Selecting Capacitors to Minimize Distortion in Audio Applications

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
As I shared before the only tantalum I used intentionally was in a side chain time constant. Because Tantalum has poor DA (dielectric absorption) they can behave slightly differently in such timing circuits.

Warning: this is about to get a little esoteric. DA or soakage is a characteristic of capacitors related how the energy storage is distributed within the physical capacitor structure. The typical model for DA is bunch or RCs in parallel. The typical example showing DA in action is taking a charged up capacitor, short it out momentarily, and then when you release the short, the capacitor recovers to a fraction of the initial charged voltage state. This is caused by residual charge inside the capacitor equalizing or normalizing over time.

In practice for most typical audio applications, like DC blocking where the capacitor is loaded with a resistive termination the DA does not show much evidence (while the audiophools tried to make a big deal about it back in the 70s/80s). However if loaded with a nonlinear termination that charges and discharges with different impedances, DA can express it's ugly head. The obvious example of this bad behavior is in a sample and hold circuit. We can easily imagine the high DA capacitor after being quickly charged by a low impedance sample circuit, followed by the high impedance unloaded hold circuit, where like the un-shorted capacitor seeks out its longer term equilibrium. We rarely see electrolytic capacitors used in sample and hold circuits. In side-chain time constants, and that meter example being asked about, the high DA capacitor being charged with low impedance fast attack time constant, then released to decay with higher impedance termination can see some influence caused by DA.

I expect this DA caused error to be small, but substituting a low DA capacitor for the tantalum "could" introduce a subtle tracking difference. If you want the meter behavior to be the same use the same dielectric capacitor.

JR

PS; I'm glad to see I am not the only one overthinking capacitor dielectrics.
 
Such substance should be built in the studio, but how many recordings are actually done in a studio today?

In the big scheme, probably very few. I'd guess they're mostly a bunch of isolated overdubs done to click tracks. Possibly in small producer/home/bedroom studios. Maybe a couple of days in a bigger room if there are strings, brass, whatever... involved.

Substance in the studio trumps anything else for sure. GarageBand and a Fireface would suffice for capturing that.
 
Last edited:
Yes, definitely for stability. There's almost zero risk of dramatic failure in this position so it's a good choice. Now, as I'm cheap, I would try one of those miniature MLCC that cost nothing and pack high density. Tolerance is about 20% but a pack of 100 yields a good dozen of 2%. Beware of tempco, though. Now, how important is the time-constant in a meter?
https://fr.aliexpress.com/item/3284...9b365ac-25&pdp_ext_f={"sku_id":"65142055195"}
 
Now, how important is the time-constant in a meter?
It depends, how important is a meter?

I am not a fan of mechanical VU meters but some customers are fans. I had one picky customer who could see the difference between driving a VU meter from a single 12v rail, vs +/- 12V.

I will repeat myself the impact of DA in that meter drive circuit will be subtle. That doesn't mean somebody won't notice.

JR
 
It depends, how important is a meter?
Yes, that was the essence of my comment. In this particular case, the meter is a LED bar, so the fast attack time (510r/1uF) allows catching transients in a useful manner. However I'm not sure of the practical consequences of a -50/+100% variation. Fastest display of transients does not necessarily implies that the operator's reaction will be significantly faster.
I was the distributor for REW. Some models had switchable attack/relase time. I haven't found it made any practical difference. Indeed the fastest setting resulted in a slightly higher indication but users expected it, so allowed the signal to be a little more in the orange zone.
I am not a fan of mechanical VU meters but some customers are fans.
Neither am I. IMO they belong to tape recording and are inadequate for digital. The best demonstration of their inadequacy is when recording claves.
I had one picky customer who could see the difference between driving a VU meter from a single 12v rail, vs +/- 12V.
I don't get it; you mean applying 12V or 24V on the meter? Or the circuit that drives it powered from either 12 or 24V?
I will repeat myself the impact of DA in that meter drive circuit will be subtle. That doesn't mean somebody won't notice.
It's quite possible, but I don't think a conclusion can be drawn regarding suitability. Of course, objectively, the effects of DA on the TC's are undesirable (or not). Actually, isn't it what the auto-settings in compressors try to emulate?
 
Yes, that was the essence of my comment. In this particular case, the meter is a LED bar, so the fast attack time (510r/1uF) allows catching transients in a useful manner. However I'm not sure of the practical consequences of a -50/+100% variation. Fastest display of transients does not necessarily implies that the operator's reaction will be significantly faster.
I was the distributor for REW. Some models had switchable attack/relase time. I haven't found it made any practical difference. Indeed the fastest setting resulted in a slightly higher indication but users expected it, so allowed the signal to be a little more in the orange zone.
don't get me started on meters... I have two patents for variants on simultaneous peak & VU.

I did some work on what makes a difference and what doesn't. I even coded a true RMS VU meter digital, and it didn't make squats difference vs a simple average.
Neither am I. IMO they belong to tape recording and are inadequate for digital. The best demonstration of their inadequacy is when recording claves.
agreed... but there are fans for VU even here
I don't get it; you mean applying 12V or 24V on the meter? Or the circuit that drives it powered from either 12 or 24V?
yup the drive circuit PS rails. 6 dB less Vp-p driving the mechanical meter made a difference in ballistics to a customer used to true VU response. Sometimes the customer schools us (me?). :cool:
It's quite possible, but I don't think a conclusion can be drawn regarding suitability. Of course, objectively, the effects of DA on the TC's are undesirable (or not). Actually, isn't it what the auto-settings in compressors try to emulate?
The issue is about making X like Y... if X used a tantalum, and Y wants to track the same as X, use a tantalum there too.

I have little use for mechanical meters myself... but I haven't recorded anything but spam calls*** on my answering machine this century.

JR

*** how come they always say final notice, then call again?
 
I did some work on what makes a difference and what doesn't. I even coded a true RMS VU meter digital, and it didn't make squats difference vs a simple average.
There's a difference between a fast meter and a slow one; slow ones, average or rms, are just slow.
agreed... but there are fans for VU even here
I know. Some even admit their preference is based on aesthetics, which is commendable, as long as it's not disguised in techno babble.
yup the drive circuit PS rails. 6 dB less Vp-p driving the mechanical meter made a difference in ballistics to a customer used to true VU response. Sometimes the customer schools us (me?).
Unfortunately I cannot reproduce the experiment, but it's an interesting subject.
The issue is about making X like Y... if X used a tantalum, and Y wants to track the same as X, use a tantalum there too.
I agree. I wouldn't want to have the right meter acting differently than the left one.
I have little use for mechanical meters myself... but I haven't recorded anything but spam calls*** on my answering machine this century.

JR

*** how come they always say final notice, then call again?
Here they have bots that disconnect when they detect an answering machine. I get dozens of aborted calls everyday, but I don't need to clean any messages.
 
Here they have bots that disconnect when they detect an answering machine. I get dozens of aborted calls everyday, but I don't need to clean any messages.
and they have bots that accidentally (or on purpose) let the machine pick up and leave a message.

I generally let it ring and they generally drop the call, except for this one passive-agressive son of a can opener (bot) that is trying to extend my warranty on a 24 year old car. They offer for me to press #2 to get on their do not call list, but I wasn't born yesterday.

They also keep saying this is the final notice, but they f'n lie (like a politician).

JR
 
Last edited:
and they have bots that accidentally (or on purpose) let the machine pick up and leave a message.
Ah, OK. Here they would risk a lot if they did. First they would be very quickly cut off the network, then get a severe fine (several €€€ per proven violation).
Occasionally, there's an exception, with an international number.
 
I have an old friend who actively engages with spam callers, taking them to small claims court and charging them money, but he wastes more time on that than I am willing to. I think he also uses some aftermarket call screener.

Sorry about the veer but apparently this comes down to how the phone system works where spam/robo callers operate in the fringes of legality with some, kinds of calls exempted.

IMO this is just another sign of government ineffectiveness. Of course they could stop these if they really wanted to.

JR

PS: Sorry about the veer...
 
I have an old friend who actively engages with spam callers, taking them to small claims court and charging them money, but he wastes more time on that than I am willing to. I think he also uses some aftermarket call screener.

Sorry about the veer but apparently this comes down to how the phone system works where spam/robo callers operate in the fringes of legality with some, kinds of calls exempted.

IMO this is just another sign of government ineffectiveness. Of course they could stop these if they really wanted to.

JR

PS: Sorry about the veer...
I actually have fun with it. I have a spam alert program on my phone called Hiya, and when a call comes up SPAM, I answer it with a made up language, sort of like African bush meets Amazon rain forest meets Chinese. They try to communicate but hang up. I have found that they never call back. It's quite fun. Only once did anyone respond, and that was an American guy that answered "rom-a-loma-ding-dong" and then hung up.
 
I see caller ID on my TV screen... Yesterday I got an rare phone call from an actual old friend and I was able to see his name on the screen.

To engage a human from a robo call I'd need to press a button, admitting that there is a human at this end of the line.

I think they know they are leaving messages and are just being even more annoying. I have to delete their messages so I don't miss a real one ignoring the blinking light.

JR
 
Did Rupert keep on using them in his later endeavours?

Is it the only reason for keeping them tants? I know acquired taste is a strong motivation, but surely, some products that do not "feature" the "tant sound" have been largely successful...

Interesting point about how 'original' designers regard their previous designs.
This is not to negate the validity of those cloning or re-creating vintage designs but an observation.
Back in the 90s (yeah - it's frighteningly long ago now I think about it) I had a relatively brief professional involvement with Dick Swettenham (RIP /Helios) when working in pro-audio in South Wales and he lived in Bristol. At the time he was involved with 'overseeing' the assets of the financially failed 'Novation' Digitally Controlled Analogue system and also promoting various audio technology solutions to the industry.
Now while he had various interesting anecdotes about eg Olympic Studios / Rolling Stones / Island Records etc etc, one thing he never (to my memory) went on about was the Helios circuits and their 'resurrection'. He was all about the new stuff - ergonomics of assignable control surfaces, audio semiconductor solutions etc.
 
Back
Top