Author Topic: Computer Power Supply - Salvageable?  (Read 15427 times)

0 Members and 1 Guest are viewing this topic.

Offline allikat

  • Contributor
  • Posts: 30
  • Country: gb
Re: Computer Power Supply - Salvageable?
« Reply #25 on: August 12, 2014, 06:48:01 am »
The unit in question is a 2007 vintage design.  It has the guts to power a modern machine if you have all the cables and so on.

Review of the unit from the day, complete with ripple shots and internals pictures.
http://www.jonnyguru.com/modules.php?name=NDReviews&op=Story&reid=41
Scored well enough.  And repair is almost certainly viable if only the caps have gone, especially if you used better caps than they did originally.
It's not a piece of junk, and can hold up rated output on a load tester, but if you want to power a modern machine, you'd be better to buy a modern unit as that thing has way more 3.3v and 5v capacity than you'd ever use on something from after 2009.
Any engineer can readily identify 3 smells:
1: Coffee, 2: Escaped magic smoke, 3: Bullshit
(from an original post by John Coloccia)
 

Offline cybertronicify

  • Regular Contributor
  • *
  • Posts: 70
  • Country: us
  • Electrical Engineering Student
Re: Computer Power Supply - Salvageable?
« Reply #26 on: August 12, 2014, 06:08:05 pm »
I have fixed a couple of those cheapy china psus. Usually they use cheap knockoff power mosfets that have all pins shorted. Just replace them and it will work, but before you do that take a multimeter and measure the pins, if they are broken, they will have close to 0 ohms or a dead short. Btw on your next power supply, add a filter and dont forget to clean them or they will overheat.

Edit: Also dont forget, if you are going to replace them, make sure they are isolated from the heatsink and make sure the voltage and amps are correct.
« Last Edit: August 12, 2014, 06:14:46 pm by cybertronicify »
 

Offline Tothwolf

  • Regular Contributor
  • *
  • Posts: 95
  • Country: 00
Re: Computer Power Supply - Salvageable?
« Reply #27 on: August 12, 2014, 11:59:22 pm »
When I do this I select the capacitors based on lead spacing and diameter first and then capacitance which may be higher.  If I can get a higher ripple current rating with a higher voltage unit which will still fit which is usually the case, then I do that as well since the increase in price is small.

There is little reason not to change all of the capacitors except for the input filtering capacitors unless their is reason to suspect them as well.  Those are usually selected based on hold up time and maximum voltage ripple so their ripple current rating is irrelevant.

Right, it just doesn't make any sense to skimp on preemptively replacing an $0.11 5mm or $0.20 6.3mm capacitor (usually a 1uF and a 47uF in many common smaller PSU designs).

With the main input filter caps though, if the PSU has a lot of hours on it, has been run very hot, overheated, or has very low quality no-name capacitors to begin with, to me it doesn't make any sense not to spend another $1.50 (for a single 47uF 400V) or $2.50 x 2 (for a pair of larger 160-200V snap-in caps). I find the biggest hassle with repairing PSUs is just getting the board out of the case and getting access to the smaller components on a densely packed board, so if I'm in there anyway, I might as well replace the input filter caps and not have to worry about them later. Many of the smaller PSUs I service are also in devices such as network switches, some of which are passively cooled. In those applications, a failure is a much larger hassle than it would be with a single PC's power supply.

According to the review for the power supply in question, the two input filter caps in this supply are Teapo brand, so if I were rebuilding it, they would certainly be candidates for replacement.

Be careful with substituting parts with a much larger working voltage. If you substitute a replacement with a much higher working voltage, you might shorten the life of the component since it won't be able to fully reform its dielectric when operating at a lower voltage. Over time you could end up with a higher ESR, more internal heating, and reduced component life. This isn't a problem for small 5mm caps, but I generally try to avoid using say a 25V rated 10mm cap at 5V or 3.3V and would instead stick with ~2x the working voltage (a 16V or 10V part). I wouldn't even think twice about replacing a 16V rated cap on a 12V rail with a 25V rated cap (and it is pretty common for lower end PSUs to cheap out and use 16V parts on the 12V rail). In many older high end power supplies (25+ years old), I often see very large 6.3V snap-in caps on the 5V rails too.

A note for other readers...much of what I mentioned previously is more specific to PSUs vs motherboards. More modern motherboards (Pentium III/4 and later) with lots of ultra-low ESR electrolytic caps are a completely different animal. Those type of motherboards /require/ those ultra-low ESR caps to function properly and it is best to replace like for like on the working voltage and value. Nichicon HE, HV, PW, PM, etc are not suitable replacements for those ultra-low ESR parts (but PW parts work just fine for replacing general purpose bypass caps on those motherboards). Unfortunately, both Nichicon and Rubycon have discontinued their ultra-low ESR motherboard type electrolytic capacitors without any sort of replacement product lines.
 

Offline Rasz

  • Super Contributor
  • ***
  • Posts: 2616
  • Country: 00
    • My random blog.
Re: Computer Power Supply - Salvageable?
« Reply #28 on: August 13, 2014, 02:03:34 am »
main filter caps (the big ones) almost _never_ go bad in pc psus
Who logs in to gdm? Not I, said the duck.
My fireplace is on fire, but in all the wrong places.
 

Online NiHaoMike

  • Super Contributor
  • ***
  • Posts: 9029
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: Computer Power Supply - Salvageable?
« Reply #29 on: August 13, 2014, 02:28:09 am »

I haven't seen a true dual rail design in a long time. Even server PSUs are running just one huge 12V rail.
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline allikat

  • Contributor
  • Posts: 30
  • Country: gb
Re: Computer Power Supply - Salvageable?
« Reply #30 on: August 13, 2014, 11:09:56 am »
Yeah, while primary input filter caps very very rarely fail, if you're going to the effort to recap the whole thing, you may as well replace them as well.  It's not much extra cost or effort to do so.  You should also check the board for any signs of overheating, and check all the caps, not just the electrolytics.

Free extra rant!
True multi-rail units are still out there, especially at the higher power levels, where 100+ amps at 12V is plenty to do arc welding on your motherboard if you get a short.
Way too many people have signed up to the multi-rail=bad idea.  |O   If you like your machine, then at higher output levels then multi-rail is safer, as the protection ICs currently in use can't do a whole lot to stop nastiness once the rail current passes a certain amount.  Even the best designed single rail 1KW units can weld things so long as you don't hit the over-current protection levels.
Any engineer can readily identify 3 smells:
1: Coffee, 2: Escaped magic smoke, 3: Bullshit
(from an original post by John Coloccia)
 

Offline KekenTopic starter

  • Contributor
  • Posts: 15
Re: Computer Power Supply - Salvageable?
« Reply #31 on: August 14, 2014, 09:05:09 pm »
Hello again everyone! Thanks for all the feedback!

As I said before I am still learning a few things. The past few days I bought a DIY kit and started to practice my soldering skills. Also tried de-soldering, tougher at first, but then I got the hang of it.

So I looked back at my PSU and tried to de-solder the two visibly bad e-caps.

SUCCESS!

SUCCESS!

Anyway, from reading all the suggestions, I believe I will definitely replace all the caps.
But I'm not sure if I have everything right.
So here goes..
In the image we have an e-cap that has 1,000 uF, 10v.
To get the diameter, height and lead spacing, just use a ruler correct? Then round UP to whatever the closest option is on digikey for the size?

Also from what I've read, choosing a replacement e-cap, I would choose it by these characteristics in this order(?);

1. Lead Spacing (3.9mm) used a ruler (was in inches)
2. Diameter (7.9375mm) used  a ruler
3. Value (1,000 uF - Choose higher or same?)
4. Voltage (10v - Do I go up to 12v or use the same?)
5. Height (14.2875 - Would not matter if all the other criteria is met?)

Is that correct?

btw, how do you find out the ripple current and esr? using a meter?
Should temperature be the same? On digikey.com it gives me an option to choose operating temperature Which range of temp should I choose?
Then there's also Impedance and Surface Mount Land Size and others that I'm confused about.

This is what I've chosen on digikey so far:
(With the numbers rounded UP)


I wont be ordering anything yet until I'm done removing the rest of the e-caps and I still need to get a meter (just don't know which yet). Maybe I'll just get the one Rasz mentioned earlier http://www.ebay.com/itm/DIY-Kit-Capacitance-ESR-Inductance-Resistor-LC-Meter-Tester-NPN-PNP-Mosfet-M168-/281262386306 and it'll be nice to get more practice :)

Not sure if I forgot to ask something, but I need to get going and I will be back again!

Thanks everyone!


 

Offline mariush

  • Super Contributor
  • ***
  • Posts: 5035
  • Country: ro
  • .
Re: Computer Power Supply - Salvageable?
« Reply #32 on: August 14, 2014, 10:20:18 pm »
Lead spacings, diameters etc are generally either in multiples of 0.1"  (2.54mm)  or millimeters.

Your capacitor is made by Teapo, the series is SC, it's rated for up to 105 C, 1000uF, 10v

You can search for the datasheet and that will tell you the actual lead spacing and the diameter most often (sometimes manufacturer makes custom batches of a particular series, or a whole custom series for a company).
In this case, the datasheet for Teapo SC is here.

If you go down on page 2 and look at the three columns valid for 10v, then going down at the 1000uF row you find there: size  = 8x20 ripple = 1080mA Impedance  = 0.065

Now, I know you said ~ 15mm for height, so it's quite possible that this capacitor is a custom order for the psu maker, making it a custom 8 x 15 mm capacitor, but since the datasheet lists only 20mm tall for 8mm diameter, I'm going to go with those specs.

Page 1, center, you have a graphic of a capacitor with several dimensions... you're interested in P, which depends on the diameter of the capacitor.
Your diameter is 8 mm, so in that table you go to the  8mm  column and see that  P equals to 3.5mm.   Note that P is the distance between the centers of the two leads, your value was larger probably because you included the whole thickness of the leads and the leads were probably not exactly straight.

With these low ESR capacitors it's important to solder the capacitors as close as possible to the PCB. It's really not a good idea to leave even half an inch of leads between the capacitor and the boards.
Therefore, it's a good idea to match the distance between the leads so that the capacitor would fit perfectly in the original location, but it's really not that much of a big deal if you'd have to bend the leads just a bit and have 2-3 mm of space between the capacitor and the pcb. So look for a 3.5mm spacing between leads but don't rule out capacitors just because you can't find 3.5mm spacing.

The diameter is 8 mm in your case. Here, you can buy capacitors that are larger in diameter but you simply have to make sure they'll fit inside the area where the old capacitor was located. Sometimes the psu is designed in such a way that there's a row of capacitors and they basically touch each other, so there's really no way to use something larger in diameter. In other cases, it's very easy to use something a bit larger.
See if you could fit a 10mm capacitor, if not you'll have to look for 8mm.

The voltage rating is the maximum voltage that the capacitor can tolerate before it gets damaged.  Since the power supply outputs 3.3v, 5v or 12v, it's obvious this 1000uF 10v capacitor was used either with 3.3v or 5v.
These voltage ratings are somewhat standardized... for electrolytic capacitors the most common voltage ratings are 6.3v, 10v and 16v.
The psu manufacturer went with 10v rating even though he could have gone with 6.3v rating without any risk. You could use a capacitor with rating of 6.3v or you could use one rated for 16v - as long as the voltage rating is higher than what actually is there when psu runs (3.3v or 5v), you're good.

Now why did the manufacturer use 10v rated capacitor? Can be for lots of reasons...  Often, the price difference between a 6.3v rated capacitor and a 10v rated capacitor is 0, so they just go with 10v.  Also, if there's no difference in capacitor diameter, psu manufacturer may choose to go with 10v rated capacitor because they may have better specs (longer hours @ 105c, better impedance etc)

The height... you can use capacitors that are larger, you just have to make sure the new capacitors won't touch a heatsink, or block the fan at the top and so on.

Two specs that you missed and are very important for power supplies are those listed in datasheet.. Ripple and Impedance.  This 1000uF 10v capacitor is rated for 1080mA ripple and 65mOhm impedance (this is practically the same as the ESR value)

When you replace a capacitor, the new capacitor must have a ripple value as close as possible or higher than the old value. The impedance must be as close as possible or less than the old impedance value - but note that too low impedance/ESR can be bad, the power supply may be fine tuned to rely on this capacitor's esr to work properly.

So drawing the line, you're looking for

1000 uF
10v rated  ( but can use 6.3v or 16v as long as ripple / impedance are good)
1080mA ripple
65mOhm impedance
8 mm  diameter (or more if the psu has room)
15 mm (or taller if there is room)

Let's see if your Nichicon HE is good... here's the datasheet.  Scroll down to the 10v column, go down to 1000uF and  you have dimensions: 8x20 mm (good) , ripple 1050 mA (a bit lower but close enough) and impedance 20c @ 100kHz = 69 mOhm (close enough to 65mOhm) so it should work fine.

But you can just go on Digikey and select 1000uF , 6.3-16v, filter, the select 8mm diameter and filter, then sort by ripple or impedance and filter again, and then see what capacitors have the specs as close as possible as the above... and you may come up with something like this:



As you can see there's some capacitors that have impedance a bit lower than 65mOhm which may be a problem, but as long as it's not a big difference it should be ok. I mean you can find capacitors with impedance of 20-30mOhm which is really far from the 65mOhm value your Teapo capacitor has, so those won't do. But these with 40-60mOhm impedance should be fine.
 
If you can't find anything, you can go back and allow 10mm diameter capacitors if you have room on the PCB and search again.

In the case of the 220uF 16v capacitor, as this may be used for 12v or -12v, you have to stick to that 16v rating, or you may go with something a bit higher, like 25v. But, you still have to respect those parameters written in the datashet:

220uF
16v rated (or higher if there is room)
550 mA ripple
140 mOhm impedance
8mm diameter (or more if there is room)
11mm height (or more if there is room)

« Last Edit: August 15, 2014, 09:51:54 am by mariush »
 

Offline Tothwolf

  • Regular Contributor
  • *
  • Posts: 95
  • Country: 00
Re: Computer Power Supply - Salvageable?
« Reply #33 on: August 15, 2014, 02:34:19 am »
In addition to 8x20, 8x16 is also a standard size. Nichicon offers a 1000uF 10V capacitor in 8x16 in their HV series, but HV isn't as widely stocked as HE. Rubycon also has a 1000uF 10V capacitor in 8x16 in their ZLH series, which is commonly stocked. A 10mm or 12mm diameter capacitor would have a 5mm lead spacing, so it would be best to stick to an 8mm capacitor to match the 3.5mm lead spacing of the original.

The 220uF 16V capacitor in the photo looks like a 6.3x11mm (2.5mm lead spacing) but that combination also isn't available in either Nichicon HE or PW. It is an option in Nichicon HV, but again isn't widely stocked. It is also available in Rubycon ZLH though. Since it looks like it is a tight fit in your PSU, a Rubycon ZLH series capacitor is probably going to be your best bet, and you could just go with Rubycon ZLH for both the 220uF 16V and 1000uF 10V capacitors.

You might want to try to add a little room between the replacement capacitors and the components attached to the heatsinks. Leaving a few mm of extra lead on the capacitors so you can angle them slightly away from the hot components won't really hurt anything and will extend the life of the replacement capacitors.
 

Offline Tothwolf

  • Regular Contributor
  • *
  • Posts: 95
  • Country: 00
Re: Computer Power Supply - Salvageable?
« Reply #34 on: August 15, 2014, 02:47:41 am »
I wont be ordering anything yet until I'm done removing the rest of the e-caps and I still need to get a meter (just don't know which yet). Maybe I'll just get the one Rasz mentioned earlier http://www.ebay.com/itm/DIY-Kit-Capacitance-ESR-Inductance-Resistor-LC-Meter-Tester-NPN-PNP-Mosfet-M168-/281262386306 and it'll be nice to get more practice :)

I would suggest you skip the meter for now unless you are going to be testing capacitors on a regular basis. For a PSU such as yours, most of the time you wouldn't even bother testing the ESR of the caps, you'd just replace them wholesale.

As for ESR meters specifically, the Bob Parker meter seems to be the most common and is probably one of the better non-commercial meters. I used to have one of the Dick Smith kits, but it "walked off". I bought one of the Blue ESR kits to replace it, but haven't yet built it. I do have a few issues with the way it was kitted up by AnaTek (CMOS parts in a clear plastic bag, for starters) but I'm going to save that up for a build/modifications/review post later. IMO, Dick Smith did a /much/ better job choosing parts for their kits.

http://members.ozemail.com.au/~bobpar/esrmeter.htm
http://shop.anatekcorp.com/products/component-analyzers/blue-esr-kit
 

Offline KekenTopic starter

  • Contributor
  • Posts: 15
Re: Computer Power Supply - Salvageable?
« Reply #35 on: August 15, 2014, 03:54:58 am »
Great big thanks to mariush and Tothwolf!

Thank you both for the great detailed information. As I am a noob, I'm going to have to read this more than once to have a good grasp on it all.

I really appreciate your help and I will keep everyone updated on this project I have going on.

Thanks again!

-Keken
 

Offline Shock

  • Super Contributor
  • ***
  • Posts: 4225
  • Country: au
Re: Computer Power Supply - Salvageable?
« Reply #36 on: August 15, 2014, 09:49:49 am »
Those cheap Chinese ESR testers work fine and are a very useful tool for ~$20.

Before you spend money buying a Blue ESR tester check out the DER EE DE5000 which is about the same price but is a full multifunction LCR meter and can also test ESR at multiple frequencies. In the last few years prices have dropped dramatically on tools like this.

There are several threads on Eevblog about both testers.
Soldering/Rework: Pace ADS200, Pace MBT350
Multimeters: Fluke 189, 87V, 117, 112   >>> WANTED STUFF <<<
Oszilloskopen: Lecroy 9314, Phillips PM3065, Tektronix 2215a, 314
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16630
  • Country: us
  • DavidH
Re: Computer Power Supply - Salvageable?
« Reply #37 on: August 15, 2014, 02:45:29 pm »
Before you spend money buying a Blue ESR tester check out the DER EE DE5000 which is about the same price but is a full multifunction LCR meter and can also test ESR at multiple frequencies. In the last few years prices have dropped dramatically on tools like this.

How well does the DE5000 work when making in-circuit measurements?

I just have an old style impedance bridge which works great but not for in-circuit measurements and only at 1000 Hz.  I have been considering a DE5000 or similar for capacitance and dissipation measurements at different frequencies and because it would be more portable.
 

Offline mariush

  • Super Contributor
  • ***
  • Posts: 5035
  • Country: ro
  • .
Re: Computer Power Supply - Salvageable?
« Reply #38 on: August 15, 2014, 03:16:09 pm »
I second Shock's opinion.

Don't waste your money on Blue ESR, it's an outdated esr meter, expensive, not really any better than any other 10-20$ esr meters on eBay.   I already gave you some links in the other thread along with link to a more "professionally" looking meter (compared to the eBay ones).

If you feel like spending money, that DER EE DE5000 is a good product.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16630
  • Country: us
  • DavidH
Re: Computer Power Supply - Salvageable?
« Reply #39 on: August 15, 2014, 05:05:48 pm »
Don't waste your money on Blue ESR, it's an outdated esr meter, expensive, not really any better than any other 10-20$ esr meters on eBay.   I already gave you some links in the other thread along with link to a more "professionally" looking meter (compared to the eBay ones).

I would not spend money on any cheap ESR meter that does not include full documentation because without that, it is difficult to know what is actually being measured under different conditions.
 

Offline Shock

  • Super Contributor
  • ***
  • Posts: 4225
  • Country: au
Re: Computer Power Supply - Salvageable?
« Reply #40 on: August 15, 2014, 05:25:48 pm »
Quote from: David Hess
How well does the DE5000 work when making in-circuit measurements?

If you want discharging and in-circuit testing I believe the newer Peak Atlas model is the preferred ESR meter.  Perhaps even for input overload protection.

I own two Chinese $20 ESR meters and a much more expensive Philips PM6303 benchtop LCR meter so I have most bases covered already. But if I was to buy another I would get a DER EE DE5000 until something better comes out. 

The Cyrustek chipset is used in dozens of products now and the DER EE DE5000 is owned by many Eevblog members. It is able to take sense (4 wire) measurements which is another good feature. Data logging may also be useful to some. Optional power supply may be useful as well. It measures way more than just ESR all for ~$80.

There are many threads discussing the pros and cons. All depends on how many faulty caps you need to measure in-circuit per day. If you want audio/pass fail beeps or want automatic discharging etc. At the end of the day most people agree in-circuit is not 100% reliable and caps have to be removed to be certain it's out of spec.



Soldering/Rework: Pace ADS200, Pace MBT350
Multimeters: Fluke 189, 87V, 117, 112   >>> WANTED STUFF <<<
Oszilloskopen: Lecroy 9314, Phillips PM3065, Tektronix 2215a, 314
 

Offline mariush

  • Super Contributor
  • ***
  • Posts: 5035
  • Country: ro
  • .
Re: Computer Power Supply - Salvageable?
« Reply #41 on: August 15, 2014, 05:49:19 pm »
I would not spend money on any cheap ESR meter that does not include full documentation because without that, it is difficult to know what is actually being measured under different conditions.

He doesn't need to know exactly what the value shown represents or if it's 100% accurate.  He just needs a simple tool to tell you capacitor is BAD or GOOD.  If you see 50-100 mOhm on the screen, and the datasheet shows 70 mOhm for ESR, then the capacitor is probably OK.  If the meter shows 2-200 mOhm, then the capacitor is probably bad.

He doesn't have a multimeter if I remember correctly what I read in his threads, he doesn't  have anything, he's just starting up or trying to fix a power supply. He wouldn't even have a clue about how to properly use such a LCR meter, and what those options mean. No point spending so much money on a tool he won't use often.  Better spend little on this and buy a multimeter or something he'll use more often if he starts playing with circuits.

Let him spend 10-20$ on something that tells him the esr of capacitors as a "close enough" value to the real esr, and optionally spend the rest of the money on a proper multimeter, not a lcr meter with minimal protections he may accidentally damage.

 
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf