It's completely true, but, as you say, it's not the whole, very exhausting, story. I am commenting at the very basic level of 'What about tolerance?' Your input is at a second level of concern. I could add that it's not necessarily a good idea to use a wide-tolerance component, because the wide tolerance is needed because the dielectric has large temperature and voltage coefficients of capacitance. But to keep things simple, I won't add that. (;-) However, your assertion about 200 pF +/- 1% and18 pF +/- 10% is not quite true, I think: it gives extreme values of 221.8 and 214.2, whereas 218 +/- 1% gives 220.18 and 215.82.

On 2024-10-28 09:16, Tony Casey wrote:

On 27/10/2024 11:20, John Woodgate wrote:

It's not clear what you are doing about component tolerances. You can get 218 pF with a 150 pF and a 68 pF in parallel. That is just within +1% of 216 pF. A single 220 pF is within +2%. If all your four capacitors have +/-1% tolerance, your 216 pF will also have +/-1% tolerance, but there are many additional stray capacitances with this arrangement. Fewer capacitors, fewer strays and simpler construction.

What you're saying about tolerance is only half-true. Paralleling multiple capacitances doesn't change the worst cases. But it does change the standard deviation. Assuming 4 equal valued capacitors in parallel reduces σ by √4. If the nominal values of the capacitors are not equal the situation is much more complicated.

In general, it is a much better idea, when trying to obtain a non-standard value, to get as close as you can with one high tolerance capacitor and add a much smaller one in parallel that can be a much looser tolerance. You can get 218p with 200p and 18p. You only need to specify the 200p as a 1%. The 18p could be a 10% for the same effect on overall tolerance.

--
Regards,
Tony