Tantalum issues

[maginnovision]

So I recently picked up an R&S NGPE 800W power supply and it was listed as untested. I tested it and found the current sense circuit was acting as if it were pulling 50A so I couldn't get anything out of it. I tested a few things and found the +15V rail was at +13.8V. +13.8V and -15V obviously doesn't work well with op amps. Searched all of the boards and found some burnt tantalums and cut them out. Every single tantalum near the +15V regulators were burned. After cutting them out and replacing with aluminum electrolytics I actually had I turned it on and pop, one blew up. Good news is the original fault is gone, doesn't think it's pulling any current. The bad news is I can't turn it on for more than a few seconds without a new tantalum blowing up on me. That's my question... Do I need to replace every tantalum now with a new tantalum or are aluminum electrolytic caps usable? Also is there any good way to clean the electrolyte from those stupid things? Scrubbing with towels and alcohol is going to take a long time even for just the 3(thought 3rd time might be the charm) that have blown up.

[DC1MC]

If the tants have reached their end of life, they will all blow and also they are the spawn of the  !!!
Replace them all and use either NEW Kemets form established distributor, NOT EBAY, or hi-quality elcos.
If you have a professional repair shop around with an ultrasonic clean bath, remove all the caps to be replaced and have the board washed. Depending on where you are, some forum members also have ultrasonic clean devices and they may help you.
Otherwise, scrub, scrub, scrub... 

 Best of luck,
 DC1MC

[DaJMasta]

It is best to replace them with tantalums, the ESR can be an important factor depending on the application, and both electros and ceramics don't really match the characteristics well.


Are you measuring the rails with a scope?  You could be getting a much larger transient that settles to undervoltage that you wouldn't see with a DMM that is popping tantalums that wouldn't otherwise be popping.

[maginnovision]

No I'm not measuring with a scope. Repair area doesn't have one and I am too lazy to go grab one. I'd have to grab diff probe too. It is most likely the 7815 providing too much power and popping them. Has 3 of the regs for the same rail that I assume are to keep it regulated since the boards plug into a backplane. The caps that pop are on boards with no local regs. I'll definitely replace with proper caps, I wouldn't replace the hundred or so caps with unknown parts.

Supply is about 20 years old but my other R&S is about 30 and none of the tants have had problems so I was HOPING I'd get lucky and have a simple fix/build a mains cable for it. Oh well, thanks, I've got to go order more parts now.

Also just checked and the tantalums are rated for 16V. 

[GregDunn]

+1 on using a scope.  I've had PSU rails measure within limits on a DMM which actually had a huge ripple component exceeding the ratings of some of the components.  My HP6920B actually had caps in it from the factory which were rated below the design DC voltage of the circuit.  It happens.

I know the issue of replacing tantalums shotgun-style in a piece of older equipment is very contentious on these forums, but I've had so many of them either fail completely or develop just enough of a defect to cause issues, that I'd rather not take a chance.  It's easier to replace them than to clean up after them.     Same with electrolytics - some brands are fine, others are just ticking time bombs; and you don't want them to blow up on you either.

[LapTop006]

Replace them all and use either NEW Kemets form established distributor, NOT EBAY, or hi-quality elcos.

... and go for the polymer types so you don't need to do it again.

[maginnovision]

+1 on using a scope.  I've had PSU rails measure within limits on a DMM which actually had a huge ripple component exceeding the ratings of some of the components.  My HP6920B actually had caps in it from the factory which were rated below the design DC voltage of the circuit.  It happens.

I know the issue of replacing tantalums shotgun-style in a piece of older equipment is very contentious on these forums, but I've had so many of them either fail completely or develop just enough of a defect to cause issues, that I'd rather not take a chance.  It's easier to replace them than to clean up after them.     Same with electrolytics - some brands are fine, others are just ticking time bombs; and you don't want them to blow up on you either.

Yea, I ordered the 100$ in tantalums and I'll just do them all in one go since the supply is all in pieces anyway. The aluminum electrolytics are rated at least 50% over the circuit voltage. I actually meant to order the regulators too. One of them was discoloring a board from heat. Makes sense when you consider the 3 of them couldn't supply the current to keep the rail at 15V and only 1 has any sort of heatsink. Once I replace the caps I'll scope and see if it's ok. I ordered 35V rated caps so hopefully even if the regs aren't doing so hot the caps can manage.

Replace them all and use either NEW Kemets form established distributor, NOT EBAY, or hi-quality elcos.

... and go for the polymer types so you don't need to do it again.

Unfortunately they're all through hole .1" lead spacing and I couldn't find polymers like that.

[bd139]

This is typical daily life for a Tektronix 4xx owner. Literally the last 5 dead ones I’ve bought are the same three tants gone short. It’s so common I bought a ton of the things and the associated fuses for the power supply.

Fortunately tek designed the power supply to gracefully pack in rather than nuke the tants.

[maginnovision]

This is typical daily life for a Tektronix 4xx owner. Literally the last 5 dead ones I’ve bought are the same three tants gone short. It’s so common I bought a ton of the things and the associated fuses for the power supply.

Fortunately tek designed the power supply to gracefully pack in rather than nuke the tants.

Interesting, I have a 475A but no problems yet. Until I power it on next time, I'm sure.

[bd139]

Also have 475A. HT will break first followed by vertical amp

[DC1MC]

You should also check some of the beefyer elcos for ESR, as some other poster said, another cause of failure for the tants is to be feed with a high ripple voltage and because of low ESR they sink all the time high-current in a high-ripple situation, tants are really good for HF filtering and short pulses but die fast in case of 120Hz high ripple..

 DC1MC

[Dr. Frank]

So I recently picked up an R&S NGPE 800W power supply and it was listed as untested. I tested it and found the current sense circuit was acting as if it were pulling 50A so I couldn't get anything out of it. I tested a few things and found the +15V rail was at +13.8V. +13.8V and -15V obviously doesn't work well with op amps. Searched all of the boards and found some burnt tantalums and cut them out. Every single tantalum near the +15V regulators were burned. After cutting them out and replacing with aluminum electrolytics I actually had I turned it on and pop, one blew up. Good news is the original fault is gone, doesn't think it's pulling any current. The bad news is I can't turn it on for more than a few seconds without a new tantalum blowing up on me. That's my question... Do I need to replace every tantalum now with a new tantalum or are aluminum electrolytic caps usable? Also is there any good way to clean the electrolyte from those stupid things? Scrubbing with towels and alcohol is going to take a long time even for just the 3(thought 3rd time might be the charm) that have blown up.

These tantalum capacitors may be in there for ESR, so you can't easily replace them with other types, especially not electrolytic ones, only by high capacitance MLCCs, because these also have low ESR, probabaly lower one.
But pay attention, the latter could also pose a stability problem.

Get the schematics, and you may decide, what to use. LDO regulators for exmple example very well selected ESR and capacitance, also several switch mode chips .

Tantalum capacitors don't have a special end-of-life endurance, normally they last forever, if nothing is wrong. In contrast to electrolytic cap's, which have a  rated life time over operational temperature.

So that systematic failure indicates possible failures in the circuit.

Ta cap's fail due to overvoltage, spikes, excessive ripple, latter is equivalent to excessive inrush current.

If you design Ta cap's directly at the low ohmic battery or behind a regulator, you have to limit that inrush current, i.e. artificially increase its ESR by a series resistors in the ballpark of several ohms.

Please check the location of the failing cap's!

2nd is overvoltage, so you should measure the steady voltage (and the ripple) at the intended location, before assembling new ones.There should be enough voltage margin, best is a factor of 2.
If even  a new electrolytic capacitor blew up, which is much more forgiving, then you have to check ripple current, and select  the appropriate quality also.
In an 800W PSU, heavy ripple currents @ 60Hz are expected but only under load, so there maybe a short circuit somewhere, which causes ripple w/o load, and especially highr ripple than in normal use.

Therefore check current consumption w/o load, first.

Frank

[David Hess]

It is best to replace them with tantalums, the ESR can be an important factor depending on the application, and both electros and ceramics don't really match the characteristics well.

An aluminum electrolytic with about 4 times the capacitance to get about the same ESR is pretty close.

... and go for the polymer types so you don't need to do it again.

Polymer electrolytics have a much lower ESR which can be a problem in circuits which depend on a controlled ESR.

Tantalum capacitors don't have a special end-of-life endurance, normally they last forever, if nothing is wrong. In contrast to electrolytic cap's, which have a  rated life time over operational temperature.

Except in high voltage parts, that is my understanding also however I have a hypothesis about what may be going on.

Surface mount epoxy packaged solid tantalum capacitors suffer from stress fractures when heated during soldering.  Running them through the procedure used after manufacturing to clear faults after soldering solves this.  My hypothesis is that epoxy parts continue to suffer stress fractures over time due to temperature and/or humidly variations.  Axial sealed solid tantalum capacitors and wet tantalum capacitors do not seem to suffer from this problem but they are constructed such that the packaging cannot stress the tantalum pellet.

Voltage derating epoxy packaged solid tantalum capacitors to 2/3 or even 1/2 may completely alleviate this issue but was not always done.  I remember old marketing and application notes which said that unlike aluminum electrolytic capacitors, tantalum capacitors did not need to be voltage derated at all.