Yesterday I built the Mimosa kit (sans decorating the box, which is not my long suit) and I believe it is working. However it's quite subtle to my ears (the only other compressors I have used are optical types, which is perhaps the explanation?). In getting it going, however, I stumbled a bit on the "adjust the trimpot to put Q2's source at 1.5 - 1.7 V". When that failed to produce a working compressor I read through 4 or 5 years' worth of Mimosa troubleshooting threads here (though not the Mimosa Jr ones I will confess). I also found some useful discussions from years past on the diystompboxes forum. I found that others have discovered the same sorts of behavior I found.
With Q2(S) at 1.6 or 1.7 V and Blend knob at 100% wet, there was barely any signal coming through. (Others have written "the Blend control is more like a reverse volume control".) Going instead on "adjust it by ear" I found in half a dozen blind trials that tweaking that knob CW past 1.7 V and toward higher Q2(S) just until the overall level was more or less constant throughout the range of the Blend control, the Q2(S) value was always within 0.10 V of 2.80 V. That is also the condition under which the compression effect, though subtle, is most noticeable. Turning the trimpot further CW (--> higher Q2(S) ) increases somewhat the loudness of the wet compared to the dry, and to my ears makes the "effect" become unnoticeable.
(Along the way, I think I was able to confirm what I saw in one of the threads here, that in general the effect-switched-in volume through the pedal matches that of bypassed when the Level control is at about the 2 o'clock position w/r to the usual 7am-to-5pm interpretation.)
I know even less about FETs than about BJTs. But one thing I think I know is that the population of a given type (2N5457 say) may actually show a wide range of some of its parameters when observed in the field. Does it seem reasonable that my particular specimens of 2N5457 include outliers in this way? In "the literature" (BYOC, DIYstompboxes online discussions) there have been plenty of opportunities for others to say that their bias voltage wasn't in the 1.5 - 1.7 V range, and no-one says so, but if a lot of people are setting it "by ear" then they may not know. I don't understand the FET-as-voltage-controlled resistor schemes well enough to know how a spec sheet "typical Vgs" offering (or knowledge of a particular specimen's Vgs(off) ) informs where the sweet spot of operation is. So unless someone recommends otherwise I am not inclined to start rummaging through some alternate 2N5457's--maybe this behavior is normal and there isn't really anything wrong.
Another characteristic I found, mentioned more over at DIYstompboxes than here, is that it's a painfully narrow range of the trimpot setting that runs you through the sweet spot. I could imagine putting a lower-R-value trimpot in series with the 10k trimpot--but don't worry, these are just thought experiments. I will save the surgery for other projects.
|