Anybody got a layout for this? Bissell envelope detector.

[rockgardenlove]

http://www.edn.com/contents/images/265499f1.pdf
There's the schematic...
I'm looking for a PCB layout...I thought I'd ask before slugging it out.  I'm terrible at layouts.

[Processaurus]

Some forumites had mentioned having trouble getting it to work.

[Mark Hammer]

This is about as close as I've ever gotten.  http://www.wiseguysynth.com/larry/guitar.htm

[mills]

Sorry to bump an old thread... but would there be much of a difference in running this at +/- 9V rather than +/-12 V, or shold I just deal with the fact that if I ever intend to use this I'll need a separate 24V Wallwart?

[earthtonesaudio]

I was just thinking about the clock speed of this.  Would there be any advantage to changing the clock speed as a function of the envelope's output level? 

Instead of the simple schmitt trigger clock, one could use something that accepts a control voltage to speed up the reset sequence as the output voltage decreases.

[HarryBBD]

This env gen is best used for a monophonic signal  (with a single note, or a hex pickup).  With multiple notes
you need to set the clock speed really low, which kind of defeats the purpose (the circuit would start to work
like much simpler ones).

The clock rate should be set so that there are not more than three clock cycles during one cycle of the input
(guitar) frequency. If you set it faster, you will get a lot of ripple in the output. Slower, it will take longer to
detect a note-off (stopped string).

In hex guitar use, you can set the clocks to an optimum point for each string, higher strings can have faster
clocks.

I don't see any benefit in a dynamic clock, unless the clock varied with the frequency of the guitar string.

The circuit should run OK on +/-9V btw

H^)

[Mark Hammer]

harry, is that you?

Nice to hear from you at this site.

Mark in Ottawa

[earthtonesaudio]

I don't see any benefit in a dynamic clock, unless the clock varied with the frequency of the guitar string.

Is the primary benefit to this approach the fact that you can decrease the minimum decay time?


I had this vague idea of using a zero crossing detector to advance the envelope generator's counter, with the envelope output also used to compress the signal going into the zero crossing detector.  I don't know if it would be useful, but it seemed interesting to have a frequency output and an amplitude output, each modulated by the other, crisscrossing feedback paths and so forth...

And I'll second Mark's welcome.  It's nice to hear advice from the man himself.

[R.G.]

Is the primary benefit to this approach the fact that you can decrease the minimum decay time?

Harry is of course the expert with this, but as far as I can tell, the problem solved by this is the issue of needing decay to be different for high frequencies than low frequencies. With ordinary rectification techniques, you're stuck with the time between peaks. The integrator cap has to "fill in" the time between peaks from zero volts to the next peak. The sampling on peaks approach here requires far less smoothing and effectively follows the peaks with no longer time between peaks than the sample time. It avoids making you choose between good high frequency smoothing and ripple with lows and good low frequency smoothing but poorer decay time for highs.

I had this vague idea of using a zero crossing detector to advance the envelope generator's counter, with the envelope output also used to compress the signal going into the zero crossing detector.  I don't know if it would be useful, but it seemed interesting to have a frequency output and an amplitude output, each modulated by the other, crisscrossing feedback paths and so forth...

It might be fun to look at, but the cost of complexity starts getting pretty big. It's hard to do that kind of stuff with little enough circuitry for inclusion in a practical stompbox. Semi-synth applications, great, go wild. The triple peak sample and hold is already tricky.

If I were approaching this fresh, I'd probably use a uC to do A-D, and do the decay envelope in programming  instead of with multiple sample/hold units, or use the peak detectors to tell a uC to sample it now. This forces the complexity into programming and out of analog and mixed circuits. Cheaper in both complexity and probably money than multiple cross connected and interacting analog/mixed circuit paths. But that's just my suspicion. Go for it if you like.

[HarryBBD]

Is it me ?  ehhh.... ~might~ be :^)

The problem with using zero crossings is that it just about guarantees that you will have massive ripple. For those
who like to play with SPICE, try summing 1V @ 80Hz, 0.5V @ 161Hz, 0.33V @ 243Hz.  This is an "ideal" guitar
wave... your results could be worse. The sharpening of the harmonics is due to string stiffness.

If any of the harmonics are larger in amplitude than this, you will get multiple zero crosses... and places where there
are very unequal positive and negative peaks.   Track this with just about any follower and you are looking for trouble.

The zero-cross technique only works well on very regular waveforms imho.

Of course if you chose to use a microprocessor, you could get a lot fancier....

Here is the idea. You want fast response (attack), very low ripple (which shows up as IM distortion in just about any
effect) and be able to detect that the string has been suddenly stopped, and shorten the decay as much as possible.

Detecting the string stop promptly, without getting ghost pulses (glitches) is the hard part. My env follower trades the
decay being 'late' by some number of milliseconds to avoid the trouble of detecting a true string stop.

hi Mark and all :^)

H^) harry

[Mark Hammer]

It was the BBD in the handle that gave it away.