VOGONS


First post, by Ozzuneoj

User metadata
Rank Oldbie
Rank
Oldbie

Okay, I'm going to try to be brief. I'm not an engineer or even a formal student when it comes to electronics and circuits. I dabble with old PCs enough that having a basic understanding of the practical aspects of circuits and components is beneficial. I like to pick up rare or useful old parts from scrap lots, so I do a fair amount of repairs with the basic knowledge I have now, but I have a lot to learn.

Being able to identify and replace swollen or leaky caps can be tedious but is a common skill for people who are into retro PCs (specifically from the early 2000s). The area where I seem to find a lot of disagreement is when to replace caps that aren't leaking or swollen. Some say that any aluminum electrolytic cap that's more than 15 years old is basically garbage and should be replaced before it causes problems. Others are more of the "if it ain't broke, don't fix it" school of thought and will leave them basically forever. I can understand both lines of reasoning, but since the cost of replacing caps (not to mention the time) probably exceeds what I spend on the components themselves many times, I needed something more concrete to work with.

I bought a DER EE DE-5000 LCR meter and started playing with it today. I have a box of caps I've "borrowed" from boards over the years. I kept these around because for a while I was finding leaking caps very frequently, so it made sense to harvest all of the intact Rubycons, Nichicons, Panasonics and Sanyos from boards that had other problems, so I'd have some on hand to do a quick patch job on a cheap old board.

Anyway, most of these caps are 5-20 years old and were harvested from old boards. Some are far far older and were found in estate sale lots.

Rather than try to copy tons of numbers and data from spec sheets that people really don't care about, I'll simply write out my results with each cap based on the reading from my DE-5000 compared to the spec sheet (if available). The meter was calibrated using the alligator clip attachment. Capacitance readings were made at 120Hz, ESR readings were made at 100KHz. Caps that were harvested from another device will be in italics. Most of the harvested caps are from Socket A, Socket 604 or Socket 478 boards from the early 2000s. The non-italic ones are unused but still purchased 5-10 years ago and probably purchased cheaply as old stock. If I have multiple caps of the same type, I'll separate their values by commas. If the readings are out of spec, I'll simply mark them with an * (like 100uF*, 1.4ohm*). If I can't find the spec, I'll mark them with a "(?)". Any unmarked readings are within the specified tolerance of the datasheet as far as I can tell. Feel free to correct me if I'm wrong.

Panasonic FJ 6.3v 3300uF = 2690uF, 2720uF / 0.004ohm, 0.007ohm

Sanyo WG 6.3v 2200uF = 2280uf, 2260uF, 2230uF / 0.007ohm, 0.012ohm, 0.011ohm

Sprague 30D 6v 150uF = 146uf, 158uf, 162uF, 165uF / 0.194ohm, 0.169ohm, 0.44ohm, 0.164ohm(?) (used, axial lead, 30+ years old, can't find spec sheet)

ELNA RJ4 6.3v 1000uf = 802uF, 790uF*, 796uF* / 0.49ohm*, 0.45ohm, 0.47ohm (datasheet says +-20% (800-1200uF) and a max of 0.47ohm are within spec when manufactured)

Rubycon MCZ 6.3v 1800uf = 1597uf, 1565uf/ 0.009ohm, 0.008ohm

Rubycon MBZ 16v 1500uf = 1414uF, 1401uF / 0.009ohm, 0.008ohm

Rubycon MBZ 10v 1000uf = 897uF, 902uF, 900uF / 0.02ohm, 0.021ohm, 0.021ohm

Rubycon YXG 10v 1000uf = 986uF, 984uF/ 0.042ohm, 0.043ohm

United Chemicon KZG 16v 1000uf = 955uF, 974uF / 0.013ohm, 0.011ohm

That's all for now. I plan to pull caps off of boards and test them soon so we can get a more accurate picture of how they have degraded with age.

All I'll say about the results is that I didn't cherry pick what to post. This is all of the caps I measured, and all (except the cheaper ELNA caps which were never low-ESR) were within the original manufactured spec for capacitance and with ESR that was far below the maximum spec, most were below half. Keep in mind that many of these were likely used for 10+ years and then also went through being desoldered and stored for 5+ years, and yet these are the measurements I'm getting.

I hope to do more eventually, but it certainly takes a lot of time. I'm just glad to finally have an ESR meter and to finally be able to test capacitors and know something about them.

Now for some blitting from the back buffer.

Reply 1 of 13, by dionb

User metadata
Rank l33t
Rank
l33t

Interesting topic!

I recently received the ESR meter I ordered online a while ago - a BSIDE ESR02 Pro. I bought it for the exact same reason - being able to test non-bulging caps. This tester has pads on the device itself to test with, and connectors for leads.

Now, with the pads I'm getting very similar results to yours. A few examples from my cap box:
- brand new Matsushita 10V 1000uF: 921.9uF measured, .11Ohm ESR
- United Chemicon 6.3V 100uF (almost new, used once on board that was still dead after full re-capping): 1042uF measured, 0.08Ohm ESR
- de-soldered Nichicon 16V 2200uF cap from a 1997-vintage MU440EX board: 2309uF measured, 0.03 Ohm ESR
- a Teapo 16V 1000uF cap that's visually OK (that's a rarity...) from a 2001 board: 925.3uF, 0.04Ohm ESR

According to the 'typical ESR' table on the back of the tester, a 10V 1000uF cap should have an ESR of 0.14Ohm, a 16V version of the same 0.12Ohm. It's clear that that table isn't referring to low ESR caps, as most of mine (with the exception of that Matsushita) clearly are.

So far so good. But these are all loose caps. To be useful I need to be able to use it with leads to measure parts still soldered to motherboards. I have three pairs, two I bought with this tester, one 'known good' from my multimeter. Unfortunately, measurements go completely to pot when using any of these leads.

Example: that United Chemicon 6.3V 1000uF cap that measured 1042uF with 0.08Ohm ESR on the pads on the tester itself.
Lead 1 (tongs, designed for SMD work): 1027uF, but .55Ohm ESR!
Lead 2 (classic 'probe', so sharp-ish metal point): 1045uF .23Ohm ESR
Lead 3 ('known good', also probe): 1041uF, 0.66Ohm ESR

In other words, the capacitance is only slightly influenced by the leads, but the ESR shoots up, probably due to the resistance of the leads. My first assumption was that each lead would have a fixed resistance that could be subtracted from the measurement to get the correct reading. That sort of seems to play out - with the Chemicon cap I measured 0.08Ohm ESR directly on my tester, but 0.23Ohm via the leads. That would suggest that the leads add 0.15Ohm. So if I measure the Nichicon (0.03Ohm direct on tester) with them I would expect to measure 0.18Ohm. I got 0.19Ohm. That sounds close enough.

But...

I got some completely different values when caps were soldered. That 0.08Ohm cap that does 0.23Ohm via the cables meaured 0.45Ohm via the same cables when it was still soldered to my dead MS-6199VA board. I can't explain that one yet, let alone easily calculate some factor to subtract. Any ideas on what's going on?

Reply 2 of 13, by BloodyCactus

User metadata
Rank Oldbie
Rank
Oldbie
dionb wrote:

I got some completely different values when caps were soldered. That 0.08Ohm cap that does 0.23Ohm via the cables meaured 0.45Ohm via the same cables when it was still soldered to my dead MS-6199VA board. I can't explain that one yet, let alone easily calculate some factor to subtract. Any ideas on what's going on?

you cant test esr in circuit, there can be other caps in parallel etc. gives false readings. only test out of circuit.

--/\-[ Stu : Bloody Cactus :: http://kråketær.com :: http://mega-tokyo.com ]-/\--

Reply 3 of 13, by dionb

User metadata
Rank l33t
Rank
l33t
BloodyCactus wrote:
dionb wrote:

I got some completely different values when caps were soldered. That 0.08Ohm cap that does 0.23Ohm via the cables meaured 0.45Ohm via the same cables when it was still soldered to my dead MS-6199VA board. I can't explain that one yet, let alone easily calculate some factor to subtract. Any ideas on what's going on?

you cant test esr in circuit, there can be other caps in parallel etc. gives false readings. only test out of circuit.

There definitely were other caps in parallel here - although I'd expect that to reduce the reading, not increase it.

If you can't measure ESR in circuit, is there any way you can measure the quality of a cap in-circuit. If you have to de-solder all of them anyway to test, you might as well not bother with the ESR tester and just replace the lot every time...

Reply 4 of 13, by canthearu

User metadata
Rank Oldbie
Rank
Oldbie

My guidelines:

a) it is fairly rare to find capacitors that look perfectly fine, but are broken. It happens, but I wouldn't go looking for this problem unless you are explicitly experiencing problems.
b) You can't make a blanket statement on a capacitors state regarding the age of capacitors. I see plenty of 20 year old or older capacitors that are running just fine.
c) With that said, don't futz around with capacitors that look fine when the component is working fine. Sure, recap if the component is not working properly, or if the capacitor is bloated or leaking, but otherwise, recapping is probably a waste of time (there will be exceptions to this rule of course, for components with known capacitor problems)
d) You can test in circuit, but be aware of what other components in parallel or series can do to your readings. Capacitors in parallel would normally reduce the resistance your ESR meter returns, but you can still see if the capacitors on a VRM are way out of spec using the ESR meter.

Reply 5 of 13, by Jo22

User metadata
Rank l33t++
Rank
l33t++
canthearu wrote:

You can't make a blanket statement on a capacitors state regarding the age of capacitors.
I see plenty of 20 year old or older capacitors that are running just fine.

I second that. As an eletronic hobbyist, I often use old caps -except electrolytics- my father has collected when he was young(er).
To my surprise, the average of them worked just fine in my eletronic projects (I checked them with our LC meter beforehand).
And I'm talking about really old types, such as tubular capacitors and metal foil types from the 1960s/70s.
Same goes for old resistors that predated the colour codes and were human-readable (resistance written in plain text).
That beeing said, I can also understand if someone prefers *new* caps for restoration projects.

Attachments

  • oldcaps.gif
    Filename
    oldcaps.gif
    File size
    379.45 KiB
    Views
    962 views
    File license
    Fair use/fair dealing exception

"Time, it seems, doesn't flow. For some it's fast, for some it's slow.
In what to one race is no time at all, another race can rise and fall..." - The Minstrel

//My video channel//

Reply 6 of 13, by Ozzuneoj

User metadata
Rank Oldbie
Rank
Oldbie
dionb wrote:

If you can't measure ESR in circuit, is there any way you can measure the quality of a cap in-circuit. If you have to de-solder all of them anyway to test, you might as well not bother with the ESR tester and just replace the lot every time...

I like to look at this a different way. If you have all the proper tools to desolder and then measure the capacitance and ESR of a cap, you could greatly reduce the amount of time (and risk) spent working on a board. Unless the board is a particularly huge pain in the butt, it may not be that hard to remove one or two caps and if you've got an ESR meter, it only takes a minute to measure it. If you're intended to replace every single one regardless of whether they need it or not, then you have to set aside time to look for ideal replacements, ordering them, waiting for them and then removing every cap and soldering on new ones, with all of the risk that entails. Not to mention the cost of ordering and shipping just the ones you need every time.

If I had an infinite stock of all the capacitors I could ever need for any board and it was within arms reach, then sure, it wouldn't make sense to remove one and put it back on. But if a board is having issues or is extremely old to the point that electrolytics may be questionable and I take some time to remove and test a few caps to see if they are aging poorly, why go to all the trouble and expense of ordering replacements and redoing the entire board if the few I test are still well within spec? I can just put them back on and be done with it.

Yes this is a cheap way to do things. No, I'm not an EE, I'm a dirt poor PC hobbyist with not a lot of spare time. 🤣

Last edited by Ozzuneoj on 2018-06-30, 01:59. Edited 1 time in total.

Now for some blitting from the back buffer.

Reply 7 of 13, by canthearu

User metadata
Rank Oldbie
Rank
Oldbie
Ozzuneoj wrote:

Unless the board is a serious pain to work on (many layers and lots of mass that absorbs heat)

That describes about every PC motherboard that was made in the last 20 years. Especially around the VRMs where the caps usually die.

The other risk is that you start damaging vias and pads on your board when you are removing, replacing capacitors.

Reply 8 of 13, by Ozzuneoj

User metadata
Rank Oldbie
Rank
Oldbie
canthearu wrote:
Ozzuneoj wrote:

Unless the board is a serious pain to work on (many layers and lots of mass that absorbs heat)

That describes about every PC motherboard that was made in the last 20 years. Especially around the VRMs where the caps usually die.

The other risk is that you start damaging vias and pads on your board when you are removing, replacing capacitors.

I agree, I guess I understated the difficulty of doing this. I was hurrying with my previous post, so I'll edit that portion. I certainly wasn't trying to make it sound like a trivial thing that everyone should do. Many recent boards are a pain in the butt to work on and you do have to be very careful with the vias and pads on any board. I've damaged some myself due to inexperience, improper tools and general stupidity. That said, since getting better tools and more experience I have had much better results. Still, if you try for five minutes on one cap and it just won't come off, quit while you're ahead before you damage something. A change in method is likely needed, or it might just be too big of a pain to bother with.

This is a big part of my reason for creating this thread. If we can gather and record some actual data to show how common capacitors on PC hardware have aged, we may be able to minimize the number of clumsy repair attempts that are made unnecessarily. There are lots of highly experienced electronics gurus out there that make it sound like a beginner should dive right in with a Radioshack desoldering bulb and a $10 iron to recap an entire board simply because its 20 years old and the caps "can't possibly be good anymore". This kind of advice has probably led to a lot of ruined boards and frustrated tinkerers.

Now for some blitting from the back buffer.

Reply 9 of 13, by canthearu

User metadata
Rank Oldbie
Rank
Oldbie
Ozzuneoj wrote:

This is a big part of my reason for creating this thread. If we can gather and record some actual data to show how common capacitors on PC hardware have aged, we may be able to minimize the number of clumsy repair attempts that are made unnecessarily. There are lots of highly experienced electronics gurus out there that make it sound like a beginner should dive right in with a Radioshack desoldering bulb and a $10 iron to recap an entire board simply because its 20 years old and the caps "can't possibly be good anymore". This kind of advice has probably led to a lot of ruined boards and frustrated tinkerers.

I absolutely agree with you there. You can't just look at a 20 year old cap and say it is bad because it is 20 years old. Because it is rarely so.

In practice, I have found that if the caps look good, and the device runs fine (and I mean under heavy load), there is more risk messing it up by recapping it than by leaving it alone.

Pretty much all the caps I remove that looked good also tested good (using my proper Peak ESR70 meter) and replacing them usually won't fix the problems I was having. Even the caps you measured above that were slightly out of spec would have never caused problems if they are left alone, as nearly all electronics have a safety margin built in for component aging.

And given most vintage hardware only gets used occasionally, these parts are not taking significant new abuse over time, so even less chance of caps going bad.

The only exceptions I will make are for caps that are known to be extremely notorious, for example the surface mount Su'scons electrolytic, or the Fuhjyyu capacitors in old Antec PSUs, or the capacitors on some of the old Dell optiplex motherboards. Some brands and batches of capacitors are known to be dodgy, in either fabrication or application, and should be replaced during refurbishment of a vintage item.

Reply 10 of 13, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
BloodyCactus wrote:
dionb wrote:

I got some completely different values when caps were soldered. That 0.08Ohm cap that does 0.23Ohm via the cables meaured 0.45Ohm via the same cables when it was still soldered to my dead MS-6199VA board. I can't explain that one yet, let alone easily calculate some factor to subtract. Any ideas on what's going on?

you cant test esr in circuit, there can be other caps in parallel etc. gives false readings. only test out of circuit.

Just remember these rules for in circuit testing:

Cmeas = Ctest // Cext therefore Ctest = Cmeas - Cext
The capacitance of a component when tested in circuit will be less than indicated.

Rmeas = Rtest // Rext therefore Rmeas = RtestRext/(Rtest+Rext) and Rtest=RmeasRext/(Rext-Rmeas)
The ESR of a component when tested in circuit will be greater than indicated, although the reading will tend to be dominated by the lesser resistance, be that the ESR of the capacitor or the equivalent parallel resistance of the balance of the circuit.

All hail the Great Capacitor Brand Finder

Reply 11 of 13, by dionb

User metadata
Rank l33t
Rank
l33t

My problem with replacing only the visibly or measurably dead caps is that that usually results in the next dubious cap along dying soon after.

Armed with my ESR meter I attacked an Abit BE6-II I had in the 'needs TLC' box. 2/3 of the caps are visibly bulging, one has leaked. Amazingly it did actually POST when tested, but there can be no doubt it needs work to ensure its continued existence. It has three kinds of bulging caps. The worst are Jackon 1500uF 6.3V beasties. The leaking cap measured 648uF, 4.4Ohm. Electrically it wasn't the worst though, one other bulging cap as 423uF, 7.2Ohm, and a third wouldn't even measure as cap, my meter kept reading it as a diode. Now, these three and a few others were obviously dead. I measured one though that still looked flat: 1473uF, 0.10Ohm. So it was well within spec, even though it's identical to the utterly failed caps.

If I was just basing my work on the ESR meter, I'd happily leave that one in place. But it's no better than the bulging, leaking ones. They were just stressed more and by failing protected this peripheral cap in the same parallel group. If I were to leave it there and replace the dead ones with good new caps, a few months down the line that seemingly good Jackon would be in the same dying/dead/bulging/leaking state as the ones I replaced. Been there, done that - even lost a board to it.

So the relevance of testing individual known-bad caps on a board where some of the same type have already failed seems a little thin.

As I see it, the biggest use of an ESR meter is to test seemingly good pulls from dead boards (like those 1990s Nichicons) to be sure they are as good as they look, and perhaps to check samples of caps on a board where one type has clearly failed, but other types from the same band haven't. Even in those cases I'm a bit doubtful, as you're testing what the cap is doing now, not how long you can expect that measurement to hold. If I compare my brand-new United Chemicon 6.3V 1500uF caps with that (so-far) good 1999 Jackon with the same specs, they give almost identical results. But the one can (hopefully) be trusted for years in a demanding role, whereas the other will bulge and leak as soon as anything demanding is done with it.

Reply 12 of 13, by dosgamer

User metadata
Rank Newbie
Rank
Newbie

I've had a DE-5000 for a couple of years now. Excellent meter and very useful. And it's a real LCR meter (which is a vector impedance meter). Unlike:

The "BSIDE ESR02 Pro" is not an ESR meter. It's a type of meter of which there is a buttload of these out there (I call them shitmeters, like in shitposting). They measure something that may or may not have any relation to ESR. Usually some form of Xc.

Do this test: measure any small value ceramic capacitor (like 100nF or less). It will show a stupid high ESR, even though these capacitors have an ESR that is in the milliohm range.

But it's even worse, the true test of a capacitor is not the ESR, it's D (also called DF or tan delta) which is the dissipation factor. It's the capacitor equivalent of Q for a coil. ESR alone doesn't tell you anything, because what might be a good ESR for one cap might be bad for another type of cap, whereas D is an independent measurement of how lossy a capacitor is. Shitmeters can't measure that, because to do that you need to measure both amplitude and phase shift to get the real and imaginary parts of the impedance, which is quite complicated. To measure D you need a true LCR meter, like the DE-5000.

Another option is to use a frequency generator and an oscilloscope, not very handy, but it works too and gives accurate results.

Coppermine Celeron 800 @ 1.12GHz (8x140) - Asus P2B Rev. 1.12 - 256MB PC133 CL2 - Voodoo5 5500 AGP - SB AWE64 CT4520 - Roland SCC-1 - Intel Pro/1000GT - 1.44MB Floppy - ATAPI ZIP 100 - 120GB IDE - DVD-ROM - DVD-R/RW/RAM - Win98SE

Reply 13 of 13, by Ozzuneoj

User metadata
Rank Oldbie
Rank
Oldbie
dionb wrote:
My problem with replacing only the visibly or measurably dead caps is that that usually results in the next dubious cap along dy […]
Show full quote

My problem with replacing only the visibly or measurably dead caps is that that usually results in the next dubious cap along dying soon after.

Armed with my ESR meter I attacked an Abit BE6-II I had in the 'needs TLC' box. 2/3 of the caps are visibly bulging, one has leaked. Amazingly it did actually POST when tested, but there can be no doubt it needs work to ensure its continued existence. It has three kinds of bulging caps. The worst are Jackon 1500uF 6.3V beasties. The leaking cap measured 648uF, 4.4Ohm. Electrically it wasn't the worst though, one other bulging cap as 423uF, 7.2Ohm, and a third wouldn't even measure as cap, my meter kept reading it as a diode. Now, these three and a few others were obviously dead. I measured one though that still looked flat: 1473uF, 0.10Ohm. So it was well within spec, even though it's identical to the utterly failed caps.

If I was just basing my work on the ESR meter, I'd happily leave that one in place. But it's no better than the bulging, leaking ones. They were just stressed more and by failing protected this peripheral cap in the same parallel group. If I were to leave it there and replace the dead ones with good new caps, a few months down the line that seemingly good Jackon would be in the same dying/dead/bulging/leaking state as the ones I replaced. Been there, done that - even lost a board to it.

So the relevance of testing individual known-bad caps on a board where some of the same type have already failed seems a little thin.

As I see it, the biggest use of an ESR meter is to test seemingly good pulls from dead boards (like those 1990s Nichicons) to be sure they are as good as they look, and perhaps to check samples of caps on a board where one type has clearly failed, but other types from the same band haven't. Even in those cases I'm a bit doubtful, as you're testing what the cap is doing now, not how long you can expect that measurement to hold. If I compare my brand-new United Chemicon 6.3V 1500uF caps with that (so-far) good 1999 Jackon with the same specs, they give almost identical results. But the one can (hopefully) be trusted for years in a demanding role, whereas the other will bulge and leak as soon as anything demanding is done with it.

If one cap of a certain type has visibly failed, yes, you should always replace all of them. If you pull a couple from a board and test them and one is way out of spec, then the rest are probably not far behind, so they should all be replaced. Generic garbage caps that are bad or are known to go bad should always be replaced.

There's a reason why I only have good name brand caps laying around to test. And I would never suggest leaving one garbage cap on a board if others of the same brand and size are being replaced. 90% of the work will be done if you have to order and replace the other caps.

Now for some blitting from the back buffer.