Capacitors in parallel

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

pinebox

Well-known member
Joined
Mar 11, 2015
Messages
117
I have been working at a studio for a few years where I always find equipment with a 0.47u capacitor in parallel with other capacitors. I am pretty certain that a previous tech modified all of these since they are always the same caps, and the more things I open the more things I find with them. Was this a trendy thing to do at one time, or were they overzealous with the technique, or whats up? From what I understand this will lower ripple, can anyone explain or point me to some reading about how this improves sound and if there is a certain type of cap that suits this mod best?
 
I've lost count how many times I have answered similar questions and shared this one anecdote but I couldn't find it in a quick search so one more time for the cheap seats.  :'(

a- the theory is that smaller capacitors with better HF impedance characteristics in parallel with larger, slower caps can make a composite super cap.  ;D

Back in the 70s I looked deeply into this on my own bench while dialing in a phono preamp when I measured too much phase shift in a phono preamp gain stage at 20 kHz caused by the ESL (equivalent series inductance) of a typical aluminum electrolytic capacitor.  My bench tests revealed that the parallel capacitor needs to be no less than 10% of the capacitance of the larger cap to make an effective improvement (so unless that 0.47 is across a 4.7 uF you are wasting time and money). 

Back then I came up with completely different strategies (for new designs) than parallel caps.

Learn how to read and understand the data sheets...  Its just physics...

DO NOT EXPECT SIMPLE ANSWERS for complicated problems.  8)

JR
 
I recall reading somewhere that there was no significant benefit to paralleling an electrolytic with a film. It might seem like a good idea - the electrolytic handles the large currents and the film adds a low esr path for transients. But in practice it was found that the film cap had to be almost as large as the electrolytic to measure an improvement.

And in a properly designed circuit, a little extra supply ripple should have zero effect on the audio output anyway. Even in a high gain class A stage there should be some extra local filtering. Differential stages have very good common mode ripple rejection.
 
abbey road d enfer said:
It is still a trend in audiophool circles. Some have made a business of selling "upgrade" kits.
;D ;D ;D

A lot of myths I wrote about back in the 80s are still around and kicking... For some reason people want to believe that audio designs can be easily improved by using more expensive components. Better is always better but not always audible.

About capacitors specifically, sometimes just swapping out 20+ year old aluminum electrolytics with fresh aluminum can be audible. Some standard advice for people considering such an overhaul on a larger console, perform some bench testing on each channel to look for weak sisters... Something as simple as LF frequency response can reveal tired capacitors in the path.

Nulling two channels against each other at a time can identify outliers (while it won't tell you which one of the two is out to lunch).

Trust objective measurements, more than here-say, or even me.  8)

JR
 
I´ve been using bypasscaps extensively. Some things I found out in blind tests I did with friends and collegues:

Bypasscaps make barely any audible difference parallel to new electrolytics. They make a difference parallel to old electrolytics in a good way, though. Bypassed old electrolytics sound closer to new electrolytics.

It´s not audible to have a single electrolytic bypassed when up or downstream are more electrolytics without bypass caps. If applied on all coupling caps it get´s audible.

Everybody involved liked the the soundquality better when bypass caps were used.

And just for the record, the theory behind bypass caps can be read here, e.g.: Jerald Graeme, Optimizing Op Amp Performance.
 
jensenmann said:
I´ve been using bypasscaps extensively. Some things I found out in blind tests I did with friends and collegues:

Bypasscaps make barely any audible difference parallel to new electrolytics. They make a difference parallel to old electrolytics in a good way, though. Bypassed old electrolytics sound closer to new electrolytics.
There is no doubt that old electrolytics suck. Just like I wouldn't power an apparatus with half-dead batteries, I wouldn't let a piece of equipment run with dry 'lytics, for more than one reason. They suck the lows by having reduced capacitance, they suck the highs by having increased ESR, they suck the mids by having increased Dielectric Absorption, AND they will die one day; best case they don't allow signal to pass, worst case they explode and release gunk all over the place.

It´s not audible to have a single electrolytic bypassed when up or downstream are more electrolytics without bypass caps. If applied on all coupling caps it get´s audible.
That would be true if all caps had the same life cycle. Not always.

Everybody involved liked the the soundquality better when bypass caps were used.
A valuable evaluation would be to compare with a freshly recapped channel with modern 'lytics and nothing else.

And just for the record, the theory behind bypass caps can be read here, e.g.: Jerald Graeme, Optimizing Op Amp Performance.
Sorry, this book is too expensive to refresh what I learned in school and 50 years of subsequent involvement in audio design.
Paralleling caps is well documented; many audiophool sites claim benefits that can never be proved in proper experimentation conditions (i.e. comparing the same new cap with and without by-pass). However, for power supply decoupling, there is a case for paralleling caps; but PCB trace impedance becomes a significant factor, and caps are not strictly in parallels anymore.
 
abbey road d enfer said:
A valuable evaluation would be to compare with a freshly recapped channel with modern 'lytics and nothing else.

That´s exactly what I did in one of my tests. I had four console channelstrips prepared with 1. in original condition (25yrs old caps), 2. original condition with bypass caps, 3. recapped, 4. recapped with bypasscaps.
 
jensenmann said:
That´s exactly what I did in one of my tests. I had four console channelstrips prepared with 1. in original condition (25yrs old caps), 2. original condition with bypass caps, 3. recapped, 4. recapped with bypasscaps.
That sounds like an ideal opportunity to perform a null test to determine what the actual difference is "objectively".  Flip the polarity on one strip and then sum two outputs together... The null test won't tell you which channel is different but with 4 channels you can probably determine the outlier.

Paralleling capacitors is commonly done for power supply rails, but that is a far more rigorous application than coupling caps in DC blocking circuits that are (or should be) lightly loaded.

The subject of my bench testing back in the 70s was a simple RIAA phono preamp gain stage. As I recall the gain resistor was something like 360 ohms in series with a 22uF aluminum capacitor. Not exactly lightly loaded. I was able to measure (NOT HEAR) tens of degrees of phase shift at 20 kHz cause by the ESL of the aluminum cap.  Since my design goal even back then was to design out all deviation from ideal transfer function, that undesired phase shift had to go.  8)

I am repeating myself but with a console first benchmark all your channels to first confirm they are working to spec... old electrolytic caps are notorious for losing electrolyte and delivering lower than predicted capacitance (in DC blocking this generally results in reduced low bass).  In old console strips there are many things that can degrade over time (like dirty pots and switches, etc).

Measure twice, cut once.....  Use you ears for mixing, not circuit design, test equipment is superior for that.

JR
 
JohnRoberts said:
Measure twice, cut once.....  Use you ears for mixing, not circuit design, test equipment is superior for that.

In my tests back then I wanted to determine if time and money are well invested in bypasscaps. I´ve read Graeme´s book and from an engineering standpoint it seemed to be solid knowledge and easy to apply. But all theory doesn´t help much if it doesn´t translate into soundquality. I´ve seen/heard gear that in theory and on paper should be great but sounded mediocre and vice versa. That´s why I wanted to hear the proof before going through all that work and expenses. (I didn´t mention before that the context was preparations for refurbishing a 40 channel console)

Beside that it was striking to me how much bypasscaps helped old electrolytics to sound more natural in that test. So when I have any piece of gear and want to preserve it´s "new and fresh" sound from aging over the years, then bypasscaps are the solution for sure.
 
jensenmann said:
Beside that it was striking to me how much bypasscaps helped old electrolytics to sound more natural in that test. So when I have any piece of gear and want to preserve it´s "new and fresh" sound from aging over the years, then bypasscaps are the solution for sure.

This matches my experience.  I improved the sound of a 10 year old console vastly with the addition of bypass caps.  A recap would be impractical (600+); large monolithic PCB with easily damaged traces.  Parallel tack-ins were really easy to do.  They were more on the order of 1% or less in value.  Do it again in 10 years?  Haven't had to.  Yeah, I know, subjective......I made a thread about it, those who want can go find it. 
 
jensenmann said:
In my tests back then I wanted to determine if time and money are well invested in bypasscaps. I´ve read Graeme´s book and from an engineering standpoint it seemed to be solid knowledge and easy to apply. But all theory doesn´t help much if it doesn´t translate into soundquality. I´ve seen/heard gear that in theory and on paper should be great but sounded mediocre and vice versa. That´s why I wanted to hear the proof before going through all that work and expenses. (I didn´t mention before that the context was preparations for refurbishing a 40 channel console)

Beside that it was striking to me how much bypasscaps helped old electrolytics to sound more natural in that test. So when I have any piece of gear and want to preserve it´s "new and fresh" sound from aging over the years, then bypasscaps are the solution for sure.
BUT you might have heard the same improvement from replacing the cap with a modern electrolytic.

If you can hear the difference you should be able to measure the difference (in my experience I have).

JR

PS: I repeat.... from my empirical bench measurements the smaller parallel cap needs to be no less than 10% of the total capacitance, and posses superior dielectric (or why bother).
 
JohnRoberts said:
PS: I repeat.... from my empirical bench measurements the smaller parallel cap needs to be no less than 10% of the total capacitance, and posses superior dielectric (or why bother).

That is a great rule of thumb John, thanks!

BTW. In Douglas Self's book he empirically proved that if you keep the signal voltage on a coupling electrolytic capacitor below around 80mV@20Hz, the distortion of the cap is indistinguishable from the noise floor of an AP unit. In other words, if the capacitor is big enough, there is no significant distortion. So for instance, with a typical 10k load, any capacitor above 150uF is probably overkill but would (according to Self) be distortion free.
 
user 37518 said:
That is a great rule of thumb John, thanks!
That was my observation for a very specific circuit. After recognizing that capacitors are not ideal and have multiple moving parts, I worked to finesse or engineer the problems out.

In simple DC blocking circuits I set the pole frequency several octaves lower, so the terminal voltage is squat inside the audio band. 

I addressed this in my old magazine column (Audio Mythology) back in the 80s. I won't repeat the entire column but a few highlights.

#1 non-ideal cap behaviors (like dielectric absorption and voltage coefficient ) can be minimized by keeping the terminal voltage across electrolytic caps small...(thus the low pole frequency example above). If you can't keep the terminal voltage small (like in band audio filters) use the best dielectric you can. These days NPO/COG ceramics are very good, back in my day I used good film caps.

#2 Capacitor characteristics like ESR and ESL (series resistance or inductance) express as a function of the current they are passing.  The example I gave of a phono preamp was effectively loading a 22uF cap with 360 ohms. Passive loudspeaker crossovers are absolutely worst case for capacitors... If you must, and can, film caps like polypropylene suck less in that application... That said active crossovers completely avoid putting capacitors through that much pain.

A classic design approach to completely avoid electrolytic blocking caps is the DC coupled mic preamp topology.

Transformer mic preamp front ends trade one non-ideal component for another, while opinions surely vary about that.

For scratch designs we have the luxury of avoiding the obvious component weaknesses. To restore old designs I will repeat until people listen... measure, measure, measure...  confirm the path performance, and if pursuing upgrades, measure the before and after so you know what actually changed or didn't.

There is a lot of expectation bias masquerading as empirical evidence around DIY tweaks. These were pretty popular decades ago and even called "Pooge" (Progressive Optimization of Generic Equipment a phrase attributed to Walt Jung). I am not a fan of second guessing design engineers, but some people make a career of that (just spend some time reading gearslutz).

JR
 
> even called "Pooge" (Progressive Optimization of Generic Equipment a phrase attributed to Walt Jung).

I first read of POOGE in Ed Dell's column, Audio Amateur. You think he got it from Walt?
 
PRR said:
> even called "Pooge" (Progressive Optimization of Generic Equipment a phrase attributed to Walt Jung).

I first read of POOGE in Ed Dell's column, Audio Amateur. You think he got it from Walt?
TAA was where I heard it used, but an internet search of "pooge" today turns up the Walt Jung attribution...

Walt was a contributor to TAA back in the day... among many others. 

JR
 
Tubetech like the parallel cap idea as evidenced in the pic below,

if you integrate the time domain with respect to the semblance factor, you get a better spatial field at all phase angles when transients of more than 3 dB are rectiied by the incoming CMMR along the longitudal force field which arises when the arctan (x) reaches it's limit as y approaches infinity during the delay cycle,

integrate the equation attached to the image with respect to y(z) and you will get the reciprocal of the hi frequency drop of plus a constant C.
 

Attachments

  • tt.JPG
    tt.JPG
    215.3 KB · Views: 43
CJ said:
Tubetech like the parallel cap idea as evidenced in the pic below,

if you integrate the time domain with respect to the semblance factor, you get a better spatial field at all phase angles when transients of more than 3 dB are rectiied by the incoming CMMR along the longitudal force field which arises when the arctan (x) reaches it's limit as y approaches infinity during the delay cycle,

integrate the equation attached to the image with respect to y(z) and you will get the reciprocal of the hi frequency drop of plus a constant C.
It is common practice to parallel caps to make odd values, and you don't need integral calculus, simple addition will work.

JR
 

Latest posts

Back
Top