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

[MrStab]

ahh right, i think i'm starting to get it. i had both signals just converging at the inverted input and vbias to non-inverted, which i thought seemed a bit off. you did say there might be one or two errors in the schem, wasn't sure what they might be but maybe that was one of them.

sorry, i was using the same numbering as the schematic (https://dl.dropboxusercontent.com/u/11996927/dual-band_ET_schem.jpg). i used all 12Ks instead of 10K as "placeholders" for the record (all i had on hand), which i figured isn't ideal, but given that all is equal shouldn't it still work to some degree? maybe not if such precise calculations are being emphasised. if you're still bothered:

R1 = 10M
R2 = 1K
R3 = 1M
R6 = 1K
R7 = 1K
R8 = 12K (should be 10)
R9 = 12K
R10 = 12K
R11 = 12K
R15 = 1K
R19 = 2.2M (was 1M on schematic but no apparent issues with vref/current)
R34 = 12k
R35 = 12k
R36 = 12k
C1 = 100nF
C4 = 22nF
C5 = 10nF
C10 = 100nF
C13 = 100nF
C14 = 100uF
C15 = 100nF

hopefully i got em all. i'll try fixing the mixing opamp's inputs and let you know how it goes. if that works, the issue still stands with the Frequency/xover pot not doing anything and the low-pass output being crap.

thanks again!

[MrStab]

i made the mixer opamp as per the wiki schem (albet with Rg going to vref). i get signals the way through the compressors and back, but it's weak and distorted (even with crossover section isolated, ie. lo/hi filter outputs going straight to output jack). ground-like (but not quite) hum, no obvious lack of ground


btw Sam my bad if i've accidentally referred to you in the third-person, i should really double-check usernames.

[samhay]

No worries.

Some thoughts. The ET must have a stable Vref, which is why you need to buffer it with an op-amp. You aren't doing that here - which is fine if Vref is only feeding the crossover circuit. However, you will need to drop R18 and R19 to 100k or less, or else Vref will not be stable and you will get distortion etc. In this case, you also need to make C15 much larger - say 47u or more. This is how I did it in my Anharmonic Tremolo - http://www.diystompboxes.com/smfforum/index.php?topic=102736.0.

I have worked through the first op-amp of your latest layout. You still don't have R8 connected to the output of U1A. After that, you may or may not have problems, but I guess I need to know which op-amp is supposed to be U2A U2B etc.

[MrStab]

i appear to have mixed up R8 and R9, my bad - i think R9 is connected to U1A output, R8 to vref. although i've since altered that part to separate that signal from U2B+ on pin 5 of the TL074, in order to fix the mixing opamp as described earlier ( i'd originally just thought that pin needed vref and nothing else, so thought i could kill 2 birds with 1 stone by sending vref there and not worrying about signal from R9 flowing back into U2B. i tried 4148 diodes and a trace cut to prevent that, hoping only vbias could get through to either). basically, this is my fix since i realised i do in fact need that input:



ive only recently picked up the gist of using different values for voltage dividers - i thought it was all arbitrary so long as they were even, but some reading about the Causality Phaser seemed to suggest otherwise. what you're seeing is just the vref for the crossover section, though - i've been using separate vrefs, with just the stock 1M ET dividers in place as always. which at least seems to work in my lone-ET build. should i still change things and consolidate all 3 vrefs into 1? could the slight voltage or current difference between the vrefs of the ETs and the Crossover be a problem?

i won't be able to work on this until sunday, but it might be a good thing to have a bit more of a plan formulated before going into battle again. i should really get a breadboard, but as i've mentioned it shouldn't be too hard to separate the crossover section from the compressor section (as they're glue-gunned together lol. well-jumpered, though.)

cheers

[samhay]

The schem I posted DC couples the crossover to each ET, so would need a common Vbias. This is how I like to do things, but you could put caps after the LP and HP ouputs then bias up the ETs with a different Vbias if that is more convenient.
The resistor values you choose for your voltage divider determines how much current you can pull out of the middle of it before it swings too far away from where you want its voltage to be. The ET buffers Vbias with an op-amp, which is probably the gold-standard method of generating a Vbias. If you are not going to do that for the crossover, then the next best approach is to significantly reduce the 2 voltage divider resistors R18/R19 - I often go down to 10k for these - and add a big cap. If you haven't read this, then it might help - http://www.geofex.com/circuits/biasnet.htm.

[MrStab]

i've just realised that the Engineer's Thumb does what you're saying by default. i'm a bit slow.
could i just sever the crossover's vbias section and take one from either ET, and if i did, would it be preferable to bias the other ET from that one or could i just have ET+xover and ET, if that makes sense? i guess all it'd involve would be pulling the 4.5v side of the 1M resistor in the ET that's to be synced & hooking that up to the one above it.

i'm confident this will sort some stuff out, particularly on the compressor-return side. i'm still a bit apprehensive about the first section, where the tone splits in the first place. i'll have another look around the frequency pot but i've spotted nothing obvious so far (i think i reversed the wiring right, since you mentioned the pot was backwards in the schem).

[samhay]

I just saw the picture in your previous post. Those diodes aren't going to help the cause - get rid of them.
If you are going to DC couple everything, then a single source of Vbias is probably the best idea. However, if you are going to have 2 or 3 sepearate boards, then seperate Vbias' and AC coupling between the stages might be a bit easier.
If you are worried abou the frequency pot, then take it out and replace it with 2 e.g. 10k resistors for testing.

[MrStab]

it was a really sketchy fix, now that i plan on reconfiguring the bias all across the board, i'm just gonna separate and fix that area as it should be. the pot still reads as working on a multimeter on both sets of lugs. the whole bias issue around pin 5 does affect U1A/U2A though so maybe the problem is all related.

...i just realised both R8 and R9 might be connected to vbias. damnit. in a rush so cant double check yet tho

[MrStab]

hi, i've fixed up a load of stuff and i'm reporting in.

what i've done tonight (if 5am counts as "tonight"):

• synced all vref to that of one of the compressors. seems to work ok. ~4.3V readings over all 3 sections (this seemed to load down the circuit from my power supply's solid 9V to about 8.7)
• replaced all 12k resistors with appropriate 10k resistors
• rewired frequency pot as it was in the schematic again (ie. backwards), in case i'd gotten it wrong
• fixed signal between U1A and U2A
• ditched all the diode nonsense

i get a working output, variable by the frequency pot, on the High Pass output, but only when the TL074 isn't in the socket properly. the signal becomes unaffected by the pot if the opamp is pushed in all the way. intuitively, i'd think the socket's just not in right - but there is continuity, so it seems (to me) that either opamp U1A or U1B (both at the top half of the IC) is preventing U2B from doing its job. absolutely no signal on the Low Pass output.

U1B's voltages are up at about 7-8V. that particular opamp has no connection to vref. i currently have two TL074s at my disposal, both repeat the issue. any ideas? i'll have another look tomorrow.

it's progress, at least - i've synced the vref, put in the right resistor values and cleaned things up a bit, so here's hoping!

cheers

[samhay]

U1B is DC coupled to U1A via the frequency pot. The voltages on U1B suggest this pot may still not be wired correctly. Try removing it and replacing it with 2 10k resistors or 2 100 k pots wired as variable resistors.

[MrStab]

i've tried two separate 100k pots with the same result unfortunately. i cleared every trace of solder off the lugs of the dual-gang pot and double-checked resistance last night, so at least the pot itself seems to be working. does this look right?



i'm just arty like that.

[samhay]

If that's the shaft coming out the top and the lugs toward us, then that's correct if you are using an A100k pot. If you are using a linear or reverse log pot, then wire it the other way round.
If that's wored correctly, then can you post the latest layout?

[MrStab]

i forget this is a three-dimensional world - the lugs are indeed facing towards us. i had it wired round pretty much a mirror image of that, but that was before all the fixes. it's still just a linear pot - i'm aware of the taper issues we discussed previously, but i just set it back to this in case i'd screwed up the reversal. should just be temporary - unless maybe the taper itself could be somehow causing the no-change issue with the opamp fully-in?

DIY Layout Creator and/or my perpetually-small C: drive have corrupted my layout, but a fair bit has changed since the last iteration anyway - i'll have a new one drawn up within the next hour or so

[MrStab]

right, i think this is more or less how i have it now. there are a coupla sketchy bits as a result of trying to salvage what i already had, but the real thing actually looks a bit neater now. there are some leftovers from where the old voltage divider used to be.



R1 = 10M
R2 = 1K
R3 = 1M
R6 = 1K
R7 = 1K
R8 = 10K
R9 = 10K
R10 = 10K
R11 = 10K
R15 = 1K
R34 = 10k
R35 = 10k
R36 = 10k
C1 = 100nF
C4 = 22nF
C5 = 10nF
C10 = 100nF
C13 = 100nF
C14 = 100uF
C15 = 100nF

[samhay]

R10 should go to the (-) input of U2A, not the output.
R34 should go to the (-) input of U2B (along with R35) and the (+) input should go straight to Vref - the 10k R is not necessary.
You don't have an output cap.
C15 is probably not helping - op-amps don't like directly driving caps to ground.

[MrStab]

damn, i forgot to add the output cap to the layout - there is one, it's a 100nF. i'll correct R10 just now.

re. R34 and R35: that's the mixer setup i used originally, which didn't work, so i went with the wiki schematic posted earlier. that's what prompted the whole shake-up (as well as the frequency pot not working well)

[samhay]

Good luck, and I hope it doesn't keep you up to 5am tonight (unless you get it working and spend the hours noodling).

[MrStab]

lol i got a noise complaint last week for working on this at 4am and accidentally letting oscillation run through a 100W combo. headphones from now on. i'm usually up late anyway.

changing R10 didnt seem to do much, didn't bother with R34/35 yet as i wanna get the crossover itself working first. High-pass output works as ever, low-pass is still silent.

with the opamp fully-in, there's now a difference in the rotation of the pot, but it's very slight compared to when it's sticking half-out. also quite quiet, not unity.
my vref has screwed up down to 1.7V for no clear reason, i'll look into that. i can imagine that affecting the intensity of the frequency pot, but i'd still expect some signal coming out of low-pass. it's not that complicated/populated a section, so i'm really unsure what's going on there.

i have had bad luck with killing opamps lately, but the odds just don't seem to stack up if the inconsistency is across 3 of em.

[PRR]

In a parallel universe.....

In the 1980s (and before), spectral compressors became a favorite tool to pump-up-de-volume in AM and FM radio.

Inovonics was a strong player, and they have plans for their 1987 "255 :: Triband / PWM FM Processor" online:
http://www.inovonicsbroadcast.com/model/255
http://www.inovonicsbroadcast.com/wp-content/uploads/catablog/downloads/255_Manual.pdf

Note: simple fixed single-pole (6dB/oct) band splitting, even though super-steep would add little to the price.

True, this is not an artist's tool. It has to make everything sound better, 24/7, without constant human intervention.

[MrStab]

that seems like some heavy reading for a sub-novice such as myself, but despite the mixed success of this project so far, it's been interesting reading up on the theory in-between attempts. i guess the general musician-come-techie approach is taking these clinical tools (eg. for broadcast/mastering use) and applying some "i wonder if...". that thing you mentioned about that 700Hz (iirc) sweet spot in dividing the spectrum encouraged me to at least keep trying, Paul.

at one point i had a really dysfunctional, noisy signal coming through the whole setup (before getting my arse into gear to tidy things up), yet there was something captivating about what little guitar i could hear. just wish this project hasn't been plagued with loads of what should be trivial issues. here's hoping.