A Definitive Fix for the Down Mode on the Phuncgnosis

[Paul Marossy]

First of all, a few days ago I found an apparent fix in this thread: http://www.diystompboxes.com/smfforum/index.php?topic=116769.0

This seems to be a definitive fix for getting a working down mode on the "Phuncgnosis" circuit. But the 2N3904 didn't work for me... I only had to try like 50,000 different NPN transistors before I found one that worked.  (The problem I was having is that it either didn't work at all or after 25-35 seconds of no input it would semi turn off creating a loud white noise, until I played something and it would turn on again, sometimes with a pop too)

Anyway, hopefully this helps someone else in the future.

[Mark Hammer]

Thanks for that Paul.  I will, however, point out that you show the sidechain, the biasing, and the filter section, but your drawing shows no path for the audio input to actually GET to the filter.  You may want to include that, since invariably, someone will mistakenly assume that is the complete circuit.

That said, I hope to try that out later today on my own Phuncgnosis (whose downward sweep was never really soul-satisfying).

[Mark Hammer]

Well, semi-confirmed.  Can't hear a downward sweep at all, however the filter gradually drifts back upwards, at a rate set by the decay control I added to my original build and to a starting point dictated by the Range control.  I'm wondering if the attack time is too fast to hear.

With a fresh 9v battery, I'm measuring 1.05V at the collector of the C945 I'm using.

[Mark Hammer]

I should add that the C945 has an hfe of around 260, though I doubt that matters.

I have several kinds of C945s, some that are EBC and some that are ECB. I was careful to verify which type before installation.

[Mark Hammer]

A-ha!  Listening more closely, and picking more carefully now, I hear the downward sweep.  But it is too fast if I don't pick ultra-carefully.

So I think what I'll do is add some resistance to the 50k (51k) for a longer attack time during downward sweep.  Alternatively I may stick another diode or two in series with the existing ones, which is what EHX did for the Zipper to change the Attack time.

[Paul Marossy]

Thanks for that Paul.  I will, however, point out that you show the sidechain, the biasing, and the filter section, but your drawing shows no path for the audio input to actually GET to the filter.  You may want to include that, since invariably, someone will mistakenly assume that is the complete circuit.

That said, I hope to try that out later today on my own Phuncgnosis (whose downward sweep was never really soul-satisfying).

I'm not sure why I don't get email notifications from here anymore even though I have checked the box to receive them. 

Anyway, this circuit has two inputs as originally designed. That cap you refer to was on the "standard" input. I'm using the auxiliary one with the 1M pot and not even using the other one with the .002uF cap, I left it out.

Are you saying that I should have a connection between the input to the 1M depth pot on the input and the junction of R6 and VR-2 on my schematic? I have good envelope filtering happening without that so I am confused...

[Mark Hammer]

In the drawing you show in the video, I don't see any path between the node marked "In" and the filter, other than the sidechain itself.  And clearly, after passing through the various diodes in the sidechain, there is no way to get a clean signal for filtering.  So, while I don't doubt that there may be two ways to get to the filter, whatever they are, you haven't included them in the drawing shown.  In the grand scheme of things, the drawing is still more accurate than what Maestro used to portray as "factory schematics". 

I tried monkeying with the attack characteristics of the circuit, and did not manage to get the attack time compatible with downward sweep, so I just left the circuit as is.  I'll keep at it, eventually, just not now, unless someone is diligent enough to come up with a foolproof change to the attack time.

I forgot to mention earlier that mine uses a larger-value electrolytic to ground instead of the 100nf value shown for the first cap to ground after the first series diode.  I run a pot in parallel with that cap to get variable decay, and use a 3-position toggle for "Range" instead of the pot you show in your drawing.  I don't think any of that gets in the way of the workability of the down-sweep mod you provided.

I think the mod will work, eventually.  My unit just sweeps too fast.  When sweeping up, fast attack works fine, if not for every tune/feel.  But I find that downward sweep on envelope-controlled filters needs a slower "descent" time or else it just ends up sounding like picking up a piece of paper with the vacuum cleaner ("sshhLOOPP!").

[Paul Marossy]

In the drawing you show in the video, I don't see any path between the node marked "In" and the filter, other than the sidechain itself.  And clearly, after passing through the various diodes in the sidechain, there is no way to get a clean signal for filtering.  So, while I don't doubt that there may be two ways to get to the filter, whatever they are, you haven't included them in the drawing shown.  In the grand scheme of things, the drawing is still more accurate than what Maestro used to portray as "factory schematics". 

Hmm... I may have misinterpreted the schematic when I built it 15 years ago. I guess I ought to try to adding this connection just to see what happens.

[Paul Marossy]

Well it turns out that I DO have that .002uF cap connected in the circuit. I was going off memory and pencil sketch perfboard layout from something I built 15 years ago. 

So I'll put up another video pointing out my error on that.

[Mark Hammer]

Nuthin' I haven't done myself a thousand times before. 

[Paul Marossy]

I'm still not totally happy with this down mode thing... it stays stable when the effect is on but after being in bypass mode for a few minutes it goes into that white noise mode which is audible when turning it on again. I wish I knew what the heck it's doing... seems like it's some kind of oscillation that's happening but I don't know where it's coming from. It's definitely an improvement from where I started but I want to get rid of that, period. I had a thought that maybe it was because I had a grounded circuit input bypass wiring scheme, but removing that connection to ground has no effect on it.

Perplexed. 

EDIT: I have no pull down resistor on this build. I wonder, would that be a factor maybe?

[Mark Hammer]

The resistor values are pretty huge, compared to many other builds.  If you're using a bifet op-amp, those values shouldn't generate much noise.  But if you're using a 4558 or something else non-bifet, that may be the issue.

[Paul Marossy]

The resistor values are pretty huge, compared to many other builds.  If you're using a bifet op-amp, those values shouldn't generate much noise.  But if you're using a 4558 or something else non-bifet, that may be the issue.

I originally had an LF353 in it. Now it has a TL072 in it. Same difference.
By huge resistor, do you mean that 5M resistor in the feedback loop of one of the opamps?

[Mark Hammer]

Yeah.  I'm just going by the op-amp comparison found here:  http://hammer.ampage.org/files/Device1-8.PDF  (see page 10)

[Paul Marossy]

I forgot about that opamp series in Device. I have all those printed out, I should read them again. I've been not doing much DIY the last 2-3 years, guess I'm a little rusty...