Console PSU recap - What about those tantalums?

[JohnRoberts]

Tantalum caps were popular back in the 70s for their relatively low ESR and small size for capacitance. Modern low ESR aluminum caps can cover any modern circuits that need low esr, while none come to mind immediately.

In the decades since those days I have only specified tantalum in one design where the cap was part of a dynamic (vinyl NR decoder) processor's time constant circuit where DA could actually make a sonic difference (for the rest of that story search CX record decoder).

JR

 

[RoadrunnerOZ]

Having seen the amount of heat generated in close packed recording consoles and the catastrophic failure of tantalum caps in those environments I am reluctant to use them at all. The SMD ones seem to suffer the same if not worse.
I serviced an MCI console not so long ago that was originally in a studio I was involved in building, wiring up and commissioning and working as an engineer back in 1977 - there were cap failures on almost every channel card, burnt resistors in the DC feeds and IC sockets crumbling. I pretty well replaced every tantalum with quality electros and the desk came up sounding great.

 

[pinebox]

I hear a lot of conflicting opinions on tantalums in signal path, curious what the consensus is here about them aging out like electrolytic caps? Would you include them in a preventative maintenance recap? (I know the idea of shotgunning caps just from age can have strong opinions too)

 

[k brown]

Tants are one of the few things I would shotgun without hesitation. The 'Neve' color they provide is not enough to make it worth keeping them.

 

[JohnRoberts]

I hear a lot of conflicting opinions on tantalums in signal path, curious what the consensus is here about them aging out like electrolytic caps? Would you include them in a preventative maintenance recap? (I know the idea of shotgunning caps just from age can have strong opinions too)

I am not a big proponent of significant audible differences between tantalum and aluminum electrolytic caps. In a console you can prove or disprove to yourself with null testing. Upgrade caps in one channel then subtract that channel from an unmodified channel by inverting one and summing into a bus. That should reveal what difference there is/isn't.

For re-capping in general I advocate benchmarking channels by measuring frequency response. LF loss is a typical marker for dried out electrolytic caps. If you find one bad cap, replace all from the same likely date/batch.

JR

 

[Whoops]

Like others have said, tantalum are known to fail short circuit in PSUs, which can make a big mess.

Modern Aluminum Electrolytic caps are very good and small, can withstand high temperature and are low ESR.

Replace those tants with Panasonic Lytic capacitors rated at 105 degrees and you really well sorted

 

[metalockpick]

The low ESR and low dielectric absorption aspects of tantalum capacitors are why they were/are selected in some circuits. For ESR, imagine that every capacitor has a small resistor in series (internally). The tantalum cap has a smaller resistance. The dielectric absorption spec of tantalum means that the capacitor is absorbing less of the signal than other types.
The problem with tantalum caps is that even though these specs are great, they don't handle ripple current well (they explode). So, even though they get used for power supply bypass caps where there is low ripple in the circuit, after a few years when the chinese brute force filter capacitors in the PSU start losing capacitance, it's only a matter of time before the tantalum caps fail in dramatic fashion. Also, because of the cost, they are not properly derated in circuits in the interest of saving money.
LIke many things, to get a few nths better in quality, you have to spend exponentially more money. Consumer products aren't typically built like that, and don't have the good stuff (merely the lowest spec "Tantalum" cap that can be had).
In some circuits, the low ESR and Dissipation Factor of the caps used are audibly noticeable.
In an EQ, you will get sharper max Q and less loss (and you can always add series resistors to mimic a lower spec capacitor). That said, it's very slight, and whether or not you like the difference is up to you.

 

[Admiral-dk]

I guess the best question to ask is: why use them at all? Just 'cause they're small?

I can think of one BIG reason - they usually have less then 1% of the Leakage current of a Electrolyte Capacitor ...!!!

So if you for one reason or another have a High Impedance Circuit where Leakage is a NO, NO, and you need a 'Big Capacitance' they are the only way to go ....
But much better if you can modify the circuit so they aren't needed at all ....

Per

 

[kidb]

I have been servicing a vintage Neve console for 11 years now, and have never had a single issue with tants in signal path, in my opinion they are part of the neve sound, which you may or may not enjoy. But absolutely tants don't belong in PSUs, only in signal path in my experience. Electrolytics for everything else.

 

[Whoops]

I have been servicing a vintage Neve console for 11 years now, and have never had a single issue with tants in signal path, in my opinion they are part of the neve sound, which you may or may not enjoy. But absolutely tants don't belong in PSUs, only in signal path in my experience. Electrolytics for everything else.

+1
I never seen them fail in any Neve amp card ever, and I serviced a lot of them.

But in PSU's I had already to replace a lot of them that were shorted in many vintage, varied, PSU units.

 

[RoadrunnerOZ]

In a console you can prove or disprove to yourself with null testing. Upgrade caps in one channel then subtract that channel from an unmodified channel by inverting one and summing into a bus. That should reveal what difference there is/isn't.

I’ve done this years back on a console with identical can electrolytic recapping done on two channels and new tantalums in audio pass of the line, fader and bus out on a third. When, with one inverted, the two identically capped channels were different - no absolute null. When using one capped with can electros and one with tantalums, inverting one, they were also different.
I tried at various generator frequencies on both scenarios and it was difficult to see the difference between the two sets of differences. Bottom line is no two channels in a console are equal - so I reduced the path and just looked at the output of the line input stage and the difference was much reduced in both cases - still no absolute dead null - still hard to tell if the tantalum channel was “more different”. I pulled the tantalum caps and continued with low esr electrolytic caps. Saved money for the client as well.

 

[JohnRoberts]

I’ve done this years back on a console with identical can electrolytic recapping done on two channels and new tantalums in audio pass of the line, fader and bus out on a third. When, with one inverted, the two identically capped channels were different - no absolute null. When using one capped with can electros and one with tantalums, inverting one, they were also different.
I tried at various generator frequencies on both scenarios and it was difficult to see the difference between the two sets of differences. Bottom line is no two channels in a console are equal - so I reduced the path and just looked at the output of the line input stage and the difference was much reduced in both cases - still no absolute dead null - still hard to tell if the tantalum channel was “more different”. I pulled the tantalum caps and continued with low esr electrolytic caps. Saved money for the client as well.

deep nulls are difficult because even tiny frequency or phase response errors will compromise the depth of null.

I would focus on small errors in the strip eqs to get the null deeper.

Listening to the null product will generally tell you if there is something nasty going on or not, probably not.

JR

 

[Newmarket]

+1
I never seen them fail in any Neve amp card ever, and I serviced a lot of them.

But in PSU's I had already to replace a lot of them that were shorted in many vintage, varied, PSU units.

Yeah - but I have (literally) seen relatively modern electrolytics in PSU type apps burst at random for no good reason on occasion. Tant TAJ series - no failures afaik. YMMV obvs.

 

[RoadrunnerOZ]

Listening to the null product will generally tell you if there is something nasty going on or not, probably not.

I did try the nulls with a music source as well and got marginal very thin HF pass of audio in both cases at very low level.

 

[RoadrunnerOZ]

Yeah - but I have (literally) seen relatively modern electrolytics in PSU type apps burst at random for no good reason on occasion. Tant TAJ series - no failures afaik. YMMV obvs.

You’ll always get failures whether can electrolytic or tantalum especially in situations where tantalum caps get over voltage or electrolytic caps get too much heat stacked next to voltage regulators - the surface heating by proximity I’ve seen cause electrolytic failure - that’s where the cap lifts off its base and spills electrolyte all over the board or the top splits open and blows. It’s more down to design I think

 

[GeorgeToledo]

An article I once read in REP stated that older production tantalum caps were more prone to failure than newer due to issues with obtaining needed purity for the tantalum.

 

[RoadrunnerOZ]

An article I once read in REP stated that older production tantalum caps were more prone to failure than newer due to issues with obtaining needed purity for the tantalum.

Yeah probably true - servicing old gear I always come up against total failure of tantalums and drying out of can electrolytics, resistance values drifting, IC’s and transistors failing or going noisy, switch contacts oxidizing, PCB connectors oxidizing, fader tracks wearing out, LED’s going short and killing 5V lines so bad caps are just another headache in a long list of things that go bang in the night…….

 

[k brown]

Yeah probably true - servicing old gear I always come up against total failure of tantalums and drying out of can electrolytics, resistance values drifting, IC’s and transistors failing or going noisy, switch contacts oxidizing, PCB connectors oxidizing, fader tracks wearing out, LED’s going short and killing 5V lines so bad caps are just another headache in a long list of things that go bang in the night…….

... or slowly whimper into dust......

 

[RoadrunnerOZ]

... or slowly whimper into dust......

No - that’s the old IC sockets that crumble like the walls of an ancient ruin hiding underneath the perfectly good op-amp that cooked them. Pull the chip and the socket falls to bits

 

[k brown]

So much for "The Graduate": " . . . plastics!"