<div dir="ltr">I've sat down and matched 1000+ caps of the same value (in factory packaging, so I'm assuming the same batch) more times than I'd like to remember. Alway, I always got picture perfect gaussian distribution but its has always been skewed off centre from the nominal value.<div><br></div><div>My assumption was that inside a single batch the tolerance is typically quite tight, and the overall 5% / 10% / etc tolerance is to accommodate batch to batch variations.</div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Fri, 9 Apr 2021 at 19:04, Florian Teply <<a href="mailto:usenet@teply.info">usenet@teply.info</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Am Thu, 8 Apr 2021 23:45:42 -0700<br>
schrieb David G Dixon via Synth-diy <<a href="mailto:synth-diy@synth-diy.org" target="_blank">synth-diy@synth-diy.org</a>>:<br>
<br>
> Hey Team,<br>
> <br>
Hey Dave,<br>
<br>
> So, I've now got a ton of orders for my little two-Dome-filter<br>
> frequency shifter, which I've figured out how to build in euro format<br>
> (12HP, if you're curious). I hand select all the film capacitors for<br>
> the Dome filters so that they all have the same (or nearly so)<br>
> mantissas -- I try to get them either all within 0.2%, or 0.4% at<br>
> worst. My phase displacement simulation tells me that this gives<br>
> very nearly perfect results. <br>
> Anyway, I ran low on capacitors to the point where I couldn't put<br>
> together decent collections for the 18-stage filter (I need 4 102s, 4<br>
> 103s, 4 104s, and 6 105s for it). I bought 250 each of the smaller<br>
> values and 200 of the 105s. Tonight I measured all of the 105s,<br>
> thinking that they would ideally form a Gaussian distribution around<br>
> 1.000uF. Well, they were nearly all clustered between about 1.030uF<br>
> and 1.050uF, and in fact, about 20% of them were off spec (higher<br>
> than 1.050uF, and these are 5% caps). Not a single one of them was<br>
> less than 1.020uF. They are Kemet caps. So now I've got to hope<br>
> that the other values also tend on the high side of their tolerance<br>
> ranges, or else I'm going to have to buy even more caps. <br>
> I think I've given myself a fool's errand here.<br>
> <br>
You're not the first one to observe the rather skewed distribution of<br>
tolerance in passives: Even though manufacturing yields aproximately a<br>
gaussian distribution of values, what ends up in distribution pretty<br>
commonly has the central part of the distribution missing. The<br>
reasoning behind is pretty simple: those parts, that qualify for a<br>
tighter tolerance are sold as such, as they can charge more for tighter<br>
tolernce groups. And you also get economies of scale by not running<br>
separate lines or manufacturing batches for different tolerance groups.<br>
On the other hand, if you're buying larger quantities (quantities way<br>
beyond what one would need for DIY, that is), you sometimes end up with<br>
a batch of tighter tolerance parts thrown in depending on what they had<br>
in stock.<br>
<br>
So, when one needs very tight matching between parts, there are (at<br>
least) two solutions: a) buy excess quantity and select parts (excess<br>
here meaning usually upwards of three times the number you want to end<br>
up with), or b) divide critical parts into parallel and/or series<br>
combinations of at least two devices. This could be either an<br>
approximate 50/50 split or 90-95 % plus one for fine-tuning. For a<br>
50/50 split you'd need either some luck with your purchase (as you<br>
mention nearly all your parts are on the upper end of the tolerance<br>
range) or again sufficient excess, while for the fine-tune approach<br>
(95/5 split) you'd need a decent stock of smaller value parts to meet<br>
the final value. Matching to specific target values becomes easier when<br>
you have more devices to work with (either in form of stock or in form<br>
of several devices connected together to form the target value), but of<br>
course you quickly reach a point of diminishing returns as sorting and<br>
selection is both money- and time-intensive such that in most cases<br>
it's moe economical to buy tighter tolerance parts in the first place.<br>
<br>
Doesn't help much in the situation you're facing right now, but worth<br>
considering in the future nonetheless...<br>
<br>
Best,<br>
Florian<br>
_______________________________________________<br>
Synth-diy mailing list<br>
<a href="mailto:Synth-diy@synth-diy.org" target="_blank">Synth-diy@synth-diy.org</a><br>
<a href="http://synth-diy.org/mailman/listinfo/synth-diy" rel="noreferrer" target="_blank">http://synth-diy.org/mailman/listinfo/synth-diy</a><br>
Selling or trading? Use <a href="mailto:marketplace@synth-diy.org" target="_blank">marketplace@synth-diy.org</a><br>
</blockquote></div>