MXR Blue Box question

[SonicVI]

I put in the switch and I think the two octaves sounds better than the single octave in this pedal. Of course you have to play high up on the neck for it to sound good. Mark Hammer offered some filtering mods to possibly improve that in some older threads.   You wire the switch as follows, lift R11 on the 4013 side and connect the pole of the SPDT to that, then wire the throws to pins one and three.

[oskar]

strange, there are a bunch of build reports on the tonepad site that say to connect them. here is a quote from the most recent build report:

I'm not going to try and repeat this. It can damage the chip, you just don't connect outputs to eachother... It's called a short!       
It should give something like an 'uggly' sinewave. When both outputs are different (50% of the time) they are at war.
Put a finger at the chip, perhaps you can feel it running warm?
What you can do is put a panpot between the two outputs.  100k linear should do it!   

[Barcode80]

i think i see, perhaps they just misworded how the switch is wired. so EITHER 1 or 3 will yield one octave or two octaves down, depending on which switch throw/output is connected. correct?

initally it read to me as though the switch was to make or break connection between those two pins.

[oskar]

i think i see, perhaps they just misworded how the switch is wired. so EITHER 1 or 3 will yield one octave or two octaves down, depending on which switch throw/output is connected. correct?

Yes! (As long as you understand that pin 13 is doing the job...not pin 3!)





Let the pot be something like a 50 - 100k linear...

[SonicVI]

Thanks, that illustrates it clearly.   It's a dual flip flop, the ins are pins 11 and 3 the outs are pins 13 and 1.  The first one divides by two then feeds the second which divides again, so the switch (or pot) bypasses the first division.

[SonicVI]

sorry, double

[oskar]

The tracking ought to be better, in principle, with a more consistent signal source.  In theory, the amount of gain applied to IC1a (x471) and IC1b (x100) ought to accomplish that by maxing out the chip, but a little bit of "help" might not be such a bad thing.  I'm wondering if the following might not be such a bad idea to make the circuit behave a little better:

1) Stick a feedback cap in parallel with R6 (Tonepad diagram) to keep the lower harmonics attenuated for octave tracking.  C3 already provides a 6db/oct rolloff above 1129hz, but a second cap/stage might not be such a bad idea.  With R6=1M, try a 120-150pf cap.  Note that this will change the tonal quality of the same-pitch distortion signal.  If the loss of sizzle bugs you, just lift the ground side of C9 or replace it with a different cap value.

2) R7/R9 form a fixed voltage-divider circuit that takes the maxed-out signal from IC1b and drops it down a bit, where it can be suitable buffered by Q1 before going to the 4013 flip-flop.  But how much does it need to be attenuated to provide optimal tracking?  There's the rub.  I'm wondering if replacing R9 with, say, a 6k8 fixed resistor in series with a 10k trimpot might provide for adjusting/fine-tuning the to-be-divided signal used for generating the octaves.

IC2b is set up as a schmitt-trigger. It's not a x100 amp!   

[ambulancevoice]

i heard from alot of people who use a blue box (todays and the old script blue box, not there own builds) that there is a volume drop
and Shaun from Sirkut Electronics posted a tutorial about how to increase the volume of the blue box
http://www.noisefx.com/article/mxr_blue_box_mod
does anyone know what the cap value is for C11?

EDIT: wait i found it on the page... hehe, its value is 0.01uF

[Mark Hammer]

IC2b is set up as a schmitt-trigger. It's not a x100 amp!    

You know, if you completely ignore which input pins are inverting and noninverting, it looks just like a noninverting op-amp.

Can't believe I missed that one for something like 5 years running.     Apologies in bushel baskets to anyone who followed my comments during that time as if they were accurate.

That certainly changes a lot of my thinking about the circuit, and shifts any and all recommendations about tweaking gain squarely on the input stage.  It also shifts my thinking about where the normal-pitch distortion signal comes from.  I always thought it was from an op-amp simply being asked to provide more gain than it was capable of, when it was actually a comparator-derived square wave coming off the output of IC1b, not unlike the Anderton/EPFM Ultra-Fuzz.  Of course, now it makes even more sense why the gating action provided by Q2 and Q3 is applied to both the octave and same-pitch signals, since there is risk of sputtering from both of them as the decaying input signal starts to veer into intermittent triggering territory.  I always thought that it was an attempt to enhance the image of the sub-octave as co-occurring with the fundamental when it was really nothing of the sort.  You learn something new every day.

You know, though, if we're really dealing with 2 different square-waves here (same pitch and dropped pitch), then there is some latitude for a little rise-n-fall waveshaping, don't you think?

[Barcode80]

yeah, i've definitely noticed the gating. kind of annoying, but not a deal-breaker.

[Mark Hammer]

The gating at the end is actually the deal-maker.

Remember that the flip-flop needs an absolute minimum-level signal to generate a divided square wave.  The comparator also needs an absolute minimum input level to result in a square-wave output when comparison against the reference voltage is attempted.  At the beginning of the note (pick attack), that level is generally assured, but as the note starts to die out,  the signal and the fundamental in particular, starts to get inconsistent.  Sometimes it's above threshold, sometimes not.  The flip-flop does not actually divide pitch directly.  It simply reverses state on each successive input pulse.  To make it seem like the pitch out is pitch-in/2 or 4, every single positive half-cycle of the fundamental must appear at the iput of the flip-flop.  Okay, now imagine you have a 100hz tone and as it decays, positive half-cycles 7, 13, 24, 25, 31, 37, 39, 48, 50, 51, 52, 57 out of the first 60 pulses in a row are missing.  Just what exactly is the pitch out?

Clearly the decay portion is the most susceptible to mistracking and erroneous octave-division, using the simple flip-flop method.  If we were using something much more sophisticated that had averaging over many more pulses, maybe we could cope with the decay better, but the humble flip-flop brings a heavy price for its design simplicity, and that is that the input signal has to be VERY consistent or else get the hell out of the way.  The gating of Q2 and Q3 is the part that  removes the signal before it gets too wonky.  The PAiA Rocktave uses a slightly more sophisticated approach and expands the signal at a different rate than it compresses, resulting in a more gradual and less noticeable noise-prevention action.

Now, having said all of that, IF you are in a priviledged position to provide a spectacularly consistent signal, then it might be worth your while to tinker with the time constants of the envelope follower that controls Q2/Q3.  You will note (using the Tonepad schematic as reference) that D2/R15/C8 provide a basic rectifier circuit not at all unlike what you see on a Dr. Q and many other pedals.  C8 and R15 set the attack/decay parameters.  As C8 gets bigger, the time to charge up increases and the time to discharge also increases.  R15 sets how quickly the discharge takes place.  So, if one wanted to, and could afford, to aim for longer and less abrupt decay/gating, then you could consider increasing C8 to maybe 2.2uf, and increasing R15 or maybe adding a 100k-250k pot in series with the 56k resistor to tweak the slowest gating action out of it that was tolerable.

Incidentally, another way to prolong the decay is to goose the input gain-stage a bit such that the envelope signal remains sufficiently robust for a while longer, and the comparator yields a supra-threshold output longer as well.  I don't know much aboout comparators, but if the threshold (reference voltage that needs to be superceded) in IC1b was reduced a wee bit, then that should also result in flip-flop triggering for a longer period.

[Barcode80]

what about building a squeezer into the front end?

[Mark Hammer]

Compression generally helps in setting the triggering/tracking threshold point more accurately or reliably.  That's part of why the Rocktave tracks so well; there is a compressor ahead of the octave division.

In some respects, however, it almost makes sense to have a noise gate before the division process, such that octave-division will only take place IF and ONLY IF the signal is above threshold.  Strategically, either way works.  You can try and do everything you can to assure the signal is always above threshold, OR you can make a point of only feeding the flip-flop a signal that you know will be above threshold and exclude everything else that's sitting on the fence or lower.

Of course, the overarching goal is to have an audio output that feels natural to the player and listener, and that includes a more natural decay.  In the case of the Bluebox, the entire audio output is the result of a comparator and flip-flip, such that you only get audio signal out IF the input meets the threshold criteria.  In the case of the Rocktave, there is both a clean signal and a fuzz, such that even if the octave part cuts out early, you still have the notes you played decaying gracefully.

All of which leads me to think that maybe the Bluebox needs to have a third output from pin 1 of IC1a that provides a clean signal that can be blended in with the comparator fuzz and sub-octave.  Keeping in mind that:
a) with a gain of x471, that output will not be hiss-free
b) the square wave outputs of the fuzz and octave provide all the top end you need or want
c) it's the illusion of continuity between sub-octave and clean you're really after in trying to get a natural decay

...it would make sense to seriously lowpass the clean output of IC1a prior to blending, in addition to attenuating it a bit.  Actually, I think a lot of folks wouldn't mind being able to blend in some clean signal with their sub-octave, as opposed to having just a choice between fuzz and sub-octave in their output.  (Francisco, if you're reading this, Ithink it's time to work on another Bluebox/Caja Azul PCB layout! )

Gotta say this has been an extremely satisfying thread.  I've learned a lot and been spurred to think about a lot.  I hope it has been as useful for others as it has been for me.

[oskar]

I wonder if it would make sense to introduce a new OP-stage with a little less amplification to drive the envelope part. I haven't fooled around with
the Blue Box, but IC1a doesn't look like a provider of good dynamics...

[oskar]

You know, though, if we're really dealing with 2 different square-waves here (same pitch and dropped pitch), then there is some latitude for a little rise-n-fall waveshaping, don't you think?

Using the 4016/4066 as a substitute for Q2,3 in the Blue Box.
--- I can't think of a simple way to trigger an Attack/Decay generator (which you can build also with a 4013) though.

[Mark Hammer]

Appreciate the effort, but I'm not sure if that's the way to go.  The Q2/Q3 arrangement provides for a less abrupt shutdown than a 4016 switch would.

As well, seems to me that a simple passive mixer could do the trick as far as blending clean with fuzz+suboctave.  Run the output of IC1a to a DC-blocking cap, a fixed resistor, a pot and a 56k resistor from the pot output going to the junction of R18/R19/C10.  The fixed resistor before the pot is simply to drop the level down appropriately when paired with the pot.  For that matter, use two fixed resistors in series and run a cap to ground from their junction to provide a little LP filtering of the clean signal.  If one used a 50k log pot for the clean blend-in, then I suppose you could use maybe even as much as 150k or so ahead of it.  Let's make that 100k+47k in series.  If the 100k resistor is the first one, then maybe a 2200pf from the junction to ground would provide the requisite filtering (corner frequency of around 720hz at 6db/oct rolloff).  Some experimentation would be needed but those values seem like a reasonable starting point off the top of my head.

As for waveshaping, I'm thinking something VERY simple, like two reverse-polarity diodes going to the outside lugs of a linear pot, with a cap to ground, so as to provide variable/complementary adjustment of rise/fall time.

[oskar]

The Q2/Q3 arrangement provides for a less abrupt shutdown than a 4016 switch would.

But isn't Q2/Q3 just switches? I don't think the shutdown speed matters at all in this case. And since it is at audio speed if anything is altered it would be musical anyway... ( I'm off to my dungeons for further CMOS-circuit abuse. I'll bump this thread somewhere around 2009 moahahaha...  )

As for waveshaping, I'm thinking something VERY simple, like two reverse-polarity diodes going to the outside lugs of a linear pot, with a cap to ground, so as to provide variable/complementary adjustment of rise/fall time.

Ah!

[Mark Hammer]

But isn't Q2/Q3 just switches? I don't think the shutdown speed matters at all in this case. And since it is at audio speed if anything is altered it would be musical anyway... ( I'm off to my dungeons for further CMOS-circuit abuse. I'll bump this thread somewhere around 2009 moahahaha...  )

I haven't played with mine recently, but my recollection is that shutdown is relatively quick but not so quick that it feels like a switch.  Certainly a longer "tail" to the note makes it feel more natural, but it doesn't take that much of a tail to start to feel at least just a little natural, and I think that is what MXR went for.  Besides, anything faster and there was the risk of envelope ripple and "gating chatter".  If a true on/off switch is what was wanted, then I think MXR would have opted for something different than simply using an envelope follower with a 1uf/56k combination.  Probably some sort of comparator would have been involved to provide a gate-on/gate-off function.

[Barcode80]

Are we on the verge of a redesign of the blue box into something new?

[Mark Hammer]

Yeah, a box you can actually use.