistance', so I believe you're talking about Reluctance. Reluctance is loss-less.

Secondly, magnetic properties of all materials (including ferrites) are temperature dependent. Do a Google search for 'Curie Temperature'. There are numerous other temperature-dependent effects on ferrites in particular, including the effect of temperature-cycling on the binding-material, which can increase or decrease the material spacing of the ferrite particles.

Thirdly, this distinction of 'permanent' and 'non-permanent' magnetic materials is for primary-school students. In reality, all magnetic materials have components of both - that's what the B-H curve is all about. The magnetization H never drops to zero completely even when the Magnetic Flux Density B returns to zero, in all magnetic materials. One of the consequences of this hysteresis is that there's always a small energy loss in every magnetization/demagnetization cycle for any magnetic material - the only 'material' that doesn't have this free space or vacuum.. The best tranformer ferrites attempt to minimize core losses by keeping the shape of the B-H curve like a thin S.

Finally, ESL is a property of the capacitor/leads, and doesn't change for each capacitor regardless of the circuit topology (it is very dependent on the lead lengths, however). When you use multiple capacitors in parallel, the series inductance of the parallel combination decreases. Your claim that it increases is plain bullshit.

[larrymoencurly]

Re: low or super low ESR + impedance caps

I doubt it. ESR too low? Hmmm. That would only mean the circuit is designed to use the limited current from normal caps. To put it simple, the normal cap discharge longer (because he can produce only less current that the low ESR one and therefore it took longer to discharge) and one can think about way what this could be usefull, however the ESR in normal caps aren't too precise thing - so when one want discharge cap slower, then one use a resistor in serial - witch negate the ESR cap value. (quietly assuming that the resistor will have several times higher resistance that the cap, so the cap resistance will play neglecting role there)
But since there is only two infinite things in the world - the universe and human stupidity and I'm not sure about the first, then there should exist circuit, where the ESR of a cap could play a role. Could you show us the circuit?

No. I asked this question last year in the newsgroups in this thread: http://tinyurl.com/mxfbg and two responders said that it could matter in some circuits. One of them, Budgie, said:

"I have found instances of a simple switchmode power supply that used standard electros on the input. Replacing with low ESR type actually increased the switching frequency spikes on the output."

The other person, Bob Parker, designer of a very popular ESR meter, said:

Some small switchmode power supply chips want the filter cap ESR to be between certain limits, but they're not very common. Generally there shouldn't be a problem with new caps having lower ESR than the originals.

[davmax]

Re: low or super low ESR + impedance caps

Trodas. Linuxguru very carefully explained earlier that there are core losses. Let's visit it again. The inductance (not to be confused with inductor, the component) itself is lossless but the following losses do occur in an inductor:

1: ESR. You measured 0.1 ohm. Looking at most of the VRM inductors I would say that reading is high. If it is correct that could mean 1.5 watts ESR loss in each inductor for 15 amps DC.

2. Hysteresis loss adequately defined by Linuxguru. The higher the magnetic flux excursion, the higher the loss.

3. Eddy current loss. With ferrite cores this does not exist because it requires conductive material in the core and ferrites are ceramic.



Take a look at this link that briefly deals with the inductor in a switched mode power supply.



http://www.mag-inc.com/software/inductor.asp



You see from the picture that the ripple voltage is the smaller component flowing through the inductor. The largest component is the DC current which would normally be about 15 amps for a three inductor design. So from this it can be seen that the DC component plus the RMS ripple current will generate heat in the ESR. The ripple voltage being small will not cause large magnetic excursions and therefore not large hysteresis loss or heat.



You debunk theory and say in practice it is very different because the inductors are hot. In fact the heat comes from several sources.

• The losses defined above that are not very high.
• From the heat in the motherboard ground/power planes. These and the inductor coils are copper that conducts heat very quickly.

It is therefore highly likely that most of the heat is coming from the heat of other devices such as MOSFET's through the copper planes. This can be checked with an infra red thermometer. Even so it can be very difficult to establish exactly where the heat found in the inductor is being generated. Have you investigated the temperature source with an IR thermometer? Real solid facts are required before risking misleading others.



I see that Linuxguru has already dealt with your ESL bullshit.



We try to give you good sound advice. You do not seem to understand and from what you have written it is not due to language, it is your insufficient knowledge of how these things actually work. Understanding theory will get the right result. That is how electronic systems have to be designed, with theory and then fine adjustment in the prototype stage.

[davmax]

Re: low or super low ESR + impedance caps

I made a real bad mistake in ESR loss. For 15 amp current and 0.1 ohm resistance the loss is 22.5watts (IxIxR). It would get real hot with that ESR loss. So maybe the heat in the inductor is basically ESR loss. Oh this assumes we have a processor drawing 45 amps through three inductors. This again is not intended to represent any particular processor, just an example of ESR loss.

[PeteS in CA]

Re: low or super low ESR + impedance caps

Here's another webpage dealing with losses in inductors. The author isn't identified, but if it's who I think it is, he's been a magnetics engineer for over 25 years that I know of. Magnetics Inc. (the site davmax linked) and Micrometals are leaders in their fields of magnetic core design.

BTW, the page I linked deals with one of the things, core aging, that makes me nervous about VRM O/P inductors and O/P inductors in cheap P/Ss. If the mfr picks the wrong core or their purchasing buys a cheap substitute, the cores will have a very short life, and the O/P capacitors are going to get cooked (as, very quickly, will replacement caps).

[davmax]

Re: low or super low ESR + impedance caps

Good input PeteS. The question is. Are the toriods in these inductors powdered iron? Or. Are they ferrite? Ferrite has a problem if the operating temp gets up to about 130C ( the curie point). Ageing and changing of inductance with temp are not severe.

Can anyone identify what toroid core material is used?

Take a look at this link on page 5 and also the characteristics of some of the different ferrite materials.

http://www.tdk.co.jp/tefe02/e130.pdf

[davmax]

Re: low or super low ESR + impedance caps

After consideration I believe that I can possibly answer my own question.

It is most likely that VRM inductors are powdered iron, because this material has the much higher saturation flux density needed where large DC currents are involved.

I also thought that PeteS would not have offered the link to a powdered iron cores if it were not relevent.

Powdered iron cores do have eddy current losses where ferrite does not.

[PeteS in CA]

Re: low or super low ESR + impedance caps

Ferrite isn't usually used for O/P inductors, for the reason you cited. An exception would be gapped E-cores and planar E-cores. Discerning core vendors and material types isn't easy. They usually aren't labelled (Mag Inc. MPP cores are an exception, but are too pricey for PC use). Micrometals uses the paint colors to identify the material type, but this is easily duplicated/counterfeited.

[gonzo0815]

Re: low or super low ESR + impedance caps

At the moment most VRm and PSU circuits are using powdered iron cores. The main reason ist due to their slow and well controllable saturation point and the distributed air gap.If frequencies will futher increase well beyound the current 300khz, the ferit cores kick in. At those low frequencies there is no way to use ferite.
i think trodas, you can belive what there is stated, noting wrong.
If anyone will see, what mobo manufacturer has done with the VRM circuit, lock at the controler chip an google for the datasheet. I `ve read the datasheet of st`s L6917D, as for locking which replacement caps are suited for my k7s8xe+.
It showed me klearely, that Asrock saved some buck on caps, regardles they are using good ones. In that datasheet are al things explained with only the needed equations.