Burn in of Tantalum caps

We are seeing around 1 in 100 Tantalum caps fail shorted when initially pow ered up on the test fixture. There was a discussion here about such failur es being an initial short from fabrication or created when reflowed on the board. An article was linked which discussed a real world app where they p owered up the board with a voltage ramp and enough series resistance to pre vent the cap from creating a hard short and failing. But no information on how slow is "slow" and how much resistance is needed to prevent damage whi le clearing the short.

On the test fixture we can add a resistance to be bypassed with a jumper. We can't control the ramp up time of that voltage rail other than adding mo re capacitance to slow it down. I suppose that is an option if we know a t arget ramp rate. I've asked the manufacturer of the regulator (CUI V7812-1

000) what the ramp rate is and how it varies with capacitance. We'll see i f they respond.

Any suggestions on improving this problem? It's not a big deal, but we did have one cap that failed catastrophically and actually burned the board, b ut not beyond repair.

Rick C.

- Get 6 months of free supercharging - Tesla referral code -

formatting link

Reply to
gnuarm.deletethisbit
Loading thread data ...

Could it be the polarity is reversed relative to the marking and the orientation in the reel?

Reply to
Tom Del Rosso

dV/dt

formatting link

Regards,

Boris Mohar

Got Knock? - see: Viatrack Printed Circuit Designs (among other things)

formatting link

void _-void-_ in the obvious place

--
This email has been checked for viruses by Avast antivirus software. 
https://www.avast.com/antivirus
Reply to
Boris Mohar

What rating and test voltage, out of curiosity?

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Design 
 Click to see the full signature
Reply to
Tim Williams

on

e

a

'll

id

but

It's a 20 volt cap on a 12 volt supply. We only see failures in test, not in the field. One engineer suggested the issue was moisture in the cap fro m improper handling, but he also is working on some invention where he used the term, "over unity" before I stopped listening.

I'd like to modify the test fixture to not have these failures if they can be prevented without too much work. As I said, I have a current sense resi stor with a jumper so I can put any value there and jump across it once the circuit is powered up. It would add time to the test procedure though. N ot interested in doing much rework on the thing. It's only 1 in 100 failur es if that high. Might be 1 in 300.

Rick C.

  • Get 6 months of free supercharging + Tesla referral code -
    formatting link
Reply to
gnuarm.deletethisbit

Tantalum caps are bombs-by-the-reel. The tantalum pellet is the fuel and the MnO2 is the oxidizer. It's prudent to not detonate them.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

ed on

e

nce

r.

g

ow a

We'll

did

, but

t in the field. One engineer suggested the issue was moisture in the cap f rom improper handling, but he also is working on some invention where he us ed the term, "over unity" before I stopped listening.

n be prevented without too much work. As I said, I have a current sense re sistor with a jumper so I can put any value there and jump across it once t he circuit is powered up. It would add time to the test procedure though. Not interested in doing much rework on the thing. It's only 1 in 100 fail ures if that high. Might be 1 in 300.

How many tant's on your board? Did you change manufacturer? Tant cap qualit y seems to be continually eroding. I was fixing a ~20 year old instrument the other day. Problem was traced down to a tant. The board had a bunch of 16V tant's on the 15V power rails... these days I'd use a 50V one.

George H.

Reply to
George Herold

ly

ch

owed on

ere

tance

o

per.

ing

know a

We'll

we did

rd, but

not in the field. One engineer suggested the issue was moisture in the cap from improper handling, but he also is working on some invention where he used the term, "over unity" before I stopped listening.

can be prevented without too much work. As I said, I have a current sense resistor with a jumper so I can put any value there and jump across it once the circuit is powered up. It would add time to the test procedure though . Not interested in doing much rework on the thing. It's only 1 in 100 fa ilures if that high. Might be 1 in 300.

ity

Just the one tantalum. It may well be rather overkill but it's too late no w. The board has been in production for 10 years and I don't want to make any changes that might require new EMI testing. No change to the manufactu rer.

I was making these units at a facility which did everything including shipp ing. They were not reporting failure information it seems. Once when I wa s there someone mentioned that when they had any problems the first thing t hey did was replace this part. I thought that was an odd thing to do. The y didn't mention that they saw some number of shorts on the power rail. Th at is one of the things the test procedure tests for, shorted supplies.

Rick C.

-- Get 6 months of free supercharging -- Tesla referral code -

formatting link

Reply to
gnuarm.deletethisbit

More like it's always been continually erratic. You can switch manufacturer and be satisfied that you have solved your reliability problem. Then after a while the good ones turn bad.

It's really a design problem: don't detonate them and they will be very reliable.

Right. Tants on a supply rail should be avoided (which is usually where you want them) or at least derated 3:1 on voltage. Keep dv/dt under control too.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Yeah, I've read that they blow up if reversed. But what if the voltage is low and derated?

Reply to
Tom Del Rosso

When you read how they come up with the various derating factors it turns o ut the real issue is surge current. So rather than derate the voltage whic h really doesn't solve the problem necessarily, the surge current should be calculated and accounted for.

Applying a derating factor to the voltage as a way to account for surge cur rent is lazy engineering.

Rick C.

-+ Get 6 months of free supercharging -+ Tesla referral code -

formatting link

Reply to
gnuarm.deletethisbit

Bypassing a power rail, voltage derating by 3:1 seems safe, but we never use them on a big input power rail... usually on a regulator output. I prefer aluminum polymer caps, but some voltage regulators like the ESR of tantalum caps.

We don't deliberately reverse bias tantalums; I'm not sure what's safe there. I've tested some aluminum polymers for bipolar use, and selected some that seem reliable at serious reverse voltage.

--
John Larkin         Highland Technology, Inc 

lunatic fringe electronics
 Click to see the full signature
Reply to
John Larkin

Fig 4 is interesting, especially for "one or more current surges."

Fig 8 is crazy. Add all that stuff to every cap?

Their math is pretty flakey.

--
John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  
 Click to see the full signature
Reply to
John Larkin

Yes, dv/dt is the key. I recall 1990s datasheet reccomending series R of

1 ohm per applied volt - in other words tantalum capacitor surge current must be limited to one ampere max. Rather limits application as bypass on beefy power rails!

piglet

Reply to
piglet

lly

uch

lowed on

here

stance

to

mper.

ding

know a

. We'll

we did

ard, but

, not in the field. One engineer suggested the issue was moisture in the c ap from improper handling, but he also is working on some invention where h e used the term, "over unity" before I stopped listening.

y can be prevented without too much work. As I said, I have a current sens e resistor with a jumper so I can put any value there and jump across it on ce the circuit is powered up. It would add time to the test procedure thou gh. Not interested in doing much rework on the thing. It's only 1 in 100 failures if that high. Might be 1 in 300.

uality

nch

So what do you use on power rails? Al electrolytics? I guess I should 'scope the power rail turn on of all my circuits.

I use a series inductor/ multi-turn ferrite bead sometimes.

George H.

Reply to
George Herold

powered up on the test fixture. There was a discussion here about such fai lures being an initial short from fabrication or created when reflowed on t he board. An article was linked which discussed a real world app where the y powered up the board with a voltage ramp and enough series resistance to prevent the cap from creating a hard short and failing. But no information on how slow is "slow" and how much resistance is needed to prevent damage while clearing the short.

. We can't control the ramp up time of that voltage rail other than adding more capacitance to slow it down. I suppose that is an option if we know a target ramp rate. I've asked the manufacturer of the regulator (CUI V781

2-1000) what the ramp rate is and how it varies with capacitance. We'll se e if they respond.

did have one cap that failed catastrophically and actually burned the board , but not beyond repair.

NiobiumCapacitors.pdf

Author of that article is an idiotic moron from hell. Who in hell would thi nk the voltage rating of the capacitor could possible affect peak current s urge. So either that fool is an idiot or something is being lost in transla tion.

.ca

Reply to
bloggs.fredbloggs.fred

owered up on the test fixture. There was a discussion here about such fail ures being an initial short from fabrication or created when reflowed on th e board. An article was linked which discussed a real world app where they powered up the board with a voltage ramp and enough series resistance to p revent the cap from creating a hard short and failing. But no information on how slow is "slow" and how much resistance is needed to prevent damage w hile clearing the short.

We can't control the ramp up time of that voltage rail other than adding more capacitance to slow it down. I suppose that is an option if we know a target ramp rate. I've asked the manufacturer of the regulator (CUI V7812

-1000) what the ramp rate is and how it varies with capacitance. We'll see if they respond.

id have one cap that failed catastrophically and actually burned the board, but not beyond repair.

The switching regulator is guaranteed to output 1A under fairly wide range of environmentals and circuit operational parameters like differential volt age i/o being a big one. This DOES NOT mean it limits output current at 1A. It may allow limits as high as 3A at low duty. All you know, or should kno w, is you don't know anything about the limit. It's the wrong part to use i n this kind of test setup. But don't you worry about that. Don't deprive yo urself of wasting thousands of man hours blithering about every conceivable peripheral issue that has nothing to do with fixing anything.

Reply to
bloggs.fredbloggs.fred

Specially using a single JFET for a high-current totem pole. ;)

Or just limit them to the outputs of linear regulators and stuff like that. It's the supply input that has the issue. (Our stuff has to be able to handle being connected to two car batteries in series, because people actually do that.)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Take a look at Eq. 2 and the surrounding discussion. Their resistance appears to be constant for any given series.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Ceramics and the occasional Alpo.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.