Replacement caps for Antec SP-450?

[mockingbird]

Toasty, I salute thee.

PCBONEZ:
Thanks. I didn't have to take anything apart (Apart from desoldering the wires from the case plugs to the PCB) to get all the caps. Just go to the dollar store and buy a 9-LED flashlight for a couple of bucks. I was able to see everything. In fact, a few posts up I took a picture of those two nasty little caps EC3 and EC4 and you can even see the markings on them. I got them out buy melting the solder with the PCB turned upside down and then pushing down on each pin simultaneously while in fluxed state until they fell out. I think I can even get the new ones in without the angled forceps as Toasty suggested because I'm going to wick away the solder so there's just holes (Something I usually never do), and then slide the really long leads from the new caps through them.

What I don't like about KG was the reduced size (Don't ask me why), so I did a little digging. Take a 50V 0.22uF Fuhjyyu TNR for example. So I'm looking at the KG spec sheet and an equal part that is only 4 x 7mm has a rated ripple of 3.5 and cost 48 cents. A 100v 0.22uF Nichicon VR has a rated ripple of 4.7 and is 5x11mm and cost only around 33 cents if you buy 1. Which is the better buy?

Regarding the filter in/outs, I think I've finally comprehended the idea here. The thing is, what forces filter in/out caps to sit right near the coils? Can't they ostensibly put them anywhere they want on the PCB? Isn't the only way to locate them is by tracing them? Isn't the only other alternative of doing it safely simply matching the ESR of ALL the PSU caps (Which isn't easy).

Finally I have a question that I've wanted to ask you for a while now. Is voltage rating on a cap really just a ripple rating? And if this is the case, wouldn't you agree that it would be safe to put a very highly ripple rated 10V cap in place of a 16V cap (I'm only asking theoretically, I have never done this). Something interesting I saw today while looking at a Panasonic FR spec sheet was that the ripple and ESR of certain caps which are same dimensions of a FM cap but had a lower voltage rating were the same! (Or something like that).

[Toasty]

>>what forces filter in/out caps to sit right near the coils? <<

ESR. Remember that we are dealing with milliohms here. An inch of circuit trace could easily add one or 2mΩ of ESR and throw the whole filter off. Even mounting a cap that won't fit by elevating it off the board on longer legs will have the same effect.

>>Is voltage rating on a cap really just a ripple rating?<<
Ripple rating is (high frequency AC or 'chopped' DC, 30kHz-100kHz) current the cap can handle without overheating, for the caps we are dealing with here. In the mains caps section, it is low frequency 50-120Hz.

>>wouldn't you agree that it would be safe to put a very highly ripple rated 10V cap in place of a 16V cap<<
No. You have to consider the DC voltage of the circuit with at least a 20-25% "cushion". You can not put a 10v cap in where a 16v cap would go, -unless- the 16v was chosen for size/price/ripple to be in an < 8v circuit. Many manufacturers will use 10v caps on the 3.3v & 5v lines because they are cheaper and handle a lot more ripple than the comparable 6.3v ones would.

PC (and others) have had many lengthy writings on this subject over the past few years. It takes a little digging but there is some excellent material in here all about this.

My 2¢

Toast

[severach]

Originally Posted by Toasty

ESR. Remember that we are dealing with milliohms here. An inch of circuit trace could easily add one or 2mΩ of ESR and throw the whole filter off.

Long leads and traces adds something devilish but I don't think it's ESR. ESR would be measurable on an ESR meter. Me and others have not been able to measure any ESR difference a half inch of leads causes. Even 30" leads on an ESR meter don't add much.

You'll cry when you see on a scope what a half inch of leads adds. Long leads harms a circuit a lot more than than ESR. A 16" wire doesn't have much ESR but solder it inline with an output capacitor and to the circuit the capacitor ceases to exist as if it had 99 ohms ESR.

Well designed SMPS circuits have an array of output capacitors. There will be a few ceramics very close to control the ring allowing the bigger capacitors to be more leisurely spaced.

[PCBONEZ]

Originally Posted by mockingbird

I'm a little confused, because I often see a mix of GP and low ESR high value caps on PSUs. How can you tell what is a filter input/output?

I was going to cover this but I forgot.

The CWT based PSUs often use GP where they should use low ESR.
If you see TN in a OP filter that's what they have done.
They VERY often use TN in +5vsb OP filters which is crazy because some boards use +5vsb for RAM power while the system is starting up.

If it's an OP filter one of the caps will be connected to an OP.
Use a DMM or just follow the traces.

.

[PCBONEZ]

Originally Posted by severach

Long leads and traces adds something devilish but I don't think it's ESR. ESR would be measurable on an ESR meter.

Actually you're right. - It's not ESR, it's resistance.
And yes, you can measure it with an ESR meter.
An ESR meter 'sees' Resistance the same as ESR.
- That's why you can calibrate an ESR meter with a resistor.

Originally Posted by severach

Me and others have not been able to measure any ESR difference a half inch of leads causes.

I have seen it on every single cap I ever checked for it using a Bob Parker ESR.
I've check a variety of caps and it is consistent.
At somewhere between 5/8" and 7/8" from normal solder point of the cap leads you will see a .010 ohm increase.
At 1/2" you DO have a difference but you are below the +/-.01 ohm resolution of the meter so you will never see it.
[In other words you aren't checking properly.]
-
Since resistance of wire is linear to it's length, if you have .010 ohms difference at 3/4" then...
At 1/4" there would be .003 ohms difference
and at 1/2" there would be .006 ohms difference.
-
And if you try to tell me .006 ohms doesn't matter... I will point out...
It is the exact difference between a 10x20mm MBZ and a 10x20mm HZ.
IOW: At the lower ESR end of things a .006 ohms difference is jumping TWO Grades...

Originally Posted by severach

Even 30" leads on an ESR meter don't add much.

That's because you zero the meter with the probes connected which cancels out the leads resistance. [DUH!]

Originally Posted by severach

Long leads harms a circuit a lot more than than ESR. A 16" wire doesn't have much ESR but solder it inline with an output capacitor and to the circuit the capacitor ceases to exist as if it had 99 ohms ESR.

That is also resistance, not ESR, and it occurs for EXACTLY the same reason as it does with excessively long cap leads.
- It's the SAME EXACT THING. -> Resistance of connecting wire..
.

[PCBONEZ]

Originally Posted by mockingbird

Toasty, I salute thee.

PCBONEZ:
Thanks. I didn't have to take anything apart (Apart from desoldering the wires from the case plugs to the PCB) to get all the caps. Just go to the dollar store and buy a 9-LED flashlight for a couple of bucks. I was able to see everything.

Your eyes are better than mine buddy.
I have to get the flashlight AND the little old lady 3x reading glasses or I can't read a damned thing on a part or PCB anymore.

Originally Posted by mockingbird

What I don't like about KG was the reduced size (Don't ask me why), so I did a little digging. Take a 50V 0.22uF Fuhjyyu TNR for example. So I'm looking at the KG spec sheet and an equal part that is only 4 x 7mm has a rated ripple of 3.5 and cost 48 cents. A 100v 0.22uF Nichicon VR has a rated ripple of 4.7 and is 5x11mm and cost only around 33 cents if you buy 1. Which is the better buy?

The KG small size one of the reasons I like them.
I can use them to replace both 4mm and 5mm without stocking two sizes and additionally the small size allows more cooling air around the cap.
I wouldn't use a 100v cap myself. 50v is probably already over-kill.
-
Where did you come up with 48 cents?
The .22uF 50v KG looks like it's 22 cents to me..
http://search.digikey.com/scripts/Dk...l&name=P927-ND

Originally Posted by mockingbird

Finally I have a question that I've wanted to ask you for a while now. Is voltage rating on a cap really just a ripple rating?

No. [Volts=volts,,, Ripple=amps]
The voltage rating on a cap is the max total voltage.
That is DC volts + the positive side of the ripple.
It tells you how much it can take without breaking down the oxide layer [aka dielectric] by what amounts to arcing.
* An analogy would be the voltage rating on the insulation of a wire.
Too much volts and those trons will jump right through the insulation which will melt it making it easier for other trons to do the same.
-
[I will give the long version for Ripple]
Ripple on the other hand is how much [AC] current can pass through without overheating the oxide layer. Overheating will cause the layer to dissolve into the electrolyte faster than normal. [Chemical reactions happen faster at higher temps]. That will make the layer thinner and eventually prone to arcing as with the voltage rating discussion. [A thinner oxide layer amounts to a lower voltage rating.] It will also tend to break down the electrolyte itself.
* An analogy is the amps rating on the conductor in a wire which is also based on not overheating it,, in that case causing it to melt.

The DC component of a cap has no current associated with it because caps don't pass DC current. What -appears- to be DC current during transients [including charge/discharge] is trons going in one end and out the other but none actually go -through- from the in to the out.
[Okay there is a tiny bit called 'leakage' but I don't want to get into that...]
.

[mockingbird]

Where did you come up with 48 cents?
The .22uF 50v KG looks like it's 22 cents to me..
http://search.digikey.com/scripts/Dk...l&name=P927-ND

Nichicon VR
Panasonic KG

I'm looking at this page at Elna
http://www.elna-america.com/tech_al_principles.php

They show two types of foil etchings. One for low-voltage and one for high. Probably when they say high they mean the big input caps. But looking at the electrolyte composition there, it certainly seems like there are probably seperate factors like your insinuating which constitutes a capacitor's ripple and its voltage. I'm sure they'll never tell you *exactly* what they put in the electrolyte. Only the Chinese spies know apart from the companies themselves.

[PCBONEZ]

Originally Posted by mockingbird

Nichicon VR
Panasonic KG

I'm looking at this page at Elna
http://www.elna-america.com/tech_al_principles.php

They show two types of foil etchings. One for low-voltage and one for high. Probably when they say high they mean the big input caps. But looking at the electrolyte composition there, it certainly seems like there are probably seperate factors like your insinuating which constitutes a capacitor's ripple and its voltage. I'm sure they'll never tell you *exactly* what they put in the electrolyte. Only the Chinese spies know apart from the companies themselves.

Yes, I've read that in detail.
In fact I recently reread it for something I'm working on.
Look closely at the photos of the oxide layer.
It's actually tiny oxide lined tubes [very deep pits] that get filled with electrolyte.
That's how the surface area is increased so much with etching.

The VZ .22uF 50v is an exact match for the TN .22uF 50v. - Both are 2.9 mA.
That would do fine and they are only 16 cents for 1pc.
http://ca.mouser.com/ProductDetail/N...E1zfoqBuY9w%3d

[mockingbird]

The VZ .22uF 50v is an exact match for the TN .22uF 50v. - Both are 2.9 mA.
That would do fine and they are only 16 cents for 1pc.
http://ca.mouser.com/ProductDetail/N...E1zfoqBuY9w%3d

Ah, perfect. This will do well.

Here's what I have:

EC3 - Nichicon 50V 4.7uF "PW" 5x11 [PW]
EC4 - Nichicon 50V 4.7uF "PW" 5x11 [PW]
EC5 - Panasonic 50V 22uF "FC" 5x11 [PW]
EC6 - N/A
EC7 - N/A
EC8 - N/A
EC9 - N/A
EC10 - Panasonic 16V 1000uF "FR" 8x20 [0.030]
EC11 - Panasonic 16V 1000uF "FR" 8x20 [0.030]
EC12 - Panasonic 50V 47uF "FC" 6.3x11.2 [0.60]
EC13 - Panasonic 50V 10uF "FC" 5x11 [GP]
EC14 - Nichicon 50V 1uF "PW" 5x11 [GP]
EC15 - Nichicon 50V 1uF "PW" 5x11 [GP]
EC16 - Nichicon 50V 1uF "PW" 5x11 [GP]
EC17 - Nichicon 50V 1uF "PW" 5x11 [GP]
EC18 - ???
EC19 - Samxon 16V 3300uF "RS" 10x30 [0.035]
EC20 - Panasonic 35V 330uF "FM" 8x20 [0.030] (In 6.3mm spot)
EC21 - Nichicon 25V 1000uF "PW" 10x31.5 [0.035]
EC22 - Samxon 16V 3300uF "RS" 10x30 [0.035]
EC23 - N/A
EC24 - Samxon 16V 3300uF "RS" 10x30 [0.035]
EC25 - Samxon 16V 3300uF "RS" 10x30 [0.035]
EC26 - Panasonic 50V 10uF "FC" 5x11 [GP]
EC27 - Nichicon 50V 0.1uF "TS" 4x7 [GP]
R98 - Nichicon 50V 0.22uF "VZ" 5x11 [GP]

Does this look OK? Keep in mind I had to use certain models just because of their ESR or their availability.

For EC18 - which appears to be part of a filter output - I cannot for the life of me find a 220uF cap that has an ESR in the 0.030~ range. The only thing I can find is a Nichicon LV polymer cap. What do you suggest?

Another thing. I found a series, Nichicon "TS" which is sometimes as good as KG, but is a lot cheaper.

[Toasty]

>>I cannot for the life of me find a 220uF cap that has an ESR in the 0.030~ range. <<

Why would you want that??

The original was a TN with a ripple of 195mA and no impedance or ESR given. Just go with a match or comparable size with equal or better ripple. An FC for example is 555mA, 2X+ the rating of the original.

Toast

[mockingbird]

Originally Posted by Toasty

>>I cannot for the life of me find a 220uF cap that has an ESR in the 0.030~ range. <<

Why would you want that??

The original was a TN with a ripple of 195mA and no impedance or ESR given. Just go with a match or comparable size with equal or better ripple. An FC for example is 555mA, 2X+ the rating of the original.

Toast

Originally Posted by PCBONEZ

I was going to cover this but I forgot.
The CWT based PSUs often use GP where they should use low ESR.

It seems to be a part of a filter output with EC19 being an input cap. Since I'll be using a Samxon cap with a 0.035 ESR, I want something that matches that, amiright?

[Toasty]

God, I need medication....

It's only part of the -12v supply, IIRC. I have no foil trace photos in that area to compare to and I'm not ripping mine apart to find it.

Putting something with that low an ESR will FUBAR the filtering. You have to stay fairly close to original. Nothing in 220µF/16v in that size can is going to come close to 37mΩ.

Toast

[mockingbird]

One lead is attached to the foil which is only attached to a vertical coil, and the other lead is attached to where all the black wires meet (Which I assume you mean to be the -12v wires). Is it just a coupling cap?

[Toasty]

BLUE is -12v

http://www.aitechsolutions.net/pchwtrblsht.html

[mockingbird]

There are four primary groups of wires coming from the PCB. Orange, Red, Black and Yellow.

[Toasty]

So? Your fussing over the -12v supply line.......?

Which is BLUE

[mockingbird]

Yes, the lone blue wire that comes off the PCB (Except for the lone navy blue wire which comes from the daughterboard) does also trace back to the big blob of black wires...

Ok, so this is nothing to worry about. Will use FC or equivalent, thanks.

[PCBONEZ]

Oh gezzz,
The old cap was a flippin' TN. - Not even low ESR.

When you can stick close to original spec.
But for this one [a GP in a filter where a GP shouldn't be] something like an FC is a better idea.
.

[PCBONEZ]

Finally had time to really look at the replacement BOM

What is up with EC20?
An 8x15mm 330uF/16v SC is like 750-mA/0.100-ohm
An 6.3x15mm 330uF/16v SC would have worse specs if it was in the data sheet.
... Probably close to what they give for the 8x11mm which is 620/0.120
The FM at .030 ohms is way over-kill.
.
I think I would use...
FC 8x15mm 330uF/25v - 730/0.085
KY 8x15mm 330uF/25v - 840/0.087
.

Everything else looks fine.
.

[mockingbird]

Ah... It is sitting right next to two coils, I thought it was part of the output filter of EC22 which is going to be a Samxon RS which was 0.035 so I tried to match it as best as I could.

Looking at the traces at the back, it doesn't seem to be part of it, instead I can see one of the traces going to what Toasty has lead me to believe is the -12V line.

Thanks, I will change this accordingly.