Geofex parametric EQ problems (new to layouts)

[MrStab]

hi guys,

i decided to stop being lazy and have a go at making some layouts instead of always relying on others' work. i managed to successfully make one for a 555-based metronome the other day, complete with power amp obnoxiousness.

so i tried making the parametric EQ from the schematic here: http://www.geofex.com/article_folders/eqs/paramet.htm, and here's my layout:


(larger pic here http://i.imgur.com/QzChRej.png)

it kinda works, but there's some crazy siren-like sound that seems to be tied in with Resonance pot 1 and Frequency 2 (although it could be a broader issue than that). just when either knob is turned near full-clockwise and full anticlockwise, respectively. there's also some hissyness. and i think i have the pots wired round backwards. and Resonance 2 doesn't seem to do anything.

substitutions: i used 3K as i ran out of 2.7k resistors. i tried 2.47k in these spots as well, but no improvement. i used 100k resistors for the vbias as i ran out of 10k.

any tips?
cheers!

[mistahead]

Whenever I hear "squealing" or "siren" associated with turning a knob all the way to an extreme end I tend to think that a feedback loop has hit too low a resistance.

I don't know if this is right or wrong but it seems to match some observation.

[MrStab]

i had a wee look round all the opamp outputs, can't see anything but it's possible there's some obvious flaw going on there.

observation: it's only when i touch the lugs of the offending pots (at certain settings) that it'll go off into that crazy oscillation or whatever. maybe that's normal with gyrators and i should just keep my grubby hands off em?

[mistahead]

Hmmm... even if wired backwards in vari-resistors like that it should work with a reversed turn so that might point at an issue.

Touching lugs is fun... but aside from that can you put a tiny little 100R or something in sequence with the knobs - just break them out of ever hitting their true "end" where the problem begins? At least one of those forms part of the OPAMP feedback network doesn't it? (Freq?)

I'm not pro lol, just trying to step through - might yet be waaaaay off.

[MrStab]

the problem seems to be "activated" at about 2/3rds of the way down the Resonance pot, then set off by touching the lug of the Frequency pot, as opposed to right at the end, so whilst i originally thought it might be what you're alluding to, i'm starting to have my doubts as it seems to be a problem for more than just the end of the taper. could maybe try cutting off that whole last third? what's interesting is that the other band does have this problem but it isn't affected by its' Resonance pot (which as i mentioned doesnt seem to do much). i'll compare both bands for differences (again)...

[mistahead]

Yeah no further ideas, shifting the pot range up with an R to avoid that chunk of R values in a whole third of the pot, as well as the resonance not being responsive does push me off the mark a ways.

Hmmmm I need to play with more OPAMPs that much I can tell you mate.

[MrStab]

thanks for the ideas anyway, man - i think you could be onto something with it being feedback-loop-related, even if it's a bit more obscure an issue than might be expected

[mistahead]

Ummm... might be a silly question - have you confirmed the cut under the bottom left 10nf?

It seems like that cap and the two 51K R's are involved in the issue "area", you also have both 22nf and 20nf caps in that area - if I saw that in my shorthand I'd guess I'd probably screwed up a notation as I'd be unlikely to bother with geting 2nf closer to perfect values in a stomp box (even though its 10% and I think most of my caps are 5% tolerance... hmmm) - should one of those values be factors higher/lower?

You've seen my posts lately mate - I'm catching up but not yet there, hopefully we can shake something out.

[MrStab]

yknow, that overlap had crossed my mind, the 20-22nF thing. i figured i'd go with the example figures given on the page first as i'm crap with numbers. i used a ceramic for the 20nF and a film cap for the 22, so definitely not a concrete tolerance.

turns out the "siren of death" isn't just brought on by touching lugs - it can be brought on just by setting knobs a certain way and in some cases after strumming hard. it's probably a tall order to ask someone to examine my poor layout, but it's all i can think of to resolve that and the hissing problem. just hope it's not the 2nd opamp, i ran out of sockets for that one!

hey you've offered more ideas than i had on all this, mistahead!

[MrStab]

i'll try putting a 4.7nF in parallel with the 22nF just to try and account for that tolerance. trace cuts all seem to be in order.

[MrStab]

no improvement. the oscillation (?) is a few cents above Ab when everything is turned full. which isn't very helpful, but interesting nonetheless. not.

Edit: right, i've got rid of the oscillation, it seems. i forgot to mention that i had was using a 500k pot with 220K resistors on each outer lug for Frequency 1, as i'd run out of 1M pots. i also went over the strips for any solder bridges, which might've been it.

now i just have the hiss, which i hear could be gyrator-related. and i'm still not 100% sure i have the pots on the right way, but they seem okay according to the schematic

[mistahead]

Heeeeey - no oscillation - thats a plus, that would have done some interesting things to the taper of that pot I would think.

It wasn't so much tolerance of 2nf I was thinking, but rather I'd have been checking my notes for the pf / uf / nf / mf (or similar) stuff up in my notation - I wouldn't expect 22nf and 20nf to act very differently in your application here, but to have such similar values right by each other that way raised an eyebrow.

Can you audio probe the opamp outputs and see if the hiss is being "introduced" at one of those stages, I would guess its just stray hiss (boxed up may resolve itself) from pickups.

[MrStab]

i might actually have it working! seems i need to stop using that pile of "might work, probably f***ed TL072s". i checked with a multimeter, as i should've done in the first place, and i was reading either full V+ or nothing at all where vbias should've gone into the first opamp. i was using an OPA2227 at times too, but maybe the current drain on that (which did significantly lower vref) caused hissing as well as the dud TL072s. all this time i figured it was my layout at fault and i didn't approach debugging like i would any other circuit.

well, i think it works... lol. the 2nd section, 450Hz-2KHz,  seems a bit subtler than the lower section before it, but it does have some effect over the sound. maybe i could do with a 3rd band for those harmonics. think it'd be do-able to test this via. my sound card as i don't have an oscilloscope?

here's my current layout (a few minor changes from the last one, notably moving the 470R on the right to split off before the 10nF cap):


thanks a lot for the help & sorry for any time-wasting, mistahead!

[mistahead]

No time wasted I didn't want to mate 

Happy its working out, to be honest I have an interest in a "template" EQ vero layout that has been verified and why not help with making one - with the few skills I can offer and am trying to develop - rather than just use another tonemender layout?

[MrStab]

there does seem to be a real shortage of raw EQ layouts on the web. they all either do something else, like the Tonemender being a boost or the TC Sustainer + Para. EQ.

i studied audio engineering years ago - i was too young & dumb to really make sense of it, but i always figured if i wanted to make an EQ, it would be a comprehensive one and not just a Baxendall or whatever. feel free to use & abuse my layout, no guarantees it's right, though! lol
unless maybe i test it with a sig generator + soundcard. i'll report back if i do

[mistahead]

It kind of makes sense, boost or fake sustainer can both offer active recovery to a passive EQ / tone, the actualy EQ is more a building block - like a buffer, at least to my thinking. The thing that catches my eye is that its probably to cover the fact they are just split-filter-blend-recovery type setups (put the recovery on the front instead blah blah blah), "proper" active EQ is worth it enough to consider all of the frequency cutting tone pedals ALMOST lacking in their most fundamental design..

What is the alternative DIY active EQ favourite? Is there one? Tonemender?

I'm musing here - but lets take a buffered signal split it out five ways and filter out our EQ bands, for each of those bands put in an active boost, tune a compressor and limiter into that band and blend, then a typical one-knob highpass-lowpass-blend circuit. For each band you are now playing with around four to seven knobs depending on how much DYNAMIC you want to be able to dial in. Five EQ bands as I really don't imagine this working on a "wider" band scale, three bands MIGHT work for guitar/bass... Maybe even make that one-knob-tone deal a two-knob-with-presence deal (I suspect pointless with compressing and limiting, could be a simplified replacement that "does the trick"). I'll stop original idea now - it is fast becoming a BIG idea for a stomp box.

I'll note what I'm thinking is in low-cost BJT/JFET/OPAMP analogue - digital kit would likely have an answer to this application too but that's what off the shelf/software toys are for... 

I'll see what I can do to help verify - I have at least most of the parts, and a shopping list to push through, but time is a killer - I basically only post here while sitting and waiting for processes/exporting/importing/other people at work, hence why I usually have no notes, numbers or anything but the most basic links around when I do.

[MrStab]

well, to me that's the advantage of just a "pure" EQ circuit: you can have the adaptability to set it up as you see fit. more adaptable to your given situation. i haven't looked into the Sustainer+EQ too much, but having already got a compressor i'm satisfied with, i'd be apprehensive about adding more (intentional) compression to the path. i guess it does make sense to have a boost along with an EQ, though.

what you're describing vaguely reminds me of an idea ive had for a while for a general-purpose mixer/EQ/boost/WMD. just all the non-effect, fine-tuning you might need in one box. hell, you could even make it so both ends of the chain pass through. i could be talking crap here, but i feel as though a larger number of bands would be bordering on a convoluted graphic EQ with Q controls. there's a first for everything, though... lol. if i added any more bands to my layout, it would just be the one - leaving a free half-opamp, but i'm not sure it'd be advisable to use that if one half is being used as a gyrator.

i recall reading somewhere (might've been RG who said) that opamps offered the most guaranteed-transparent way to handle this stuff, so personally i'd go with that (unless ofc it would make more practical sense to use transistors).

urinating into the wind and hoping it doesn't spray back all over me. that's how i operate!

[mistahead]

Hahaha - sounds like a lot more scientific than my method, you are observing outcomes.

Hmmm... now I'm thinking "dual console" version of my msuing (lets call it the TwoMuchHeadQ), one for the front and one for the back of the chain - effects loop in between and an amp at the end.

Half the pedals in many chains would become... less not redundant... spoiling the fun a little I guess.

[MrStab]

tbh i'm mostly driven by having run out of stuff i wanna build. i guess on the plus side i'm using it to force me into designing layouts etc.

wait... since when can you have too much head? lol