Spring Reverb Design

[plumsolly]

I wanted to build a spring reverb that has:
-Low noise
-Balanced inputs and outputs
-A high shelf
-Highpass/Lowpass filters
-Switchable tanks for different reverb times

The attached schematic is what I came up with. Would you mind having a look at it?

I stole all the blocks from other designs.
The filters are from Harrison/Harrison Ford
The high shelf is from Douglas Self's Small Signal Design
The reverb drive and receive are from here: http://sound.whsites.net/articles/reverb.htm

I just stitched them together.

A couple of specific questions:

How do you calculate the worst-case input impedance for the low-pass filter? I need it to figure out the an appropriate value for C35.

Do you have any recommendations for minimizing hum/noise? I lashed a similar circuit up on the breadboard and it was pretty noisy. I figured I would get it all in a box and experiment - but any advice you had would be appreciated.

Thanks,

Ben

 

[JohnRoberts]

I can't easily follow that schematic,  some reverb tanks like to be driven from current sources, your circuit appears to use a low Z voltage source. I didn't look at everything.

JR

 

[volker]

Not sure I agree with that kind of elaborate EQing in a reverb unit, but my thoughts on what you have:

- Lose R1/R2/IC1A. If you want to drop the level by 6dB, use the THAT1246
- You don't need C8/R6/IC2A, but a resistor in series with C28.
- I'd be inclined to use a tank with higher primary impedance, so it can be driven directly by the opamp (or 2 in parallel)
- With R30 at zero, there is only R31=390R at high frequency, so calculate C35 accordingly.
- The output of IC1C isn't connected correctly, compare with original.
- You short the signal to ground in front of the DRV134. Pin3 should be connected to ground.
- Is it on purpose to have effect signal only? Normally you would mix reverb and clean signal.

 

[joaquins]

I can't easily follow that schematic,  some reverb tanks like to be driven from current sources, your circuit appears to use a low Z voltage source.  I didn't look at everything.

JR

  The 8Ω verb input is connected between the output of the hybrid and the non inverting, so it's working as a current source.

  You can loose C33 as the opamp is working on 10X gain (except at DC but that doesn't matter) You could leave unpopulated and use it only if having stability issues or lowering the stage gain. C23 must be there as the gain of the stage is under 2 and the NE5534 isn't stable at that low gain settings.

  You have a few misconnections on that schematic, the mentioned ground at DRV pin 4 is one, I've picked another but I can't see it now.

  You can also tide up things a little bit, nicer and easier to read schematic less prone to mistakes. Look for example C23 and C33. Isn't C23 much nicer? Consistency in shape helps a lot.
  IC2B is the only one with the inverting input on top, I see why is that but if you find a way around could be nice.
  Tip: you can use opamp symbols with power supply pins separated, and leave all the PS distribution and decoupling in a corner of the schematic, not confusing around already confusing parts.

JS

 

[JohnRoberts]

  The 8Ω verb input is connected between the output of the hybrid and the non inverting, so it's working as a current source.

Yes, thanx.... had to open the schematic in another window to see that.

JR

  You can loose C33 as the opamp is working on 10X gain (except at DC but that doesn't matter) You could leave unpopulated and use it only if having stability issues or lowering the stage gain. C23 must be there as the gain of the stage is under 2 and the NE5534 isn't stable at that low gain settings.

  You have a few misconnections on that schematic, the mentioned ground at DRV pin 4 is one, I've picked another but I can't see it now.

  You can also tide up things a little bit, nicer and easier to read schematic less prone to mistakes. Look for example C23 and C33. Isn't C23 much nicer? Consistency in shape helps a lot.
  IC2B is the only one with the inverting input on top, I see why is that but if you find a way around could be nice.
  Tip: you can use opamp symbols with power supply pins separated, and leave all the PS distribution and decoupling in a corner of the schematic, not confusing around already confusing parts.

JS

 

[plumsolly]

Thanks for the the thorough responses!

I will clean it up, make a couple of changes, and make a new post.

See you soon.

Ben

 

[plumsolly]

- Lose R1/R2/IC1A. If you want to drop the level by 6dB, use the THAT1246

Done.

- You don't need C8/R6/IC2A, but a resistor in series with C28.

I'm curious about this one. Self calculates the input impedance of a similar eq circuit that includes another 5k pot for a low shelf as 987Ω when all controls are set flat. Is that TL074 able to drive the high shelf?

Also, if I did replace it with a resistor, how would go about determining an appropriate value?

- I'd be inclined to use a tank with higher primary impedance, so it can be driven directly by the opamp (or 2 in parallel)

I have that driver circuit on a handy little daughter board from Pacific Recorders and Engineers, so it is super easy to implement.

- With R30 at zero, there is only R31=390R at high frequency, so calculate C35 accordingly.

Thanks. Switched to 470µF

- The output of IC1C isn't connected correctly, compare with original.

Fixed.

- You short the signal to ground in front of the DRV134. Pin3 should be connected to ground.

Doh!

- Is it on purpose to have effect signal only? Normally you would mix reverb and clean signal.

It will always be connected to a send, so I didn't bother with it.

You can loose C33 as the opamp is working on 10X gain (except at DC but that doesn't matter)

Lost.

You can also tide up things a little bit, nicer and easier to read schematic less prone to mistakes. Look for example C23 and C33. Isn't C23 much nicer? Consistency in shape helps a lot.
IC2B is the only one with the inverting input on top, I see why is that but if you find a way around could be nice.
Tip: you can use opamp symbols with power supply pins separated, and leave all the PS distribution and decoupling in a corner of the schematic, not confusing around already confusing parts.

I addressed some of this. I moved the power pins that were separate to a different area along with the decoupling caps. The ones that were attached, I wasn't sure how to address.

Thanks for all the insight! Let me know how it looks now.

Ben

 

[PRR]

Output from a spring is tiny.

Your Reverb Recieve {sic} stage has gain of 1.1, where I would expect 100 or more.

 

[plumsolly]

Output from a spring is tiny.

Your Reverb Recieve {sic} stage has gain of 1.1, where I would expect 100 or more.

Right. R28 = 220. Will update (and change spelling).

 

[benb]

Just looking it over, c24 is upside down, the + should be at the top.

In several other places, you've got electrolytics used for signal coupling that have 0 volts DC across them. I've seen this before, and I guess it mostly works, but it bothers me.

 

[plumsolly]

Thanks for looking it over!

Just looking it over, c24 is upside down, the + should be at the top.

Yes - I'll correct it.

In several other places, you've got electrolytics used for signal coupling that have 0 volts DC across them. I've seen this before, and I guess it mostly works, but it bothers me.

Would you mind pointing those spots out? I was trying to be cautious of the input bias currents of the 5532/4's and maybe went overboard.

Thanks,

Ben

 

[Bo Deadly]

Reverb tanks are a little sensitive about signal level. Too little and it's not clear sounding. Too much and the whole things starts to jangle. That's why fancier spring reverbs often have some kind of compression / limiting / noise reduction or some such.

If you don't want to go that route, you should at least have a level control for what's going into the tank. Meaning a pot on the input of the reverb drive. That way you can maximize signal level for the bast SNR performance and yet be able to attenuate as necessary to control how much is getting into the reverb tank.

And you should probably also have a pot on the input of the DRV134 to control overall level out.

Personally I would probably do something simpler like a cheap 600:8 transformer to easily drive an 8 ohm tank from a conventional output and that's it. Do your EQ / compression / noise reduction with outboard gear and use a vanilla mic pre for recovery. Maybe include a simple recovery circuit and a line driver with enough power to drive a really long cable (so that you can put the unit far away in a vibration free place). You might even make something that could run on 48V but it could be very tricky since you only have ~10ma to work with but then it wouldn't need any external power and the noise that goes with it. You just have the two signal cables.

I'm not speaking from experience but my guess would be that vibration reduction, shielding, grouding, and that sort of thing will be very important.

Also, reverb tanks are so noisy that using an NE5532 vs a TL082 won't make a lick of difference. Actually the TL082 would probably perform better because you could then ditch half of your electrolytics and actually reduce distortion.

 

[Bo Deadly]

Would you mind pointing those spots out? I was trying to be cautious of the input bias currents of the 5532/4's and maybe went overboard.

If you use op amps with low offset (like TL074) you can remove:

C8 (and thus R6)
C28
C12
C19
C17/C18 (w/ quality pot)
C20

Again, using TL074 vs NE5532 isn't going to make a lick of difference with a reverb tank that has a relatively high noise floor. Except you should keep the NE5534 for the recovery amp because it's high gain.

 

[plumsolly]

Thanks for looking it over, squarewave!

Reverb tanks are a little sensitive about signal level. Too little and it's not clear sounding. Too much and the whole things starts to jangle. That's why fancier spring reverbs often have some kind of compression / limiting / noise reduction or some such.

I thought about this, but I don't necessary mind the reggae-style spring clang/crash, and, as you suggested, I can use always use something outboard in the case that I need it.

If you don't want to go that route, you should at least have a level control for what's going into the tank. Meaning a pot on the input of the reverb drive. That way you can maximize signal level for the bast SNR performance and yet be able to attenuate as necessary to control how much is getting into the reverb tank.

I thought about this also, but my console has global send controls and I figured that would be enough, and the gain-staging would have less variables.

And you should probably also have a pot on the input of the DRV134 to control overall level out.

I thought about this too and figured I would be OK with using the return controls on my console, but I guess the question is, "Do I run the risk of overdriving the return inputs?", in which case I'd need it. I guess I'll have to decide how likely I think that will be.

Personally I would probably do something simpler like a cheap 600:8 transformer to easily drive an 8 ohm tank from a conventional output and that's it. Do your EQ / compression / noise reduction with outboard gear and use a vanilla mic pre for recovery. Maybe include a simple recovery circuit and a line driver with enough power to drive a really long cable (so that you can put the unit far away in a vibration free place). You might even make something that could run on 48V but it could be very tricky since you only have ~10ma to work with but then it wouldn't need any external power and the noise that goes with it. You just have the two signal cables.

I've had a completely passive setup before, which is super easy and versatile, but I did not totally love the interfacing and patching involved.

I'm not speaking from experience but my guess would be that vibration reduction, shielding, grouding, and that sort of thing will be very important.

Agreed.

Also, reverb tanks are so noisy that using an NE5532 vs a TL082 won't make a lick of difference. Actually the TL082 would probably perform better because you could then ditch half of your electrolytics and actually reduce distortion.

I do like the idea of tossing those electrolytics, but the resistor values in high shelf require the drive capability of a 5534-type opamp, and I am not keen on reworking that. And the reverb drive circuit is a little Pacific Recorders & Engineers daughter board that I have a ton of, so it's actually easier to leave that like it is.

 

[benb]

Thanks for looking it over!Yes - I'll correct it.Would you mind pointing those spots out? I was trying to be cautious of the input bias currents of the 5532/4's and maybe went overboard.

Thanks,

Ben

The big thing (an error that I don't see using other coupling caps) is IC4's + input, you have a 220uF cap in series with a 1k resistor.  Those seem to be scaled to go into a virtual ground, but the + input here is practically an open circuit.  The small input DC bias current will either charge or discharge (depending on the polarity of the opamp's input bias current) that cap eventually, and the input voltage will end up at one of the rail voltages, making it not work after "working fine" for a few seconds, minutes, or hours.

Make the cap a 0.22uF non-Electrolytic connected from IC28's pin 7 directly to IC4's pin 3, then put in a 1 meg resistor from that pin 3 to ground.  If you're worried about offset, you can add a 1 meg resistor in series with pin 2, but even without that, C27 will charge up to whatever offset voltage is at the output, and there won't be DC going through the reverb tank coil.

Then again, C27 is yet another "coupling cap," an electrolytic with no DC voltage across it.  This is the same as an electrolytic not in a circuit, sitting on the shelf - if it's not operated at a significant DC voltage, it will eventually deform and not work well.

There's a discussion of DC blocking capacitors here:
https://groupdiy.com/index.php?topic=65755.0

But then again, others have reiterated that a spring reverb is not quite a "high fidelity" device,  and so too much attention to detail may be overkill.

 

[plumsolly]

Thanks, benb.

I added a 10k resistor to ground at the + input of IC4. The new schematic is attached.

Then again, C27 is yet another "coupling cap," an electrolytic with no DC voltage across it.  This is the same as an electrolytic not in a circuit, sitting on the shelf - if it's not operated at a significant DC voltage, it will eventually deform and not work well.

I was under the impression I needed C27 to prevent DC from going through the reverb input coil...

Thanks,

Ben