GEOFEX variable stutter (cont.)

[puretube]

gez, you know it`d take only 1 leg;
but it sure isn`t rectangular at the edges...

(haven`t tried "analog" inverters for this particular purpose,
but observed the waveforms well with a scope...)

[gaussmarkov]

The idea voltage to be applied to the gate diode would be the bias voltage minus one diode drop plus one signal peak voltage.  ...

thanks, R.G. i am going to ponder for a while.  i'm just thrashing around with the simulations now and not really coming to grips with all that you have explained, here and in the slow gear thread.  and i will be travelling for the next week or so anyway.  and i will have some 2N5485s by the time i return so that i won't be trying to make this thing work with a (low Vgsoff) J201.

more later, gm

[gaussmarkov]

i should have thought of this without simming, i guess, but the RC filter also accomplishes what your trimmer does, right?

Not really gm.  ...

thanks, gez.  i appreciate your feedback!

[gaussmarkov]

back from a trip, i had a chance to breadboard the latest version of this project.  i am satisfied with the results.  my impression is that the 2N5485 (instead of a J201), the trapezoid shaper, rail rerouting, and decoupling caps on the LFO IC grounds made the most difference for reducing the click i heard in my first breadboard layout.  the lfo feed through is comparable to what i hear in a commercial pedal (Boss PN-2).

here is a link to a simple clip: with and without the low e-string of my strat.    this is straight from the breadboard into my computer through an m-audio firewire 410.

note:  when you first power this up, it takes (maybe) 30 seconds for the LFO to settle down.  during that burn in, there is a much louder click.  this is not relevant for actual use.

also, i used a 27K resistor for the slope in the trapezoid shaper with a .1uF cap--instead of the 100K initially specified--it doesn't make much difference, a 100K resistor across the drain-source of the 2N5485 instead of 1M as suggested by R.G.  there is no trimmer between the trapezoid shaper and the 2N5485 at this point.   

[gez]

GM, check the following from Mark Hammer:

http://www.diystompboxes.com/smfforum/index.php?topic=50288.msg375660#msg375660

I haven't experimented with this little trick but it might enable you to do away with the trapezoid shaper.

[John Lyons]

Mark

That sounds nice and clean. Good work!
John

[gaussmarkov]

GM, check the following from Mark Hammer:

http://www.diystompboxes.com/smfforum/index.php?topic=50288.msg375660#msg375660

I haven't experimented with this little trick but it might enable you to do away with the trapezoid shaper.

thanks!  as a twist of fate would have it, i have been waiting for mark to post that explanation here.  so i completely missed it.    while waiting, i have experimented with turning R.G.s square wave LFO into a trapezoid LFO directly, rather than shaping it after the fact.  a simple RC filter did the trick but it didn't change the feed through appreciably in simulations. 

given that there are many square wave  LFOs out there, surely someone somewhere must've made a comparison of feed through characteristics.  i wish that we could find them ...   

on a related topic, i have not been able to decipher Ton's cryptic message about hex inverters (nothing unusual there).  any hints?

Mark

That sounds nice and clean. Good work!
John

thanks again.  maybe some day i will actually have the ideas, not just implement them.    i have learned a lot in the process.

gm

[gez]

on a related topic, i have not been able to decipher Ton's cryptic message about hex inverters (nothing unusual there).  any hints?

If you check the data sheets for unbuffered CMOS inverters, plus info in The Art of Electronics, you'll see that a single stage can round off the corners of a square wave.  This might help things along, but you might as well build the whole LFO out of inverters if you go down this path, in which case it would be wiser to run the CMOS chip from a supply lower than 9V (regulator) otherwise the current spikes can be quite large (all swings and roundabouts unfortunately).

[gaussmarkov]

on a related topic, i have not been able to decipher Ton's cryptic message about hex inverters (nothing unusual there).  any hints?

If you check the data sheets for unbuffered CMOS inverters, plus info in The Art of Electronics, you'll see that a single stage can round off the corners of a square wave.  This might help things along, but you might as well build the whole LFO out of inverters if you go down this path, in which case it would be wiser to run the CMOS chip from a supply lower than 9V (regulator) otherwise the current spikes can be quite large (all swings and roundabouts unfortunately).

interesting.  thanks for following up.  i guess i will resist the temptation to build the whole LFO out of inverters.    for now.   

[gaussmarkov]

hey gez!

that boscorelli/hammer mod to the square wave section of the lfo seems to deliver a comparable feed through reduction in simulations when i use a 100K resistor and 1uF cap.      i'll have to try it on my breadboard version.  i have know idea what the optimal combination of resistor and capacitor is.  i guess that i could increase the resistor value to keep the cap small.  like 500K  and .2uF, figuring i've got around the right RC value.

gm

[gez]

i have know idea what the optimal combination of resistor and capacitor is.  i guess that i could increase the resistor value to keep the cap small.  like 500K  and .2uF, figuring i've got around the right RC value.

Thanks for the feedback, much appreciated.  It's one of those things that is so obvious when pointed out (once again, thank you Mark) but  it never really clicked with me what it was for (and I've seen this addition in a number of circuits).

Re using a larger resistor, that will indeed allow you to use a smaller cap  but be wary of things like input bias current of the amp you use (though it shouldn't be a problem with most modern amps).