Sanity check - this is likely to be RF causing trouble, right?

[anchovie]

Been messing around with some ideas for a distortion-plus-octave-down on the breadboard. Well, it was meant to be one idea but turned into a bit of a component-swapping frenzy. 

First try was 3 x CD4069 inverter gain stages (each low/medium gain) into half of a CD4013 set up to divide by 2. Output from the 4069 sounded great, output from the 4013 was quiet, thin, fizzy and at the original pitch. Thought I'd maybe got unlucky with the components I'd chosen to pair up, so starting doing some swapping which turned into far too much time spent on various configurations, summarised below:

3 x 4069 -> 4013: as described above
1 x 4069 (max gain) -> 4013: no change
1 x 4069 -> CD4020: no change
LM386 (max gain) -> CD4020: no change (tried the 386 in both inverting and non-inverting setups)
LF353 (non-inverting, gain 100) -> 4013: no change
LF353 (inverting comparator) -> 4013: silence

The last one was where I knew I had to look elsewhere, as I have that same combination already working in 3 other pedals! I had so far also tried 2 guitars, 2 amps, 4 batteries and a wall-wart. So I dug out a CD4093-based drone oscillator that I'd built and boxed-up a while back, swapped the guitar for that and...it drones away happily and comes out of the breadboard circuit an octave lower. 

Switched back to guitar and started poking bits of the circuit. Found that I could get a glitchy octave to come out if...
...I put a 1nf cap from guitar input to ground
...or a 470nF cap between !Q of the 4013 and ground
...or touched the insulation of the wire linking the comparator to the 4013

So my guess is that via a guitar-as-antenna situation I'm picking up some inaudible gremlins that don't affect the gain stages because I can't hear them, but make the flip-flop crap out because it can work with stuff at those frequencies. I had no trouble breadboarding CMOS in my old house but I think this is the first not-just-gain circuit I've tried since moving. Building and boxing something would confirm or disprove this, but I was hoping to do all the tweaking before committing to solder. 

Anyway, am I on the right path with my assumptions? I'd imagine that a few of you read the title and saw the word "breadboard" in the first sentence before nodding knowingly! 

[bool]

IDK about the particulars of your breadboard, and your setup.

I had a bass "synth" DIY-something in 80's, I have driven a "SN-something" synth chip that was popular back then via some cmos logic and a low-impedance input stage (iirc it was two small BJTs). IDK any of the details, but that input stage both "gated" and distorted, not dissimilar to the bazz-fuss and I had a fair amount of low pass before any of the chips.

So how about some pre-filtering and (perhaps a back-to-back silicon diode) "gating" before chippery?

[anchovie]

The original 3 x 4069 gain setup had plenty of low-pass filtering in it. Anyway, the main head-scratchers for me now are that a raw comparator-into-4013 setup that works in 3 boxed pedals doesn't work on the breadboard with any combination of 2 guitars or 2 amps, yet works with a drone oscillator.

[bool]

I'd suggest to try a separate path to feed the divider. Perhaps try a bazz fuss, filtered of course. Perhaps it will give you "gate" and "square" at the same time if you rig it right?

I never ever used anything else than a 4017 for "dividing" (it's fairly easy to rig and "just works" when all you need is a pulse), so IDK about your rig.

In another thread there was a couple of praises of a Schmitt trigger over a raw comparator (edit: SCHMITT, damn).
http://www.diystompboxes.com/smfforum/index.php?topic=108483.0

In either case, an opamp based Schmitt, a cmos or a discrete BJT version could make your rig more stable. Just a wild guess, you know ... I never ever used a breadboard, always dead-bugged my ratnests for added comfort, s-turdiness and a free groundplane thrown in ...

So good luck

[amptramp]

If you have clean waveforms like a sine wave or a sawtooth or something that is easily described, octave down generators have little problem.  If you have something that looks like this:



then the circuitry can be justifiably confused as to what frequency it is trying to divide.  If you use a comparator with no hysteresis, it will try to divide from the higher frequency content and in this case, may even go an octave up.  You can filter the high frequencies with a lowpass filter, but the circuit may have difficulty operating in the upper octaves if you filter it too much, so this is usable but limited on its own and should be used along with other techniques.  Using the neck pickup only increases the fundamental, so this may be advisable.  I could visualize a guitar with the normal output and a neck-only output.  A comparator with hysteresis switches higher on the upward swing of the signal and lower on the downward swing and will get rid of some of the local maxima and minima in the waveform.  If you have a hexaphonic pickup located farthest away from the bridge, the combination of separate lowpass filtering and a comparator with hysteresis will give the most reliable frequency division.  Of course, if you had a guitar like some of the MIDI-input devices that sensed what fret you were on for each string, that would be the most reliable way to go, but I assume you want something for a generalized guitar input.

A comparator with a value of hysteresis that varies with the amount of input signal may be interesting.

[duck_arse]

don't forget that the 4093 is a schmitt, whereas the 4069 isn't. and the schmitt won't do linear, of course.

[bool]

Back then I had a BJT schmitt "fuzz". I somehow modded the classic transistor shmit circuit into a fuzzing "something". Worked with a bass guitar. IDK any of the details anymore.

[anotherjim]

You can get some interesting variations using the 4017 for division. It's actually easier to wire up than 4013 -  no pesky R & S inputs to remember
Also, if there's room, an X-or plus 1 cap and resistor can give you double clock rate - but you do need that Schmitt trigger. Op-amp based not CMOS, you can't set the hysteresis with the 4093.

If you have 4069 inverters to spare, 2 in series with a feedback and input resistor are a Schmitt!

[~arph]

I"ve used a LM386 to slam the input of a 4093 schmitt. It works pretty good. See here.

http://www.diystompboxes.com/smfforum/index.php?topic=96658.0

[bool]

I'm sure it works, but it's wasteful current-wise (I can imagine a 386 being quite desirable wrt low part-count).

I'm also sure it can be done the way I hinted (I had a working design in 80's) but I can't remember any of the details. (damn)

[anotherjim]

Anchovie ,
My own workshop is noisy electrically - a garden shed a long way from the earthing pole. Sometimes I have to switch off the lights (they're all fluorescent of one sort or another) and even my soldering station radiates crap. The only good thing is that my projects are noticeably better once boxed up and in the studio.

But, without a scope of some sort, interference issues are hard to spot.

Breadboard can cause or reduce and hide some problems - there is quite a high capacitance between the contact strips. So high that when I'm audio probing, I plug into an adjacent strip by the terminal of interest and pickup plenty of signal !

~Arph,
Is the 386 a better front end for this kind of thing anyway? I think I get issues maybe due to op-amps momentarily latching out on over shoot when there is a lot of gain (needed to maintain signal level into the squarer throughout the decay tail). Does the 386 just gracefully clip and never latches?

[bool]

@arr-fx

Just out of curiosity - a "concept proof".

Could you test (or verify) a mod to your "two step" circuit?

1) Namely, make the input bjt-booster into a "bazz fuss" classical darlington with a red/green LED
2) just add a "two Si diodes" clipper in parallel to the 386 input

That's for starters, I'm sure that RC coupling/filtering would need some adjusting.


That would imho translate some of the concepts I used/described into more contemporary form.. namely, this mod would "squarify", limit the ampplitude - and "gate" - right at the very input with minimal part count. So, if it works, hey presto...

The next step would be to eliminate the need for a 4093 with some of the commonly used opamps..

[~arph]

I might try it out, but I have none of it on breadboard or in an actual pedal. The two step front end is not properly designed at all I just needed a bit of filtering and lots of gain. I have no problems with it latching up and the sustain is decent. It worked so I kept it.
I think the front end benefits from filtering the harmonics before squaring it up. The bass fuzz suggestion also squares up all harmonics. I don't want to replace the 4093. I like the simplicity it gives me. Schmitt trigger action without extra resistors. Low parts count oscillator etc. current draw is not an issue for me. I don't build battery powered pedals. What would be helpful is a simple compressor in front or a simple Automatic Level Control (with half a NE570) to keep the signal as high as possible for as long as possible.

[anchovie]

Anchovie ,
My own workshop is noisy electrically - a garden shed a long way from the earthing pole. Sometimes I have to switch off the lights (they're all fluorescent of one sort or another) and even my soldering station radiates crap. The only good thing is that my projects are noticeably better once boxed up and in the studio.

But, without a scope of some sort, interference issues are hard to spot.

Breadboard can cause or reduce and hide some problems - there is quite a high capacitance between the contact strips. So high that when I'm audio probing, I plug into an adjacent strip by the terminal of interest and pickup plenty of signal !

Thank you! I was worried this thread had derailed into techniques for improving divider tracking.
I've experienced crosstalk between broadboard rows before, so thanks for reminding me that I may have signals travelling where I don't want them to!

Guys - the fact of the matter is that a maximum-swing signal into a divider will produce a string of rail-to-rail pulses whether it tracks properly or not. I have built pedals that do this successfully. In this current mystery, I am getting a low-volume, thin, undivided signal out of the divider from more than one chip combination.

I think I'll rig up a lead and use one of those soundcard-as-oscilloscope programs to verify the theory that something ultrasonic is messing it all up.

[~arph]

Have you checked other positions on the breadboard?
I have had some weird issues too with an old breadboard of mine as some strips were either inter connected or failed to connect with components inserted. No more issues since it has been replaced.

[anotherjim]

Anchovie,
Another thing you could try with the 4069 pre-amp is separate the inverters - use 3 corner ones and ground the inputs of the unused middle ones. Oh, and do you have plenty of supply decoupling capacitance on the breadboard power rails?

[anchovie]

Have you checked other positions on the breadboard?
I have had some weird issues too with an old breadboard of mine as some strips were either inter connected or failed to connect with components inserted. No more issues since it has been replaced.

I've got four columns for DIP ICs and I've tried all of them, and also the two chips on different columns. Might move the whole breadboard to the garage and see if being behind a metal door helps!

[anchovie]

Anchovie,
Another thing you could try with the 4069 pre-amp is separate the inverters - use 3 corner ones and ground the inputs of the unused middle ones. Oh, and do you have plenty of supply decoupling capacitance on the breadboard power rails?

I've been all around the perimeter of the 4069 and, as mentioned earlier, also tried LM386 and LF353. Breadboard has currently got 220uF across the rails, but I've also tried 1000uF and put 100nF across the power pins of each chip. Might need to build my own personal Faraday cage 

[bool]

how about simply placing a sheet of tin foil beneath your breadboard and connecting it to your ground?

[duck_arse]

has anyone here tried the TL431 as replacement for the 386 in a this type circuit? only 3 pins!