Paralleling Capacitors to get Accurate Values

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

Guest

Guest
I am busy soldering away at the Pultec EQ and am using Wima Caps
Wima don't do 12nf and 18nf so can I put together two caps to get the right value
If I have problems with getting resistors to the correct value I put them in series (eg 200k is 2 x 100k)

I think I read somwhere that for Capacitors you put them in parallel to get the values and divide by 2

Have I got this right? I can';t seem to find the right search criteria for google to find out.

So for 12nf - I am going to parallel 10nf and a 15nf to give me 12.5nf
and for 18nf I am going to parallel 22nf and 15nf to give me 18.5nf

Or am I barking up the wrong tree entirely
 
connecing caps in parallel doubles the value...

but if you want correct values buy some more caps and measure the real value, the tolerance of most of the wima caps is 20%...
 
[quote author="matthias"]connecing caps in parallel doubles the value...[/quote]

If, and only if they're of equal value!!!

The total value of two capacitors in parallel is the sum of each of their capacitances. So 1uF||2.2uF=3.2uF, 47nF||47nF=94nF et cetera.

Peace,
Al.
 
Putting capacitors in parallel adds the values.

For example:
10uF//.47uF= ~10.5uF

I think CJ explained this somewhere before in relation to
getting desired values for EQ's, but I can't find it.

:guinness:
 
If you want to get amazed, try measuring those caps. I mean, you might be thinking you are done (w.r.t. accuracy) when you take the trouble to use a few caps together for good number-matching w.r.t. the schematic.

I was mildly shocked when I measured caps for a pair of G-Pultecs.
The exact values/freqs are of course not too important, but for a pair then why not trying to make them as equal as possible ?

So I measured and was underwhelmed by the accuracy. Most caps were
within stated tolerance, but weren't much better than that. So to find equal values required a modest pile of caps to select from, since the values were quite all over the place. So finding two alike wasn't easy, let alone close to the nominal desired value.
(Story holds for the red Wima's & older Philips tropical fish caps)

Bye,

Peter
 
So a 10nf and a 15nf would do the business then (assuming 20% tolerance - on the over side!)
 
[quote author="uk03878"]So a 10nf and a 15nf would do the business then (assuming 20% tolerance - on the over side!)[/quote]

I'm sorry, don't fully understand. When going for 25nF ? Yes, 10+15 would work. But with those tolerances chances are that a 27nF-type might have an actual value closer to 25 than 10+15... :cry:

IIRIC most actual values were under the stated value (WIMA), so if you can't measure caps then better overdo it a little bit.

(And yes, that meter was calibrated :wink: )
 
nope -
I am trying to get to a 12nf and a 18nf
Unfortunately wimas only come in stuff like
10, 15 ,22,33, 47, 68 etc..
100, 150, 220, 330, 680 etc..
You get the idea
I have every one of those up and down the ballpark (ie from around 100pf up to 1uf)
 
Clintrubber's point about initial accuracy is very well-taken.

Measure your candidate caps and mark them. Do the combinations to find the closest values. You want accuracy, not "this much or more is o.k.", unless in some non-critical cases like power supply filters a little more is better.

It's hard to make accurate caps. +/- 1% error is unusually tight, whereas with resistors it's very standard and easy to make.
 
[quote author="matthias"] if you want correct values buy some more caps and measure the real value, the tolerance of most of the wima caps is 20%...[/quote]

Actually, a lot of Wimas are available in 5% and 20%. But a lot of end user oriented retailers only stock the cheaper 20% selection.

I read somewhere - and I believe it is true - that 20% acually means you won't ever get the exact value. Whenever there tight and not so tight tolerances, chances are they selected all caps within the tight tolerance and the remaining ones are sold as the looser tolerance quality. Thus +/-20% is unlikely to contain any caps within +/-5% of their nominal values.

I recently measured a few Epcos MKT caps that were rated +/-10%. Although they go cheaper than Wimas, they were astonishingly close to their nominal values. Maybe that's because they don't sell any caps of tighter tolerances. At least I don't think they do.
 
You guys are going at it bass akwards.
Skip the math.
Build the box.
Leave everything out but the caps.
Run a pair of alligators to your cap under test and listen.
For bass, put on a CD that has some hip hop sub wolf action.
See what the absolute lowerst freq you can hear and adjust by altering cap values.
Make this your positition 1 frequency on your band selector.
Next, find the highest freq you want on the bass band knob.
Find the cap for this.
Then, you will have 3,4 maybe 5 more bands to pick caps for, depending
on how modified you are doing the original pultec.
If you have three bands left, pick the caps so that the sonic distance between each band is as identical as possible.

It's nice to have buckets of caps when doing this, but you can combine values.

Repeat the above for mid and high bands.
If you use just formulas and cap checkers, you will have an eq that might have three bands doing the same thing.
 
[quote author="CJ"]If you use just formulas and cap checkers, you will have an eq that might have three bands doing the same thing.[/quote]
I understand what you say and I'd value the 'organic' approach you des ribe as the more valuable indeed.
But once the 'magic'/suited/whatever value is determined, and we want to taste a different type of cap, we want to rule out the value as variable. Certainly don't want to make stuff by a meter that's giving the ultimate thruth; the things we have here (accuracy and what's actually the desired value) should reinforce each other.

[quote author="Rossi"]I read somewhere - and I believe it is true - that 20% acually means you won't ever get the exact value. Whenever there tight and not so tight tolerances, chances are they selected all caps within the tight tolerance and the remaining ones are sold as the looser tolerance quality. Thus +/-20% is unlikely to contain any caps within +/-5% of their nominal values. [/quote]
W.r.t. tolerance: right, that cap-measuring gave indeed the impression of a hole in the middle ! And I have indeed heard of that as well - I believe for various components it's simply a commonly used thing in manufacturing, for passives as well as for active stuff.

Bye,

Peter
 
Since too much capacitance is better than too little in most applications, most caps seem to carry a minus zero/plus 10 percent type of tolerance around.

Tube Tech is really into the parallel capacitor thing as this example concurs:

caps1.jpg
 
CJ said:
You guys are going at it bass akwards.
Skip the math.
Build the box.
Leave everything out but the caps.
Run a pair of alligators to your cap under test and listen.


Do the math, but be sure to actually measure as CJ suggests.  With the mid pultec, I'm finding it better to estimate to the low side on capacitance, as that pushes the freq's higher.  Then you can trim up the actual freq if necessary with additional paralleled capacitance.  You can't take away, if the value turns out too big.  Some study aids on one mid pultec variant start here; just the last paragraph and pic link on that first one.  These are all on the same thread page:

http://www.groupdiy.com/index.php?topic=27964.msg493508#msg493508
http://www.groupdiy.com/index.php?topic=27964.msg493611#msg493611
http://www.groupdiy.com/index.php?topic=27964.msg494342#msg494342
http://www.groupdiy.com/index.php?topic=27964.msg496474#msg496474
 

Latest posts

Back
Top