The Unknown Spring Reverb

[razabri]

perhaps that "electric noise from your surroundings" is because the board itself isn't in a shielded enclosure yet...

Guess that's very possible too - I really don't know how this 1904 is going to behave, once when it's all mounted. I know that lm386 can pick up various noises and even radio signal (I had this issue with an Acapulco Gold build). Perhaps I could make a sort of a cover out of tin for the circuit, if necessary.

[11-90-an]

>Also, I'm not sure about what are you saying about the 100ks' in series and trimpot of 220k

Rob means changing the value of this resistor to 220k (circled in red)

[Rob Strand]

Guess that's very possible too - I really don't know how this 1904 is going to behave, once when it's all mounted. I know that lm386 can pick up various noises and even radio signal

If you get radio and some forms of crackle adding judicious place of some ceramic caps can help.    For the reverb spring you can try adding say 1nF across the input of the recovery amp.  Maybe best done by soldering a cap on the underside of the PCB.

One thing really asking for trouble are those flying leads on the output side of the reverb tank.     At minimum you should twist these fairly firmly.     

Accutronics reverbs have very short wires.  They then go to RCA jacks, some are insulated RCA's.   From the RCA's shielded wire is used to connect to the electronics.  It's not so important for the drive side but the recovery side is very important.   You can see the short leads in this pic,



In addition to that, output side of the tank which connects to the circuit ground at the recover amp is normally connected to the outer metal casing of the tank.  That lets the metal casing act as a shield.  The way that is done is by using a non-isolated RCA jack.  For tanks without RCAs it is done with a small wire link or folded metal tab which goes to the chassis.

IMHO these measures are important to fend off noise on reverbs.   The signals are very small so you need to put in as many counter measures as possible.

As for noise in the form of hiss.   You would need to redesign the recovery amp to do any better.  It shouldn't be too bad.   The inverting op amp used on the ELKA isn't the best choice for low noise.     You can shave off a small amount of noise by low-pass filtering the output of the recovery amp.   That's done by selecting adjusting the cap across the outside of the 220k trimpot + 100k resistor; essentially across pin 6 and 7 of the opamp.    Current that value is 47pF which sets the low-pass filter frequency to be
flp = 1/(2*pi * 47pF *(220k+100k)) =10.6kHz to 1/(2*pi * 47pF *(100k)) =34kHz depending on the pot setting.  Given we gain is set high at the moment it's probably more like 10.6kHz.    If you wanted to shave a small amount of noise you could use say 100pF.
(Keep in mind if you increase the 100k resistor to get more recover gain you need to redo the calcs and that will require a smaller cap for the same filter cut-off.

Rob means changing the value of this resistor to 220k (circled in red)

That's it, thanks.

A big thanks to 11-90an for doing that schematic.   There's *no way* we would have got this far without it.

[11-90-an]

>A big thanks to 11-90an for doing that schematic.   There's *no way* we would have got this far without it.

It was fun to do, to say the least... 

what if we also add a simple R/C LPF after the output?

[Rob Strand]

what if we also add a simple R/C LPF after the output?

No problems doing that.  The RC LPF and the adding the feedback cap on the recovery amp both achieve the same result.

Normally the feedback cap has the advantage that you don't need to add a series resistor to the output
or the opamp.   However, in our case we have already added a 1k resistor.  The LPF at the output avoids
the filter cut-off changing with different trimpot settings so the LPF might be a better solution.

[razabri]

I have now successfully mounted the thing inside and it is fully functional! Tapped into the INSERT jack from the inside and took the power at the small cable connecting the power board with the amp board and it lit up nicely! Not to mention that a placing of the effect was as if it's built for it - just right above the speaker, with two screws through already made holes on the chassis. 

No more of that noise either, and yes, it was a hiss sound before, but I think it's due to the adapter used (from an old modem). Now, I do have some noise in the reference to the clean sound, but it seems to me that it's kind of normal, as I remember that when you turn the reverb knob on amps that actually have them, it gets a bit hissy. The amp itself is not that quiet and I can't seem to find its signal to noise ratio, but my guess is that it's somewhere close to >50dB for the overdrive, >65dB for the clean, as that's how they rated it with the other Crate amp that I have, and there's pretty much the same amount of noise when turned on.

Also, I did that for the shielding/grounding - a ground pin is connected to the metal chassis of the effect and it does prevent additional noise that was happening.

I have tried to connect 22k pot as the output, but it didn't wok well and I also tried 1k, as I had it at hand, but that too didn't make a good match - I guess it's that 10k would be just right. Oh, and I don't think we really need to amplify the return signal, it sound good to my ear like it is. I'll just get that 1M input resistor and set the trimpot right, I guess.

Another thing is, there's a big reverb splash when the amp is turned on. I mean, I know it's sort of normal, but maybe it could be quieter. I thought that I was doing it as I'd hit the switch and maybe disturb the springs, but no, it's still loud even if you do it real gentle.

Oh, and I'll be saving that circuit drawing, of course, and the whole discussion, as when I understand it better, I'll surely get back to this one and tweak it some more, maybe even include that 3080 somehow 

Thank you guys, you've been so great about it all and the amp now sounds way more inspiring to play, as before it was just too dry. Cheers!

[11-90-an]

Cool! As I said before, try removing the LM3080, as you can use it on other projects, like an Engineer's Thumb compressor... 

[razabri]

Cool! As I said before, try removing the LM3080, as you can use it on other projects, like an Engineer's Thumb compressor... 

Hey, that's exactly the one compressor I've considered building! Perhaps I'll give it a shot one of these days. Love the guy who's demoing it embedded at tagboard effects, great insight and he made me want to build it, although I've never really used one.

I just got me some 1M resistors, so I'll stick one in there, see how it goes...

[Rob Strand]

I have now successfully mounted the thing inside and it is fully functional! Tapped into the INSERT jack from the inside and took the power at the small cable connecting the power board with the amp board and it lit up nicely! Not to mention that a placing of the effect was as if it's built for it - just right above the speaker, with two screws through already made holes on the chassis.

Excellent!

No more of that noise either, and yes, it was a hiss sound before, but I think it's due to the adapter used (from an old modem). Now, I do have some noise in the reference to the clean sound, but it seems to me that it's kind of normal, as I remember that when you turn the reverb knob on amps that actually have them, it gets a bit hissy.

If you trim the low-pass filter cap to have a cut-off off 6kHz or so you can easily gain 2dB in signal to noise.  There's virtually no effect on signal quality.   It's good value IMHO.

I have tried to connect 22k pot as the output, but it didn't wok well and I also tried 1k, as I had it at hand, but that too didn't make a good match - I guess it's that 10k would be just right. Oh, and I don't think we really need to amplify the return signal, it sound good to my ear like it is. I'll just get that 1M input resistor and set the trimpot right, I guess.

Not sure why the 22k would affect things so much.   All I can suggest is once you pick the reverb pot 22k or 10k you need to adjust to 100nF cap at the output of the opamp to trim-up how much low-end you want from the reverb.   Perhaps you should try a 220nF, or add another 100nF in parallel with the existing 100nF (giving 200n in total).  It's easy to do A/B testing with the parallel cap.

With a 1k pot you are going to cause the filter cut-off to go way-up and start cutting considerable bass.   f3 = 1/(2*pi*2k*100n) = 800Hz; it's 2k because of the 1k at the output of the opamp.    Also, the 1k pot starts to load-down the 1k in series with the opamp output so it cuts the signal by half.   The combination of the 10k pot and 1k resistor on the opamp output doesn't attenuate much at all.

Another thing is, there's a big reverb splash when the amp is turned on. I mean, I know it's sort of normal, but maybe it could be quieter. I thought that I was doing it as I'd hit the switch and maybe disturb the springs, but no, it's still loud even if you do it real gentle.

That happens because the TDA1904 is a single supply amplifier.   Some modern power amp chips have a "soft start"/power-on mute feature.  It's hard to fix without adding more circuitry.

If you look at the schematic there is a 47uF cap on pin 2 of the TDA1904.   Basically it's across the power rails.  If you increase the value of that cap to say 470uF or 1000uF it might soften some of the "bang" when you power-up the circuit.  You can only try it.

[razabri]

Tried that 1M at input today, didn't really like how it was going - it was as if something changed as of yesterday and the device started to easily make a feedback, the whistling sound that would go forever and that didn't work well with adding that 1M, as I had to increase the output on trimpot, which made it worse, so I got it back to the 560k and to my ear, with the trimpot set to half (as it was before I even touched it), sounds good. Still clueless about why would it make a feedback like that, too much of something... When I'd touch the springs it would stop, as if they were reacting to something and vibrating, producing the whistle....

If you trim the low-pass filter cap to have a cut-off off 6kHz or so you can easily gain 2dB in signal to noise.  There's virtually no effect on signal quality.   It's good value IMHO.

This sounds good to me, but still my brain struggles with what it actually mean in praxis. 

Not sure why the 22k would affect things so much.   All I can suggest is once you pick the reverb pot 22k or 10k you need to adjust to 100nF cap at the output of the opamp to trim-up how much low-end you want from the reverb.   Perhaps you should try a 220nF, or add another 100nF in parallel with the existing 100nF (giving 200n in total).  It's easy to do A/B testing with the parallel cap.

With a 1k pot you are going to cause the filter cut-off to go way-up and start cutting considerable bass.   f3 = 1/(2*pi*2k*100n) = 800Hz; it's 2k because of the 1k at the output of the opamp.    Also, the 1k pot starts to load-down the 1k in series with the opamp output so it cuts the signal by half.   The combination of the 10k pot and 1k resistor on the opamp output doesn't attenuate much at all.

I'll try and get the 10k then - as I understand you'd recommend the use of the pot to set the effect output, rather than adjusting the input and trimpot onboard, and just do the combo of the 1M input resistor and crank the trimpot all the way?

That happens because the TDA1904 is a single supply amplifier.   Some modern power amp chips have a "soft start"/power-on mute feature.  It's hard to fix without adding more circuitry.

If you look at the schematic there is a 47uF cap on pin 2 of the TDA1904.   Basically it's across the power rails.  If you increase the value of that cap to say 470uF or 1000uF it might soften some of the "bang" when you power-up the circuit.  You can only try it.

This worked like a charm! I had 470uF at hand and with it the splash is far more tolerable! Thanks!


Just another thought - since 11-90-an mentioned the Engineer's Thumb compressor that uses 3080, could it be possible that that part of the circuit was made to compress the signal? Maybe it did, but I didn't really think of it, so I didn't notice... 

[Rob Strand]

Tried that 1M at input today, didn't really like how it was going - it was as if something changed as of yesterday and the device started to easily make a feedback, the whistling sound that would go forever and that didn't work well with adding that 1M, as I had to increase the output on trimpot, which made it worse, so I got it back to the 560k and to my ear, with the trimpot set to half (as it was before I even touched it), sounds good. Still clueless about why would it make a feedback like that, too much of something... When I'd touch the springs it would stop, as if they were reacting to something and vibrating, producing the whistle....

The 1M shouldn't affect that.    So me it sounds like some of your wires sitting on top of each other, or near each other,  and it's causing oscillations.   Try to keep all the signal wires away from each other, including the reverb wires.   You need to fix it because those types of problems mess with the sound and send you on a false path when adjusting stuff.

This sounds good to me, but still my brain struggles with what it actually mean in praxis.

All it means is the reverb sound is unaffected but the hiss is reduced.     The signal out of the reverb starts to drop around 4kHz to 7kHz but the hiss (at all frequencies) still comes through.  The low-pass filter removes the hiss part.

I'll try and get the 10k then - as I understand you'd recommend the use of the pot to set the effect output, rather than adjusting the input and trimpot onboard, and just do the combo of the 1M input resistor and crank the trimpot all the way?

I might have misunderstood your motives here.   The 10k pot is a Reverb put that you would put on a panel with a knob.  If you are only going to have one reverb level locked-up inside the unit then there's no real point having it since you can just tweak the reverb level with the trimpot and leave it.    Earlier in the thread we also spoke about having a switch instead of a pot so you could have two reverb levels but that still needs the switch to be accessible to the outside world.

This worked like a charm!

Cool.

Just another thought - since 11-90-an mentioned the Engineer's Thumb compressor that uses 3080, could it be possible that that part of the circuit was made to compress the signal? Maybe it did, but I didn't really think of it, so I didn't notice...

I'm more inclined to accept "fake" reverb time theory,  Makes sense if ELKA did that type of thing.   It could also be used as a noise gate for the reverb.  The thing about the reverb is you need to add it back to the original signal with a mixer.  The keyboard the ELKA board came from would have had a mixer somewhere.   A compressor needs to operate on the whole signal so the position of the LM3080 is in the wrong place in the signal chain.

For your amp, getting a compressor working on the ELKA has the same problem.   The raw signals are passed through to the power amp regardless of what happens to the signal through the jack.   To get a compressor to work you would need to break the signal chain.   You would also need a place to put the controls for the compressors.    IMHO, I think it's easier to steal the LM3080 from the ELKA board and build a separate Engineers thumb.    That way you can use it elsewhere as well.

[11-90-an]

+1 on the stealing of the LM3080

[razabri]

The 1M shouldn't affect that.    So me it sounds like some of your wires sitting on top of each other, or near each other,  and it's causing oscillations.   Try to keep all the signal wires away from each other, including the reverb wires.   You need to fix it because those types of problems mess with the sound and send you on a false path when adjusting stuff.

So, I've "refreshed" my wiring, as at first I used an audio cable for the IN/OUT/GND where all the wires are insulated together - it seemed like it would be a cleaner job, as I have speaker, power, switch wires in there too. Now I've used only those colored breadboard wires, so they are all sitting away from each other. The power wiring too is done the same way. I'm now using only those first 4 connectors - +V, IN, GND, OUT and I'm connecting only +V from the amp's board as it's getting ground from the INSERT jack.

Also, I've placed the 1M resistor back in there, but I still don't really like what is happening - it's not that the resistor itself is causing any trouble, but with IN signal lowered, I do need to amplify that output via trimpot and that's when I'm getting the noise.

Of course, I still don't have any clear idea on what's happening, but I'm beginning to think that maybe with the 3080 still connected, the noise was less of a problem and I think maybe to reverse it's bypassing, but I don't know if that's making any sense, or it's just me...

I think it's not the shielding either, but I'm willing to try making some sort of metal case around the circuit. Still it seems to me that the circuit itself could be quieter.

All it means is the reverb sound is unaffected but the hiss is reduced.     The signal out of the reverb starts to drop around 4kHz to 7kHz but the hiss (at all frequencies) still comes through.  The low-pass filter removes the hiss part.

What I meant to say is - should I apply something more in a sense of LPF? How to do so?

I might have misunderstood your motives here.   The 10k pot is a Reverb put that you would put on a panel with a knob.  If you are only going to have one reverb level locked-up inside the unit then there's no real point having it since you can just tweak the reverb level with the trimpot and leave it.    Earlier in the thread we also spoke about having a switch instead of a pot so you could have two reverb levels but that still needs the switch to be accessible to the outside world.

Yeah, I kind of thought that some permanent intervention on the amp would be necessary, but then I realized that I like the mod to be not so invasive and to still be able to put things back as they were. So, no drilling, etc.

I'm more inclined to accept "fake" reverb time theory,  Makes sense if ELKA did that type of thing.   It could also be used as a noise gate for the reverb.  The thing about the reverb is you need to add it back to the original signal with a mixer.  The keyboard the ELKA board came from would have had a mixer somewhere.   A compressor needs to operate on the whole signal so the position of the LM3080 is in the wrong place in the signal chain.

For your amp, getting a compressor working on the ELKA has the same problem.   The raw signals are passed through to the power amp regardless of what happens to the signal through the jack.   To get a compressor to work you would need to break the signal chain.   You would also need a place to put the controls for the compressors.    IMHO, I think it's easier to steal the LM3080 from the ELKA board and build a separate Engineers thumb.    That way you can use it elsewhere as well.

My thoughts were more like - did that part of the circuitry had an effect already? Was it noise-gating? And since I already seem to feel like the set-up could be more "clean", I thought it might be the right thing somehow... Not that I don't want the Thumb and I'll surely build it 

[11-90-an]

If I may ask, where did you get that? From a shop somewhere? Or from an attic or garage? Surely there must be some clue to where it came from...

[razabri]

Ask away  Well, there's no much to say - bought it off some guy at our local online trade platform, he has no idea on it's origin, although I think he's into some sort of folk music show biz and that over here has lot to do with accordions, keyboards... Only clue is that ELKA 1910 printed on the board that we've already discussed here and there is no precise instrument, since it's clear that ELKA was keyboard manufacturer, that we could relate this piece to. It's also logical since they made these keyboards with slotting a lot of boards, one for each note, so the board may as well be out of the keyboard unit, although they did make some guitar amps as well, with reverb... 

[Rob Strand]

Also, I've placed the 1M resistor back in there, but I still don't really like what is happening - it's not that the resistor itself is causing any trouble, but with IN signal lowered, I do need to amplify that output via trimpot and that's when I'm getting the noise.

At some point reducing the drive level (ie. increasing the 560k/1M resistor) and increasing the gain (ie. increasing the gain with the trimpot) will cause the hiss to become noticeable.   At the other end of the scale reducing the resistor and decreasing the gain will overload the reverb drive amp or the reverb spring itself, however, when the signal isn't high this scenario will have lower hiss.  There is a game of finding a happy medium.    If the reverb recovery circuit isn't designed well and is noisey that happy medium turns into a very fine line.

So the question is, if your problem is *only* hiss, then it's probably the above scenario.   And maybe the 560k is as good as you can do.

You can always reduce the hiss with a low-pass filter.  For example add a  33nF cap from the output at the connector to ground.   The 1k output resistor and the cap form a low -pass filter.    It's very common to filter the highs on a reverb spring.

If you problem isn't just hiss it could be something more sinister like oscillation.    That can seem like hiss as well but you might also get whistles, weird fizzes in the signal, even sometimes even radio.    A tell-tale sign might be the unit suddenly acts funny as you slowly increase the gain trimpot.   Like is breaks into oscillation.    It's likely to be caused by the trimpot setting regardless of 560k or 1M value.   To make life difficult these problem can run on a thin edge and sometimes the 560k to 1M can tip the balance.    An on top of that adding the 33n cap I mentioned about usually helps reduce these problems!

Of course, I still don't have any clear idea on what's happening, but I'm beginning to think that maybe with the 3080 still connected, the noise was less of a problem and I think maybe to reverse it's bypassing, but I don't know if that's making any sense, or it's just me...

Not impossible but not very likely.

[razabri]

At some point reducing the drive level (ie. increasing the 560k/1M resistor) and increasing the gain (ie. increasing the gain with the trimpot) will cause the hiss to become noticeable.   At the other end of the scale reducing the resistor and decreasing the gain will overload the reverb drive amp or the reverb spring itself, however, when the signal isn't high this scenario will have lower hiss.  There is a game of finding a happy medium.    If the reverb recovery circuit isn't designed well and is noisey that happy medium turns into a very fine line.

So the question is, if your problem is *only* hiss, then it's probably the above scenario.   And maybe the 560k is as good as you can do.

You can always reduce the hiss with a low-pass filter.  For example add a  33nF cap from the output at the connector to ground.   The 1k output resistor and the cap form a low -pass filter.    It's very common to filter the highs on a reverb spring.

If you problem isn't just hiss it could be something more sinister like oscillation.    That can seem like hiss as well but you might also get whistles, weird fizzes in the signal, even sometimes even radio.    A tell-tale sign might be the unit suddenly acts funny as you slowly increase the gain trimpot.   Like is breaks into oscillation.    It's likely to be caused by the trimpot setting regardless of 560k or 1M value.   To make life difficult these problem can run on a thin edge and sometimes the 560k to 1M can tip the balance.    An on top of that adding the 33n cap I mentioned about usually helps reduce these problems!

I hear you Rob - well, after a lot of tweaking and listening to it, I've rolled back to that 560k, as it does let me go easy on the trimpot thus reducing noise. The design is probably like you described - not so great and the use of TL082 probably doesn't help. However, I'm pretty happy on how it sounds and given that I had doubts if this is even possible to set up - everything turned out great, of course, with the huge and essential help and advises from you and the guys. Here's the "last" view at the set-up before I close it for whole eternity:



Oh, and I did place the cap at output, in similar way that we did that 1-8 pin resistor, under the board - just connected the out from underside and one of the ground pins. I didn't have 33n, but placed 47n instead and it did shave of some of the hiss, but 33n could be a better match - I'll try that when I get one. Also, I gave up on reversing the process and returning the 3080 part as it was, as it probably wouldn't make difference.

Anyway, the whole thing sounds pretty good and I can't stop playing - thank you guys so much!

[Rob Strand]

I hear you Rob - well, after a lot of tweaking and listening to it, I've rolled back to that 560k, as it does let me go easy on the trimpot thus reducing noise. The design is probably like you described - not so great and the use of TL082 probably doesn't help.

It's sounds like that's as good as you can do.

However, I'm pretty happy on how it sounds and given that I had doubts if this is even possible to set up - everything turned out great, of course, with the huge and essential help and advises from you and the guys.

Here's the "last" view at the set-up before I close it for whole eternity:

Looks great.

One thing, when you push it all together be very careful that the speaker wires don't get close to the wiring to the reverb connector, or the reverb tank wires - especially the reverb tank output to the recover amp.  That could cause all sorts of weird oscillation problems.

Oh, and I did place the cap at output, in similar way that we did that 1-8 pin resistor, under the board - just connected the out from underside and one of the ground pins. I didn't have 33n, but placed 47n instead and it did shave of some of the hiss, but 33n could be a better match - I'll try that when I get one. Also, I gave up on reversing the process and returning the 3080 part as it was, as it probably wouldn't make difference.

Honestly the 47n is probably fine and that's going to knock out the hiss as much as possible.     There's plenty of other places the design could be modified/tweaked to get a bit more twang out of the reverb.

Anyway, the whole thing sounds pretty good and I can't stop playing - thank you guys so much!

Yeah, if you like reverb, putting reverb in an amp is a big plus.  I've done this for my friends in the past.

[11-90-an]

Now I want a reverb tank...

[razabri]

Well, guys, I've spoken too soon about wrapping up as I have now entered the feedback hell. Something's not right and the amp became too prone to just getting into this continuous feedback that seems to me is coming from a bad combo between reverb tank and the speaker, as it works just fine through the headphones. Some sort of oscillation is at work and I can't seem to remove it.

I've tried many things to solve this, to give it more grounding, shield the wires of send and return, roll the set 180 degrees so that the springs are further away from the speaker (and that only made it worse, maybe it's the closeness of the circuit...), get all the wires as far away from each other as I could, but noting really helped. Only thing that works and that I don't really like is to mute one of the springs with a piece of felt, so there's less possibility for feedback, but it's kind of still there, lurking...

All I can think of now is to separate the circuit and the tank and place them further, but also shield them in some sort of metal casing.

Also, I'm not sure if anything I've done so far would make it more sensitive, as I don't really recall having the issue earlier, but I did made first test with longer cables and everything was not so tight back then...

Now I want a reverb tank...

Besides this nerve wrecking issues, it is a lot of fun 

One thing, when you push it all together be very careful that the speaker wires don't get close to the wiring to the reverb connector, or the reverb tank wires - especially the reverb tank output to the recover amp.  That could cause all sorts of weird oscillation problems.

It seems like it's happening...

Honestly the 47n is probably fine and that's going to knock out the hiss as much as possible.     There's plenty of other places the design could be modified/tweaked to get a bit more twang out of the reverb.

I did found me an old 39n paper cap and it went straight in there - I may be wrong, but it seems like a better match. What tweaks did you have in mind, once I somehow solve the issue of the feedback?