Post by thread:Surge current testing Re: [OT] From: Fire in the H

Dan Michaels wrote: {Quote hidden}> > > > > "Reliability degrades when they are operated in low-impedance > > circuits, and so they are not suitable for power supply > > reservoir operations". > > > > ref, p 83, "The Circuit Designer's Companion", Tim Williams, > > 1996, Newnes pub. > > > >Really? I see them used this way constantly, even recomended (particular > >brands and values, usually) in SPS chip app notes, and advertised for this > >purpose (ie by ESR, IIRC.) Are SOME tantalum caps specifically (re)designed > >for such applications, or are some other specific types (perhaps no longer > >very common) the ones that are unsuitable there. > > > >(come to think of it, I don't know that I've ever seen tantalum caps widely > >used for anything BUT power bypass...) > > > >BillW > > > > Someone else mentioned the issue referenced above, ie use in > "... low-impedance circuits...", in another thread recently. But I > know what you are saying. I am still fuzzy on the issue. But then > we just heard 27 'Fire in the Hole' stories about them exploding too. > > DanM Tantalum caps are notorious for exploding due to inrush, or surge current. Some parts are tested for surge and may be purchased as such. The methodology of the test is to inject a high surge current into and discharge the DUTs several cycles (usually 3). Limit the "parts exploding" problem by putting a suitable fuse in the test circuit (and a protective cover over the parts if new operators aren't easy to come by). There is even a mil spec on this test procedure. which specifies circuit resistance, charge times and Voltage source capacitance (basically, you dump a really big bank of electrolytics, through a small resistance, on to the Tantalum parts). Parts which blow the fuse are deemed bad and those which don't proceed through the process for further testing. The surge test can increase the leakage current of the part and that gets tested as well as cap and df, etc. If you buy leaded Kemet parts that are surge tested, they were run one of several testers I designed and built. Surface mount parts was some other guy. The thing which tells the true story is a quick glance at the underside of the protective cover on the testers. The fuses don't always do the job and the cover has pock marks and burns all over it from exploding parts :). -- Robert Wuest, PE Sirius Engineering Company

Robert Wuest wrote: ...... >Tantalum caps are notorious for exploding due to inrush, or surge >current. Some parts are tested for surge and may be purchased as such. > >The methodology of the test is to inject a high surge current into and >discharge the DUTs several cycles (usually 3). Limit the "parts ..... >If you buy leaded Kemet parts that are surge tested, they were run one >of several testers I designed and built. Surface mount parts was some .... Ok, for the testing part, but what would you recommend as far as use of tantalums in practical circuits? Use as P/S reservoir filters would seem to be a problem. Yes? And in use as Vcc bypass caps, they'd still be right on the main power bus in general. Yes? Would you have to put a small series R in the power bus to the chips where they are used for bypassing?

Dan Michaels wrote: > > Robert Wuest wrote: > ...... > >Tantalum caps are notorious for exploding due to inrush, or surge > >current. Some parts are tested for surge and may be purchased as such. > > > >The methodology of the test is to inject a high surge current into and > >discharge the DUTs several cycles (usually 3). Limit the "parts > ..... > >If you buy leaded Kemet parts that are surge tested, they were run one > >of several testers I designed and built. Surface mount parts was some > .... > > Ok, for the testing part, but what would you recommend as far as use > of tantalums in practical circuits? Most applications of Ta caps that I have seen are OK. Most power supply circuits cannot provide the high amp surges required for this to be a really significant issue and most Ta caps don't blow up :) > Use as P/S reservoir filters would seem to be a problem. Yes? Same thing. What kind of currents can the supply provide? Some applications may require protection for the caps. > And in use as Vcc bypass caps, they'd still be right on the main > power bus in general. Yes? > > Would you have to put a small series R in the power bus to the > chips where they are used for bypassing? If the power supply can provide mucho current, you should place some sort of inrush current limiting (like an NTC device) in series. On a linear supply, put it before the regulator. Also, you can use a higher voltage rated part because they can take more inrush. Actually performance specs are hard to get, but 1 ohm source impedance can blow parts. So for a 5 Volt supply that provides 5 amps max (and say a 6 volt cap), you're on thin ice. 2 ohms is safer. Or use a 10 volt cap. -- Robert Wuest, PE Sirius Engineering Company

In message <spam_OUT1.5.4.16.20000427133524.b9377e40TakeThisOuTlynx.sni.net>,> .....oricomKILLspam@spam@lynx.sni.net writes >And in use as Vcc bypass caps, they'd still be right on the main >power bus in general. Yes? Used in this way in TV sets they were incredibly unreliable, after a few years of use in the late 1970's they totally disappeared from domestic electronics usage!. If you have an HT short on a board, the first thing you look for is any tantalum capacitors :-). -- Nigel. /--------------------------------------------------------------\ | Nigel Goodwin | Internet : nigelgKILLspamlpilsley.co.uk | | Lower Pilsley | Web Page : http://www.lpilsley.co.uk | | Chesterfield | Official site for Shin Ki and New Spirit | | England | Ju Jitsu | \--------------------------------------------------------------/

Robert Wuest wrote: >> ...... >Most applications of Ta caps that I have seen are OK. Most power supply >circuits cannot provide the high amp surges required for this to be a >really significant issue and most Ta caps don't blow up :) ..... {Quote hidden}>If the power supply can provide mucho current, you should place some >sort of inrush current limiting (like an NTC device) in series. On a >linear supply, put it before the regulator. Also, you can use a higher >voltage rated part because they can take more inrush. > >Actually performance specs are hard to get, but 1 ohm source impedance >can blow parts. So for a 5 Volt supply that provides 5 amps max (and >say a 6 volt cap), you're on thin ice. 2 ohms is safer. Or use a 10 >volt cap. > Great, Robert, thanks for the straight scoop. This last sounds like a good rule of thumb. It looks like the big dogs like Wagner and Grif, with their 5A to 200A supplies, better keep their helmets on, but us little dogs with stuff running off wall warts can rest easy. Also, as someone else mentioned, ie PaulB, there are a couple of different types of tantalums. Can we treat them the same, viz-a-viz this "low-impedance" blow-up-in-your-face issue? best regards, - Dan Michaels ==============

Dan, Robert has provided a summary. It boils down to basic engineering. Know your environment and know the device characteristics. I have'nt had a Tantalum cap fail in decades. There is a good reason they are specified in vendor's application notes. If you use a 6V Tantalum in a 5V supply of questionable design, you are asking for trouble. This topic has come up before and there are those folks that think Tantalums are `evil'. I've seen far more problems with electrolytics in hot environments and they can be much more `explosive'... - Tom PSBS: I had more to say on the LTC2400 and Dennis' comments about `Think before you comment' and "Bit Jittering" which I assume meant "Dithering" which I don't think is practical at the sub uV level but I've been busy and I'll try to follow up... At 12:19 AM 4/28/00 -0600, Dan Michaels wrote: {Quote hidden}>Robert Wuest wrote: >>> ...... >>Most applications of Ta caps that I have seen are OK. Most power supply >>circuits cannot provide the high amp surges required for this to be a >>really significant issue and most Ta caps don't blow up :) >..... >>If the power supply can provide mucho current, you should place some >>sort of inrush current limiting (like an NTC device) in series. On a >>linear supply, put it before the regulator. Also, you can use a higher >>voltage rated part because they can take more inrush. >> >>Actually performance specs are hard to get, but 1 ohm source impedance >>can blow parts. So for a 5 Volt supply that provides 5 amps max (and >>say a 6 volt cap), you're on thin ice. 2 ohms is safer. Or use a 10 >>volt cap. >> > >Great, Robert, thanks for the straight scoop. This last sounds like >a good rule of thumb. It looks like the big dogs like Wagner and Grif, >with their 5A to 200A supplies, better keep their helmets on, but us >little dogs with stuff running off wall warts can rest easy. > >Also, as someone else mentioned, ie PaulB, there are a couple of >different types of tantalums. Can we treat them the same, viz-a-viz >this "low-impedance" blow-up-in-your-face issue? > >best regards, >- Dan Michaels ------------------------------------------------------------------------ Tom Handley New Age Communications Since '75 before "New Age" and no one around here is waiting for UFOs ;-)

I once got a windfall of electrolytic filter capacitors from a place where I worked. They were around 1,000uF at 35 volts and they were all part of a recall. The complaint was that they had a habit of exploding. My boss said that they were being operated too close to their rated voltage so I figured I'd take the chance and just use them in supplies which were going to produce less voltage than they originally got. The caps were encased in hard plastic, about thumb-sized, and had two leads coming out one end. I got my windfall of about 20 or so of these in the Summer of 1971 and used several without incident. One day, I switched on the power to a new supply I had built with one of these caps and I noticed the voltage was a bit low. Before I could react to that, the end without the wires blew off with a bang like a firecracker. There was the usual smell and the finding of bits of plastic for weeks if not months after the bang, but I generally had no trouble with most of the caps. I have some of them still in use in various projects, but they are starting to fail now by opening. I think the ones that exploded were defective to begin with and had an internal short. They would end up looking pretty much like fired shotgun shells. Martin McCormick

Tom Handley wrote: > Dan, Robert has provided a summary. It boils down to basic engineering. >Know your environment and know the device characteristics. I have'nt had >a Tantalum cap fail in decades. There is a good reason they are specified >in vendor's application notes. If you use a 6V Tantalum in a 5V supply of >questionable design, you are asking for trouble. .... Tom, is it possible your circuits have all luckily fallen onto the good side of the "border" that Robert was talking about? Ie, power supplies with effective series impedance somewhat larger than 1 ohm? BTW, for another "rule of thumb", I was going to mention that I measured the series R in a number of wall warts a couple of years ago, and they were all pretty substantial, 10 ohm range or greater, for the smaller WWs. For instance, the difference between Vopen_circuit and Vloaded was generally 3-5v. Worst cases would be the larger WWs. Say, [Voc-Vloaded]/Ioad = 3v/1A = 3 ohms, close to Robert's border. Smaller WWs have higher R. Cheers, - Dan Michaels

At 12:45 PM 4/29/00 -0600, Dan Michaels wrote: >Tom Handley wrote: >> Dan, Robert has provided a summary. It boils down to basic engineering. >>Know your environment and know the device characteristics. I have'nt had >>a Tantalum cap fail in decades. There is a good reason they are specified >>in vendor's application notes. If you use a 6V Tantalum in a 5V supply of >>questionable design, you are asking for trouble. >.... > > >Tom, is it possible your circuits have all luckily fallen onto >the good side of the "border" that Robert was talking about? >Ie, power supplies with effective series impedance somewhat >larger than 1 ohm? Dan, I'd like to think it was attributed to good design ;-) I just pay close attention to vendor data sheets and try to `bullet-proof' the design by looking at the environment. Most of my experience is in process control. I have blown Tants but it was due to `stupid human tricks' during prototyping that we've been discussing in another topic. Even in a 5V logic environment with a linear supply, I specify 16V caps. - Tom {Quote hidden}>BTW, for another "rule of thumb", I was going to mention >that I measured the series R in a number of wall warts a >couple of years ago, and they were all pretty substantial, >10 ohm range or greater, for the smaller WWs. > >For instance, the difference between Vopen_circuit and Vloaded >was generally 3-5v. Worst cases would be the larger WWs. Say, >[Voc-Vloaded]/Ioad = 3v/1A = 3 ohms, close to Robert's border. >Smaller WWs have higher R. > >Cheers, >- Dan Michaels ------------------------------------------------------------------------ Tom Handley New Age Communications Since '75 before "New Age" and no one around here is waiting for UFOs ;-)