Member
Joined 2009
Paid Member
A tantalum failure mechanism not mentioned yet is high current. Essentially, if the cap can be exposed to high transient currents, the worst being shorting a power supply rail, they will fail catastrophically, usually releasing 'magic smoke'. In some uses for tantalum, such as inside of a time constant network loaded with let's say 1 kΩ resistors or higher, in a configuration where no reverse voltage will ever happen, a tantalum cap will truly last forever.
Reverse voltage needs to be avoided as well. Ta caps are much more finicky than Al electrolytics with reverse voltage - a reverse voltage of a few percent of rated voltage is enough to kill a Ta cap.
Over-voltage also needs to be avoided, and the tradition is to use a cap rated for 2x the expected peak voltage. So, that means a 35V cap for a 15V rail.
For all of these reasons, I've decided to use the modern CMOS voltage regulators that do not need a Ta output cap for stability. Even though the CMOS regulators can be annoying, Ta is just too risky.
Someone mentioned polymer caps, and I'd like to point out that their leakage is at least an order of magnitude higher than a wet Al cap. They are also somewhat unstable and finicky - I had a 25V polymer Al cap (Panasonic OsCon) go into meltdown when driven at 22V. So, a 2x voltage derating seems prudent, if it is even available, but then you still have to deal with the relatively huge leakage. Never use a polymer cap for coupling.
For a polarized electrolytic, I find the Panasonic FR hard to beat. Sure, it's stone age technology, but they are tiny, low impedance, well made, and they last an extremely long time.
Reverse voltage needs to be avoided as well. Ta caps are much more finicky than Al electrolytics with reverse voltage - a reverse voltage of a few percent of rated voltage is enough to kill a Ta cap.
Over-voltage also needs to be avoided, and the tradition is to use a cap rated for 2x the expected peak voltage. So, that means a 35V cap for a 15V rail.
For all of these reasons, I've decided to use the modern CMOS voltage regulators that do not need a Ta output cap for stability. Even though the CMOS regulators can be annoying, Ta is just too risky.
Someone mentioned polymer caps, and I'd like to point out that their leakage is at least an order of magnitude higher than a wet Al cap. They are also somewhat unstable and finicky - I had a 25V polymer Al cap (Panasonic OsCon) go into meltdown when driven at 22V. So, a 2x voltage derating seems prudent, if it is even available, but then you still have to deal with the relatively huge leakage. Never use a polymer cap for coupling.
For a polarized electrolytic, I find the Panasonic FR hard to beat. Sure, it's stone age technology, but they are tiny, low impedance, well made, and they last an extremely long time.
Seems to me that most participants in the discussion didn't read this sentence..
Last few days I'm listening output with tantalum capacitors, didn't hear nothing wrong.. I even think that sounds better than LME49710 output on same DAC without capacitors on signal path..
DRV603 board - I2S over USB Audio
http://jlsounds.com/uploads/DRV_board.pdf
Nothing wrong with modern Tantalium caps (at least sound wise) but it seems the audiophile community is somewhat conservative. If your first girlfriend cheated on you 20 years ago that does not mean the current girlfriend will do the same. Mmmm, bad analogy 
It would help to mention brand, types/series to people that want to re-try Tantalium caps.
It would help to mention brand, types/series to people that want to re-try Tantalium caps.
Last edited:
I agree that, especially in a single ended supply situation, a Ta cap can work well as a coupling cap. But, in an output coupling application, where the cap is essentially attached to an output connector where wiring / cable problems could present a short to the output, a Ta cap could fail catastrophically.
For an interstage cap where the circuit on both sides of the cap can be a reliably controlled load, then a Ta coupling cap will work well, both sonically and reliability-wise. But, where one side of the circuit is "blue sky", it's not a safe bet.
I guess if you're a builder who will take care not to torture your gear with improper connections, then it could be OK to use a Ta cap as an output cap. But, for a commercial product, or even a device used in (for example) a professional studio, too many things could go wrong, and I'd be leery. Just the potential for a very brief output-to-ground glitch of patching a patchcord into a jackfield while signal is present could tickle a Ta output cap in the wrong way, and now that device need repair. Not worth it IMHO.
For an interstage cap where the circuit on both sides of the cap can be a reliably controlled load, then a Ta coupling cap will work well, both sonically and reliability-wise. But, where one side of the circuit is "blue sky", it's not a safe bet.
I guess if you're a builder who will take care not to torture your gear with improper connections, then it could be OK to use a Ta cap as an output cap. But, for a commercial product, or even a device used in (for example) a professional studio, too many things could go wrong, and I'd be leery. Just the potential for a very brief output-to-ground glitch of patching a patchcord into a jackfield while signal is present could tickle a Ta output cap in the wrong way, and now that device need repair. Not worth it IMHO.
Normally there are 47 or 100 Ohm series resistors in the output stages of buffers/preamps to avoid problems with capacitive loads. Those would prevent short circuiting stuff too. Most buffers/preamps also have very low offset voltage at this point. I would classify this as "ultra low risk".
Headphone amps with asymmetrical PSU so with a large DC offset voltage present (and no series resistors) are a different story as indicated.
Headphone amps with asymmetrical PSU so with a large DC offset voltage present (and no series resistors) are a different story as indicated.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.