envelope filter nightmares...arghhhhhhhhhhh...

[Gurner]

The problem is that the end device that Rob wants to drive is a LED....that mcmeat  is an improvement, but the EF section is biased to ground....& therefore the envelope cap will need to charge up to the fwd voltage of the led before the led will turn on (in other words a 'dead zone'). Same on the release side, as soon as the EF cap discharges below the fwd voltage of the led, it'll extinguish....but in this situation there's still about 1.8V still to discharge on the holding cap.....therefore you miss that 'tail' release aspect. This is why I reckon the led should be biased to 'just off' with no input signal (so for example a RED LED should be biased to 1.7V etc)

btw after a while the guitar's envelope won't satisfy you i think (i mean the guitar's envelope is generally the same shape all the time, that is why they try to shake up things with the attack/decay cap) that is when you should try triggering an ADSR to control the filter instead of an envelope follower but then there are problems with reliable triggering imo

Normally, (IMHO) we're not trying to extract/grab the guitar's 'signature' as such....but merely extract the magnitude of the 'plucked' string. Re going for an ADSR type filer, well with most guitar-esque EFs,  the 'A' is covered already, the 'D' is too (though confusingly is normally called release)....you can't have an 'S' aspect with a guitar (well, not without a sustainer at least!) & nor can you have the 'R' (release) that follows the 'S'!

[WhenBoredomPeaks]

But the McMeat's EF section drives a LED too (actually two vactrols and a "status" led or whatever they call it but they are still LEDs)

The speed of the Attack is always the same with a guitar. (basically the pick attack) You can try pick weaker but then you will just get a lower peak but generally the shape of the envelope will be always the same. So it is like having an ASR and you can only change the value of the Sustain (with the strength of picking) Or if i think a little more about it you don't even get sustain, after the really fast pick attack portion of the signal it immediately starts getting weaker till it gets to zero. So it is more like AR.  Or you can call it AD but with zero sustain and then the release becomes irrelevant because it goes from zero to zero.

So you can try whatever on a guitar but you will never get an opening filter like sound or complete silence after the attack portion or fast decay or many other sounds that you can do with an ADSR.

(edit: btw the envelope follower section of the Meatball (musicpcb.com) is biased like what you said (not around zero but to some voltage around 1V) so maybe that would be better)

[Gurner]

But the McMeat's EF section drives a LED too (actually two vactrols and a "status" led or whatever they call it but they are still LEDs)

yes, I realise that   , my last post was in response to your comment  "btw i don't see how the envelope follower section could be this much of a problem"  I'm basically agreeing that yes an EF section is easy, but a touch more involved if you seek a smooth response especially if the end device being driven doesn't turn even on until it gets 1.8V across it  & then turns off when the voltage drops below 1.8V. (i.e a RED LED)

[deadastronaut]

hi guys, i tried the mcmeat EF,

from the 4.7k onwards to led..

i didn't get any response from the led though....with guitar straight in....hmmmm.....i'll give it another go after dinner...



ive been scouring the threads and web,  and what i want seems to be a common problem (quite a few threads on led EF..) , with no definitive solution that i can find anyway...

[Gurner]

On pin 3 I'd have binned R3 (which biases IC1a to 0V ....which I don't reckon is best)  & instead placed a trimpot with one lug at ground, the trimpot wiper to the opamp pin 3 & the top lug of the trimpot to a fixed value resistor which then in turn connects to 9V. The value of the resistor depends on the value of the trimpot you use.....a good starting point would be to use a resistor just under 2x the value of your trimpot.

You can then set the DC level of the opamp output to suit (i.e. tweak it until the LED D3 is *just* off.) IMHO you can lose R17.

[deadastronaut]

YAYYYYYYYYY....whooohooo....

nice one G....

yep that works...

[Gurner]

Good.  

I'd have used a schottky for D2 as well (you might not even notice the change, but it ought to help in this circuit)...I'd have also thought that R4 is too high - I'd have gone for 470k (probably less)

[deadastronaut]

aha those schottky's  come in handy after all...

yep done those...

re: the trims actions:

attack: full on =led on/off.  halfway=slightly short swell on.(soft lattack)  full CCW led just about on..      (halfway=very short soft led fade)
 
decay: (with attack full) : fully CW =dim led,... full CCW= bright led.... on /off action.  (no fade)

intense: full = led almost off  ,   full CCW led on/off action (no fade)




edit as an experiment i swapped the c3 10uf for a 100uf to be extreme .....and yes...lovely slow attack fade on and off,  

47uf is cool...too, probably more practical/useable..

now to make it pot adjustable,  

edit:

actually the decay pot acts as an varying attack with the 47uf now, when CW its on/off, and when CCW its a nice fade...with fadey goodness in between...

i could probably lose those other pots....hmmmmm...

defo the best EF yet.....


i'll have a measure up and sort that voltage divider out...cheers G: you the man!!!

[Gurner]

Your mileage may vary, but I'd have thought R1 should be above the trimpot.

In fact, a better arrangement would be something along these lines...

9V
|
22k resistor
|
25k trimpot (wiper to opamp IC1a pin 3)
|
7.5k resistor
|
ground


.....this will allow you to set the bias on IC1a pin3 anywhere between 1.24V & 5.4V ....which should be enough range for any colour LED you put on the output.

You should also have left R16 in situ (say 100R)....this will at least yield some series resistance if your intensity trimpot is set for zero resistance (zero resistance is not good for the LED)...in other words it'll protect the led if the VR5 trimpot is set too low.

Also...a point worth noting....pesky input impedance again....at about 4.7K & not good for connecting direct to a guitar (so use a buffer or preamp to feed IC1a)

[WhenBoredomPeaks]

BTW this circuit's attack/decay controls are a bit problematic imo, for example if you set the attack higher than the decay it mutes the envelope voltage somehow. (at least this happens with the meatball and meatsphere) Also using a bigger cap instead of the 10uF for more usable decay times makes the attack too slow. I think those problems were some part of the reason why i started looking at ADSRs a while ago.

[Mark Hammer]

It,s also why, over the years, I've found that I get more return for my effort on variable decay than variable attack.  Some variable Attack functions work well - the MXR Envelope Filter comes to mind immediately - but in many implementations they just eat up too much drive current.  Moreover, one needs to eat up a LOT of drive current to yield audibly different attack times.  With variable decay, you can easily obtain a 20:1 or higher ratio of decay times without significant impact on the sweep width of the filter.  That's much harder to do with attack in simple circuits.

[deadastronaut]

in my rummaging around i found the bassballs circuit, which sounds to me like what i was originally looking for...sort of guitar faux synthy type sound..

to go with an off biased fuzzy synthy sound i have......

anyway i found this remarkably hummy but ''very informative'' video of a guy demonstrating his excellent mods.

http://www.youtube.com/watch?v=eoAIKHFGOcQ

i'll definately try this on the breadboard today and see how i get on..i'll leave off the moody fuzz though i think.....sounds great mark!!

i want to be able to pick softly and harder...dynamically...to go from gently opening, to fully open filter....like a synth...lots of delay etc..

like this clip i did with my dodgy thrown together EF....

https://dl.dropbox.com/u/7464107/SYNTHFUZZ.mp3

i'll let ya know how i get on ok...

i also found this by runoffgroove....minus the fuzz..

http://www.google.co.uk/imgres?imgurl=http://runoffgroove.com/images/attachments/bb.png&imgrefurl=http://www.diystompboxes.com/smfforum/index.php?topic%3D72307.0&h=570&w=700&sz=8&tbnid=7c1BWTGs-CJkGM:&tbnh=87&tbnw=107&prev=/search%3Fq%3Dbassballs%2Bschematic%26tbm%3Disch%26tbo%3Du&zoom=1&q=bassballs+schematic&usg=__w_NZJKFfX36E9PlZsjCYh95TRJg=&docid=EFRM7XO_xjDoNM&hl=en&sa=X&ei=AFGJUPp85crRBdiMgLAE&ved=0CCIQ9QEwAQ&dur=838

[Mark Hammer]

The webcam mic is the pits, and the videos are shot right beside the desktop computer and directly under a fluorescent fixture.  I could have only made it worse by shaving with an electric razor while talking!  That's part of why there haven't been many recent followups.

Fast decay seems to perceptually accentuate the sweep, making it sound more synth-ey.  And it's the sense of "action" in the sweep of any filter that seems to appeal to us.  Although that "action" comes in the form of both upsweep and downsweep, the series resistance needed to vary the attack diminishes the upsweep, even as it provides more sonic variation.

I've often wondered what you'd get by sweeping the two bandpass filters in the Bass Balls at different rates, with independent envelope followers.  My sense is that it would start to get a little more "talk-ey" as the desynchronized filters started to behave more like formants do.

[deadastronaut]

 ...ha ha, shaving too...that's funny.!..

yeah i guess it would end up vowel-ey...

halfway through breading.....i can't wait to tinker with those twin filters....sounds right up my street!!!..

i might try adding my own fuzz after to the out of ic1b...anyway see how it goes.

cheers mark!..some cool mods on that vid, you should do more!....but don't shave at the same time..

[pinkjimiphoton]

fwiw, rob, the escobedo phunqgnosis worked for me right out of the gate.

do you have it sweeping up, or down? if it's sweeping down, it may not be too noticeable..

just thinking on my feet.

[deadastronaut]

got it going. that fuzz is nasty in a bad way......thats gone! .(the diodes trick was much better though)

i also swapped out the 'gain' 220k for an 82k..

it has much better sensitivity to picking now.. its great on the B+E strings all over too...

i like that 330k mod...nice and short to really long..

i love how this thing 'phases' the trail off....

anyway still tinkering.....it gets very clippy on certain notes and whack, so i'll tinker with it...but i'll persevere..

https://dl.dropbox.com/u/7464107/guitarBBbollox.mp3

tried my bass for a laugh too...

https://dl.dropbox.com/u/7464107/bassbollox1.mp3

[Mark Hammer]

One of the things we tend to forget all too often is that filters need something to filter.  If you feed a VCO to a filter, governed by an envelope/transient generator, the harmonic content of the VCO is constant.  It only disappears if you filter it out, no matter how long you take to do so.

Not so with guitar.  Most of the harmonic content is at the beginning, and fairly quickly dies out.  If you come late to the party with your filter, all that's left is a bit of dip and the pieces of cauliflower nobody else wanted.  Distortions can help to keep the harmonic content "alive" for a longer period, but the harmonic content can still easily peter out.  Moreover, distorting prior to envelope detection results in less dynamics.

So faster decay helps to get in there and filter stuff while there is still stuff to filter; making it a more pleasing and musically interesting sound.

[Lurco]

The envelope follower needs something to follow, too!

[Mark Hammer]

Which is why it is a best practice to tap a clean signal to send to the envelope follower, and introduce any fuzz (which naturally steals dynamics) after that point.

[Jazznoise]

I've often wondered why envelope based pedals dont provide a Sidechain Thru or something for this reason. What if I want to run my Bluebox into an autowah, f'chrissakes!