Input/Output Capacitor Size?

[blip]

Sorry for what is probably a stupid question, but what factors go into determining the size of an input or output capacitor?

As far as I can tell it is a combination of the expected DC offset and the impedence of the next piece of equipment (headphone etc.) up the chain... But if someone could flesh that relationship out for me (or tell me that I've totally misunderstood what's going on) that would be great.

 

[blip]

Never mind... I figured I think I figured it out. It's not the expected DC is it... its the interaction with impedance which determines the high-pass. And if you are using an output capacitor you just plug in the input impedance of the next piece of equipment. Duh! I'm a bit braindead tonight.

 

[dsavitsk]

Keep in mind that in general, after the output cap you need a resistor to ground (usually in the 47K to 100K range works fine.) The R you need to consider for sizing the cap is this resistor in parallel with the input resistance of the next stage. Also, the amound of DC offset does matter w/r/t/ the voltage rating of the cap. If you are building a single ended tube amp with high B+, you need a cap that can block hundreds of volts.

The formula you need is here: http://www.ecp.cc/cap-notes.html

 

[blip]

Yeah, I was thinking solid state... I shouldn't be so narrow minded.

Oh okay... That makes sense about paralleling.

So if the resistor to ground is 100k and the impedance of the next component is 300 ohms then the R you would plug into the formula would be around 299.

Anyway, thanks for the weblink... I've got to check out those Aerovox caps!

 

[dsavitsk]

I shouldn't be so narrow either -- you are talking about for a headphone amp and I was thinking preamp. Yes, the output resistor will have virtually no effect. It really only matters if it is of the same order of magnitude as the next stage, otherwise the effect is minimal and you can safely ignore it. Caps are generally+/-20%, so being off on the calculation a bit is fine. Since you are powering phones, you can drop the resistor way lower too -- something like 1K is fine -- really just about anything is fine.

And, if you are powering phones, most of the caps on that page won't work as the are all way too small -- the ASC's might be okay, but they are really big physically. You can see 4 of them sticking out the top of my preamp here: http://www.ecp.cc/low-mu.html. Solen also makes some 100uF film caps which are also quite big. For 300ohm phones, 100uF is standard and will result in a 3db point of around 5Hz. You might move up to a 220uF, but that is plenty. For Grado's (32ohm) people usually use 470uF.

Sadly, the goldmine has raised the price on the Aerovox caps to $5. Probably still worth playing with them, but not with as much giddiness.

 

[blip]

I wasn't thinking about headphones actually... Just using known values to make it simpler for me to understand.

What I was thinking about is my Tjoeb 99 which used to produce some offset... An opamp swap brought it under control... The whole process got me thinking about how I would have calculated the value of new caps if they had been nescessary. The math of this hobby isn't my strong suit... but I'm trying to get a better hold on some of it.

 

[Tomo]

Hi,

You tell me you are not good with math ... So let's reduce the problem with coupling caps to something easier to understand.

First, input coupling cap or output coupling cap do not really matter. They are all the same; coupling caps. You know this already, but they isolate your (headphone?) amp from source/preamp. (Or, some of us have multiple stages in our amps and they prefer to isolate each stage with coupling caps.)

Look at the schematics for Szekeres amp.

Let's forget there is Rp, the POT. You see that there is C1. This capacitor forms High Pass Filter with R2 and R3. (kinda weird I know but stay with me) Look at the output. You see another capacitor C2. This forms another High Pass Filter with R5.

Now take a breathe and compare the input and the output. They are ... identical. In all cases of any coupling caps (except for tone control), High Pass Filter is involved.

The only equation you need to know is this.

f0 = 1 / (2Pi C R)

f0 is the lower corner frequency. C is the coupling cap and R is the resistor. You have to be careful about determining R. If you connect your headphones (32~300 ohms), you may end up with LOWER R. Look at Szekeres amp schematics. You see C2 and R5. R5 is 4.7kohms. But! if you connect your headphones, your headphones is connect PARALLEL to R5. The equation for paralleled resistor is:

Rtotal = 1 / (1/r1 + 1/r2)

All funky. Don't need to remember it now because 4.7kohm is soo darn big compared to your headphones. (32-300ohms) In such case, the Rtotal equals the headphone impedance. (almost anyway)

This is precisely why C1 is quite big if you have done your numerics with 4.7kohms.

Ciao!

Tomo

P.S. Don't forget that NASTY 2Pi thingie.