Matching “NSL-32SR3” opto-Couplers

good times today – spent doing measurements and writing spec numbers to a set of Silonex “NSL32-SR3” opto-couplers

I’m plonking down my notes so you can see what the selection process might look like … out of the 61 there are only 3 that I felt fell outside acceptable limits … a 4’th that was a dud, and a 5’th that was simply way off …

that 4’th one was obviously hooped, measuring around 8Meg ohm at 2mA (!!!) and the 5’th was “far off” measuring 6Meg ohm at 10uA of LED current … strangely even though the “on” resistance value was 151.8 ohms on the 5’th cell, which is excellent, still it can’t be considered for matching as I know ahead of time what values are reasonable for the two types of cells I use … to be clear, I’m not saying this 5’th cell is useless altogether (it can still be used for other purposes)

I also felt that the ones measuring 133k, 175k, and 232k at 10uA where too high (just on the outside of the range I like to see now that I’m used to it) … to be honest, the later two might be ok as a “rough” pair now that I think about it … (dang, just for fun I’m gonna try those in one of my optoVibe boards to see how the circuit reacts)

anyway …

the idea is to maximize the return on dollar by yielding the greatest numbers of matched pairs … we want two things here: matched pairs that are either very close in value // as used in Vibe LFO’s and other opto-controlled circuits, eg. Mu3 … or matched pairs in close proximity that can be combined to form reasonably close-valued quads // as used in four-stage phasors (otherwise, carefully selected shifter-cap values no longer mean too much) …




I’m actually pretty lucky on this day of testing … the majority of the cells have their “Ron” values (@2mA) lying between 100 and 240 ohms … and, most of the cells have their “mid-Roff” values (@10uA) lying between 15k and 90k … very typical values for these opto-couplers

despite relatively tights specs, the data plot (below) shows how randomly spread out the values can be (as with anything else) …

now, several approaches can be taken in choosing proximity pairs, either in a Pythagorean “circle-distance” way (ie., SQRT((X1-x2)^2+(y1-y2)^2)), or a by using a “square-distance” approach … but still,

I prefer to match pairs based on a slightly stronger “weighting” of the 10uA value

that’s why, as you’ll see below in the GIF animation, this generates wider/shorter data-pair rectangles // which are of good proximity in my mind … the idea is to make sure cells don’t start diverging too much as they head towards darkness … >> note that this is somewhat close enough to the dark zone but still somewhat bright for these guys // especially considering that 1uA still produces enough light to give about ~5 Meg of cell resistance …

as a result, this also opens up alternate options when considering other quad formations // see points a, b, c, d, and e … depending on where I want to grab my quads from … obviously, other possibilities exist beyond the ones I show here … the x-axis represents “Ron” values (in ohms) and the y-axis represents “mid-Roff” values (in k ohms) …

(click image to activate GIF animation)


notice that in some cases the difference in “Ron” values is a mere 40 to 60 ohms // so, laterally speaking, this is not much of a spread … that is why, at the beginning, some cells are discarded in order to provide this “tight” range … now, in a logarithmic sense (the natural scale that gain and filter circuits respond to) this is very tight indeed and many good pairs can be formed here … in other words, to a circuit that responds to logarithmic differences, this differences manifests itself more at higher values // hence the need to slightly favor matching at those higher numbers, and hence the rectangles …

in fact, even in the worse cases (ie., “very-wide” rectangled pairs) this is still VERY good matching when we consider that the cells will be made to climb into the hundreds of K’s and even Meg-ohms when heading towards darkness … ie., the spread of 60~80 ohms down to 100 ohms is nothing compared to what would happen if such a spread happend at higher values … yeah, we want the cells to taper off towards infinity in a similar way … this gives us a very side range of ‘smooth” operation to work with // what gives my optoVibe circuit the extended range not possible in a bulb-based unit for example …

finally, notice there are two pairs that are almost on top of each other … I might choose to reserve those for Vibe LFO circuits, not that looser pairs won’t work there // … better yet, I think I’ll reserve those for my voltage controlled opto-pot circuits

360+ Pro Fuzz/Preamp


360+ DIY Kit


Mu3 DIY Kit


IFMTA (book)




Champ (PDF)


Univibe+ PCB