"What's the deal with…" matching part values/pairs | tolerance range?

[jessenator]

So I'm not wanting to cause a riot, but I'm genuinely interested in whether there is an understood tolerance range for these parts, and if there is, in what context it's applicable.

Biggest case in point: clipping diodes. Now, I've seen there are loads of posts about types, quality, the actual value of Vf, but what I'm mainly interested is the tolerance threshold that exists for overdrives, and even distortion and fuzz circuits. Does the "match" need to be to down the milivolt? or is there some wiggle room? To begin with, a guitar's signal: is it symmetrical? If not, would there need to be a perfectly symmetrical signal clipping? Again, honest questions.

In one's own building world, what is your tolerance level? I've read many build docs that, well, come to think of it, almost every build doc I've read doesn't really touch on it; perhaps it's a known thing that a clipping pair needs its symmetry? ±1 mV? ±3 mV? ±5 mV? ±10 mV? etc. With some of these old germy boys it's quite a task to get several mV-tight matching pairs (or heaven forbid quads) out of batches of specimens. So that's what I'm curious about: where is that line for you?

[madbean]

Can't say I've ever felt the need to match Vf on any clipping diodes. I think I did it on a Klon build once. It's not a bad idea but I'm not sure that level of precision is going to be appreciated by the ear. Provided the parts in question are within their proper spec it's reasonable to assume that any random pair isn't going to sound much different than a precision matched one, IMO. Probably the better argument is for matching germanium type. Silicon, nah.

It's certainly a thing worth testing and exploring in a more rigorous setting than just "listening", though. Like, with controlled recordings and scope readings!

Transistors can be another matter. Depending on the application matching certain specs can make a big difference. Gains, leakage, Vgs for phasers, etc.

[jessenator]

Provided the parts in question are within their proper spec it's reasonable to assume that any random pair isn't going to sound much different than a precision matched one, IMO. Probably the better argument is for matching germanium type. Silicon, nah.

Gotcha. Yeah, I figured Si wouldn't be as persnickety. for Germies, I PM'd another member to ask about their process, and I've taken from that conversation is a coarse and then semi-fine batching/bagging process. Still, I'm one of those who will at least record the values I get, so they're all there. I think I'm just a tad obsessive about the whole thing.

It's certainly a thing worth testing and exploring in a more rigorous setting than just "listening", though. Like, with controlled recordings and scope readings!

I would love to go full Electrosmash on the thing. I still have a small pang when I think about pitching my old scope. It was an ancient, analogue B&K, which only went to 10-ish MHz or something, but it was a scope. Going as deep as one can (without an engineering background, anyway) is something I like doing, but I'd want to do it right.

[WonkoTheSane]

I curve trace diodes with the DCA75 when I make octave pedals, and pick 2 that are within 1%. I also do that with transistors if I'm building something with a differential pair.

For Fuzz Faces I just match them as close as possible, they sound smoother to me like that when biased up properly without any obnoxious compression like I get when they are mismatched or use that 40-50hfe spread that's been going around for decades now, tbh I think that was just nonsense to keep people from building good sounding Fuzz Faces.

[thesmokingman]

as an antiparallel pair, no need to match them up. as part of a ring mod circuit (green ringer, octave fuzz)  then matching them within 10mV makes sense. You're not likely going to notice asymmetry in clipping diodes until you over 100mV difference and even then it is harmonic content so we're down to really listening to discern a difference vs just playing and making music.

[culturejam]

I've only ever tested forward voltage drop when trying to verify unknown parts (like from surplus grab bags or whatever).  As Brian said, I don't think you will be able to hear small differences in Vf when it comes to anti-parallel clamping/clipping arrangements. In other applications, it may be more important to get them close.

[lars]

Matched output tubes. LOL.