Capacitor tolerance important?

[dustbro]

Can someone enlighten me on the importance of electrolytic capacitor tolerances? What would be the sonic impact on some of these DIY projects if a cap that had a 20% tolerance was used over a cap that had a 5% tolerance?
thanks for any info.

 

[Svart]

For audio coupling, tolerance isn't that important as long as you are sufficiently above the point where circuit impedence and capacitance form a high pass filter. If you were right on the edge, then a low tolerance might push you below that point and the circuit could affect low frequencies somewhat, but that is rare these days, at least in my experience.

Now for oscillation/clock circuits and things of that nature tolerance is much more important and the highest tolerance and quality tempco are important. Not a lot of that in analog audio though.

 

[Samuel Groner]

In about any reasonable application electrolytic capacitors simply need to be "large enough" (this is different from e.g. film capacitors which often need to have a very specific value, e.g. for an EQ to get the right frequency). Hence good designs usually overspecify the value. This means that better tolerance than 20% will not improve anything.

Samuel

 

[Svart]

my thoughts exactly, Samuel said it much better than I did.

 

[SSLtech]

[quote author="Svart"]Now for oscillation/clock circuits and things of that nature tolerance is much more important and the highest tolerance and quality tempco are important. Not a lot of that in analog audio though.[/quote]
I dunno...

Most Parametric Equalisers are just down-tuned oscillators really: look at Jakob's Calrec version. In these cases, capacitor tolerance error translated directly to center frequency error. -If you can tolerate a stereo EQ pointing to 1kHz, with one channel actually centered at 1200Hz and the other channel centered at 800Hz, then ±20% will suit you fine.

Me, I prefer a bit more accuracy, so for all such circuits I buy closer tolerance, low thermal drift caps, and then hand-select.

Keith

 

[ruairioflaherty]

Keith,

do you find you have to go through many caps to get matched sets? I'm actually hoping to build a Calrec eq with rotary switches and have been wondering about this,

thanks,
Ruairi

 

[Svart]

Keef, yes of course, I did mean that high tolerance caps should be used in gyrators and other types of tuning/oscillation circuits. I guess I should have said so, I didn't make that clear.

 

[SSLtech]

Yeah.. for anything tuned, the type of cap requirement that Svart mentioned is important.

I usually find that hand-selecting I get a 50% rejection rate. I'll surely them somewhere, as Make-up values (If I find some 10nF caps that I can't use because they're 10.5nF, that won't stop me using them to make 220nF if I measure a cap at 210nF... basically 0.5nF is a significant error as part of 10nF, but insignificant in relation to 220nF.

Basically, for most of my designs, I make all PCB layouts for timing-or-tuning-critical locations "parallel-available"

-Hope that makes sense...

Of course, my local component supplier has open bins, so taking the cap meter down to the store makes the number of 'spares' that I buy much smaller!

Keith

 

[dustbro]

Thanks for the replys guys! So I can assume from the above posts that a proper rated Electrolytic Cap with a 20% tolerance will do fine in my Gssl.. right?
thanks again!

 

[Svart]

absolutely.