LFO Clicking in a Tremolo

[kforte318]

Hey guys,

Firstly, I'm new to the forum and apologize if this is the wrong place to post!

I'm working on a tremolo at the moment with a pulse/tri LFO (the "Nicholas" LFO modified for my power needs) modulating the gain of a 386.

http://i.imgur.com/CtZfoaO.jpg?1

Here's my schematic thus far. In this (http://www.diystompboxes.com/smfforum/index.php?topic=62637.0) post, smallbearelec suggested keeping the grounds separate and only bringing them together at the negative terminal of the power source. How would I go about doing that?

[MrStab]

consider this: would you rather all your sewage go downstream into the same water people drink from, or would you rather they both go out to sea separately?

battery negative, in this instance, is the sea. basically, every ground to do with the LFO (LM324 etc.) should follow its own path, and never meet any grounds to do with the signal path (eg. VCA chip ground) until the negative terminal of the power supply. they both converge there.

BUT: you seem to be using the same Vref for both the LFO and the signal. you should make a separate Vref voltage divider for the LFO, because in a single-supply circuit, that's basically ("virtual") ground. you should stick a 47uF or so cap on either Vref, too, while you're at it.

[duck_arse]

like mrStab says, kforte, "hello and welcome". he usually leaves the sewerage to the second post, but nevermind.

I don't think I've ever seen a 386 ldr tremolo circuit before. nice one (if you can rid the tikkks).

[edit :] to your problem. if on the breadboard, build the 386 part on the left hand end. build the 324 on the right hand end of the bb. now run the V+ and ground wire from your power source to the middle of the BB, in between the two sections. done.

same basic idea applies to strip/vero board. and perf. and pcb.

[MrStab]

apologies for the lack of welcome message, wouldn't want to compromise the friendly atmosphere

[nick d]

       I have encountered this problem a few times , the cure - buffer the vref with an op-amp .

[kforte318]

That's okay MrStab, I usually prefer to be greeted with sewage talk. Makes me feel at home. You guys taking the time to help is enough of a display of friendliness, but thanks!

So basically duck_arse, should I not be using the power + and - buses on the breadboard? I will also have to try your suggestion, nick. I think I'll be adding an op-amp to this circuit anyways to get the LFO signal amplified a bit, so I can just make it a dual to include the buffer. Going to try to switch the VCA portion to a transistor-based VCA.

[MrStab]

FWIW, i was loosely referencing what RG Keen (a guru here) often calls "sewer ground". it wasn't such an arbitrary gross analogy. blame him!

[kforte318]

No worries, I actually really liked the sewer analogy. Very helpful, especially given the many analogies between water and electricity I've already read.

[duck_arse]

FWIW, i was loosely referencing what RG Keen (a guru here) often calls "sewer ground". it wasn't such an arbitrary gross analogy. blame him!

whereas, when prr talks sewers, he means sewers.

use the bb busses, yes, but put the wires to your supply into the bus at the midpoint of the board, so it goes:
lm386 end < supply wires to bb buses > lm324 end.
that way you'll form a resistor on either side of the supply wire, and current to one side won't appear in the current to the other side, more or less.

you could also try an R/C filter (? 100R and 47uF ?) in the positive line to the lm324 end.

oh, and the ldr might be better off with a linearising resistor wired in parallel, ? maybe 220k ? mrStab will know ......

[MrStab]

whereas, when prr talks sewers, he means sewers.

i have to say, i always find Paul's stories of having to manually wire and plumb his gaff quite interesting.

i generally avoid LDRs, but as illumination:resistance seems pretty linear to begin with on LDR datasheets, i'd imagine you'd end up with a logarithmic taper by paralleling a fixed resistor. that would alter the waveforms you're trying to put into it.

if attenuation is a problem:

you've probably noticed the square ("A") output is considerably stronger than the other waveforms (by 2/3rds usually, i think?), and as you said you might wanna amp those up. unless i'm looking at the LFO wrong, which is likely. if you use an additional IC to amplify instead of just using a different LFO, treat that with the same V+, ground & bias precautions as the existing chip.

maybe it's as bright as it'll go already, but perhaps driving the LED via. a NPN would work better and be less disruptive to the power supply. there are other considerations with that, though. similarly, if you decide to switch to a FET to control the signal, then you'll need to consider the not-so-linear nature and thresholds of transistors compared to LDRs. it's also worth considering that maybe power amp chips aren't really used in tremolos for a reason - i honestly don't know enough about the chip (or anything ever) to comment.

+1 on Duck's decoupling of the LM324 (the RC filter), re. the tick. i'd do that with the LM386, too (i see you already have the 220uF cap).

i'm a bit sleep-deprived atm, apologies for any more assumptions or misinterpretations than usual.

[GibsonGM]

FWIW, i was loosely referencing what RG Keen (a guru here) often calls "sewer ground". it wasn't such an arbitrary gross analogy. blame him!

you could also try an R/C filter (? 100R and 47uF ?) in the positive line to the lm324 end.

oh, and the ldr might be better off with a linearising resistor wired in parallel, ? maybe 220k ? mrStab will know ......

Provided you have SOME kind of star ground pattern going on...and NO LFO wires pass near signal wire runs...^^ This is often a very useful trick to lose the tick!!!

I have used 100R and up to ONE THOUSAND uF, in an Easyvibe that wouldn't cut the cr@p.    I would expect that something built better than that one would not need so much capacitance, tho, LOL.  DECOUPLING rules, it's easy to omit it but then we pay for it.

[anotherjim]

I question the 680ohm resistors driving the LED -  they are probably a bit too low in value causing high currents which will make ticking more likely and possibly saturate the LDR with too much light on the positive half of the LFO cycle. Maybe (guessing), a 1k trimmer (but it might need more) in series with the LED will help improve? Incidentally, as either LFO waveform each use 680R, then only the one 680R could be used on the LED side of the switch.

Nothing to do with ticking, but the 100uF size output cap is probably for speaker driving duty. If it's used with an amp, it could be much lower in value - say 100nF.

The 386 isn't really an op-amp, so probably doesn't have the supply noise rejection of a proper op-amp, so what's been said about separate power filtering should be acted upon.

The switch is carrying the square wave output of the LFO. Square waves have very fast switching which corresponds to radio frequency content. This radiates from the wires and gets picked up by the audio circuits. Keep those wires short. Use screened cable for the audio connections. Ideally, the switch would be board mounted so not requiring any wires running off the board.

[kforte318]

I am going to give all of these suggestions a try once I get the chance, and I am doing my best to keep up just reading them all (I am VERY new at this, and though I have been trying to do my fair share of theory reading, it isn't all immediately sinking in and sticking). Thanks for all of the continued help!

I actually bought an op-amp today to use instead of the 386, so I'm glad to see that might help.

[duck_arse]

I vote stay with the 386. it's about time it was put to something usefull. and I'd hate to think a dozen of us were putting this on our bredbords and you went and changed the design on us.

also, add the bypass cap to the bypass pin, pin 7, as noted in the datasheet.

[anotherjim]

This circuit is using the 386 gain adjust pins 1 & 8 to wobble the volume via the LDR. An op-amp does not have this feature. It would need a completely different circuit with an op-amp.

[duck_arse]

I hadda go at it tonight, with a single op-amp (lm358) triangle osc (you all know the one) and a led-in-the-loop led driver. the led part would have benefited from a square-wave instead I think, trem didn't get very deep, but it did work, without any tikkkk, with no particular precautions against.

and you can tune up how much distortion you want, kinda.

[kforte318]

That's awesome that you got it to work! Let's hope I have the same luck! Another question, too: I also noticed that the LFO signal was a little weak and the effect wasn't too too noticeable. Am I correct in assuming I can simply amplify the LFO signal a little to increase the depth of the effect?

EDIT: I have it working with very little clicking! However, I'm really not happy with how this sounds to be honest. Unless I can get the change to be more drastic, I might ditch the 386 and the gain-modulating idea for a simpler VCA.

[duck_arse]

the problem w/ the lm386 is that the internal resistor being bypassed is only 1k35, so you need (probably) a very low R ldr to get the big output when bright. I have a pair of different-type ldr's looking at a single led in a 35mm film can for testings. when I wired them in parallel, I got 'better' tremming.

you could, I suppose, try and wire an npn transistor in place of the ldr, and switch the base with the lfo. it might not work, it might need plenty fiddle, and it might only do chop, but it might not hurt to try, either.

also, have a try with the lfo used in things like the tremulous lune or the cardinal, and look at the pulsar or schaller trems for transistor switching ideas. maybe.

[Kipper4]

Check out the shoot the moon trem too

[kforte318]

I really need to get better at reading schematics. I can identify pieces here and there in a full schematic at times, but it's usually overwhelming. I did get a VCA set up in the sim that seems to work pretty well (you have to manually move the "LDR" pot to simulate it), but I'm worried that the LDR just doesn't have the attack time to make it sound good. Breadboarding now...

http://www.falstad.com/circuit/#%24+1+5.0E-6+16.817414165184545+66+5.0+43%0Aa+192+64+304+64+0+9.0+0.0+1000000.0%0Aw+192+48+192+16+0%0Aw+192+16+304+16+0%0Aw+304+16+304+64+0%0Aw+192+80+192+192+0%0Aw+192+192+208+192+0%0Aw+240+192+208+192+0%0Aw+208+272+240+272+0%0Aw+192+192+192+288+0%0Aw+240+272+304+272+0%0A174+304+128+320+224+0+100000.0+0.7871+Rate%0Ar+320+176+384+176+0+10000.0%0Aa+384+192+480+192+0+9.0+0.0+1000000.0%0Aw+384+208+384+272+0%0Aw+304+272+384+272+0%0Aw+384+176+384+128+0%0Ar+384+128+480+128+0+100000.0%0Aw+480+128+480+192+0%0A174+512+160+480+224+0+100000.0+0.5+Shape%0Ad+512+160+576+160+1+0.805904783%0Ad+576+224+512+224+1+0.805904783%0Aw+576+160+576+224+0%0Ar+576+160+640+160+0+1000.0%0Aa+640+176+720+176+0+9.0+0.0+1000000.0%0Aw+640+192+640+272+0%0Aw+640+272+384+272+0%0Aw+640+160+640+112+0%0Aa+720+192+800+192+0+9.0+0.0+1000000.0%0Aw+720+208+720+288+0%0Aw+192+288+720+288+0%0Aw+800+192+800+288+0%0Ar+800+288+720+288+0+22000.0%0AR+640+272+672+272+0+0+40.0+4.5+0.0+0.0+0.5%0Aw+720+112+720+144+0%0Aw+720+144+720+176+0%0Aw+192+80+160+80+0%0As+160+80+128+80+0+1+true%0Aw+112+80+128+80+0%0AR+112+80+96+80+0+0+40.0+2.0+0.0+0.0+0.5%0Ax+803+66+833+69+0+12+Pulse%0Ax+806+133+822+136+0+12+Tri%0Aw+368+64+784+64+0%0AS+848+96+784+96+0+0+false+0%0Aw+784+64+784+80+0%0A162+208+272+208+192+1+2.1024259+1.0+0.0+0.0%0A162+240+192+240+272+1+2.1024259+1.0+0.0+0.0%0Ac+640+112+720+112+0+1.0E-6+-1.4238095027726096%0Ar+848+96+912+96+0+10000.0%0Aa+912+112+1008+112+0+9.0+0.0+1000000.0%0Aw+912+96+912+64+0%0Aw+1008+112+1008+64+0%0Ar+912+64+1008+64+0+22000.0%0AR+912+128+912+160+0+0+40.0+4.5+0.0+0.0+0.5%0AO+1008+112+1040+112+0%0Aw+288+80+288+224+0%0Aw+288+224+304+224+0%0Aw+304+64+304+80+0%0Aw+304+80+288+80+0%0Aw+368+64+304+64+0%0Aw+720+112+784+112+0%0Aw+1008+336+192+336+0%0Aw+192+336+192+560+0%0Ar+192+560+304+560+0+10000.0%0At+304+560+368+560+0+1+-6.073819902729687+0.5677074008478321+100.0%0Ag+368+624+368+640+0%0A162+368+464+368+544+1+2.1024259+1.0+0.0+0.0%0AR+368+464+368+416+0+0+40.0+9.0+0.0+0.0+0.5%0Aw+1008+112+1008+336+0%0Ar+368+576+368+624+0+1000.0%0AR+288+464+240+464+0+1+200.0+5.0+0.0+0.0+0.5%0Ar+288+464+352+464+0+10000.0%0Aw+352+464+432+464+0%0Aw+432+544+432+624+0%0Aw+432+624+368+624+0%0AO+608+464+640+464+0%0Ax+385+510+417+513+0+12+VACT%0A174+432+544+448+480+0+10000.0+0.9950000000000001+LDR%0Aw+432+464+448+464+0%0Aw+448+464+448+512+0%0Aw+448+464+608+464+0%0Ao+53+64+0+34+12.528092670535079+6.103515625E-5+0+-1%0Ao+74+64+0+34+4.676805239458889+9.765625E-55+1+-1%0A

It works pretty well! No click, and it sounds better than the gain modulation. Now I just need to figure out a gain stage. Thanks for all of the help!