*gasp* Useful EDN article on tantalum caps
in [Design]
|
Prev: Call for Papers: World Congress on Engineering and Computer Science WCECS 2010
Next: Are there any 24VAC to 3.3VDC/1.5A SMPS?
From: cassiope on 4 Jun 2010 13:03 On Jun 4, 8:59 am, Jim Yanik <jya...(a)abuse.gov> wrote: > Phil Hobbs <pcdhSpamMeSensel...(a)electrooptical.net> wrote innews:4C091ABE..3090007(a)electrooptical.net: > > > > > On 6/4/2010 10:04 AM, John Larkin wrote: > >> On Fri, 04 Jun 2010 09:47:43 -0400, Phil Hobbs > >> <pcdhSpamMeSensel...(a)electrooptical.net> wrote: > > >>> Stupid title, but actual useful info on tantalum cap behaviour: > > >>>http://www.edn.com/article/509092-What_a_cap_astrophe_.php > > >>> Now that there's this existence proof, maybe there'll be a bit > >>> more...perhaps someone will write them another useful article, e.g. a > >>> compendium of all two-transistor circuits. > > >>> Cheers > > >>> Phil Hobbs > > >> Tantalums fail from excess current, namely dV/dt. MnO2 is an oxidizer > >> and tantalum is a fuel, and only a tiny ignition source sets it off. > >> They are fine for current-limited applications but firebombs if used > >> to bypass power rails. Derate them 3:1 on voltage if you must use them > >> as power bypasses. Better yet use aluminums, polymer aluminums for > >> wide-temperature appls > > >> John > > > The interesting thing in the article is that soldering makes large tants > > vulnerable to overvoltage failure well below their rating, and that they > > can be reconditioned (and made much more reliable) by putting a current > > limit on the power supply. That's more useful for repairs or fixing > > problems with a shipping product than for new designs, admittedly. > > Still, I thought it was kind of cool, and remarkable for being in EDN! > > > Cheers > > > Phil Hobbs > > TEKTRONIX used tantalums for many years in their scopes,and the failures > were not objectionably high. Usage was mostly on power rails,BTW. > > -- > Jim Yanik In the late '70s Tek decreed that its design engineers should only use tantalums with explicit series resistance (3ohms/volt IIRC), and enforced it with the design review process. Direct power supply bypassing was a no-no. This was after a lawsuit stemming from a fire started by a Tek product took out a large mobile lab. I think it was around that time that Tek also got much more serious about ensuring that connector bodies were flame retardant, and started worrying about their switch cams with all its flammable delrin. Maybe the rate of failure was not "objectionably high", but when the consequences can be serious (as with the tantalum self-immolation, or with a certain current oil leak), you have to consider implications beyond your own immediate product. -former Tek design engineer
From: Joerg on 4 Jun 2010 13:54 Phil Hobbs wrote: > Stupid title, but actual useful info on tantalum cap behaviour: > > http://www.edn.com/article/509092-What_a_cap_astrophe_.php > > Now that there's this existence proof, maybe there'll be a bit > more...perhaps someone will write them another useful article, e.g. a > compendium of all two-transistor circuits. > They could have reduced the text body to eight letters: *PHUT* ... *BANG* :-) -- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
From: John Larkin on 4 Jun 2010 14:47 On Fri, 04 Jun 2010 09:53:36 -0700, "Paul Hovnanian P.E." <paul(a)hovnanian.com> wrote: >Phil Hobbs wrote: > >> On 6/4/2010 10:04 AM, John Larkin wrote: >>> On Fri, 04 Jun 2010 09:47:43 -0400, Phil Hobbs >>> <pcdhSpamMeSenseless(a)electrooptical.net> wrote: >>> >>>> Stupid title, but actual useful info on tantalum cap behaviour: >>>> >>>> http://www.edn.com/article/509092-What_a_cap_astrophe_.php >>>> >>>> Now that there's this existence proof, maybe there'll be a bit >>>> more...perhaps someone will write them another useful article, e.g. a >>>> compendium of all two-transistor circuits. >>>> >>>> Cheers >>>> >>>> Phil Hobbs >>> >>> Tantalums fail from excess current, namely dV/dt. MnO2 is an oxidizer >>> and tantalum is a fuel, and only a tiny ignition source sets it off. >>> They are fine for current-limited applications but firebombs if used >>> to bypass power rails. Derate them 3:1 on voltage if you must use them >>> as power bypasses. Better yet use aluminums, polymer aluminums for >>> wide-temperature appls >>> >>> John >>> >> >> The interesting thing in the article is that soldering makes large tants >> vulnerable to overvoltage failure well below their rating, and that they >> can be reconditioned (and made much more reliable) by putting a current >> limit on the power supply. That's more useful for repairs or fixing >> problems with a shipping product than for new designs, admittedly. >> Still, I thought it was kind of cool, and remarkable for being in EDN! > >That might make a good argument for a properly thought out burn in >procedure. There's too much pressure from management to just build it and >ship it (airplanes included). We've had tantalums fail months or years after a product was shipped. On one of our VME modules, which we do burn in, well over half the field failures were tantalum caps. Field MTBF increased by about 3:1 when we fixed that problem. This is more an engineering issue than a test/burnin thing. Dry tantalums have to be designed in very carefully, and the manufacturers mostly don't want to tell you what's safe or not safe. John
From: Joerg on 4 Jun 2010 16:25 John Larkin wrote: > On Fri, 04 Jun 2010 09:53:36 -0700, "Paul Hovnanian P.E." > <paul(a)hovnanian.com> wrote: > >> Phil Hobbs wrote: >> >>> On 6/4/2010 10:04 AM, John Larkin wrote: >>>> On Fri, 04 Jun 2010 09:47:43 -0400, Phil Hobbs >>>> <pcdhSpamMeSenseless(a)electrooptical.net> wrote: >>>> >>>>> Stupid title, but actual useful info on tantalum cap behaviour: >>>>> >>>>> http://www.edn.com/article/509092-What_a_cap_astrophe_.php >>>>> >>>>> Now that there's this existence proof, maybe there'll be a bit >>>>> more...perhaps someone will write them another useful article, e.g. a >>>>> compendium of all two-transistor circuits. >>>>> >>>>> Cheers >>>>> >>>>> Phil Hobbs >>>> Tantalums fail from excess current, namely dV/dt. MnO2 is an oxidizer >>>> and tantalum is a fuel, and only a tiny ignition source sets it off. >>>> They are fine for current-limited applications but firebombs if used >>>> to bypass power rails. Derate them 3:1 on voltage if you must use them >>>> as power bypasses. Better yet use aluminums, polymer aluminums for >>>> wide-temperature appls >>>> >>>> John >>>> >>> The interesting thing in the article is that soldering makes large tants >>> vulnerable to overvoltage failure well below their rating, and that they >>> can be reconditioned (and made much more reliable) by putting a current >>> limit on the power supply. That's more useful for repairs or fixing >>> problems with a shipping product than for new designs, admittedly. >>> Still, I thought it was kind of cool, and remarkable for being in EDN! >> That might make a good argument for a properly thought out burn in >> procedure. There's too much pressure from management to just build it and >> ship it (airplanes included). > > We've had tantalums fail months or years after a product was shipped. > On one of our VME modules, which we do burn in, well over half the > field failures were tantalum caps. Field MTBF increased by about 3:1 > when we fixed that problem. > > This is more an engineering issue than a test/burnin thing. Dry > tantalums have to be designed in very carefully, and the manufacturers > mostly don't want to tell you what's safe or not safe. > Sounds just like LDOs :-) -- Regards, Joerg http://www.analogconsultants.com/ "gmail" domain blocked because of excessive spam. Use another domain or send PM.
From: John Larkin on 4 Jun 2010 17:10
On Fri, 04 Jun 2010 13:25:20 -0700, Joerg <invalid(a)invalid.invalid> wrote: >John Larkin wrote: >> On Fri, 04 Jun 2010 09:53:36 -0700, "Paul Hovnanian P.E." >> <paul(a)hovnanian.com> wrote: >> >>> Phil Hobbs wrote: >>> >>>> On 6/4/2010 10:04 AM, John Larkin wrote: >>>>> On Fri, 04 Jun 2010 09:47:43 -0400, Phil Hobbs >>>>> <pcdhSpamMeSenseless(a)electrooptical.net> wrote: >>>>> >>>>>> Stupid title, but actual useful info on tantalum cap behaviour: >>>>>> >>>>>> http://www.edn.com/article/509092-What_a_cap_astrophe_.php >>>>>> >>>>>> Now that there's this existence proof, maybe there'll be a bit >>>>>> more...perhaps someone will write them another useful article, e.g. a >>>>>> compendium of all two-transistor circuits. >>>>>> >>>>>> Cheers >>>>>> >>>>>> Phil Hobbs >>>>> Tantalums fail from excess current, namely dV/dt. MnO2 is an oxidizer >>>>> and tantalum is a fuel, and only a tiny ignition source sets it off. >>>>> They are fine for current-limited applications but firebombs if used >>>>> to bypass power rails. Derate them 3:1 on voltage if you must use them >>>>> as power bypasses. Better yet use aluminums, polymer aluminums for >>>>> wide-temperature appls >>>>> >>>>> John >>>>> >>>> The interesting thing in the article is that soldering makes large tants >>>> vulnerable to overvoltage failure well below their rating, and that they >>>> can be reconditioned (and made much more reliable) by putting a current >>>> limit on the power supply. That's more useful for repairs or fixing >>>> problems with a shipping product than for new designs, admittedly. >>>> Still, I thought it was kind of cool, and remarkable for being in EDN! >>> That might make a good argument for a properly thought out burn in >>> procedure. There's too much pressure from management to just build it and >>> ship it (airplanes included). >> >> We've had tantalums fail months or years after a product was shipped. >> On one of our VME modules, which we do burn in, well over half the >> field failures were tantalum caps. Field MTBF increased by about 3:1 >> when we fixed that problem. >> >> This is more an engineering issue than a test/burnin thing. Dry >> tantalums have to be designed in very carefully, and the manufacturers >> mostly don't want to tell you what's safe or not safe. >> > >Sounds just like LDOs :-) Lots of LDOs need tantalum output caps to be stable. It's a conspiracy. John |