Substituting tantalum capacitors

[Steve Jones]

I spend so much of my time swapping out tantalums in old synthesizers, they are diabolical when they get old, often going short circuit. In the intrests of authenticity I usually replace them with new tantalums in some machines, but in others where it is clear that they are only there because of size, I often put in a small modern electro.

Does anyone know of any reason not to use low-leakage electrolytics as a general sub except in the most exotic of circuits?

 

[bcarso]

That power supply bypass cap article series I posted a link to recently had some information about tantalums I hadn't been aware of, namely that they have rather higher ESR than I had thought. This can be important when there are resonance effects afoot.

For coupling cap applications I can't see the harm in using low-leakage electrolytics, if they are indeed low enough leakage. I would test candidates, if not each individual cap then at least a few out of a lot. As you may notice the guaranteed specs on Al lytics are usually much worse than the typical.

Because of the messy failure modes of tantalum, multilayer high K ceramics are now available with pretty amazingly high values of capacitance. However they are microphonic and quite temperature sensitive, as well as expensive.

 

[JohnRoberts]

I'm not aware of any new info on tants, but in general see little problem with low leakage aluminum in modest impedance blocking apps, and in PS decoupling perhaps low impedance switching aluminum electrolytic, or standard aluminum with ceramic disc in parallel.

In synths if the tants are used to define att/release envelopes or such timing circuits there may be subtle differences between tant and other dielectrics.

JR

 

[thermionic]

Hi Steve,

Not sure how much use this will be to an experienced synth tech such as yourself, but I started a thread a few days ago that touched on this subject: http://www.groupdiy.com/index.php?topic=25005


Justin

 

[Steve Jones]

[quote author="thermionic"]Hi Steve,

Not sure how much use this will be to an experienced synth tech such as yourself, but I started a thread a few days ago that touched on this subject: http://www.groupdiy.com/index.php?topic=25005


Justin[/quote]

Hi Justin, wish I had seen your thread before. As an owner of a Yamaha CS-80, (I have had several) I do agree with Old Crow as far as that particular synth goes - I replace every 4000 series, particularly on the aftertouch board, and add decoupling caps. As for replacing them out of hand in a Prohet V, I would only do that if the machine was going to be used on a major tour and had to be made as bullet proof as possible. It isn't really practical to swap out all the old CMOS in most machines.

I have just recapped an AMS reverb, and I have an AMS delay to do soon, and both these units are full of tantalums, and fairly high voltage ratings, so they are very expensive to re-cap, tantalums don't come cheap. But in a device like this I don't mess with the original design.

Where tantalums are a real problem is in ARP synths, where each board usually has 2 10uF tantalums where the +/- 15V rails come in, and they go short all the time as they are so old. ARP Omni's have dozens of them in the keyboard circuit, and I have had no problem replacing them with electro's.

Got a Moog Sonic 6 in today, the customer says there is smoke.... bet you it is a tantalum.

As JR points out, they are often used in timing circuits in older synths, and also on high-impedance keyboard CV memory circuits, so best to leave the tantalums in there.

 

[Gus]

I suggest that if replacing tants you measure a few things.

First measure write down the value and ESR and DA and leakage of a known good cap on a meter that you can use all the time. Different meters might give different ESR readings.

I have noted over the years that the ESR of tants that start to fail in a certain circuit I repair often seem to have higher ESR than ones that work.

After seeing the insides of a old ARP syth I would get a good cap meter and measure all the caps that are removed and installed even the Al electros.

Google motherboard cap failures.

 

[bcarso]

Apple banned tants altogether at one point IIRC.

 

[Freq Band]

Apple Tarts mmmm
.....oh, you said tants...nevermind.

I have many old (but un-used) axial Wet Tantalums...military spec.
Some with "burn-in" testing, individually numbered, intended for Titan missiles.
Despite all that spec fluff, are these 1960's parts destined for their own mini-explosions ....and I should avoid them ?



'smoke is what makes these parts work...you will find that when the smoke leaves them, they stop working.'

 

[bcarso]

I think the rubber seals crack and leak, but I don't know how they behave when they fail.

I have some of those too, and I always find them too weird-looking to use even in a breadboard.

OTOH I used a lot of mil-grade tantalums in a big instrument for UCLA circa 1978, and the only one that failed was one I had managed to install backwards. It did not cause a fire, probably because I had a decoupling resistor in series.

Golly some of those parts were pretty, gold-plated and all, with hermetic seals. The halcyon days of aerospace...

 

[mikep]

I agree, but let me say something else.

reading this thread it would seem tantalum caps are all nightmares. I dont think modern tantalums are that bad at all. there are some very low ESR types available, as well as "Surge-robust" designs that are made to withstand use in PSU decoupling without series limiting resistors, and then there are hi-rel versions: I recently designed in a part with 500,000 hour MTBF. (not a $$$$ mil-spec part, it was a pretty cheap one). IMO, for certain apps, like super-compact SMD switching regulator designs, tantalum is unbeatable. (Ive replaced a whole lot of SMD al-electo caps in audio gear, lets not go there, please.)

mike

 

[bcarso]

Isn't there also supposed to be something politically incorrect about using tants? Something about the oppression of the workers in about the only place it is mined?

 

[JohnRoberts]

As I recall the design movement away from tantalums was due to cost and availability. I believe tantalum powder is only available from a few regions of the world.

Their is/was a secondary criticism from audio purists who may know more terms than what they mean, who avoided tantalum because of their dielectric absorption. While they do have more DA, that is little expressed in common DC blocking or PS decoupling applications.

In my very first published phono preamp (actually a DJ mixer) I made a post design substitution from aluminum to tantalum to reduce phase shift at 20kHz due to the caps series inductance. I even had one semi well known customer advise me about how bad the tantalums "would" sound, while my bench measurements and ears told me otherwise. Since then I prefer to just not use electrolytics at lower impedance loads.

JR

PS: Note aluminum caps circa the 70's were indeed lesser parts than what is available today. I gather this from modern advice to parallel with 1/100 value film caps. In my bench testing back then I determined the need to parallel with 1/10 value, to get film like performance overall, which rendered the technique ineffective for larger values.

 

[thermionic]

You can find documentaries on Youtube. Tantalum is derived from Coltan. There are large coltan mines in unstable parts of Africa, ruled by criminals. The electronics OEMs use middle-men to try and keep their hands clean, but the bottom line is that a significant percentage of all coltan (an essential component in mobile phones) comes from criminal organisations who forcibly use slaves to work in the mines.


Go here: http://www.youtube.com/watch?v=1UN44Lj4KwY


Justin

edited to remove inaccuracies.

 

[CJ]

yeah, for a while there, we had a six month lead time on tants because Nokia stole all the inventory.
Finally enough people panic argued and we got some stock the next week, go figure.

There is an AES article on using tants in audio circuits.

 

[Kid Squid]

Whoa, Justin,

Thats some bad ****, I had no idea....

Steve

 

[jensenmann]

When I was in Ruanda some years ago I visited a Coltane mine. It was a small mine which was under control of a German company. This mine was a big gain for the community. Everybody there made exceptionally much money compared to the surrounding communities (which are basically doing agriculture). From their wages they could even afford to have a doctor and their own school in their village. There was no army around at all. So my expierience seems to be totally different to what the movie shows.

 

[JohnRoberts]

That's another inconvenient truth... :wink:

Only truth I knew was they got expensive and hard to get...

JR

 

[mikep]

Im curious, which ones are "hard to get"? I know about the crunch that happened a few years back, but is it really continuing?

don't forget about the input coupling cap in vintage 24V Neebler circuits. if you replace with an al-electro cap its not the same. has anyone tried to theorize why it sounds good in that app? I bet low signal level has something to do with it.

Im not a champion of tantalum caps I swear.

 

[JohnRoberts]

my expensive and hard to get was tens of years ago... don't know now..

JR

 

[JohnRoberts]

Not to open up a general capacitor rant but this requirement for (AC) terminal voltage to be present for voltage coefficient to express as distortion or charging and discharging from different impedances for DA to express, makes many of the common criticisms of tantalum for audio moot. If you are not making a sample and hold circuit, and have the pole in your DC blocking application down at 2 Hz, don't expect to hear distortion.

Capacitors aren't perfect and big electrolytic are less perfect than most, but properly used and lightly loaded they aren't necessarily all that bad.

I have a short list of places I really don't like them, like passive speaker crossovers. YMMV

JR

PS: If I had bench time to throw at this, and actually cared, I might be tempted to research if voltage coefficient related distortion only occurred when the terminal voltage was due to AC signal below pole frequency, or if it also expressed at high end when terminal voltage is caused by ESL. In the latter case it might be a source of in band IM grunge if significant.