Weird thing in a Klon clone: clipping diodes, sound "choking out"

[KarenColumbo]

Just made a Klon clone for a friend, based on the "Refractor", added a bass cap switch, a true/buffered bypass switch and a switch between antiparallel germanium/LED-4148 antiparallel switch. This would be a small red 2mm (?) LED with Vf 1.77V and the classic 4148 with, dunno, 0.7V.

When I crank up the "Gain" and strike a power chord in the LED/4148 setting the sound has an audible delay (some 100 millisecond or more) before setting in. I hear a very short attack, then the delay (silence), then the sound. It doesn't do that in the "normal" setting, i.e. Germanium antiparallel. Sounds like a tube "choking out".

What can that be? I'm flummoxed. Did I align them the wrong way? Is one of them buggered?

[antonis]

Andreas, plz post a schematic with cap/resistor values (especially on NFB/Gain circuit) 'cause it seems to me like a instantly big current demand "somewhere"..

[KarenColumbo]

Hey, Antonis 
The schematic is in this PDF, page 3: http://www.mrscolumbo.com/uploads/aion-refractor-centaur-documentation-v2.pdf

The only additions are a toggle switch that alternates the C7 value between stock 82 nF and 150nF and said toggle between a Ge diode pair and a LED/4148 pair. And the suggested 3PDT toggle between true/buffered bypass. The choke effect happens on both bypass settings and both cap values.

Apart from the weird choke effect it sounds perfectly normal. I built another earlier this year without any mods, an A/B comparison shows almost identical sound and output.

I will desolder LED/4148 pair and put in another 4148 instead of the LED, I think, and post my findings tomorrow (in about 16 hours from now).

Very strange, this delay.

Seems another member of this forum had the same problem: https://www.diystompboxes.com/smfforum/index.php?topic=116322.0
A thing called "reverse gating" - but I can't for the love of it find out if the problem had been solved in the thread.

[dschwartz]

Are you 100% sure they are in opposite direction?
I had mistakenly put clipping diodes in the same direction and the result is a choking effect, specially on lower strings.
I actually liked the effect and called it "broken" mode on my Drive maker design.

[Rob Strand]

Assuming there's no fault, I think the problem is caused by the asymmetrical diode setup.   You might need to add a resistor across the diodes (at least in that mode).  Try say a resistor value in the zone 10k to 100k.
I was pondering and the more I think about that circuit the less likely the problem is around the diodes.

Then I noticed this key info:

The choke effect happens on both bypass settings and both cap values.

How can it be happening on bypass!
How are you doing the bypass exactly?

EDIT:

One area which looks a bit dodgy is the presence of R7 and R19.   When IC2A is not clipping everything is probably OK.  But if IC2A clips the point at IC2A pin 2 is not longer pinned to Vb by feedback.  This opens up the potential for positive feedback from the output of IC1B back to the input of IC1B.   IC1B has can be set to a high gain so small input changes due to signals coming back down the R7 + R19 path might cause some positive feedback.  C3 is not large enough to block the signals.  C9 and C10 provides AC coupling which prevents a permanent lock-out/ lock-up situation due to DC positive feedback.

[EDITED]
My first guess at going down this path would be to lift one end of R7 and R19 to see if it fixes the problem.  If it does or doesn't then that's a starting point to work out a fix.  Another experiment  along this part would be to a back-to-back LED clipper across R20 to stop it clipping.  Another experiment would be to add say a 47k to 100k resistor from IC2A pin 2 to VB with the aim of creating a divider with R16 which activates only when IC2A clips.   The circuit already has a divider path from R16 via R17 but one end of R17 goes to the gain pot and that makes the divider less effective when the gain pot isn't at the extreme settings.

The root cause is the high DC gain of IC1B.  To reduce that we'd need to add a cap between R10 and the gain pot.

[dschwartz]

Sorry double post

[willienillie]

How can it be happening on bypass!

He has a switch for true bypass or buffered bypass, like the KTR.

Daniel might be on to something.  On the Refractor board I built, it had the diodes oriented the same physical direction.  Though connected correctly (anti-parallel), it still looks wrong, and I can see somebody flipping one of them.  No idea if that would cause the symptoms described, but it certainly wouldn't sound right.

[Rob Strand]

He has a switch for true bypass or buffered bypass, like the KTR.

Do you have a link to a schematic at all?

Daniel might be on to something. 

Very well could be.  I guess I assuming there's an intrinsic problem since some people see it an some people don't.   It's like the circuit has a problem and it just needs a push one way or the other to make the problem show up or disappear.  [PS: I added stuff to my previous post while you typed.]

[willienillie]

He has a switch for true bypass or buffered bypass, like the KTR.

Do you have a link to a schematic at all?

The Aion Refractor schematic is in the .pdf linked above, but doesn't include the bypass option switch.  He said he added it though.  FWIW, I wired my Refractor as true bypass only.  Also FWIW, I sold it to my bandmate yesterday, sounds better with his Les Paul than with my Telecaster anyway.  It never exhibted any weird behavior at high gain settings.

[KarenColumbo]

Yeah, the true/buffered bypass switch is not included in the schematic. I tried to track down the three "insertion points" where the 3PDT switch would be connected, to no avail so far, haven't had the time.
What I meant is: It's of no consequence so far - the "delay choke" happens only when the clipping is set to the LED/4148 position and the gain is fairly cranked up (90% or so).

I have a pending visit to a dermatologist, so I'll be back in about 4 hours' time to saw off the LED/4148 and put in a pair of 4148s instead and try it again.

I'm - as usual - overwhelmed by your helpful suggestions. I need to put out this build todaym though, so if the replacement diodes work as intended, I'll ship it out without questioning the integrity of the build. My mate who "ordered" this thing for his birthday won't crank it all the way up anyway, he's using it as a kinda "dirty boost".

But I just ordered another kit from Aion/Musikding.de 5 minutes ago and I will try all the suggestions - so to say, to build a fail-safe version of this. There's a lot to learn for me here!

Would be a bummer to have wasted your time if it's only a wrongly oriented diode in my case ...

[Rob Strand]

I had a quick look at the response of the circuit from the output of IC1A
to IC2A pin 2 *when IC2A is clipping* (not really clipping but disconnected as if it's not there).

The results are very suspicious.

So here's the response for various gain pot settings.


(To me) The big peaks are an indication of a positive feedback problem.   The resonant frequency is very low, around 5Hz, so it's not as if you will hear an oscillation.  However what you might hear is the sound being blocked while the oscillation clips over a half cycle then after that the oscillation might dampen down.

For kicks I added a 10k resistor from IC2A pin 2 to ground and the response shows a much tamed peak -  perhaps not solving the problem entirely.



I tried a few other fixes and it's not so easy to fix without changing the circuit in a way which *might* affect something.  I'd need to look closely at the details of what is affected by the fix.

Anyway from the response I think the positive feedback theory is showing a potential issue.  Things like cap tolerance and the ability to overload IC2A at various gain settings may actually be variables that make this problem occur or not.

[KarenColumbo]

So ... interestingly, it's the Ge-diodes that exhibit that "choking" behaviour :/
I wired them like this, directly on the toggle switch to save space. The switch is soldered in instead of D1 in the Refractor schematic, K-side of D1 (GND side in the schematic) being the "output" of the switch.



Again: Silicium-side (SI1, SI2 and SI3 in this schem) takes full throttle without a glitch, GE-side choking out from about 90% of the pot rotation onwards.

[Rob Strand]

So ... interestingly, it's the Ge-diodes that exhibit that "choking" behaviour :/

Interesting indeed.  I don't really understand it ... whaaa 

Maybe the 1k clip resistor is driving the germanium so hard the output is going *higher* than silicons?

So one idea is to leave the silicons in circuit and switch the germaniums in and out in parallel.

Is there any setting gain where it doesn't block?

[FlyingWild]

The schematic for the Refractor version that has the buffer/true bypass switch has been available on Aion's site for a while:

https://www.dropbox.com/s/aiyxb45ccgsncdb/refractor_documentation.pdf?dl=1

[KarenColumbo]

So ... interestingly, it's the Ge-diodes that exhibit that "choking" behaviour :/

Interesting indeed.  I don't really understand it ... whaaa 

Maybe the 1k clip resistor is driving the germanium so hard the output is going *higher* than silicons?

So one idea is to leave the silicons in circuit and switch the germaniums in and out in parallel.

Is there any setting gain where it doesn't block?

There is: Germanium diodes setting, Gain below 85%.
Asked friend: he said he wont misuse the thing as a hard rock distortion unit anyway, so it's okay. but I ordered another PCB anyway and will give it the DIYSB treatment with all the failsafes just to see if there's some possible leak in the system

[Rob Strand]

There is: Germanium diodes setting, Gain below 85%.

OK so low gain saves it.

So high gain would lend itself to IC2 clipping and increase the the low frequency gain of IC1.
It doesn't explain why the silicon diodes are OK.

Asked friend: he said he wont misuse the thing as a hard rock distortion unit anyway, so it's okay. but I ordered another PCB anyway and will give it the DIYSB treatment with all the failsafes just to see if there's some possible leak in the system

Yes, you might get away with it.  Others have worked around the same issue.

What about the buffered case,  is that blocking as well?

The schematic for the Refractor version that has the buffer/true bypass switch has been available on Aion's site for a while:

Thanks for the link, so it's just one opamp buffering the signal.  It would be weird for the buffer to be affected by rest of the circuit, something nasty would have to be happening.

[willienillie]

Maybe one of the Ge diodes is bad?  Destroyed by soldering heat or something?

[pinkjimiphoton]

sounds to me like a diode backwards, or shorted. or maybe open. i dunno. the description of the tone sounds familiar.
leds are weird sometimes. they can blow out, and still somehow pass signal sometimes.


you got this bro... its gonna be something stupid!

[KarenColumbo]

Yeah well ... I soldered in new Ge-diodes and, whoda thunk, it works as intended. Weird stuff going on when you blow one in a pair of diodes, it seems.
Thanks @ all!

[Rob Strand]

Yeah well ... I soldered in new Ge-diodes and, whoda thunk, it works as intended. Weird stuff going on when you blow one in a pair of diodes, it seems.
Thanks @ all!

Amazing.  I wonder if everyone that has that problem has damaged the Ge diodes!