Hollis Frobnicator revisited

[cmdrfun]

I tried to build the Frobnicator a few years ago when I was still a noob at building, and had no luck. I had even paid someone to make some etched single-sided boards for me (which were very good quality) but I couldn't even get the oscillator to work.

It's about three years later, I quit building for a while and lately needed something to do, so I decided to try and finish off all the half-done/aborted/never attempted projects I still have laying around.

Anyhow - I decided to breadboard this up and see if I couldn't get it working now. After some internet research I found that a guy in Germany  had success adding an additional LED in series with the other two to adjust the bias voltage. I tried that and sure enough, my oscillator started working. My guess is that the newer red LEDs that I have (clear package with red light) have quite a different voltage drop than the red LEDs available in 2001.

However, I was still not getting any modulation of my input signal, it was just passing through the CA3080E chip I'm using without modification. Based on my success getting the oscillator to work, I started trying some other combinations of current-regulating diodes besides the red LED + 1N914 that was in the original Hollis design. I found that for the tremolo section, two red LEDs worked and gave great results, however for the ring mod section, the bleed-through from the oscillator was terrible and the trimmer seemed to do nothing (however, it was modulating). Experimenting some more, I found that with three red LEDs plus a 1N914, I could get good ring modulation and reduce the bleed-through quite a bit with the trimmer, but not completely. (I did get it to a level where using a noise gate could nearly eliminate the bleed-through output signal, but this also attenuated the overall volume somewhat.) This made the tremolo somewhat crappy though, as it lost all the attack of the initial dry signal.  I decided one way to fix it is to use a DPDT instead of a SPDT for switching between tremolo and ring modes, and using the second pole to switch between the appropriate current-regulating diodes that worked best for each mode.

As I said, I think this is mostly due to the differences in LED voltage drops between now and 2001, but the original Hollis design also used a CA3080A while all I could find was a CA3080E, and I'm not sure if that difference had an effect. I'm also using a TL072 for the oscillator. Anyhow, I thought this might be somewhat useful for people trying to make this project work.

I just read the year-old thread from Mark Hammer where he mentioned changing R6 to a smaller value and getting better nulling results with the trimmer, I'm going to try that as well and see what happens, will add that info later on.

[cmdrfun]

I'm finding somewhere between 2.2M and 3.3M for R6 seems to allow me to get the lowest bleed-through, in agreement with what Mark found. I don't the exact value matters much.