Denis, the pinhole diam that gives the best result has a factor of wavelength multiplied by focal length in it. There are a few different versions of the criteria; Ralph pointed out in this thread two particular cases, from which I quote:

So if you pick aOriginally Posted byRalphdand then collimate different wavelengths through it, the optimal FLs will be different. I.e. if you keepdconstant across the wavelengths red, green, and blue, you'd need for the product wavelength * FL to be constant. Which means that a shorter (bluer) wavelength requires a longer FL, and a longer (redder) wavelengths requires a shorter FL.

These formulas also show that if you want a larged, you need to go to a large FL.

That is my (latest) understanding of it, and I'd welcome further clarification from the experts.