GGG univibe has a very hard "knee"?

[MikeH]

Not really sure how else to describe it... The turnaround at the top of the sweep goes really fast, and it creates an almost tremolo-like pulsing.  It's more obvious when in vibe mode and at faster rates.

Built to GGG specs, all 3904 transistors. Photocells and bulb are the ones that smallbear sells. Light shield is made from sheet metal and photocells are mounted flat on the board (although it sounds the same with the light shield on or off if the box is closed up). I did have to make a slightly taller light shield because the bulb from smallbear was too tall for the shield template in the build doc. Maybe the photocells should be mounted so that they face right up to the bulb?

It's hard to describe but I do have a sound sample recorded. Is there anyway to post it here?

Thanks!

[Digital Larry]

I'm not that familiar with the GGG univibe but I trust that it uses some form of phase shift stages based on some form of variable gain element driven by some form of LFO.  A fast turnaround at one end of the LFO sweep is an indication that "something" in this chain is changing too rapidly for a pleasing audible result.  I personally do not know what the desired objective goal is here - but with VCFs something to shoot for is "equal octaves for equal sweep voltage" which is an exponential relationship.  On a phase shifter/Univibe presumably you can look at the notch positions in the spectrum based on the gain control voltage (LFO waveform).

I made a Paia Hyperflange many years ago (kinda wish I still had that) where I first learned about the concept of warping LFO waveforms to get a perceptibly smoother sweep.  In the DSP world I often inhabit, warping LFO waveforms is fairly easy because you can multiply an LFO by itself (one or more times) to squish it towards asymmetry and then try to dial that in by ear.  When dealing with a stock analog circuit that gets a little more difficult.

If you have a scope I'd start by checking the shape and amplitude of the LFO.  Then, disconnect the LFO from the gain element and substitute a linear pot connected in such a way that its output voltage range matches your LFO.  Move that back and forth by hand to try to get a feel for what the range and "feel" of the control curve is.  It may be that the best thing to do is simply to limit the sweep amplitude (perhaps with a resistor divider) so it doesn't go that far and then you simply avoid the whoop-de-doo at that end of the LFO sweep.  Or you can put a low pass filter on the LFO output so its amplitude goes down the faster it goes.

All of this diatribe betrays no specific knowledge about the GGG circuit in question, and it's possible that what's really needed is to add/remove a special component in a special place to solve a well known issue.

[PRR]

It's an incandescent(?) lamp onto LDRs. I sure would play with LDR aiming and lamp bias.

[idy]

It's a mostly stock Univibe. Did you play with the lamp bias trimmer?

[Digital Larry]

It's an incandescent(?) lamp onto LDRs. I sure would play with LDR aiming and lamp bias.

It's a mostly stock Univibe. Did you play with the lamp bias trimmer?

That's exactly what I would have suggested had I had any idea what I was talking about.   

[idy]

yes, generally the LDRs are mounted to face the bulb, not "stare at the ceiling." Some make the light shield mirror-shiny inside.

[blackieNYC]

http://www.runoffgroove.com/tri-vibe.html
In this article for a different vibe circuit, ROG goes out of their way to make the waveform an imperfect pseudo-sine wave.  The only explanation offered is the avoidance of "seasickness".  I may have picked up elsewhere that the univibe has a slightly asymmetrical LFO. Maybe Geofex. Anyhow, I'm suggesting it might be normal. If you can post a clip, there are folks who could tell you in an instant.
Or half your LFO could be clipping perhaps?