True, but both are so low that the effect is irrelevant for anything except an ultra low output MC.
In most uses the leakage current noise will be far more significant
Solid Ta electrolytics usually have much larger ESR than modern wet Al electrolytics. There are a few regulators from the '70s such as the 79xx series that were designed to use their relatively high ESR to stabilize the regulator's feedback loop. Specifically, modern Al or ceramic caps are not equivalent because of their low ESR.
FWIW, I do not use these and other similar 'old' regulators, simply because it's an annoyance to have to use a specifically high ESR load cap to stabilize the regulator. Modern regulators can use MLCCs or high performance wet Al or polymer Al caps.
That said, solid Ta caps can do some really nice things as long as their peak current and polarity requirements are not exceeded.
FWIW, I do not use these and other similar 'old' regulators, simply because it's an annoyance to have to use a specifically high ESR load cap to stabilize the regulator. Modern regulators can use MLCCs or high performance wet Al or polymer Al caps.
That said, solid Ta caps can do some really nice things as long as their peak current and polarity requirements are not exceeded.
I've just made an attempt to estimate the magnitude of both effects and find that they are of the same order of magnitude around 120 Hz, but the leakage-related effect decreases faster with increasing frequency (and increases faster with decreasing frequency). Maybe some of my assumptions are wrong, though, particularly the assumption that leakage noise is shot noise without any 1/f. Anyway, this is my calculation:
Regarding the thermal noise of the ESR, you have to be a bit careful with very low-valued electrolytic capacitors. The tan(δ) of an aluminium electrolytic capacitor is typically between 0.1 and 0.3 at 120 Hz, depending on the voltage rating (63 V or 100 V typically being the best). The resistance is tan(δ) times the reactance, which boils down to a resistance between 132.6 Ω and 397.9 Ω at 120 Hz for a capacitor of 1 µF. The resistance tends to decrease at higher frequencies, but not very quickly (assuming that the maximum ripple current ratings are related to the dissipation in the ESR, manufacturers rarely specify directly what happens to the ESR or the tan(δ) over frequency).
Of course a straightforward solution is to use larger-valued aluminium electrolytic capacitors, like 10 µF and higher. That also helps to prevent extra noise due to the input noise current of the amplifier giving a voltage drop across the reactance of the capacitor. More importantly, it doesn't pay for any Congolese wars.
About leakage, I'm not sure how to model the noise associated with the DC leakage current. The leakage is related to the aluminium oxide layer restoring itself, so I guess it's a random phenomenon that has shot noise associated with it. If you can regard it as shot noise without any 1/f part, the power spectral density of the noise current (number of squared amperes per hertz) is 2qIleak. The leakage current is in parallel with the impedance of the capacitor, which is roughly 1/(2πfC).
Using a Norton-Thévenin transformation, you can then transform it to a noise voltage source in series with the capacitor with a power spectral density (squared volts per hertz) of 2qIleak/(2πfC)^2. This is equivalent to the thermal noise voltage of a resistor of (2qIleak/(2πfC)^2)/(4kT) = qIleak/(8π^2 f^2 C^2 kT). Assuming a maximum leakage current of 4 µA and a capacitance of 1 µF, the result at 120 Hz and 293.15 K is 139.3 Ω, of the same order as the noise due to tan(δ).
It again gets better with higher capacitance values. The maximum leakage is usually specified as being proportional to capacitance with some lower limit for very small capacitances, but the capacitance occurs quadratically in the denominator.
As this noise term falls of with 20 dB per decade of frequency, this effect can only be dominant at low audio frequencies, unless my assumptions are all wrong. For example, the story may change altogether if there are lots of 1/f noise associated with this current, with a corner frequency above the audio band.
By the way, in my experience the leakage currents of aluminium electrolytic capacitors usually drop to many decades below the specified maximum if you apply the voltage for much longer than the specified number of minutes. That's not something you can rely on for mass produced equipment, of course.
Tantalum replacements in older gear?
Hi there -
I have a unit for a client - a lovely Sony TA-88e preamplifier. it has two types of tantalums in the signal path.
I am being overly cautious with the client as I havent come across that many tants before. There are two types in this - one I assume is the bulbous wet type and the other is (many) smaller dry pellet types (?). Pics of both are attached.
Should I replace either of these? I would incline towards "yes" for the first type, but not for the second? I cant seem to find consensus anywhere on audio forums I am a member of to answer the clients questions.
thanks!
Hi there -
I have a unit for a client - a lovely Sony TA-88e preamplifier. it has two types of tantalums in the signal path.
I am being overly cautious with the client as I havent come across that many tants before. There are two types in this - one I assume is the bulbous wet type and the other is (many) smaller dry pellet types (?). Pics of both are attached.
Should I replace either of these? I would incline towards "yes" for the first type, but not for the second? I cant seem to find consensus anywhere on audio forums I am a member of to answer the clients questions.
thanks!
Attachments
You won't find a consensus, but at this age, replace parts that normally degrade in use,
like all of the tantalum and electrolytic capacitors. Also, older solid tantalums were prone
to shorting and burning up, often in a spectacular fashion. Not the way to impress an anxious
customer. If you must use tantalum replacement types, derate their voltage rating by 50%.
Last edited:
Tantalum caps don't always degrade with use. Once they have settled into their DC bias voltage, their leakage approaches zero and stays that way forever, unless they are damaged.
Yes, solid Ta caps are dangerous, since when the oxide layer is ruptured, a reaction takes place where the manganese dioxide gets consumed to oxidize the metallic Ta into tantalum pentoxide, restoring the oxide layer. As long as these reactions are few and far between, a solid Ta cap can form to the applied voltage, and once the oxide layer is stable, very few of these reactions take place. These reactions are the result of tiny holes in the oxide that cause a small short circuit with a reaction that re-grows oxide, so a solid Ta cap's leakage current is more of a scintillation of many of these tiny reactions, rather than a steady leakage. Again though, after the cap has stabilized, these scintillations get less common, and can eventually go to zero, resulting in a perfectly stable cap with no wear out mechanism.
So, as long as the Ta cap is not exposed to unlimited surge currents, it can be an extremely long life part, and will not dry out like a wet Al electrolytic. If the cap is surrounded by relatively high valued resistors, the peak currents will be limited, making the part less susceptible to failure.
An old Amber 4400 I'm now restoring uses many solid Ta caps, including some used as AC coupling for the output, terminated at the output jacks. Needless to say, with no established load on these output caps to put bounds on the transient current through the cap, they have all failed. However, other solid Ta caps tucked inside of circuit nodes with light loads such as 20KΩ will never fail, and have not, so there is no need to replace them with an aluminum electrolytic that _will_ eventually dry out and that will always have higher leakage than a formed solid Ta cap.
Solid Ta across the power supplies is a dangerous proposition, since transient currents are usually not limited to small values. Worse, steady state currents are not limited to small values either, so these are places where a fire can be started. Replacement with a wet Al electrolytic is a good idea, especially given the nice parts like Panasonic FR that we have now.
Wet Ta caps are extremely reliable parts, as long as they are de-rated to 50% of their faceplate voltage. In the NASA paper "Leakage currents and gas generation in advanced wet tantalum capacitors", they state:
Currently, military grade, established reliability wet tantalum capacitors are among the most reliable parts used for space applications
When the oxide layer gets ruptured, hydrogen gas is generated in an electrolysis reaction that re-grows the oxide, which could cause the case to swell and fail if enough hydrogen builds up. This is similar to the reaction in a solid Ta cap, but a wet Ta caps generates hydrogen, and the case is often hermetically sealed, trapping the evolved hydrogen gas.
I don't have much experience with wet Ta caps, since they're usually too expensive for audio equipment. Still, if they work now after decades, they will probably last additional decades. It's probably best to remove them from power supply applications, but for signal applications, I'd leave them in, again as long as they have defined resistances around them to limit the peak current through the cap, and they're de-rated to 50% of their faceplate voltage.
Thanks all - some very good information there.
Im certainly not an expert at this - and I dont feel prepared to take the risk on, for working on this preamp. Im letting him know that I would rather pass on it, there are more experienced guys out there who woud feel more confident about working on this, Im sure.
Im certainly not an expert at this - and I dont feel prepared to take the risk on, for working on this preamp. Im letting him know that I would rather pass on it, there are more experienced guys out there who woud feel more confident about working on this, Im sure.
I would suggest that you just replace all the tantalums (and all the electrolytics)
with new electrolytics., no risk there. Film types probably won't fit.
35 years ago I used wet tantalums and the rule was 1 Ohm per Volt series resistor.
I know that tantalum is found concentrated in old tin mining waste in Malaysia, but is not recovered as it is contaminated with thorium. This is an alpha particle emitter, so there is a possibility for a short to be generated by radiation, even with a stable voltage across it. I bet other tantalum sources have similar problems.
I thought I'd never use them either, but for a circuit that is generating a very low noise DC voltage, I find that an RC filter network built with solid Ta caps is not only extremely reliable, but much more compact and cost effective compared to MLCCs.
Further, many ceramic caps can be microphonic, but a solid Ta cap is not. So, the total amount of "garbage" expected on this voltage can be minimized using solid Ta caps.
Oh oh an opinion not based on experiences from decades ago 
Before we know it some one will mention tantalum drop type caps exploding.
Todays tantalum caps can be very good in audio. Now the point is that the differences between brands and series can be quite large. So instead of keeping mentioning tantalums past the question could be "which tantalum cap is best in audio?"
With regards to environmental and political issues.... I see the tendency is to speak about that after all the bad stuff has happened.
Before we know it some one will mention tantalum drop type caps exploding. Todays tantalum caps can be very good in audio. Now the point is that the differences between brands and series can be quite large. So instead of keeping mentioning tantalums past the question could be "which tantalum cap is best in audio?"
With regards to environmental and political issues.... I see the tendency is to speak about that after all the bad stuff has happened.
Last edited:
I'll have to bow out for an audio coupling recommendation. As far as manufacturers, I find the Kemet and AVX parts work really well.
Regardless of this, using solid Ta reliably requires a well defined input and output current, and/or a relatively high resistance surrounding the Ta cap, in order to put an upper bound on the peak current that can be driven through the cap. Coupling cap uses tend to be a little too loosely controlled to work well with solid Ta caps, unless they're buried well inside of a device, away from the vagaries of random loads and voltages. The traditional "blameless" uses for solid Ta are for large sample and hold caps or gain reduction caps, places where large C values are helpful, with little chance of aging or decay. However, some designers have used them for coupling uses, with better or worse results.
IMHO, as long as you can make sure that they don't catch fire, then sure - why not use solid Ta. Still, that constraint really does limit their use significantly.
Regardless of this, using solid Ta reliably requires a well defined input and output current, and/or a relatively high resistance surrounding the Ta cap, in order to put an upper bound on the peak current that can be driven through the cap. Coupling cap uses tend to be a little too loosely controlled to work well with solid Ta caps, unless they're buried well inside of a device, away from the vagaries of random loads and voltages. The traditional "blameless" uses for solid Ta are for large sample and hold caps or gain reduction caps, places where large C values are helpful, with little chance of aging or decay. However, some designers have used them for coupling uses, with better or worse results.
IMHO, as long as you can make sure that they don't catch fire, then sure - why not use solid Ta. Still, that constraint really does limit their use significantly.
Yes I also use AVX TPS series and like these. Not for coupling but for decoupling. But ... why would one choose tantalum caps for coupling when today small film caps can be found in the same value? The thread was started in 2005 (!) and things are different now.
I think issues regarding tantalum caps tend to be mixed up as the subject to use modern tantalum caps in 2021 is something different than replacement of old tantalum caps in older devices (and question their merits and/or to replace them). As can be seen in the very thread those subjects get tangled.
I think issues regarding tantalum caps tend to be mixed up as the subject to use modern tantalum caps in 2021 is something different than replacement of old tantalum caps in older devices (and question their merits and/or to replace them). As can be seen in the very thread those subjects get tangled.
Last edited:
FWIW, back in the 80's and 90's I maintained two Neve sidecars that a local studio had attached to their main console. Prized for their sound quality, those things were single-supply and used tantalums everywhere; interstage, decoupling, outputs, etc. The only lytics were at the DC power entry. I don't recall ever having to replace a tant.