General Advice for Decapitators - Tripp Litely

[SgtRock]

Greeting EEVBees:

--I had pretty much decided that the capacitor plague was over after not having replaced a capacitor for lo these many months. Then, four projects in a row in the last couple of weeks have bad capacitors;  1) A Westinghouse 22" CFL LCD monitor, repaired. 2) A one year old LG Air Conditioner, under repair. 3) An 8 year old Dell motherboard, scrapped. 4) A Tripp-Lite 900LCD UPS, just now repaired and working.

--Below is some general advice for those who want to do multiple, many cap recapping projects. Also for those who want to work on expensive equipment belonging to someone else, without embarrassing themselves.

1) Use a good professional desoldering tool that has a vacuum pump, Hakko, Dennon or even one of the less expensive knock-offs. I use a Hakko 470, with the bench top enclosure holding the pump and PS, and the light weight Hakko 802 gun, all for less than $150, shipping included. It is a bear and it never picks up anything it is not supposed to. I have yet to buy a new  tip. Hakko USA offers all parts at very cheap prices. Smart! Trying to use the old $5 bang stick and wick technique on crowded expensive boards with small through holes and unleaded solder is going to cause you to pull up a trace, suck up a via, short or ruin an SMD, or otherwise kill the patient. Also it can take forever going back and forth with the stick & wick and the soldering iron to re-wet the leftover unleaded with leaded. Unleaded solder can be a bit of sticky wicket. It does not like all come out. Using a DS gun you just hold the gun tip against the lead and hole an re-wet with leaded and immediately re-vac, done. Often components just fall out. One of these will definitely improve your confidence. 50 Capacitor Motherboard, no problem. Even Dave had to finally breakdown and buy one of these. Here is the link to Dave's Video;  http://www.eevblog.com/2013/11/01/eevblog-542-zd985-desoldering-station/. As you can see by Dave's video the Rhino works as well as the Hakko, but is more cheaply made, and the spring filter has a large hole at the pointed end, unlike the Hakko, and consequently drops solder on the ceramic fiber rosin filter, ruining it much faster than the Hakko. Also the Hakko has a red telltale on the side of the gun, that lets you know when the vacuum begins to drop. Still the Rhino indeed "works a treat", and Dave aptly demonstrates the utility of the device.

2) You will need a rigorous procedure to make sure you can get the replacement caps back into the correct locations. Photographs are helpful, but usually cannot see everything. You could just replace the caps one at a time, but often you have to remove some to identify them. And then what are you going to do? Re-install?? If you have only a few caps, a sketch will suffice, but it must be checked three times or disaster will result. If the cap numbers are marked on the board, my method is to remove the caps one by one and write C16 or whatever on the body of the cap using an indelible black marker, or white paint pen (see picture below), and one by one drop them into a labeled pill bottle. Also make sure the polarity of the caps are clearly marked on the board, or mark them yourself. I just blew one up this week, being inattentive.

3) When ordering replacement caps, I prefer Digikey, or Mouser if Digikey does not have. In addition to making sure that the capacity and voltage are correct, you will need to make sure that the replacements can physically fit. Sometimes you can go wider or taller, or lay the cap on its side, and sometimes you cannot, so you need to measure size. Digital Electronic Calipers (carbon fiber, see picture below) are way cheap on eBay and make things very easy compared to using a ruler. I usually just take the already labeled bad caps, one by one, and measure them for height and width as I am ordering them from Digikey. This way the Digikey order is the list, need not be put on paper. Then, one by one the bad caps are put back into a labeled pill bottle. If I have a mix of 25V and 16V caps of say 22uF, I often just order all 25V if they will fit, to shorten the line item list, and avoid errors. Often I order one more than I need of each cap value, just to cover screw ups, build an inventory, and spread the postage cost per item.

4) Once I discover a bad cap, I generally just replace all of the caps, except the large power supply filter caps, which are almost always OK. Sometimes I measure the ESR (using the AnaTek Blue ESR Meter which I love) of all the caps and make a chart noting also the Brand, Capacitance, Voltage, and Temperature, as I have done below with the 9 caps from the Tripp-Lite OMNI900LCD UPS. You can easily see the virtue of replacing all the caps. What is amazing is that the TRIPP-LITE UPS worked for the longest time with nearly all of the capacitors out of spec. This is a very typical situation, so if you have a non-working device, and you see a swollen cap, you can just replace all but the power filter caps, and Roberto es su tio.

NR = No Reading.
C1   Gemcon   85C  100uF  25V  ESR  2.8    Bad.  NR = No Reading. New Nichicon ESR is 0.09.
C2   Gemcon   85C    10uF  25V  ESR   NR   Bad, the new Nichicon ESR is 0.07
C3   Gemcon  105C  680uF  25V  ESR 0.12  Bad, the new Nichicon ESR is 0.01
C4   Gemcon  105C  680uF  25V  ESR 0.24  Bad, see above
C10 Gemcon    85C  470uF  16V  ESR  NR   Bad, the new  Rubicon ESR is 0.04.
C16 Gemcon    85C    22uF  16V  ESR  NR   Bad, the new Nichicon ESR is 1.6.
C17 Gemcon    85C    22uF  16V  ESR  NR   Bad,  see above.
C19 Jamicon  105C 1000uF  25V ESR  0.03   Good, but the new Panasonic ESR is 0.01.
C23 Gemcon    85C    22uF  16V  ESR  NR   Bad, see above.

--Moral of story is Gemcon is Wun Hung Lo. Other than the bad caps the TRIPP-LITE OMNI900LCD UPS had only one small potential problem. One of the 3W resistors was stood on end to save room, and its inertial mass was causing it to rock and lift up the traces, creating a potential open. I traced out the connections and made the lead and trace more mechanically secure by bending over the soldering leads of a cap it is in parallel with, as it was too crowded to use glue on the top side. As a general matter I find that the quality of construction of the APC brand UPSs to be superior to the TRIPP-LITEs. In addition to using better caps, APC cleans their boards after soldering. With the TLs you have to clean the flux off of the board in order to inspect for bad joints.  As uncle Wun Hung Lo once said on eBay "Happy for you to desoldering please".

"To learn who rules over you, simply find out who you are not allowed to criticize."
François-Marie Arouet - Voltaire
1694  -  1778
 
Best Regards
Clear Ether

[NiHaoMike]

On the A/C unit, if the original cap is a 370V, use a 440V cap as a replacement. It will last longer.

[TheWelly888]

Oh! I thought this thread was going to be about making guillotines! 

[BravoV]

"And there's the cane from Citizen Kane. Wait, there was no cane in Citizen Kane."

— Lisa, The Simpsons, "

 

[codeboy2k]

I like to make voltage deregulators with my decapitators.

BTW... good advice overall, Sgt.

Thanks!

[Bryan]

Depends on the ESR meter, but usually it's better to remove the cap from the board.

[SgtRock]

Dear 3roomlab:

--As a general matter when I open a device and look at the circuit board, and I see a swollen or leaky cap, I then remove all of the one by one, marking them and recording their location in a fool proof manner so I can get the new replacements back into the correct locations. I already know that I am going to replace them all, but usually if I am not pressed for time I try to measure and record the capacitance and the ESR of each cap, just to add to my general knowledge of cap behavior and create records that can be compared later.

--If none of the caps look bad under visual inspection, I start testing ESRs. If and when I find one with an abnormal ESR (High Reading, No Reading or 0.0 Reading) I remove the cap and retest. If it is bad or even marginal I proceed as above and replace all except sometimes the larger filter capacitors.

--Sometimes when I am repairing a device, for reasons other than bad caps, I will replace all of the caps anyway if they are cheap no names or known Wun Hung Lo, prone to failure brands. This is very common in virtually all brands and kinds of equipment, anything with caps in it.

“Don't look at me in that tone of voice.”
Dorothy Parker  1893  -  1967

Best Regards
Clear Ether

[The Electrician]

SgtRock sent me all the bad caps he pulled out of the Tripp Lite.  I swept them with the Hioki impedance analyzer and I'm going to post them here.

To help the beginners understand these sweeps, let me explain.  Impedance generally is made up of two parts--the real part which is resistance and the imaginary part which is reactance.  An important difference in these two things is that when a current flows through a resistance, heat is generated and energy is lost.  When a current flows through a reactance no heat is generated and no energy is permanently lost.  Some energy is temporarily stored in the reactance, but it is later returned to the circuit later in the cycle.  The real part of the impedance of a capacitor is called ESR (equivalent series resistance), and on the analyzer is denoted Rs.  It's this heat generated in the ESR of a capacitor that degrades it and eventually leads to failure if there's enough heat generated.  When a capacitor is used in some particular circuit, a certain AC current probably flows through it (especially if it's in a power circuit).  If the capacitor ESR increases, then more heat is generated by that current, causing the cap to dry out faster, and thus increasing the ESR even more.  You can see where that leads.

Electrolytic capacitors generally aren't as good as other types such as film, ceramic, mica, etc., and this shows up as a higher ESR for the electrolytic compared to a film of the same capacitance, for example.

Here's a good reference that shows (Figure 4) the kind of curves you get when the impedance of a capacitor is plotted on an impedance analyzer:

http://www.murata.com/products/emicon_fun/2013/02/cap_en14.html

The sweeps I'll be showing display the impedance magnitude as green and the ESR as yellow.  Impedance (Z) can be written as Z = R + jX, where R is the real, or resistive, part and X is the imaginary, or reactive part.  The magnitude of the impedance is SQRT(R^2 + X^2).  This means that the impedance magnitude is never less than R or X by itself.  On the sweeps, the green curve (impedance magnitude) never goes below the yellow curve (real part or Rs or ESR).

Here's a sweep of an OSCON electrolytic capacitor.  These are probably the lowest ESR electrolytics you can get.  Notice how the green curve (impedance magnitude) decreases until its minimum just touches the yellow (ESR) curve.  When the green curve has a relative narrow minimum and only touches the ESR curve at one place, this is a really good capacitor:



Here's a sweep of a Panasonic HF (high frequency) series electrolytic.  Notice how the green curve touches the yellow over a wider range, and the minimum of the green curve is wider:



Here's a sweep of a more typical small radial lead electrolytic, not particularly low ESR, but perfectly adequate for ordinary use.  Notice that the minimum of the green curve is broader yet, and remains in contact with the yellow (ESR) curve over a wide frequency range:



Finally, here's a sweep of an electrolytic that is marginal; probably on the way to failure.  As the ESR of a capacitor increases, the yellow (ESR) curve moves up in the image, and this causes it to be coincident with the green curve (impedance magnitude) over a wider frequency range.

[The Electrician]

Note that the ESR shown at 100 kHz for all the sweeps I'm going to show are typically higher than SgtRock measured.  This is because the temperature at SgtRock's place in Florida is higher than where I am; it has been quite chilly here lately.  The lower temperature where I am causes the ESR to be higher.

Here's the sweep of C1 from SgtRock's list:



SgtRock measured an ESR of 2.8 ohms, whereas my sweep shows (the last yellow number at the top right of the screen) 4.38 ohms; a higher value due to the chilly temperature here.

Notice how the yellow curve is coincident with the green curve for a very wide range.  This is a sign of an ESR higher than a new cap would exhibit.

[The Electrician]

Here's the sweep of SgtRock's C2.  This is a good example of a totally failed cap.  SgtRock's meter was unable to measure the ESR, but on the sweep we see that it's 2.36k ohms.



I also did a sweep showing the D (dissipation factor), and at 120 Hz it's 3.2; definitely a failed cap.



Incidentally, none of SgtRock's bad caps were swollen; even the ones with very high ESR, such as this one showed not the least bulging of the top of the cap.

[The Electrician]

Here's the sweep of SgtRock's C3.

[The Electrician]

Here's the sweep of SgtRock's C4.  C4 is a little bit worse than C3:

[The Electrician]

Here's the sweep of SgtRock's C10.  He was unable to measure ESR, but we see that it's 196 ohms.

This cap has the characteristic sweep of a totally dried out cap.  The green curve has no minimum followed by a rise; it's just a steady downward slope, with a high ESR.

[The Electrician]

Here's the sweep of SgtRock's C16.  This one is really dried out.  At low frequencies the ESR is off scale, greater than 10k ohms.

[Galaxyrise]

Interesting reading, both of you

the green (ESR) curve

But I was getting confused at first because the colors are consistently backwards in your description.  Once I noticed that the colors were labelled in the image, and that the labels didn't match your colors, the writeup made a lot more sense

[The Electrician]

Interesting reading, both of you

the green (ESR) curve

But I was getting confused at first because the colors are consistently backwards in your description.  Once I noticed that the colors were labelled in the image, and that the labels didn't match your colors, the writeup made a lot more sense

Sorry about that.   

Should be fixed for the next people who read it.

[The Electrician]

Here's the sweep of SgtRock's C17.

[The Electrician]

Here's the sweep of SgtRock's C19.  Just looking at this, I would have said that it was a good cap; of course, one should check leakage and capacitance too.

[The Electrician]

Here's the sweep of SgtRock's C23.

[Galaxyrise]

Here's the sweep of SgtRock's C19.  Just looking at this, I would have said that it was a good cap; of course, one should check leakage and capacitance too.

Well, he did say "C19 Jamicon  105C 1000uF  25V ESR  0.03   Good, but the new Panasonic ESR is 0.01.", which seems pretty consistent with your measurement.

[SgtRock]

Greetings EEVBees:

--Well this on is on the bench again. It turns on when you plug it in, and turns off when you unplug it, and the voltage display is blinky. I believe that the 4 pin device marked V833Y with F847C below is a Fairchild Optocoupler. Pins 1 and 2 appear to be a diode and 3 and 4 appear to be a transistor output. When the input voltage across 1 and 2 reaches about 0.2 Volts the resistance between 3 and 4 goes from infinity to somewhere between 0 and several hundreds of ohms, but jumps around and is "blinky". Just now it has completely given up the ghost with pins 1 - 2 now showing a resistance of 370 Ohms both ways.

--My question is: What should I replace it with. I cannot find a data sheet. When I Goggle it I only get teasers from AliHungLo. Any suggestions for a jelly bean replacement would be sincerely appreciated.

“The main difference for the history of the world if I had been shot rather than Kennedy, is that Onassis probably wouldn't have married Mrs Khrushchev.”
Nikita Sergeyevich Krushchev 1894 - 1971

Best Regards
Clear Ether

[ovnr]

--My question is: What should I replace it (F847C) with. I cannot find a data sheet. When I Goggle it I only get teasers from AliHungLo. Any suggestions for a jelly bean replacement would be sincerely appreciated.

Well, the 847 part is immediately recognizable as a jellybean optocoupler, so I'd just look for a PC817. The 847 is supposed to be 4-channel but yours is 1-channel; hence the PC817 is a better fit. The C suffix refers to the Current Transfer Ratio - basically how sensitive it is. Just find one with a CTR of 200% or higher - I'd probably go with one of these.


Disclaimer: This is only an educated guess, and I may not be held responsible for having a laughing fit if something catches on fire.

[SgtRock]

Dear Ovnr:

--Thanks, I was pretty sure that someone with a little more experience, would recognize the part. I just ordered some from DigiKey.

"Happy for you to desoldering, please."
Wun Hung Lo 1948 -

Best Regards
Clear Ether

[TerraHertz]

Those Hioki cap impedance analyzer results are really interesting. Hadn't seen such things before.

SgtRock, seems like you're encountering the Ebola of bad caps plague. Reading this thread yesterday I thought "I've been lucky lately, no bad caps."

Well that was unwise. Literally an hour later... I'd powered up an old Sony studio monitor that had been sitting idle for 'a while', and was still setting up the color bar generator when odeur de fumée magique made itself known. Quickly turned mains off at the wall. Smoke continues, followed by a POP about 5 seconds after power removed, and more smoke.

Starting to reach for the fire extinguisher (there's two right by the bench) when the smoke trailed off.
Removed the monitor outer case, couldn't see anything burned or exploded looking.

Here's a 'stink finder' trick probably not many techs use. Maybe because it makes one look stupid, but still...
Get a couple of feet of plastic hose, approx 1cm dia. Hold one end against a nostril, move the other end around the suspect area while breathing in through nose. It works pretty well.

Anyway, it was a 4.7uF 250V electro, on the CRT gun driver board. From a quick track trace, it's just a bulk filter cap on a 150V supply.
Rats, I don't think I have any small HV electros in stock. Unless I go hunting through a box of dead camera flash boards I have, um.. somewhere in the upstairs middens. Well that's.... oh wait!

There's a small pile of accumulated mess on one end of the bench atm. Including a dead 'green light bulb' (what a joke) that carked it recently. Broke the case seam open, and yes, it has a 6.8uF 400V electro, the same physical size as the dead cap. Checked the fluoro's leads, and one of the filaments is open circuit, so this cap will be fine. And it was; the monitor now works.

How serendipitous! So those bulbs are good for something besides making rooms hopelessly dim.
 

[SeanB]

I would still be wary of that green electrolytic, it will have a short life due to it's construction and also having been cooked to medium rare in the dead light. I often find them having not failed, but within a short while after you reuse it it fails.