Show us your ugly repair

[TerraHertz]

Go on, let us see the most ugly hacks you've ever done. A cathartic admission of shame. Or something.

Here's one of mine. A spontaneous repair attempt, still in progress.

It began when DavidG asked about a HP 8161 dead part: https://www.eevblog.com/forum/repair/what's-this-part/
I have a HP 8160A and manual, with that same CPU board. It's been on the shelf 'awaiting repair' a long time. I opened it for pics to post in that thread, then after putting it back together powered it up to evaluate whether to have another attempt at repair. (Now that I have the manual.)

The power supply had infamous Rifa polyester mains caps; one blew up immediately. So suddenly I'm *definitely* attempting repair. Replaced all 7 Rifa caps, then noticed another card in the power supply has 10 wet foil tantalum aluminium-can caps. Checking those, they are ALL utterly kaput. Supposed to be 650uF 20V and 180uF 30V, all axial.

For the 650uF I could use some recently bought low ESR 1000uF 16V (yes, lower voltage is OK, the circuit is actually around 7V.)
But the others...

Urrgh. Due to general lack of cash I don't want to buy from local distributors (and it's the weekend) but I also don't want to wait 6 weeks for an Aliexpress delivery. Hmm.... Oh wait!  I recently acquired a huge stock of 'old junk' in nicely sorted containers.
https://www.eevblog.com/forum/buysellwanted/free-(aus)-world's-largest-electronics-parts-collection-as-seen-on-eevblog/
I'd thought I would probably throw out all the old electrolytic caps, but haven't yet. Are there any suitable for this?

Found two tubs, one 220uF 35V, and 470uF 35V. But they are all old, used pigtail, with short desoldered leads. Sigh. But OK, what the heck.
Measuring caps from the 220uF tub, they are (as expected) a range of values less than 220uF. Normal electro aging distribution, see 1st pic of some sorted into value ranges. Most still above 180uF and usable for this, but I thought I'd add some extra margin and go for the 470uF ones. These also exhibited an aging distribution, but picking best ones allowed lots of margin.

They are of course much bigger than the old wet foil tantalums, and the wrong lead format. Hence the nasty kludge in the 2nd photo. But who cares!

While doing this it occurred to me that one use for a *lot* of used, old electro caps, all sorted into value and voltage bins, would be to do a statistical review of exactly how bad they are. By type, manufacturer, etc.

[xrunner]

Uhg - Glee! 

[Cubdriver]

Just for grins, have you taken a crack at re-forming any of them?

-Pat

[vk6zgo]

"Fit,form & function"------ the most important of these is "function".

The soldering looks OK,it seems mechanically rigid,it works---------OK,it isn't pretty,but Electronics isn't about aesthetics!

Most of my time in Electronics was spent in Radio & TV Broadcasting,where equipment was needed "yesterday".
A good "bodge" which restored full "in spec" performance  was often superior to waiting for the "correct" part to arrive.

Three fine "bodges" come to mind:-

(1) A 6v supply in an NEC TV transmitter was supplied by a 200v AC rail,which had separate fuses for each module connected to the rail.
We inadvertently blew the fuse,which was hidden in the "guts" of the Tx.

To get out of trouble,my homemade "VK Powermate"was pressed into service.
Various other priorities intervened,& it stayed that way for a couple of weeks!

(2)Same place,but I wasn't there at the time.
The  *28v supply for the FET RF driver stages of a Thomson TV Tx "croaked". 
* Note edit:-I somehow deleted the "power supply" bit earlier,so it didn't make much sense!!

Time for a replacement from France:----------months!
Time for local acquisition  of parts,& repair.probably months as well.

As luck would have it,there was a (very much larger) NEC 28v power supply sitting at the  Station I was at,so they borrowed it,set it out on the floor behind the Thomson,made the requisite connections,& away it went,so the people in the area didn't miss their Commercial TV.

(3) Later,when I was at the TV Studio,an Electrohome BW Monitor came in for repair.
As part of the H output circuitry it used a non-polarised 22uF 400v electrolytic,which was as "dead as a doornail"

Looked in the RS catalogue,& there they were-------much smaller physically,too!

Marvelling at the advances in technology,I fitted one of these.
You've guessed it!-----it lasted 10 seconds!!.
It turns out non-polarised low ESR capacitors are quite large.

They were also unobtainium,so the "E'home" ended up with a "Christmas tree " of Polyester "greencaps".
Ugly?---Certainly!
Fully functional?---Definitely!!

[alsetalokin4017]

This was before I knew anything about SMD repair and only had a cheap Weller soldering station to work with.   

The repair works fine and is still in use today.

[edpalmer42]

But sometimes the ugly repair is the right way to do it.

I worked for a telephone company.  One day, one of the young engineers was working in the lab on a circuit that just wasn't cooperating.  Nothing was making sense.  This was a commercial board, we had full documentation and the board was believed to be in good working order.  I happened to be there, saw that the young guy was getting frustrated and suggested that he pull the card out of the shelf and give it a good rap on the bench.  With a very puzzled look on his face he pulled out the board and gently tapped it on the bench.  I said "No, no - like this"  BANG, BANG, BANG.  "Okay, try it now."  It worked.  The look on the guy's face  was priceless.  Another guy that worked for me saw all this and was practically rolling on the floor laughing.

We had both seen multiple mercury-wetted reed relays on the board.  Unless the board is kept upright, the mercury will ball up and short the contacts.  So the correct, but ugly, repair is to smack it on the bench to shake the mercury out of the contacts!  Once I explained it to the young guy, we all had a good laugh. 

Ed

[TerraHertz]

Just for grins, have you taken a crack at re-forming any of them?

Hmm... No not really. The ones I ended up using, I first charged up to rated voltage, left to sit unconnected a few minutes then measuring the retained voltage. Purpose being to compare leakage. They were all similar, a few volts down but OK. It was just a test to catch any bad ones.

What is the recommended profile for reforming old electrolytics?
I certainly have lots of old caps to play with, and it would be interesting to see the results.

Oh, and those caps added to the PCB are now hot-melt glued down for extra ugly, and the power supply runs OK. Still to reinstall in the instrument and try to fix that.

[mathsquid]

I replaced a ribbon cable in a TI-83+ using stripped wire-wrapping wire.  It worked, but the results were ugly.

I covered the wires with tape before reassembly, but after this awful picture was taken.

[wraper]

Repairing broken off micro USB connectors.

[wraper]

more pics

[128er]

Not my work. But it realy is an electronics nightmare.

Got this calibrator nonfunctional from Ebay. The first look at the PCB was like "What the hell? Which Plumber did that?" 

I only replaced a few defective chips and all the leaked eletrolytics. Yes, they used radial leaded caps with surface mount pads. The thing is one big bodge.

[128er]

More pictures

[Urs42]

Don't try this at home (or at work ;-)

Before and after:

[AlfBaz]

This hack is of an FE5680B. A 74F74, dividing the 20Mhz signal down to 10Mhz. Found a bit of space and glued it down, dead bug style.

[Cubdriver]

Just for grins, have you taken a crack at re-forming any of them?

Hmm... No not really. The ones I ended up using, I first charged up to rated voltage, left to sit unconnected a few minutes then measuring the retained voltage. Purpose being to compare leakage. They were all similar, a few volts down but OK. It was just a test to catch any bad ones.

What is the recommended profile for reforming old electrolytics?
I certainly have lots of old caps to play with, and it would be interesting to see the results.

Oh, and those caps added to the PCB are now hot-melt glued down for extra ugly, and the power supply runs OK. Still to reinstall in the instrument and try to fix that.

If possible I use a current limited supply*, set the open circuit voltage to the voltage rating of the cap (or a bit below), and the current limit to 30 or so mA.  Connect the cap and turn it on.  Depending on value and how leaky the cap is, it may sit at the current limit for a while as the voltage slowly builds, but if it successfully re-forms you'll eventually see the supply hit the voltage limit setting and then the leakage current will drop off.  Let it hang out till the leakage current stops falling then let it sit at its rated voltage for at least a few minutes (longer is better, just for testing's sake if nothing else), then discharge, cross your fingers and test.  Datasheets may or may not give acceptable leakage current limits for a given cap, but if this info is available it's definitely something to use to determine the cap's health.  There are also some leakage rule-of-thumb numbers in the Sprague TO-4 manual that if I recall correctly apply more to higher voltage electrolytics; I can try to dig them out later if you'd like.

*for higher voltage electrolytics used in tube power supplies I get to use my Sprague TO-4 and babysit it, slowly bringing up the voltage manually while monitoring the leakage current.  That's a PITA.  I need to get my HV current limited supply up and running.

This of course will not work for caps that have dried out, and there is no guarantee that it will work for any, but it's relatively easy to do and worth a shot, especially for caps that no longer have easily available replacements.  FWIW, if you have several of the exact same type and any are noticeably lighter than the others, the light ones are likely dried out.

I've had reasonably good success re-forming caps on HP old test equipment this way, but they were high quality parts to begin with.  With consumer grade, all bets are off, but all it costs is a bit of time to try.

-Pat

[T3sl4co1l]

Does this count?



Tim

[joeqsmith]

Here I will demonstrate the proper way to decouple and terminate a high speed digital system.  Jumper on A2 is a connection for the scope probe. 

[Samogon]

Here I will demonstrate the proper way to decouple and terminate a high speed digital system.  Jumper on A2 is a connection for the scope probe.

Like a headcrab from HalfLife.

[guido]

Only have the 'before' picture, but got this one working...

[joeqsmith]

Sometimes things are more critical and you need to do some planning before turning on the iron.   Here I show how to properly clock some high speed logic. 

[TerraHertz]

@joeqsmith
They are both beautiful. I don't see what anyone could complain about ... electrically.

@Cubdriver: "There are also some leakage rule-of-thumb numbers in the Sprague TO-4 manual that if I recall correctly apply more to higher voltage electrolytics; I can try to dig them out later if you'd like."

I'd appreciate it. Pic below is my existing electros stock *before* adding the ones in the recently acquired collection. (And I'm pretty sure I missed a cube or two more. Terribly disorganized, due to non-completed storage space shelving.)

I won't mix the sets, just add the boxes of 'new' caps that I decide to keep into the storage cube pile these live in.
About half of mine are 'new', as in never used. One place I used to work manufactured in large volume, and had a stores/purchasing dept that sometimes chose to throw out large volumes of surplus components. And I was the resident dumpster diver.  So, lots of caps but a very limited range of values.

[SeanB]

Reforming capacitors I used a power supply and a 10k resistor in the one lead, with a voltmeter across the resistor to measure the drop as they charged, then using it to measure leakage. Just put all the ones of the same voltage in parallel and you will reform the lot all at once. Should drop down in current to a low reading after a few minutes, if not divide to see which one is leaking the most and bin it after ripping the leads out of the rubber bung, so you do not try it again.

Had to do that with a batch of wet tantalums, which were rejected because of high ESR. Inspector on the other side sent them back, saying " visual inspection passed" so they went back again sans any leads.

[AF6LJ]

This thread is fun.

[Cubdriver]

Electrolytic leakage limits rule of thumb from the Sprague Tel-Ohmike TO-4 manual - these are, I'm guessing, from the mid to late 50s time frame, so take them with a big grain of salt.  I'd say for sure if a modern electrolytic doesn't meet these specs it's toast, as modern ones are likely better than what existed when this was written.

The leakage limit depends on the capacitance and voltage rating of the cap, and is calculated from a constant (k) that's based on the cap's voltage rating, the cap's value in uF (C) and a fixed allowance that's added.  It says that the caps should sit at rated voltage for at least 5 minutes before the reading is taken, longer if they are still re-forming and the leakage current is decreasing.

The formula is:  leakage limit (mA) = kC + 0.3

k is:
0.01 for 3 to 100WVdc caps
0.02 for 101 to 250WVdc caps
0.025 for 251 to 300WVdc caps
0.04 for 351 to 500WVdc caps

Using this, an 80uF, 450V cap would have a leakage limit of 3.5mA (0.04 x 80 + 0.3)
a 470uF, 35V cap would have a limit of 5mA (0.01 x 470 + 0.3)

The TO-4 manual that this formula came from also notes that caps that have been unused for a year or more may take 30 minutes + to re-form, and suggests limiting the forming current to 10mA or less to prevent overheating.  I've thus far had no issues with around 30mA, for what that's worth.

-Pat

[Excavatoree]

My repair of my "stand fan" is pretty ugly.  I did do a thread about the problem (microcontroller died), but I never described the final repair.  No pictures, but it's just an 8 pin PIC 12F509 in a 16 pin spot for some obscure Chinese micro controller.    The power and ground pins for the 16 pin were at 5 and 12.  (pins 1 and 8 when the 8 pin package is inserted in the right half.)  Fortunately, that worked out perfectly for the PIC.   

 It gets ugly because I didn't want to bother implementing the timer, and I only used medium speed.  So, I bodged up a simple de-bounced switch program and set the output for the medium speed Triac to 5V.  I was disappointed to see the pin corresponding to the "power" switch didn't have an internal pull up on the 12F509. I couldn't be bothered with bodge wires, cutting traces, or adding a pull up, so I just used the "fan speed" switch.  The result was a simple drop-in microcontroller replacement, with the elimination of every feature but the medium speed, and using the wrong button.

I spent more time tracing the PCB to find out what pin position on the controller did what than I did actually repairing it.  The previous repair was a jumper wire from 5V to the medium speed triac.  That required me to remove the power plug from the wall to turn it off, so that was a bit too ugly.