Square wave bleed through into noise gen

[etaoin]

Hi all

I've recently finished my Soundlab and it appears to be working well, apart from a problem with the noise generator. I get bleed-through from both the LFO and VCO2 square wave output into the noise generator.
Check this for a sample: http://www.dompselaar.org/Soundlab/noiseproblem.mp3
This is plain noise output with LFO output at zero. First you hear a slow LFO which I then turn up and about half way you hear the VCO2 frequency being turned up.
If I pull out IC2, VCO2 bleed-through is gone, so it's only the square wave that is causing problems. Also, there is nothing from VCO1 in there, nor from the AR.
The oscillators themselves appear to be fine. I can even get them tuned over an octave or two without any temp compensation.

Some notes about my build that may be relevant: I've used TL072 and TL074 op-amps (they draw more power - can that be the problem?). I needed to replace 2M and 3M resistors with two and three 1M in series because I couldn't get 2M and 3M; the 500 ohm ones are two 1k parallel for the same reason Likewise, 2uF capacitors are two 1uF parallel (all of the correct type though). All other components are as specified. Some resistors are 1% where only 5% was needed. I think I've checked the wiring and component values at least half a dozen times and the batteries are brand new Duracell Plus.

Any suggestions as to what might be causing this bleed-through?

A related issue (possibly?) might be that I think the LFO and AR sweep range of the filter is quite small, but as I cannot compare it to anything, it might be that it's just as it should be.

[etaoin]

I should add that I see about 20mV drop in the +9V over an LFO cycle. Someone suggested that this would be caused by the high current TL07x combined with the high internal resistance of the batteries and that I should use TL06x instead (which use one tenth of the current), but Ray says TL07x should work fine. The -9V is stable.

[Wild Zebra]

I know this will probably be redundant and not really advice, but check for a solder bridge on your traces or maybe a little jump in your wiring. I had a little trouble with my noise source and I popped in another tranny and it started to work fine. There doesn't seem to be anything wrong with your subs. I can't speak for your chips though. hopefully smeone else will chime in.

[etaoin]

I just tried the lower power TL062 and TL064 replacements but (as expected by Ray) they don't make any difference at all. The hunt goes on...

[Wild Zebra]

When I first got my Soundlab running I had the same problem, but more with the LFO. I changed the tranny and it started to work fine. Well last night when I was messing with the fine tuners I noticed it was sort of doing it again. I tried swapping trannys and some were worse and some were better. The weird thing was with the power on(yeh I know this probably isn't good) I was swapping trannys and if there wasn't one in the socket I could here the bleed through. I checked all over and saw no bridges or could solder. So I don't know if its just like that or what. Mine is tolerable not much bleeding. I guess what I'm trying to say is get a handful of 2n3904s and experiment. Sorry for the long post.

Let me know how it goes because now I'm wondering

[etaoin]

I tried about 15 different 2N3904 for the noise. Some were more noisy than others, but it didn't affect the amount of LFO and VCO2 getting into the noise amps at all.

I also tried removing the LFO LED without any success.

Also, I purpose built a 500 mA +/- 9V mains power supply to see if maybe the batteries are the problem, which it appears they aren't. It's slightly better on mains, but only slightly.
Obviously, it [b]is[/i] a power problem. Last time I checked my Soundlab drawed around 35 mA and in another thread on this forum I read it should be around 10 mA. Although I haven't checked after I swapped my TL07x with TL06x.

The only thing that definitely removes VCO2 from the noise is removing IC2. Since LFO and noise share IC7 I can't perform the same check there.

I'm moving house next week so the Soundlab is on the shelf for a bit (or rather, in a box). Will pick it up again after that.

[Wild Zebra]

Sorry to hear that. I know how frustrating it can be. I build a Rat Distortion box for a friend. Worked great so another friend wanted one.
Even gave me the money up front. Its done, but for the life of me I can't figure out whats wrong with it and its really p@ssing me off. Its usually
something really simple in the end.

[mosc]

Oh, boy, sometimes these things can be very difficult to cure. Sounds like you are on the right track for a methodology, try everything.

One thing I noticed once may not have anything to do with your signal leakage problem but interesting nonetheless.

I built a ring modulator using the AD530 chip. There was a professionally made circuit board with all the electronics, including the power supply and transformer, and a front panel for the jacks and knobs. The grounds for all the jacks and knobs were all bussed together and one wire connected the panel ground to the circuit board. I connected it as close to the power supply as possible.

Well, there was a lot of noise - hum some would call it. To make a long story relatively short, I discovered that if I moved the point I connected the wire from the panel ground to the board, the noise would be reduced. By trial and error, I could get the noise to almost completely vanish. I couldn't find a good explaination of what that point had the minimal noise, but I used it anyway obviously.

I made twelve of these boxes to sell. On every one, the point of attachement of that ground wire was different.

[etaoin]

Since a picture says more than a thousand words (or so they say), I thought I'd try and grab one from the scope. The top one is the +9V supply at 20mV/div, the bottom one is the LFO at 5V/div. You can clearly see what I mean here with drops in the power supply. It is these drops that are amplified and made audible in the noise generator. Now the question remains... as you can see, the LFO looks near perfect, so why would it pull down the +9V?

http://www.dompselaar.org/Soundlab/slwaves.html

[etaoin]

By the way, is there anyone out there with both a fully working SL and a scope who can tell me if the LFO square wave is supposed to look like this, or if that may be a sign of what's wrong with mine?

[etaoin]

It's a Sign that tells me to switch the scope to DC coupling

[Wild Zebra]

I got fully working SL part, but no scope. I have a friend who had one I'll see if he still does and borrow it.

[etaoin]

Wild Zebra. Thanks, but changing the scope to DC coupling was the solution. That show's the square waves as they should look. So it wasn't a problem related to the SL, just my ignorance of scopes in general

[Wild Zebra]

Oops missed that little thread there. I'm pulling for you, man. I sure wish you'd get that thing running right. Its alot of fun. I still feel like I havn't unlocked its full potential.

[etaoin]

I sure hope I can find it. I have all the parts for the S&H ready to go as well, but since that requires a decent noise source........

[Wild Zebra]

I recently got my S&H working, but it starts to not work as soon a my batteries die a little. I always buy cheap batteries though so I'm thinking I need to start getting some higher quality ones. Usually first sign of dying batts LFO goes out. I don't know if thats an issue with anyone else.

[etaoin]

I've been using the same alkalines since I started. They seem to last pretty long. I also did a small +/-9V mains power supply to power the SL (based on the 12V supply diagram on the musicfromouterspace website), however, since that introduces a slight hum into the sound, I usually keep it on batteries.

[Wild Zebra]

Yeh, I like the batteries. That way I can lay on the couch like a lazy American and play with the SL. I don't know if you guys have these over there, but here we have Dollar Stores where like it says everything costs a dollar. Usally overstock or common household items. Thats where I get my batteries and I have a feeling there old.

[etaoin]

We have a chain of "Euroland" stores, which are probably much the same.

[v-un-v]

[etaoin]

v-un-v: it was just the scope that was connected through the wrong coupling. And I run my SL on mains power.

[v-un-v]

okay :)

[etaoin]

Well, after a moving home and building an x0xb0x (which worked 100% from first power-up ) I decided to give the SL another spin. As I cannot find anything wrong with either the LFO or VCO2 (despite their bleeding about 20mV of noise into the +9V rail), I decided to check the noise source more thoroughly. I found that where the instructions say there should be at least a couple of volts above ground at the emiter of the noise transistor, mine actually has about -1.5V there. Can that be the clue I'm looking for?

[Uncle Krunkus]

I reckon it is. Mine has the same problem, but I haven't checked the voltages yet.
For what it's worth, here's my theory.

My thoughts were that maybe the de-coupling of the noise transistor and/or the LFO and OSC2 isn't sufficient. If you think about it, the LFO and the SQR wave on the Osc2 both being the only square wave sources in the circuit would both induce spikes in the power rails because the square wave transients are fast voltage changes. Now if even a tiny spike gets into the power rails it could contaminate the noise source as the noise transistor is right across the +-9V lines (thats what breaks down the transistor junction to produce the noise)
So, I think that maybe what we need is better de-coupling on either the LFO, OSC2 or the Noise transistor or all three.
I haven't run this theory past Ray yet but I think I will.
It's possible to check it out by connecting a .1uf greencap across the +ve input on a chip and it's ground, and another one from it's -ve (if it uses one) and ground. Do it as close to the chip as you physically can. You can't do any damage by trying it out. This is what I'm going to try as soon as I get some parts together to add some other stuff while I'm in there.

[adhdboy]

The Sound Lab Mini-Synth board contains 4 caps for bypassing the power
supply. Two larger electrolytics, C16 and C18. And two smaller ceramics (or equivalent) C19 and C20. Adding more bypass caps is never a bad idea and using alkaline batterys will lower the power supply impedance as much as possible. This will help to reduce bleed through if the power supply is the problem.

Wire dressing can also contribute crosstalk since wires that are tie wrapped together over any length act as if a small capacitor were connecting them together. The longer the run the higher the capacitance and greater the chance for bleedthrough. This will be in the pF range though and only fast edged signals (square waves or high frequency) would bleed through. So you might check any bundled wires and unbundle them to see if that helps.

If the bleedthrough is in the millivolt range I suspect the above culprits.
However if there is significantly more than that I suggest that the panel
wiring be thoroughly checked. Missing a ground wire will wreak havoc as far as signals showing up where you don't expect or want them.

If you add more bypass caps add them carefully to the copper side of the
board between +9V and ground and -9V and ground dispersed throughout the board. In the end the Sound Lab is only going to be so good and it will
never be perfect. Its a small project to have fun with and will never
replace its far more elegantly designed (albeit expensive) commercially
available modular big brothers.

Cheers and I hope this helps.

Ray