Dano's Big Bag of Capacitor Questions

[dano12]

So I'm working on a little write-up about capacitors in stompboxes. The more I write, the more questions come up. I think I am getting close to grasping a lot of the basic concepts, but some things are still hanging out there.

Apologies in advance for the long list, but I'm sure I'm not the only capacitor dunderhead in the crowd.

Output capacitors: Stompbox designs seem to be divided into two camps here: film caps for output vs. polarized electrolytics. In camp A, a 100nf or so film cap is on the output. Camp B uses electros, typically 1uf.

- Is it correct to assume that a 1uf is a good "neutral" value in that it
  will pass the full frequency range typical in electric guitar applications?

- Is the choice of camp A vs. B simply a function of attenuating a
  certain frequency range at the output? Or does output impedance
  come into the equation?

Non-polarized vs. polarized electrolytics: Non-polarized electrolytics seem to be a rarity.

- What advantages, if any, do they have in domain of stompboxes?

- Why did that Ibanez engineer chap use 1uf non-polarized electrolytics
   when he designed the Tube Screamer? He had 'em on the bench,
   or some deeper purpose?

Cap choices for 1uf and beyond: So I want to use a 1uf or so value in a de-coupling application.

- Are the design considerations, besides cost, that would point me towards
  non-electro caps like tantalums?

Ceramic vs. Poly Film:

- If I can find the correct capacitance and voltage rating in a film cap, and
  cost is not an issue, can I assume that there is no good reason to
  use ceramic caps.

Tantalums: I hate to keep harping on the tubescreamer, but
- why did the Ibanez chap choose a .22 tant after the first opamp stage?

Power supply filtering: I know that a biggish electro to ground after my DC input jack will reduce ripple, which I understand to be the superimposition of residual AC which I don't want on my DC line.

- Is ripple suppression all I am accomplishing here? Or does a stompbox 
  have the need for a reservoir current? I know that some amps need
  a reservoir, but do stompboxes?

- Should I assume that all AC-DC wall-warts that my stompbox may see
  are unfiltered affairs and that is why I need to do power supply filtering?

- Ok, so what is going on in the following Rat power supply snippet? I understand C1—a typical higher value electro to ground.


- But what about C2—does it get rid of high-frequency noise that may be
  present on the power supply?

- And what the heck is going on with C3? Why would I need another
  electro to ground there?
   

[Barcode80]

that looks to me like the AMZ led pop filter in that schem.

i can discern no audible difference in ceramic and film caps, though some others claim otherwise. IMHO films are easier to work with though, as they are more square and therefore in my designs and boards they fit a lot better. of course, i can't stand components being free at all on a board, so the fact that i can't get a ceramic cap leads completely flush with the board makes me only use them for really small values.

tants can introduce some noise to my ears, though miniscule. i assume most likely that "that ibanez chap" chose the tant for space design constraints. a .22 in film is fairly large to fit in that enclosure.

[MartyMart]

Some answers as I understand it - others may differ/offer a more "EE" response ! :
1uf input/output cap - will pass everything down to below a 6 string bass  !
I try to use poly's at this value and often replace a 10uf electros in the audio path with a 1uf poly
Output caps where an electro of 10uf is required can be paralleled with a 100n poly to retain the high end.
Typical "guitar" stompers have perhaps 100n I/O which is fine for the freq response, reduce for less low end
mud - in particular with OD's/Distortions - values such as 10n/22n/47n input and then perhaps 100n/220n at
output to get some bas content back after clipping.
NP electros are rarely used, in the TS-9 it's at a point where the voltage can swing either way, so a standard
polerized cap would have failed - also a size issue - getting 1uf NP poly's can be hard ... but they are around.
0.22 uf as tants are quite small, loads of MXR pedals use lot's of tants for this reason - to fit in a 1590B !!
tantalum and electro caps in the signal path can cause some " dirt n grit " which seems fine in an OD etc, but
I'd rather have them out of a clean booster etc.
Some people dont like changing the tants from the TS-9 to poly for this reason.
PSU filter caps, electro's have low ESR so paralleling a 100n poly or ceramic helps with high end noise filtering
we also need to de-couple the point where the VB 4.5v rail is created ( usually ) so hence the extra cap here.
Hope that helps some ?
MM.

[dano12]

Thanks for the feedback so far.

Still wondering on power supply filtering questions...

Here is what I have so far:
http://www.beavisaudio.com/techpages/Caps/index.htm

[Rmanen]

Nice read, I knew most of it already, since I study electrical engineering, am a DIY pedal builder and self proclaimed audiofile .
A nice read nonetheless.

Rob

[edster]

The .047 is to reduce RFI.

[calculating_infinity]

Man this is a great article!  Finished or not!  Thanks Dano for your time and effort.

[edster]

Yeah seriously man..... do you sleep?

You are one ambitious mofo.

I mean that, of course, in the best possible way.   

[ultar]

awesome article. i have all of your tech pages saved in a folder because they are the most non-boring easy to understand articles to read as a newcomer. very helpful stuff!!

[pjwhite]

So I'm working on a little write-up about capacitors in stompboxes. The more I write, the more questions come up. I think I am getting close to grasping a lot of the basic concepts, but some things are still hanging out there.

Apologies in advance for the long list, but I'm sure I'm not the only capacitor dunderhead in the crowd.

Output capacitors: Stompbox designs seem to be divided into two camps here: film caps for output vs. polarized electrolytics. In camp A, a 100nf or so film cap is on the output. Camp B uses electros, typically 1uf.

- Is it correct to assume that a 1uf is a good “neutral” value in that it
  will pass the full frequency range typical in electric guitar applications?

It's not really correct to assume anything.  But if you do assume that your guitar amp has a 50K input impedance and your stompbox has a relatively low output impedance, a 1uF output capacitor will pass frequencies down to about 3Hz.

- Is the choice of camp A vs. B simply a function of attenuating a
  certain frequency range at the output? Or does output impedance
  come into the equation?

Yes and yes.

Non-polarized vs. polarized electrolytics: Non-polarized electrolytics seem to be a rarity.

- What advantages, if any, do they have in domain of stompboxes?

I'm not an expert on NP electrolytics, but I would think that they would have somewhat lower distortion characteristics than
polarized electrolytics.

- Why did that Ibanez engineer chap use 1uf non-polarized electrolytics
   when he designed the Tube Screamer? He had ‘em on the bench,
   or some deeper purpose?

See my previous answer.  And they were probably cheap and in-stock.

Cap choices for 1uf and beyond: So I want to use a 1uf or so value in a de-coupling application.

- Are the design considerations, besides cost, that would point me towards
  non-electro caps like tantalums?

Tantalums are polarized and would have the same drawbacks as aluminum electrolytics in audio path applications.

Ceramic vs. Poly Film:

- If I can find the correct capacitance and voltage rating in a film cap, and
  cost is not an issue, can I assume that there is no good reason to
  use ceramic caps.

Film caps are generally preferred over ceramic in audio path applications.

Tantalums: I hate to keep harping on the tubescreamer, but
- why did the Ibanez chap choose a .22 tant after the first opamp stage?

I'm not familiar with the circuit so I couldn't say.

Power supply filtering: I know that a biggish electro to ground after my DC input jack will reduce ripple, which I understand to be the superimposition of residual AC which I don’t want on my DC line.

- Is ripple suppression all I am accomplishing here? Or does a stompbox 
  have the need for a reservoir current? I know that some amps need
  a reservoir, but do stompboxes?

I think that most stompboxes have no need to handle large current spikes.  Ripple suppression is the main reason for this capacitor.

- Should I assume that all AC-DC wall-warts that my stompbox may see
  are unfiltered affairs and that is why I need to do power supply filtering?

If you plan on using random wall warts, you can assume they will have minimal internal filtering.  Some wall warts are better than others though.

- Ok, so what is going on in the following Rat power supply snippet? I understand C1—a typical higher value electro to ground.


- But what about C2—does it get rid of high-frequency noise that may be
  present on the power supply?

Right.  Electrolytics are not so good with high frequencies, so a smaller capacitor that can handle high frequencies in parallel with the electrolytic will suppress high and low frequencies.

- And what the heck is going on with C3? Why would I need another
  electro to ground there?

The combination of R2, R3 and C3 form an RC filter which stabilizes the 4.5 volt reference against voltage changes caused by small transient currents.
Without C3, the 4.5 volt supply would wander around as small variations in current occurred.

[/quote]

[rockgardenlove]

Hey I had an idea for another section:
Emitter bypass capacitors?  Lets the voltage swing further down to get more amplification.

For the "Variable Capacitors" section, you might also add that you can synthesize your own by just panning between different caps with a pot.

You could also note that tropical fish capacitors are really just poly film with a nice colouring scheme.

Under the "Ceramic Capcitors" section you've got a picture of a silver mica cap it looks like, not sure if this counts as ceramic.  They might be considered a special kind of ceramic or something, I dunno.

You could also mention can capacitors–the electro multi-caps often used in amps.

Great work!

[Fret Wire]

On the "tant" section, you might want to add that tants are very, very sensitive to reverse polarity. It doesn't take much to blow one when they are reversed. Also, it's good that you worded it as "some people say" when it comes to tants adding grit. It's not definitive. Personally, I've found that replacing the 10uf electro with a tant on the output of an 808 makes it smoother sounding. I usually sub tants for electro's everywhere it's practical in most ckts for a smoother sound. And besides distortions, MXR and Ross also used the tants in flangers, phasers, and delays. Replace them with electro's, and I doubt you'll notice much of a difference, sound wise.

One thing that is pretty much accepted as fact is that in 9v ckts, cap types will have less influence on sound than in amps with their larger voltages. Most well designed distortions will sound just fine with the cheapest type caps and resistors available. Therefore, in your acticle, I wouldn't get too much into claiming a certain type cap having a certain type sound. You'll be really just be promoting internet myth and mojo.

For instance, in a particular ckt, Marty may hear more grit with a tant, I may hear more smooth. Neither of us is right or wrong, different ears hear different things. So, we're both right. Yet, if either one of us says that "cap type X sounds like "this" in this ckt, we are really just a knat's hair from promoting myth. Just because it sounds like that to us, it may not be that way for others.

So, the truth is that with the miniscule voltages and currents that stombox's run at, cap type does less than most people think. Why do we still have our personal cap preferences? Because it's just plain fun to experiment with. Even though I know the sound differences run from slight to none, I still have my cap and resistor type preferences. My money and my time.

Another thing you might want to impress upon people is that the hallmark of a good sounding ckt design is that it will sound good with the cheapest cap type in your parts bin. Same with resistors. What makes or breaks a great ckt is the values of the components, not the type. That's 99% of the battle. Then, you can have fun with the other 1%, and experiment with cap types.

[blanik]

my dad is a (now retired) EE for >30 years in audio applications, he told me that they used to remove every tantalum cap in audio equipment at the tv station he was working, to replace them with regular caps 'cause the tants were too unreliable and would easily blow up and could mess a broadcast (in a time were live tv with new technology was actually the norm...  )

R.

[Fret Wire]

As long as a tant doesn't receive any reverse voltage, they outlast electro's by far. Nothing to dry out.

That's interesting. Given the large amount and relative complexity of the analog equipment in a TV or radio station back then, they must have had to go off the air for over a week to disassemble and replace every tant. Can't even imagine the pile of schematics, manuals, and caps laying around on that job.

[MartyMart]

For instance, in a particular ckt, Marty may hear more grit with a tant, I may hear more smooth. Neither of us is right or wrong, different ears hear different things. So, we're both right. Yet, if either one of us says that "cap type X sounds like "this" in this ckt, we are really just a knat's hair from promoting myth. Just because it sounds like that to us, it may not be that way for others.

That's true, I did say "some people" prefer those .22 tants in a TS and that I dont 
There's not much in it at all, as you say the circuit is everything NOT part type.
As for electro's in the signal path .........
Dano should be factual and not mythical .....of course
MM.

[dano12]

Wow, this little cap article has generated more emails, PMs and feedback than most of the other stuff I've done.

Great feedback, viewpoints and ideas. I updated the page this AM to incorporate as much of the feedback as possible. I'll polish this one over the coming weeks.

Again, wow. And thanks!

[Sir H C]

So I'm working on a little write-up about capacitors in stompboxes. The more I write, the more questions come up. I think I am getting close to grasping a lot of the basic concepts, but some things are still hanging out there.

Apologies in advance for the long list, but I'm sure I'm not the only capacitor dunderhead in the crowd.

Output capacitors: Stompbox designs seem to be divided into two camps here: film caps for output vs. polarized electrolytics. In camp A, a 100nf or so film cap is on the output. Camp B uses electros, typically 1uf.

- Is it correct to assume that a 1uf is a good "neutral" value in that it
  will pass the full frequency range typical in electric guitar applications?

It is an impedance thing.  What is the next stage loading.  You have an RC high pass with the output capacitor.  You can calculate the roll off frequency if you know the output loading with xc = 1/(2*pi*f*c) which is the impedance of the cap, and then find the frequency that it is equal to the Rload.  This is the 3db rolloff. 

- Is the choice of camp A vs. B simply a function of attenuating a
  certain frequency range at the output? Or does output impedance
  come into the equation?

Non-polarized vs. polarized electrolytics: Non-polarized electrolytics seem to be a rarity.

- What advantages, if any, do they have in domain of stompboxes?

Laziness.  You don't have to worry which node is at a higher voltage than the other with a NP electrolytic.  In some circuits the voltage of the two sides of the cap might be ill defined and so you can not say for certain which is the higher voltage, so NP is needed.  Usually you know, and polarized are cheaper so they are used.

- Why did that Ibanez engineer chap use 1uf non-polarized electrolytics
   when he designed the Tube Screamer? He had 'em on the bench,
   or some deeper purpose?

Cap choices for 1uf and beyond: So I want to use a 1uf or so value in a de-coupling application.

That works well for high impedance applications, impedance goes down, you can see much higher values used (100uF in some audio applications).

- Are the design considerations, besides cost, that would point me towards
  non-electro caps like tantalums?

Not anymore.  It used to be that the ESR was much better for tantalums, now they have made huge strides in electrolytics so they are much closer if not better than tantalums.  I have had to replace many tant's in old circuits, I don't like them.

Ceramic vs. Poly Film:

- If I can find the correct capacitance and voltage rating in a film cap, and
  cost is not an issue, can I assume that there is no good reason to
  use ceramic caps.

Ceramics can be microphonic too.  You can hear some in charge pump supplies, it is really weird.  Ceramics can give distortion to the circuit, if you want that, it can be good to use.

Tantalums: I hate to keep harping on the tubescreamer, but
- why did the Ibanez chap choose a .22 tant after the first opamp stage?

Power supply filtering: I know that a biggish electro to ground after my DC input jack will reduce ripple, which I understand to be the superimposition of residual AC which I don't want on my DC line.

- Is ripple suppression all I am accomplishing here? Or does a stompbox 
  have the need for a reservoir current? I know that some amps need
  a reservoir, but do stompboxes?

Depends on the circuit.  Lots of digital, resevoir right by the chips, analog some get ring-y without a good decoupling.  You are looking at supply impedance, too high and you can get glitches on the supply rail and all hell breaks loose.

- Should I assume that all AC-DC wall-warts that my stompbox may see
  are unfiltered affairs and that is why I need to do power supply filtering?

Yes.

- Ok, so what is going on in the following Rat power supply snippet? I understand C1—a typical higher value electro to ground.


- But what about C2—does it get rid of high-frequency noise that may be
  present on the power supply?

ESR.  If you look at a real cap, it has a series resistance.  This "equivalent series resistance" means that at higher frequencies the cap is no longer right there in the circuit, it has a large R in series, and so it is useless.   Smaller caps have better ESR, so you often see the 2 in parallel.

- And what the heck is going on with C3? Why would I need another
  electro to ground there?

Since you have 100k in series with that node, you have a high impedance.  So a glitch on that line will go to everything using that voltage reference.  The cap makes the impedance there low for AC, so that you can concider that an AC ground.

Jim Williams of Linear Tech has some good articles on decoupling, and Walter Jung has a great cap article out there.