Decoupling cap selection

Frequently asked questions on capacitor types, ratings, brands, use and abuse.
Post Reply
User avatar
boblob
Breadboard Brother
Information
Posts: 85
Joined: 27 May 2010, 20:53
Completed builds: I've built clones of these:
Si and Ge Fuzz Face
Ampeg Scrambler
Vox Repeat Percussion
Big Muff triangle
Mosrite Fuzz Rite
Maestro Super Fuzz
MXR Distortion +
Rebote Digital Delay
Shin-Ei Companion
Has thanked: 5 times
Been thanked: 10 times

Post by boblob »

Hello.

Just finished a Vorg Warp Sound (op-amp based eq. pedal) build and it made me think of this...

The Vorg has two 10k resistors forming as voltage divider between +9V and ground to provide a reference voltage, and a 33uF electro for decoupling at the junction. I didn't have any 33uFs on hand, and so subbed with a 100uF.

This got me thinking - what design criteria do you consider when selecting decoupling caps in general? I'm guessing the sub of a 33 for 100uF will work fine, but most designs have a smaller cap there. Why is that?

User avatar
Nocentelli
Tube Twister
Information
Posts: 2222
Joined: 09 Apr 2009, 07:06
Location: Leeds, UK
Has thanked: 1152 times
Been thanked: 954 times

Post by Nocentelli »

From a design point of view, decoupling caps need to be big enough capacitance to sink AC hum down to very low frequencies (e.g. 60Hz mains hum), so a large electro cap of 100uF should be plenty to remove ripple from the +9v, but not be so physically big it's a problem finding space for it on the pcb. I assume most designers reason that since vref is usually around half the voltage of supply, half the capacitance is needed to remove ripple, so 100uF across supply and 47uF across vref is common. The cap also has a role in acting as a reservoir of charge to stabilise the circuit as the demand for power fluctuates, lower values down to 10uF are common for vref.
modman wrote: Let's hope it's not a hit, because soldering up the same pedal everyday, is a sad life. It's that same ole devilish double bind again...

User avatar
boblob
Breadboard Brother
Information
Posts: 85
Joined: 27 May 2010, 20:53
Completed builds: I've built clones of these:
Si and Ge Fuzz Face
Ampeg Scrambler
Vox Repeat Percussion
Big Muff triangle
Mosrite Fuzz Rite
Maestro Super Fuzz
MXR Distortion +
Rebote Digital Delay
Shin-Ei Companion
Has thanked: 5 times
Been thanked: 10 times

Post by boblob »

Yes, I get that the 10k and cap form a low pass filter, but plugging in the numbers gives a corner freq of < 1Hz - not sure if that is going to help much with mains hum @ 50-60Hz.

Doing a bit more reading and playing with the circuit, it certainly removes some high frequency hiss, which suggests it's doing more of the stabilisation thing. I see a lot of digital guys use a smallish value electro (2.2uF) and another 100nF ceramic/film in parallel to deal with the higher freq content, but not a lot on why those values are chosen.

User avatar
Nocentelli
Tube Twister
Information
Posts: 2222
Joined: 09 Apr 2009, 07:06
Location: Leeds, UK
Has thanked: 1152 times
Been thanked: 954 times

Post by Nocentelli »

boblob wrote:Yes, I get that the 10k and cap form a low pass filter, but plugging in the numbers gives a corner freq of < 1Hz - not sure if that is going to help much with mains hum @ 50-60Hz
It's a lowpass filter so it only allows AC of lower than 1Hz to remain, 50-60Hz is above the cut-off so gets filtered out. Think of it as a maxed-out treble control that rolls off any AC signal in the vref right down to sub-audible frequencies.

I think smaller value ceramic or box caps are supposed to be better than electros at killing potential high frequency oscillations (even though the electro on it's own *should* filter them out), so the two are sometimes used in combination.
modman wrote: Let's hope it's not a hit, because soldering up the same pedal everyday, is a sad life. It's that same ole devilish double bind again...

User avatar
boblob
Breadboard Brother
Information
Posts: 85
Joined: 27 May 2010, 20:53
Completed builds: I've built clones of these:
Si and Ge Fuzz Face
Ampeg Scrambler
Vox Repeat Percussion
Big Muff triangle
Mosrite Fuzz Rite
Maestro Super Fuzz
MXR Distortion +
Rebote Digital Delay
Shin-Ei Companion
Has thanked: 5 times
Been thanked: 10 times

Post by boblob »

Duh! Yes, of course. Being dumb.

Apparently the 100nF is useful in logic circuits - stops switching transients interfering with the power supply, potentially adding noise to other parts of the circuit that share the power supply. Probably not massively useful in analog pedals, but sounds like a good idea to have built into a power supply. I have a cheap power supply with non-isolated grounds that does a marvellous job at transmitting an unpleasant digital wash from my Boss TU-3. Wonder if adding a 100nF would kill that. I know the Beavis Dehuminator has just that.

Anyhow, I think this pretty much answers my original question - the 100uF will be fine to sub for the 33uF.

User avatar
DrNomis
Old Solderhand
Information
Posts: 6807
Joined: 16 Jul 2009, 04:56
my favorite amplifier: Self-Built Valve Amp Head :)
Completed builds: Dallas Arbiter Fuzz Face,Tone Bender Professional Mk 3,Tone Bender 3-Knob,Baja BK Butler Tube Driver,Baja Real Tube Overdrive,Roger Mayer Octavia,EH Soul Preacher,Tech 21 XXL Distortion,MFOS Weird Sound Generator.
Location: Darwin,Northern Territory Australia
Has thanked: 98 times
Been thanked: 278 times

Post by DrNomis »

boblob wrote:Duh! Yes, of course. Being dumb.

Apparently the 100nF is useful in logic circuits - stops switching transients interfering with the power supply, potentially adding noise to other parts of the circuit that share the power supply. Probably not massively useful in analog pedals, but sounds like a good idea to have built into a power supply. I have a cheap power supply with non-isolated grounds that does a marvellous job at transmitting an unpleasant digital wash from my Boss TU-3. Wonder if adding a 100nF would kill that. I know the Beavis Dehuminator has just that.

Anyhow, I think this pretty much answers my original question - the 100uF will be fine to sub for the 33uF.

Providing that you have room for it on the circuit board substituting a 100uF cap for a 33uF cap will be perfectly fine, the rule of thumb to remember is that the bigger the value of cap, the better the filtering, the only caveat is that the working voltage of the cap needs to be equal or greater than the expected supply voltage..... :thumbsup


Just thought I'd add my two cents worth...... :thumbsup
Genius is not all about 99% perspiration, and 1% inspiration - sometimes the solution is staring you right in the face.-Frequencycentral.

Post Reply