So, Netlist,
Did you continue the test on the Panasonic and Rubycon caps? Did you put them at rated temperature? electrical conditions?
Need.... more.... info!
Did you continue the test on the Panasonic and Rubycon caps? Did you put them at rated temperature? electrical conditions?
Need.... more.... info!
Re: Where is netlist?
Why, did I say Netlist? Oh dear.... this is proof I must really be losing my mind..
I was referring to you Pereanders. Any news?
peranders said:
Why, did I say Netlist? Oh dear.... this is proof I must really be losing my mind..
I was referring to you Pereanders. Any news?
... and who is he?Bakmeel said:
If we talked about Spice models it would have been natural to mention Netlist
The news is no news, it works good.
With tens of thousands of amps, I imagine even a few nH can generate some serious kV... Enough to obliterate a cap. Not to mention wire and PCB inductances?
Perhaps I am wrong since no one else has said something like this.
- keantoken
Perhaps I am wrong since no one else has said something like this.
- keantoken
Perhaps this paper from Cornell Dubilier may help shed some light on the mode of failure you have been experiencing.
http://ecadigitallibrary.com/pdf/CARTSASIA07/1_4a Macomber-CDE.pdf
regards,
Keith
http://ecadigitallibrary.com/pdf/CARTSASIA07/1_4a Macomber-CDE.pdf
regards,
Keith
Interesting. This must be why I used to have to design boards with a a few Ohms in series with each tantalum decoupler 30 years ago.
Similar result, but caused by a somewhat different mechanism.
Tantalum capacitors are made by sintering (compressing) granular tantalum powder to form a solid matrix. The contact areas between the particles of the matrix are where the dielectric is formed.
Due to their low ESR, the high inrush current during initial charging can cause hot spots to develop in the interstices of the matrix, leading to dielectric breakdown, excessive current flow, heating, and ultimately catastrophic failure. The process proceeds very quickly, much in the same way as secondary breakdown in semiconductors and generally results in a dead short. The effect is especially evident when the capacitor is used at or near its’ maximum rated voltage, which is why it is always recommended to voltage overate tantalum capacitors by 100% or more. Inserting a several of ohms resistance in series helps to mitigate this phenomenon, however, it also mitigates much of the reason for using a tantalum capacitor in the first place.
Personally, I learned of this tendency back in the days of mini-computers, after experiencing first hand the catastrophic failure of several tantalum bypass capacitors in a military computer system. The result of a dead short on a 100 amp 5 volt bus can be rather unpleasant to say the least. For this reason, I have avoided using sintered tantalum capacitors in bypass and filtering applications ever since. In my opinion, their liabilities outweigh their advantages in all but a few circumstances.
Tantalum capacitors are made by sintering (compressing) granular tantalum powder to form a solid matrix. The contact areas between the particles of the matrix are where the dielectric is formed.
Due to their low ESR, the high inrush current during initial charging can cause hot spots to develop in the interstices of the matrix, leading to dielectric breakdown, excessive current flow, heating, and ultimately catastrophic failure. The process proceeds very quickly, much in the same way as secondary breakdown in semiconductors and generally results in a dead short. The effect is especially evident when the capacitor is used at or near its’ maximum rated voltage, which is why it is always recommended to voltage overate tantalum capacitors by 100% or more. Inserting a several of ohms resistance in series helps to mitigate this phenomenon, however, it also mitigates much of the reason for using a tantalum capacitor in the first place.
Personally, I learned of this tendency back in the days of mini-computers, after experiencing first hand the catastrophic failure of several tantalum bypass capacitors in a military computer system. The result of a dead short on a 100 amp 5 volt bus can be rather unpleasant to say the least. For this reason, I have avoided using sintered tantalum capacitors in bypass and filtering applications ever since. In my opinion, their liabilities outweigh their advantages in all but a few circumstances.
My guess is that the roughness you are seeing is indicative of them using substandard parts in this level of supply. Likely they are sourcing from a cheaper manufacturer of their metals. As such it is possible that the internal foils in the EL cap are not as pure as they should be (Rubycon, UnitedChemicon, Nichicon, Panny, etc. have known about this for some time). The impurities in the aluminum (I seem to remember copper was particularly bad, could be wrong though) result in H2 generation from the electrolyte which then causes the cap to go POP!
tantalum caps are notorious for dead short failures. NCR had a CPU board for a Proof Machine (a check processing machine) with about 10 SMT tantalum caps on it, and there were a lot of failures of those on the +5V rail. the epoxy case of the device used thermally sensitive color dyes in it so that the laser etching of the component markings would show up better. when one of the caps failed, half of the otherwise yellow epoxy case would turn brown making it easy to identify the failed device. regular aluminum electrolytics worked just fine as replacements
I have seen more examples of capacitor banks in commercial products so it seems to work but I haven't seen any board bigger than I have uploaded in the first post in this thread. Can anyone beat the record of 250 caps in parallel?
- Status
- Not open for further replies.