Capacitor Mods 101

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.

---

Books are great, but there is no substitute for in class instruction. throughout my schooling I learned absolutely no new concepts i didnt know about before, but class was invaluable in tightening up my grasp on things that were a little fuzzy. Also the hands on experiance of class is more helpful than any book for people with no background in electronics. People say the stuff I post on these forums is technical but compared to some of my text books from college, my posts are as easy to read as "The Cat in the Hat" by Dr. Seuss.

If you can afford it, I would highly recommend classes, who knows you may find a career out of it (a very high paying career too :D )

Quote:

---

Originally Posted by SPL15

who knows you may find a career out of it (a very high paying career too :D )

---

Is there that much money to be made in electronics (in the US)? :D

Not that that should be your main reason, if any, for getting into electronics, I'm just inquiring :)

Quote:

---

Originally Posted by kristos

Is there that much money to be made in electronics (in the US)? :D

Not that that should be your main reason, if any, for getting into electronics, I'm just inquiring :)

---

Depends on where you are. In my area, there are few talented electronics engineers and few good electronics technicians. Means good money for people that are talented around here, but also means companies are more likely move to somewhere there is lots of cheap talent..... like China

I just did a caps mod on my DFI lanparty. I added 13 caps in total and when I went to fire it up I got a quick flash from my led fans and that was it so after messing around with various wires and stuff it still didn't work so I startesd to take of caps. The first two I took off where right next to the 2 12v lines going into the board after this it worked. This left me with 11 caps. When I look at tutorials on caps mods on the 478 they only use 11 caps is this strange or is there something behind it. I thought maybe one of you might have the answer. Peace!

Quote:

---

Originally Posted by yasoumalaka

I just did a caps mod on my DFI lanparty. I added 13 caps in total and when I went to fire it up I got a quick flash from my led fans and that was it so after messing around with various wires and stuff it still didn't work so I startesd to take of caps. The first two I took off where right next to the 2 12v lines going into the board after this it worked. This left me with 11 caps. When I look at tutorials on caps mods on the 478 they only use 11 caps is this strange or is there something behind it. I thought maybe one of you might have the answer. Peace!

---

It's just a coincidence, there's nothing significant with the number 11 except that it is one more than 10 and one less than 12 :D .

The reason your board wouldnt power up could be caused by several things or a combination. Your power supply has a current limit. If it is sensing a massive current rush when you power up, it will shut down in order to save itself and connected equipment. Also, you've got to charge those caps, if your power hasnt stabilized and fully charged the capacitors in say 300mS (Spec'd for PCP&C 510) the "Power Good" (AKA PG) signal will not go to 5 volts and the motherboard will unlatch the "Power On" signal to the power supply which stops the oscillator in the PWM IC. The "Power Good" signal from the power supply releases the motherboard's internal reset latch thus starting the boot-up process. There is a Monitor IC in every AT and ATX power supply that monitors voltage/current conditions. If something is not kosher, this Monitor IC will stop the PWM IC from PWMing and drop out the "Power Good" signal. If you look closely I bet you'd find a 16 DIP IC that is manufacturerd by ASTEC in a whole poop ton of power supplies. This will be the monitor IC.

Also, your motherboard could be sensing an out of spec voltage drop after 250 mS (Industry standard) which is a function of current draw. More capacitance creates a larger initial current draw when you power up, If the motherboard monitoring hardware senses that there is a large voltage drop after the predetermined stabilization period (250mS), it will unlatch the "Power On" signal that goes to the power supply (Green lead from power supply). Larger capacitances take longer to charge up due to voltage drops across parasitic resistances in the power loop which effectively lengthens the RC time constant which determines how fast a capacitance charges. If the world was perfect you could magically get rid of all the parasitic resistances in your system and you'd have no Vcore droop, Vdimm Drop, no voltage drops period and much less heat as long as your power supply could provide the amperage needed which woldnt be hard since the power supply would be 100% efficient which means no heat.

I'm glad we dont live in a perfect world though cause I'd be out of a job and you guys would be out of a hobby :D .

After playing around with the bios the highest stable clock I'm able to achieve is 3.82. Does anybody think it would help to try CAPs with different uf.

Quote:

---

Originally Posted by yasoumalaka

After playing around with the bios the highest stable clock I'm able to achieve is 3.82. Does anybody think it would help to try CAPs with different uf.

---

After a certain point, no matter how many caps you use, you will start to reach a point where you will see no benefit from adding more caps, and you might even start to see a decrease in performance. You might have reached this limit if you are getting boot problems with 13 capacitors. Adding capacitors will never be able to make up for a power supply, it will just help to improve it slightly. Are you sure that it is the Vcore regulator which is limiting the overclock and not some other factor?

No I'm not sure, but I know others have had sucess with different combimations of capacitors. I solved the boot problem. My overclock is still relatively weak for this processor, mem, moded mobo and psu. In addition I have a processor that nebver gets over 44c. I'm thinking of posting my bios screens somewhere. Maybe I've got a setting wrong somewhere. After ruling everything out I migh just see if the proc can handle 1.5v or more and whether this helps. Thanks for any suggestions. Peace!

You could try putting some smaller value caps (tantalum caps maybe) as close to the CPU's Vcore pins as possible.

Quote:

---

Originally Posted by yasoumalaka

After playing around with the bios the highest stable clock I'm able to achieve is 3.82. Does anybody think it would help to try CAPs with different uf.

---

Capacitors are not magic band-aids that fix limitations in your system. If your CPU is running at its max at a particular temp/voltage, caps will do little if anything. Spend your money lowering temps, not increasing capacitance.

3.82Ghz is not bad if your running air. I'm assuming its a P4 since 3.82 for AMD would be pretty ridiculous.

Until you start getting into high bus/CPU speeds, I wouldnt bother with voltage mods and capacitor mods.

Here's an analogy, Doing major mods to a motherboard running a relatively low bus would be like putting a full ultra performance race suspension on an Acura Integra with stock rims/tires and a motor putting out stock horses for an 80 year old lady. It doesnt make much sense to spend all that money/time on an ultra performance suspension when your performance is being limited by a lack of tire traction/sidewall flex and a puny 110 Horsepower engine rated at the flywheel AND a driver who doesnt quite know how to manipulate the car to bring out performance..... Also if that ultra performance suspension is not tuned correctly, you'll have worse performance than the stock suspension and could very well cause a deadly accident. Slapping on a performance suspension will not make up for limitations in other parts of the car and if not tuned right, it can cause major stability problems.

SPL15 - first - thanks for very very informative posts there, and keep them up. I have only one correction. When you elaborated the chils heating-up, you mentioned that the heat is generated by the resistance and impedance thing. There is also 3rd factor, and that is the magnetic resistance of the used material - versus used frequency. It could not be underestiminated, because these coils heat up to about 60-70C pretty fast after start, so... BTW, most of the ferite materials have the peak performance (minimal amount of energy need to flip the magnetic field) at around 100C, witch is very bad for the mobo components (mainly caps) anyway :lol: The magnetic resistance rised over 100C relatively very quickly, so that is why they are designed to run at about 60-75C in most cases. When mine hit 126C, I know that there is something very very bad :(
But after all, this just support your warnings about number of heat dissipiated out there :)

Second:

Quote:

---

Capacitors are not magic band-aids that fix limitations in your system.

---

This should be carve to rock and people should be slaped with this rock frequently. Including myself.

After years of OC and heavy HW modding, I lost only there components. Epox 8RDA+, eVGA FX5700U and DFI LP B. All there was lost by trying to do capacitor mod in order to enhance the voltage regulator performance (Epox, eVGA) or decrease the produced heat out of the mosfets...
All there attempts failed. Except eVGA, witch refuse to even start, after adding two caps on the Vgpu rail in order to ease the voltage regulator suppliing 1.5Vgpu for 610Mhz, the two mobos worked. Epox for 20minutes and DFI LP B for about 6 hours. Then both died.

I will try to resuscitate them, tough, but the chances are IMHO slim.

I just learned the hard way, that the discharging speed/power of capacitor is the most important thing ever and adding normal quality caps will kill the board almost instantly... :(
I can't buy low ESR caps out there (Czech republic), so Im restricted to what I can get from other dead mobos.

Thanks to SPL15 is also for explaining the coils problem, and why just replacing the coil used on mobo with other, similary looking one with more threads is far from wise thing. Increasing impedance should provide better spikes filtering (the coild on DFI LP B has very few treads - only 5 (!), while old ECS K7S5A (also AXP mobo, SiS 735 chipset) got coils with 10 threads) - but it, of course, also decrease the frequency of the pulses, witch might do more harm that good... :(
And it sounds as such nice idea, to just replace these coils! Even in the docs for the mosfets are mentioned that increasing the impedance in the LC filer part could reduce the mosfets temp up to 30%! ...however this is probably valid only for systems with stable current requirments, witch CPU's simply are not :stick:

Hi SPL15,

I'm curious as to the mobo 'Never before Seen Mods/Tweaks that you speak of. Care to give a hint on what you have done or tried to achieve.

In my opinion, highly technical information is exactly what I want from forums, and what I find most useful. There is waaaay too much mumbo jumbo out there, and a distinct lack of high quality and qualified information, particularly for mobo mods. People scared away by things they dont understand perhaps shouldn't be dabbling in these areas in the first place. Hence I appreciate you input thus far....

Its also interesting how you seem to differ in opinion to pretty much everything suggested by Hipro5 in 'the great IC7 mobo mod' thread, i.e cap/inductor mods. You both exceed my knowledge in these areas so its not appropriate for me to make detailed technical comments, I am more interested in what your opinion of that is, considering the speeds he has attained.

Quote:

---

Originally Posted by the phantom

Hi SPL15,

I'm curious as to the mobo 'Never before Seen Mods/Tweaks that you speak of. Care to give a hint on what you have done or tried to achieve.

In my opinion, highly technical information is exactly what I want from forums, and what I find most useful. There is waaaay too much mumbo jumbo out there, and a distinct lack of high quality and qualified information, particularly for mobo mods. People scared away by things they dont understand perhaps shouldn't be dabbling in these areas in the first place. Hence I appreciate you input thus far....

Its also interesting how you seem to differ in opinion to pretty much everything suggested by Hipro5 in 'the great IC7 mobo mod' thread, i.e cap/inductor mods. You both exceed my knowledge in these areas so its not appropriate for me to make detailed technical comments, I am more interested in what your opinion of that is, considering the speeds he has attained.

---

My mods on my P4C800e-dlx so far are invisible and look 100% stock. I've got the Vdimm and Vdroop mods, but instead of grafting on some wires and pots onto my motherboard, I've replaced the resistors that control these things in the first place and added common sense design ideas to it. The Voltage that comes out of the DC-DC converter is created by a Voltage divider, I've stabilized the voltage divider with small tantulum caps in order to reduce spurous noise transients and voltage sags. But again you cant even tell its there, unless you have another stock P4C800e-dlx to compare it with and a good pair of eyes. I've added caps where they are missing in stock locations, I've rewound the Inductors on the DC-DC converters for the CPU and for the 4pin 12V input line. There are 5 places where ASUS decided they should leave the surface mount tantalum caps out so I added them there. I've added better quality capacitance in places that had just an electrolytic cap. I've basicly refined what ASUS already did, or did cheaply, or decided not to do. There are still many areas where I could do a mod, but it's not worth the hassle anymore on this old platform. There are spaces on the backside of the motherboard for MORE mosfets to help with the CPU Vcore that no one has said anything about....

If you look at Hipro's systems, he uses major voltmods to achieve his great results. Cap and inductor mods dont gain you the kind of results like Vmodding does. The cap mod for the IC7 that puts the cap right at the Northbridge could be LOTS more effective by simply using tantalum caps.

The things I say about electronics I didnt conjecture and didnt discover, the laws of physics determine what I say to be true. The issue's I've brought up on this thread and other threads are very well documented throughout the industry and have proven to be true in application and in theory. These are not new concepts by any way shape or form. Understanding electronics has many depths of knowledge levels. Lots of Engineers can do the "Programming" or design of a circuit, lots less understand the real world component interactions, tolerances, functions, limitations within the actual real world circuit. I've designed amplifiers where just by switching to a tantalum cap in the input stage reduced high frequency distortion at 25Khz by a magnitude of 10000%. They are the same capacitance wise, but FAR different regarding the "other" parameters ignored/not understood by lots of engineers and most design programs. ESR with caps matters depending on application, It takes lots of thorough understanding to know why ESR matters in one situation while raw capacitance matters in another.

OK I'm done for now, it's late and I'm about to get a good long...... foot massage :D .

I have to say, this has been one of the Most Useful posts I have ever read. in the last hour I think I learned more about caps, then I ever have in any other single post.

this area isn't my area to make stickies, but I think this post definatly deserves to be a sticky, cause theres alot of great information and very informed people in this post.

Since I have a lot of power drops in my house, what about adding two 200v 1000uf caps to the main driving circuit of the psu + the original 200v 470uf I've got :confused:

I don't want it to improve anything but the tolerance to power drops...

Another trick I've been using is to put diodes on the 5v and 12v lines then combined choke from an old psu.
Then adding 2200uf 10v cap on the 5v and 3000uf 16v on the 12v.
It protects my hard drives from experiencing any power drains due to swapping some 160GB swappable SCSI drives I have.

i managed afterwards to shortcut 3.5 fuse on another line without any voltage drops in my hard drive lines :toast:

Quote:

---

Originally Posted by TheBlackTulip

Since I have a lot of power drops in my house, what about adding two 200v 1000uf caps to the main driving circuit of the psu + the original 200v 470uf I've got :confused:

I don't want it to improve anything but the tolerance to power drops...

Another trick I've been using is to put diodes on the 5v and 12v lines then combined choke from an old psu.
Then adding 2200uf 10v cap on the 5v and 3000uf 16v on the 12v.
It protects my hard drives from experiencing any power drains due to swapping some 160GB swappable SCSI drives I have.

i managed afterwards to shortcut 3.5 fuse on another line without any voltage drops in my hard drive lines :toast:

---

It wont do much of anything for that situation. Caps are not batteries. 2470uF is a peasely amount of capacitance for power backup. LOTS LOTS more would be needed to make any difference for a power drop out situation while the PSU is under load. I have an amplifier with 1 Farad of capacitance (that's 1,000,000uF) and if there is any load on the amp, it will discharge to an unacceptable level within milliseconds if I shut off the power. Adding caps in your situation would make the mains input rectifier diodes work much much harder and will create a much larger charging current transient which creates a tremendous amount of RFI and EMI which is just plain not good when dealing with high clock frequency digital devices. The only thing adding these caps will do is make your standby 5 volts voltage last a second or two longer when you disconnect the plug with the PSU off. In the worst case, you'll blow the rectifier diodes or you'll get electrocuted by the added capacitance if you ever decide to open the PSU after it was just plugged in.

What you need is a good UPS (Uninterruptible Power Supply) I recommend a higher quality unit that converts the 120/220 VAC into 12/18/24 DC volts and then converts the 12/18/24 volts back to 120/220VAC. This type isolates the load from the main's circuit and the output of the unit is 100% unconditional to the 120/220VAC input. This type filters out any and all harmonics, frequency deviation, spikes, sags, etc. The cheaper ones (That 99.9% of consumers have) connect the 120/220VAC directly to the outputs until it senses bad voltage. This type uses passive filtering (Maybe) and has a wide acceptable voltage variation (90-135vac TYPICAL). Also, the cheapers ones have no frequency variation protection. But in reality, any decent quality UPS is better than none. Especially if your in an area or country with crappy power.

In your filter that you described above, eliminating the Diode will not hurt the LC filter. The diode is taking probably .7 to 1.4 volts away from the rail voltage. If you are using low voltage schottky or germainium Diodes, you may be getting only a .3 volts of drop, but still not exactly good. The Diode on you configuration with that amount of capacitance with a series inductor isnt doing anything helpful, but it IS dropping voltage. Unless your overvolting to compensate for the Diode drop, I'd get rid of them. If you trying to use the diodes to limit voltage spikes from removing the Hard Drives, you'd be better off using transorbs (Transient voltege suppressors) across the lines. a 15 volt 600watt unidirectional transorb across the line (Reverse bias) would solve voltage spike issues, and is what I use personally on lots of electronics i build.

i noticed that the voltage is 11.86 and 5.02 which is not that bad...
i used diodes from psu

Are those readings before or after the diodes?

If your using diodes from a PSU, they are just regular silicon diodes which will have about a .6 to 1.4 volt drop across them. Also, depending on what diodes your using, the current through them may be excessive which will short them out in little time. If your running multiple Hard Drives, your going to need a large TO-220 or TO-245 package with a small heatsink in order to dissipate the heat. Also, start-up current is high with hard drives.

If you are getting small voltage drops across the diodes or no voltage drop at all, your diodes are shorted. I would check them out, and probably just get rid of them as they are doing anything much for noise/voltage spike supression.

This is all assuming that the diodes are in series with the load. How do you have them configured?

(the voltage readings is after the diodes..)

i used fast respond diodes which is Ge made of (those small gray ones with Cu heatsink installed)
they were on main lines of 5 and 12 volts from old dvd psu which is 5v -5A and 12v - 4A written on
i think it shold be more then enough
they isolate the hard drives line from drain cuased externally loading on of the line (like blowing 3.5A fuse or connecting large scsi drive)
as i said they do the trick so i dont have drains any more and the drives arent effected any more...

what about adding the choke on the scsi line - then power drain will be limitted to some delta since a choke is just another word for inductor right???...

they are like this

in--5vLine---|>--cap+---out
GND-------------cap----out

Holy thread resurrection. Finally got interested into doing something to my motherboard, and this looks like the perfect thing. Could I use this type of capacitor: http://www.mouser.com/index.cfm?hand..._pcodeid=01420
in parallel (cause hellfire said parallel>single big one) with other caps by the cpu??

any capacitior mods for dfi nf4 ?

Quote:

---

Originally Posted by Revage

any capacitior mods for dfi nf4 ?

---

Ditto x1000!!!

Oh, i missed this thread last year. I will supply good quality ultra low ESR caps for this project if i know this thread last year.:cool:

sexy core ;)

Where can I get Ultra Low ESR caps from a known reputable source? I am looking for Rubycon or Sanyo if possible.