Capacitor Mods 101

Quote:

---

Originally Posted by LostInSpace278

I think the caps you are refering too are like these

I did a google search and there are many types of tantalum caps, most of the varieties you see used on mobos. I am just making sure I am following correctly.

---

Umm... no I think your lost in space :D .

If you're refering to the small brown surface mount caps on the backside of the motherboard under the CPU, those are just cheap surface mount ceramic capacitors and probably have a voltage rating of at least 50 volts probably much much higher so you dont need to worry about them.

I'm talking about surface mount tantalum caps that are found sometimes in the center of the CPU socket on 478 socket P4's. They are black and one side has a line acrossed it or is white to signify postive.

I'm really not too sure why you would want to put 2 volts or more into a P4 anyways.

Well not sure if I would more than 2v into a P4 but mine is an 2600+ mobile. I am running air and have already put over 2v into it. For a brief period of time any way.

I chose that picture, because when I did a google search for tantalum caps, I found a site that had the ones you refered to above as well as the orange looking caps. I do know the caps you talking about, next time I have my cpu out, I'll take a gander in the socket. Guess I had better do some more research first.

usually just ceramic caps on AXP boards.\

SPL15, I have been wondering if the reason nF2 cannot do high FSB could be related to the fact the northbridge is fed both by CPU voltage and by the VDD power circuit, but one or the other is not supplying clean enough voltage to it? Would a tantalum cap in parallel with the stock electrolytic caps (6.3v 1500uf x1 from cpu vcore supply, 6.3v 1000uf x3 from vdd) be worth trying maybe?

Quote:

---

Originally Posted by STEvil

usually just ceramic caps on AXP boards.\

SPL15, I have been wondering if the reason nF2 cannot do high FSB could be related to the fact the northbridge is fed both by CPU voltage and by the VDD power circuit, but one or the other is not supplying clean enough voltage to it? Would a tantalum cap in parallel with the stock electrolytic caps (6.3v 1500uf x1 from cpu vcore supply, 6.3v 1000uf x3 from vdd) be worth trying maybe?

---

I guess you could try it, but no cap regardless of quality can fix design flaws/Limitations of a PCB, Chipset, CPU, Power Supply, Layout, Software, etc.

Capacitors aren't bandaids and shouldn't be used like one. If your getting astronomical gains from adding caps, then I would move to different hardware because that clearly shows a really crap design. Also, if the filtering was that bad to begin with, it is likely that you'd see problems at stock settings.

Now caps do deteriorate over time especially with switchmode power supplies so if your caps are going downhill, then new caps would make a difference. Hopefully your motherboard maker is using good caps...... Abit

I'm thinking the NF2 doesnt do ultra high FSB is because of an Architecture reason, not because of noise and ripple on voltage lines. The AMD platform is not known for it's headroom like Intel is. NVidia designs a chipset that is good enough and bit more because that is what AMD tells them to do. Intel designs one that has massive headroom because they design for rock solid stability. Notice that VIA and SiS Chipsets for Intel dont do as well as Intel's chipsets. kinda the same thing with AMD and their chipset makers. You care the most about yourself, so if your building yourself a chipset, your going to make it a good one.

Also It may not even be the chipset, it could very well be that the Athlon XP 's architecture CPU wise doesnt handle higher FSB's well. The NF2 chipset may do 2.6 bagigaherts 110% stable, but if the CPU's FSB interface can only handle 420Mhz, then that is the limiting factor.

people have had AXP's to 300mhz but the northbridge cant handle it. nVidia's nF2 chipset does ok on 2x256 but once you throw 2x512 on there its all downhill. Bumping up VDD can help sometimes, more for some than others.

We already know its the chipset primarily, but I am wondering if the voltage regulation isnt part of that "build it good enough to work and dont worry about overhead" since it wouldnt be the first time nVidia or an AMD board was built like that.

Guess i'll have to see what i've got around for tantalum caps maybe.

Quote:

---

Originally Posted by STEvil

people have had AXP's to 300mhz but the northbridge cant handle it. nVidia's nF2 chipset does ok on 2x256 but once you throw 2x512 on there its all downhill. Bumping up VDD can help sometimes, more for some than others.

We already know its the chipset primarily, but I am wondering if the voltage regulation isnt part of that "build it good enough to work and dont worry about overhead" since it wouldnt be the first time nVidia or an AMD board was built like that.

Guess i'll have to see what i've got around for tantalum caps maybe.

---

If voltage regulation problems on NF2 boards ARE in fact the reason for mediocre FSB (Although I'd say 300Mhz is dam good) Caps arent going to solve the problem. Again, Caps dont make power, they are helping to filter the DC pulses from the Switching mosfets. The biggest component of the filtering is the output choke. In series, a choke acts like a capacitor in parallel to the load so you dont want to go real low with its value, BUT a choke is also a series resistance, so you dont wanna go to high. TO the power supply, a choke in series with the load acts as a high impedance to high frequency while a cap in parallel to the load acts as a short to high frequency. Do you see how they are both acting as low pass filters??

Ya never know though, a cap may gain you a few extra MHz, but I would not expect anything large from it.

Quote:

---

Originally Posted by SPL15

If voltage regulation problems on NF2 boards ARE in fact the reason for mediocre FSB (Although I'd say 300Mhz is dam good) Caps arent going to solve the problem. Again, Caps dont make power, they are helping to filter the DC pulses from the Switching mosfets. The biggest component of the filtering is the output choke. In series, a choke acts like a capacitor in parallel to the load so you dont want to go real low with its value, BUT a choke is also a series resistance, so you dont wanna go to high. TO the power supply, a choke in series with the load acts as a high impedance to high frequency while a cap in parallel to the load acts as a short to high frequency. Do you see how they are both acting as low pass filters??

Ya never know though, a cap may gain you a few extra MHz, but I would not expect anything large from it.

---

Hmm.. 300mhz is **rare**, but its known that the CPU isnt the problem, or, shouldnt be atleast... My 2600+M has handles 267fsb form my infinity, but my an7 has voltage issues and cant maintain 220 stable.

The outlet im on isnt quite the best, so Part of my problem may be jumping voltage... feels like 220 can be stable some times, isnt others... 1.65vdd->1.75Vdd does nothing..

I see what ya mean by a cap filtering instead of holding power-- AA's can't power a monitor for long :p:

I really wish Nvidia would give datasheets for the NFII stuff. :toast:

Quote:

---

Originally Posted by STEvil

We already know its the chipset primarily

---

If it's the chipset, then why does just about every DFI do 250 fsb or more whilest NF7's more often then not don't make that number stabely.

Quote:

---

Originally Posted by kristos

If it's the chipset, then why does just about every DFI do 250 fsb or more whilest NF7's more often then not don't make that number stabely.

---

PCB layout... SMD selection, BIOS?

Or just maybe special NB's? :p:

That just proves it IS the chipset--

DFI didnt engineer the LP and infinity like an AN7 or NF7.

Doing a couple mods similar to this seems to help the Abit IC7 series, so just speculating really.

http://www.xtremesystems.org/forums/...ad.php?t=28553

How do I decipher the text on SMD tantalum capacitors? Got some that say "226c p5 1xc" on them but dont know what everything stands for and google hasnt turned up much yet.

Quote:

---

Originally Posted by STEvil

How do I decipher the text on SMD tantalum capacitors? Got some that say "226c p5 1xc" on them but dont know what everything stands for and google hasnt turned up much yet.

---

Different manufacturers have different codes. I have a hard time deciphering them too, so usually I dont :D . My guess would be that it is a 22uF at 6 volts and the p5 1xc has to do with the size and series of cap. Dont quote me on this though.

A lot of multimeters have a built in capacitance meter which will tell you the value, which will help you narrow down what number will be the voltage.

Doesnt seem any of my multimeters have a capacitor tester setting on them.. hm.

The only other way, that I can think of off the top of my head, to determine the capacitance is to charge the cap to a specific voltage and see how long it takes the capacitor to discharge across a resistor.

I take it this is a capacitor that is already on a board and not something just laying around.

In any case, we can prob still that little experiment and determine the capacitance if ya really wanna. Not a hard thing to do.

Quote:

---

Originally Posted by Hell-Fire

The only other way, that I can think of off the top of my head, to determine the capacitance is to charge the cap to a specific voltage and see how long it takes the capacitor to discharge across a resistor.

I take it this is a capacitor that is already on a board and not something just laying around.

In any case, we can prob still that little experiment and determine the capacitance if ya really wanna. Not a hard thing to do.

---

That is a way to get a rough estimate on capacitance, but you'll get a lot of error because there is a huge variable in time (Using a stopwatch) and when the cap is "Discharged" (How many RC tmie constants??). Also Dielectric absorbtion will have a big impact on the measurement with this testing method.

The best way is to apply a Known AC signal to the cap in series with a variable resistor. Adjust the Pot until the voltage across the cap is equal to the voltage across the pot. The resistance of the pot will be equal to the Impedance of the cap at that particular frequency when votage is equally divided between the two. If you know the frequency and the Capacitive impedance, you can calculate capacitance. This is the same method used by real capacitance meters and is very accurate. In cap meters they use a fixed resistor instead of a variable one and a high frequency.

The best way though is to just get a meter with a capacitance measurement

Ever since I implemented a droop mod with minimal gains I've been toying with the idea of doing a CAPS mod. I've located capicitors with the same uf and voltage, but I'm not sure how to judge them on other aspects. Could somebody tell me if these are good for use on mobos. I ordered them at digikey.com. Thanks. Peace!

Part#: P11206-ND
digikey.com

Quote:

---

Originally Posted by yasoumalaka

Ever since I implemented a droop mod with minimal gains I've been toying with the idea of doing a CAPS mod. I've located capicitors with the same uf and voltage, but I'm not sure how to judge them on other aspects. Could somebody tell me if these are good for use on mobos. I ordered them at digikey.com. Thanks. Peace!

Part#: P11206-ND
digikey.com

---

Be weary of adding to much capacitance to any circuit loop. The more capacitance you add the slower that cap will charge and the higher the current will be for that loop for a longer period of time versus the stock conditions. If you are adding caps that equal the stock, you should be fine as many people have done this with success.

As for the rated voltage, as long as its rating is above the voltage that will be applied to it all will be right with the world.

Quote:

---

Originally Posted by SPL15

That is a way to get a rough estimate on capacitance, but you'll get a lot of error because there is a huge variable in time (Using a stopwatch) and when the cap is "Discharged" (How many RC tmie constants??). Also Dielectric absorbtion will have a big impact on the measurement with this testing method.

The best way is to apply a Known AC signal to the cap in series with a variable resistor. Adjust the Pot until the voltage across the cap is equal to the voltage across the pot. The resistance of the pot will be equal to the Impedance of the cap at that particular frequency when votage is equally divided between the two. If you know the frequency and the Capacitive impedance, you can calculate capacitance. This is the same method used by real capacitance meters and is very accurate. In cap meters they use a fixed resistor instead of a variable one and a high frequency.

The best way though is to just get a meter with a capacitance measurement

---

SPL15, there is no argueing your knowledge in this field, but some of these suggestions are beyond the scope of what most users here can do. Thats not saying our members are not intelligent enough to do them, but we need to provide the methods for them to do it for themselves. Asking them to start using AC signals across a circuit loop in order to estimate the capacitance is just not feasible for most I would think. This may be Electrical Engineering 101 I know, but most members here are not electrical engineers.

I would agree that using a multimeter is a way to go, but only if someone already has one, or plans to buy a new one. Our job (mine, yours, others who understand it), is to provide them ways to do simple experiments in order to figure some of this stuff out for themselves. That experiment is a very handy way to figure out a solid estimate for capacitance in conditions where it is not readily marked on the actual capacitor.

If you charge the cap to a fairly high level (5v or more), and have a multimeter that tracks to atleast 0.001, and you use a fairly large resistor, the discharge will be slow enough to get a good estimate of the capacitance. You can get to within 1% of the rated capacitance if you do the experiment correctly. I would say 3-5 runs is more than sufficient. Generating that number of RC constants will yield a solid estimate.

Unless you have a function generator, forget about applying a sinusoidal waveform...and I figure most people do not have these in their homes. And most people will not have access to these at school either.

If you have the true resistance of your series resistor, not rated, and do the experiment properly you will get a very good estimate of the capacitance. Not to mention very few capacitors are truly accurate to their rated capacitance to begin with.

Man oh man do we need a "Xtremely Techinal" forum as this topic is above most users knowledge.

Are those caps of the right quality.

Quote:

---

Originally Posted by Hell-Fire

Man oh man do we need a "Xtremely Techinal" forum as this topic is above most users knowledge.

---

Ha Ya I'm with ya there. I understand that alot of what I say is pretty technical, but from my viewpoint, when you do a mod, you should know what it is that it is doing. Maybe not exactly specifically what it is doing, but a general idea of what is happening so they can better troubleshoot problems that arise other than wasting server space with a thread that reads "I F'd up my board doing a mod and all the solder pads are burned to hell, Can someone tell me what I did wrong and help me fix this irrepairable board I screwed up because I rushed into something I had no knowledge about?"

Even if the people here can not do the test I described above, it is a good example of how a capacitor opposes AC frequency. I really doubt most people have the knowledge here to really grasp everything I'm saying but hopefully the confusion of what I say inspires them to learn more and experiment on their own to gain knowledge. I really can't stand it when people do a mod just because it supposedly gains performance and then see pics of the mod and it has so many things wrong that it renders the mod useless. The more knowledge the users here have, the better the quality of information.

I know for a fact that there are users here that want to know the in depth details of how the electronics are integrating with each other, but so far they kinda have to rely on stumbling across a rare good post. I think a dedicated area where highly indepth postings is a really good idea, but it would need to be highly moderated to insure that posts that read "Where's the link to the Vdroop mod for the P4C800 at?" will not be tolerated. It should be dealing more with the electronic theory as it pertains to computers, and not procedural topics on how to do a mod.

Look at the Phase change forum, I would say that the info there is pretty dang technical and I personally have learned just about everything I know about phase change from those forums. I now have the knowledge to do anything and everything that would be needed to set up a single stage system. Highly technical information is vital to the quality of a forums usefulness.

How do I know if a cap has a low ESR?

Quote:

---

Originally Posted by yasoumalaka

How do I know if a cap has a low ESR?

---

As far as I can recall, the ESR for a capacitor is based on the internal resistance of the electrolytic used (as well as the plates etc etc), and the very small resistance of the connecting external legs. This forms the caps Equivalent Series Resistance (ESR).

The ESR is very low for a good capacitor, and only diminishes over periods of years....or major wear and tear. There are many other things that affect the ESR, but not a topic to talk about now. One of the major things however, is temperature. High temps can either damage the rubber seal or begin drying the electrolytic out. This causes the ESR to increase. Circuits are setup with specific resistances in mind. So, if a cap initially has a very low resistance, it will not affect the circuit. However, if the ESR begins to rise in the cap, that will affect the charging time of the cap as well as its ability to subdue AC signals. What you end up with is the resistane already present in the circuit, as well as a "series resistor" inside the capacitor. Basically it cant do its job properly. Not to mention caps that take abuse over extended periods can actually explode...literally.

Power supply caps are prone to this type of failure as they have to endure large ripple currents which heats the electrolytic beyond its tolerance and it begins to break down. Thats why 6 outta 10 psu failures are due to blown caps.

Unless I am mistaken, just check the resistance between the pins of the cap.

Maybe SPL can verify what I said above.


You could not be more right SPL.

Hopefully some members will take it upon themselves to have a look around the net and their locl book store. There is no replacement for knowledge...period.

The first step in being able to figure out mods on your own, and troubleshooting mods that arent working, is know wth is going on behind the scenes. Something as simple as how resistors/caps respond once in series and parallel is lost on many people here.

Soon as I get some free-time I will talk with FUGGER to see about setting up a technical forum here. Cough:: throw your hat in for Modship ::

I agree that tech replies may force some members to do the right thing and grab some knowledge, but I would wager most just follow guides and close their eyes to learning what is truly making that wonderful mod tick.

In any event, I for on thank you for your expertise and time invested here....so please dont think I was giving you a hard time.

Quote:

---

Originally Posted by yasoumalaka

How do I know if a cap has a low ESR?

---

You have to measure the ESR using a special meter designed for reading the ESR. You might be able to find info about a caps ESR by looking at the caps datasheets, but that would only tell you the ESR that the cap is designed to have, and not its true value after hours of (ab)use.

sometimes they say "Low ESR" right on them, although its not a very technical description of the actual esr rating.. ;)

I attend UF so I could take classes in circuitry but I'd rather just read a book. Do any of you guys have suggestions? I would really like to learn a few things before my next build. That way I could mod it from the jump.

Quote:

---

Originally Posted by yasoumalaka

I attend UF so I could take classes in circuitry but I'd rather just read a book. Do any of you guys have suggestions? I would really like to learn a few things before my next build. That way I could mod it from the jump.

---

www.google.com

Google knows all. :banana4:

Added the SMD tantalum to the board, didnt help any. Even tried feeding the northbridge 2.6v and still nothin'... argh.

nForce2 have to be one of the most aggrivating chipsets there is.

EDIT

considering putting the spare SMD tantalums off my dead 9700np onto my living 9700 pro now.. :p: