EQ + (Engineer's Thumb x2+) = multiband comp.?

[MrStab]

i know it's probably tedious to expect anyone to check vero layouts, but out of interest, here's my attempt at the splitter & mixer (opamps U1 and U2 basically) on your schem, Sam:



i omitted U9, think i could get away with that?

[samhay]

VA = +9V, VB = +4.5V.
Not sure if it will help, but there is a verified vero layout of the audio section of my Anharmonic tremolo (essentially what you are trying above I think) here:
http://www.diystompboxes.com/smfforum/index.php?topic=102736.msg915228#msg915228
and the relevant schem: http://www.diystompboxes.com/smfforum/index.php?topic=102736.0

[StephenGiles]

http://soundwood.weebly.com/uploads/1/5/6/5/15656028/aphex_fm_pro_model_2020_manual.pdf

This professional broadcast unit has a 4 band compressor for your delight - not the VCAs but the filters may be of use.

[MrStab]

cheers for the links, guys. i'd say they've come a bit too late, as i've just finished populating the board for my layout based on Sam's schematic, but not so - as Paul pointed out, i shouldn't be so finite about how i split the input, and there's a good chance i could use a different approach altogether further down the line. if i'm careful, i should be able to slice the whole split/mix section right off my board with a knife if needed.

it's gonna be a tall order if any debugging needs done, and i'm not sure i have enough opamps let alone dual-gang pots (definitely don't have any of those), but i should be stocking up on parts later in the week. in the meantime i'll have a wee smoke break, then try some rudimentary offboard wiring with the parts i have. luckily i can temporarily omit most of the controls on the compressors until i score the right pots.

[MrStab]

splitting section seems to work! variable output from both high & low outputs.
high-pass filter output is a bit distorted at one extreme but i'll have a look into it, minor issue if anything.

it's not practical to test this conclusively using 2 pots instead of a dual gang, but i'll get the compressors working in the meantime. not sure if the mixer opamp works yet.

still not sure what caps i'd change in the Engineer's Thumbs themselves to accommodate this design

[PRR]

> insane like 8-gang pot would be required

A 4th-order filter has four frequency resistors. A high-low split then needs eight frequency resistors. The 4-pole is so sharp that the least mis-match of resistances will cause a major error (bump/shave) in the frequency response.

1st-order filter, only one freq-R per filter, so a dual-gang works. And the corner is so broad that major mis-match is insignificant.

I will offer 700Hz as another possible place to split guitar. Look at Fender Bass/Treb tone-stack. At most settings there is a deep dip near 700Hz. Something about guitar likes to be treated different above/below 700Hz.

[tubegeek]

FYI the Rane DC24 is a commercial product which - despite being discontinued - is still VERY popular for bass rigs. It combines a crossover and two compression sections, with flexible configuration options, in one unit. So at least there's a data point for your consideration. The bass players - I am told - LOVE being able to have the grunt and the slap separable, so they can be compressed differently.

Schematic here: http://www.rane.com/pdf/old/dc24sch.pdf and manual here: http://www.rane.com/pdf/old/dc24man.pdf

Rane documentation is never less than entertaining, and often it is extremely educational.

[MrStab]

interesting stuff, tubegeek. good to know something is not only do-able, but has a tried & tested history. i can't help thinking i've seen that in a rack somewhere, but i've probably just imagined it. bookmarked! at least now i know if this doesn't work out, i can sell it to my bassist...

funnily enough, i'm slowly grasping the concept of all this through learning "hooks" such as your explanation into the varying levels of required exactness, Paul. i'll shoot for 700 when experimenting. i'm picking up a dual-gang 100k linear pot today (and a bunch of TL072s) - i think Sam's schematic calls for a log pot, as it has a range that extends beyond guitar, but i just need something to fill the gap for now. assuming some kind of signal generator, is there some semi-reliable PC app out there for analysing the spectrum via. soundcard or USB interface? just so i know what i'm working with.

cheers guys! this project has gone from hopeless to plausible so i'm kinda looking forward to see how it pans out. i'm thinking it'll definitely need switches to maximise functionality. which makes me apprehensive about changing parts in the compressors themselves, as they'll each become less flexible, but i may have to change those caps i keep going on about. whatever they are.

[samhay]

I used a dual-gang A100k pot for my crossover frequency - frequency is proporational to 1/R, so a log pot will work more naturally than a linear one (but both will work).
My pot works backwards - clockwise for more bass - as I didn't have a C100k pot. If you use a linear pot, then you probably want to wire it backwards to that shown in my schematic.
If you want to reduce the frequency range, then you can play with the 1k resistors (R6,R7) and the value of the pot. The math is a little heavy, so you will likely find this rather useful: http://sim.okawa-denshi.jp/en/OPseikiLowkeisan.htm

[MrStab]

ah right, so we are using the same lettering system for pots. i'm never certain unless i read the type straight-up. thanks for the link, numbers definitely aren't my strong point so i'll hit it up if 10kHz is too much. i was thinking earlier: maybe that's why i was getting a screwed up signal from high-pass output when the frequency was maxed. i didn't have much time to play around with it. so are you saying you'd ideally use a reverse-log pot, and do you mean the schematic itself is actually wired backwards? i'll be using flawed-taper pots for the rest of the controls until the weekend anyway, plus this dual-gang i scored looks like a cheap POS lol (one of those ones with no grooves for a knob, either   )

cheers

[MrStab]

hooked up new pot. with one TL074 there's no change in sound over hi-pass output, with another there is, but it's heavily distorted. attempting debug.

[samhay]

The frequency pot numbering is for a log pot. If you want to use a linear or reverse log pot, then wire the pot backwards to that shown.
A (reverse) log pot would preferable over a linear pot.

Not sure how you feel about the place, but Maplin has a reasonable selection of dual-gang pots. No reverse log ones though I'm afraid.

[MrStab]

funnily enough, i scored this pot on one of the few occasions i bit the bullet and went to Maplin. being able to order and pick up the next day is handy if you need stuff fast. still a bunch of ***********. lol

debugging is proving to be a nightmare, probably because i built the two circuits in parallel, as it were (not that kinda parallel). i almost have the bottom compressor working, voltage-wise at least, but having bias issues/null voltages with the top one. problems seem to be with that half of the LM13700. but they seem to be more general issues outwith the realm of this discussion. unless there's some weird issue i'm not aware about, re. using either side of the OTA IC.

[MrStab]

upon testing the compressors in isolation, i'm getting some really weird white noise-like interference, brought upon merely by touching certain parts of the circuit, and really intermittent signal in tandem with that on one of the comps. it seemed symptomatic of a poor IC socket connection - wasn't that, doesn't seem to be the board bending, all off-board wiring is accounted for, no apparent shorts, even happens (albeit quieter) with the OTA removed. the other compressor also seems prone to this noise, but the signal is more constant. neither compresses, although i did have the worse of the two compressing but distorting at one point - then gone as i tried to push the IC in properly. but not even IC looseness is a repeatable cause. have a few different TL072s to test. all i can really think of now is a faulty cap, given the organic-like, unpredictable nature of the problem. voltages seem to check out.

[MrStab]

just letting you guys know that i've started work on this again. i got a bit frustrated before and probably needed the break - one of my aforementioned issues was caused by a fairly-obvious short. i have 1.5 of the compressors working and think i need another coupla TL074s (i seem to be quite careless with opamps lately) to get the crossover up & running. the crossover currently seems to be bypassing everything, but i hooked that up in a rush as it's approaching 7am and i'm going to bed! between this and a bass fuzz to be boxed, i have my hands full.

[tubegeek]

MrStab:

Good idea to take a hiatus. Back with a vengeance!

I had some good results (as in: easy to implement, work correctly, low parts count) recently with a pair of unity-gain Sallen-Key filters for a similar crossover application.

I'm on vacation without my own computer so I don't have the schematic to show you at the moment but check the link below for more info:

I found this site to be helpful in double-checking my filters with real-world component values:

http://sim.okawa-denshi.jp/en/OPseikiLowkeisan.htm

That link goes straight to a Low Pass Filter calculator but the links on the left include the High Pass etc. And the schematics on those pages show EXACTLY how I implemented each filter.

More useful info here:
http://sound.westhost.com/dwopa2.htm

That link goes to Part Two of a 3-part article by Rod Elliott. All 3 parts were very helpful for me but Part 2 is where the discussion of Sallen-Key filters is found.

[MrStab]

thanks for the links, tubegeek - i'll definitely need to read up on this as Sallen-Key sounds completely new to me but i may need an alternative to what i'm using, as my crossover section isn't doing well at all. Ctrl+D'd! i can't read it just yet, but would your method allow for variability?

the compressors more or less work, but i dunno if my layout's at fault or if the schem has errors. Frequency pot does nothing anymore, since successive alterations. high-pass filter output is a lot stronger than low-pass, as if the pot were "stuck", but not quite, as something similar was the case when the pot did work (always worked on the hi-pass, was quieter & undefined on low).
also, going by Sam's layout a page back: U1A (very first opamp) was leaking into U2B (final mixer) with the only obvious connection being the vbias, so i tried diodes to prevent that. it just made the sound distorted, which i can kinda understand. so for the sake of experimenting i just decided to put U2B on a separate IC - no output at all. i've only had the chance to work on this late at night, which can't help. slowly but surely... i guess it's just one of those projects where it's hard to keep track of every little thing.

my layout at the start of this page (assuming you're viewing the full version on a computer) has been modified slightly since, so i may have corrected some things on it, but could anyone give it a wee look, bearing in mind the aforementioned problems? would be really appreciated. apologies if this post is all over the place.

[samhay]

The actve crossover I posted works by subtracting the LP signal from the dry signal to generate the HP - so chosen so that you can have a 2-pole filter without using a 4-gang pot. U2A is a differential amp (http://en.wikipedia.org/wiki/Operational_amplifier_applications#Differential_amplifier_.28difference_amplifier.29), so it takes signals from U1A and U1B. If it doesn't get both these signals, in the correct ratio, your crossover will not work.
If you mix these the HP and LP signals back together - e.g. via 2 10k resistors or a summing amp, then you should geta flat response back and it will apear that the crossover is not working.

Do you have a schematic of what you have done so far?

[MrStab]

to the best of my knowledge, i'm following your schem as closely as possible - although when i briefly tried a separate TL072 for U2B, i left out the 10k resistors - i should try putting those in (they were in the original TL074 implementation, though).

i know a layout may not be helpful, but here's more or less what i had (sans diodes from vref on row 3):



in the Wiki link you posted, in the diagram, is V2 still vref (as per your schem) or is that one filter going to the + input and another going to the - ?

i'm starting to think the original layout has some flaw whereby U1A connects directly to U2B, and my separate attempt at U2B has the cancellation (if you can call it that) issue you describe. food for thought. i'll have a look at it later, for the time being i have to go be the repair lackey at a charity music shop lol.

cheers!

[samhay]

In the wiki schem, V1 and V2 are two different signals - e.g. LP and dry. Note that this is drawn for a bipolar supply, so instead of Rg going to ground, you would have it go to Vref - which is where, I think, you are getting confused.
If you look at my schem, The dry signal goes to (+) input of U2a via R8. To keep the levels constant, we use a 10k/10k voltage divider (R8/R9 with R9 going to Vref) to half the dry signal to the op-amp.

Without the resistor and capactitor values, your layout is a little hard to follow and I don't know what the 4th op-amp is doing. However, if you are using the same numbering as in my schem, then R8 should go from the input buffer output to the (+) input of U2A and none of the op-amp inputs should be directly coupled to Vref.