Multi tank stereo spring reverb

[Tubetec]

Hi ,
I have a few reverb tanks lying around here ,3 long ,2 short .
Id like to make up a stereo spring setup, Im thinking of  making a driver  a bit like the fender standalone reverb ,transformer coupled 6v6 ,maybe useing a champ style 10 watt transformer .These transformers have several taps on the secondary ,so should be easily able to drive at least two 8 ohm tanks . Ive rewired the outputs from the tanks to TRS jack ,so they can feed a balanced mic input ,this reduces any hum picked up from the output coil and is a fairly good match for the 2250 ohm transducers .
What I really fancy trying is M\S stereo sum difference arrangement on the tank outputs but no one seems to have tried this before ,If anyone out there has any ideas on how to go about doing this any info would be greatly appriciated .

Multiple tanks definately give a smoother reverb ,less like a spring and more of a  plate reverb type sound ,the clangyness of the single spring arrangement disappears when multiple tanks are combined ,Mesa have dual tanks in one or two of their higher end preamps ,and a company called Paia do a hot springs kit which also uses two tanks .
All comments /criticisms welcome

 

[JohnRoberts]

Unlikely to sound like a plate but more diffuse is good... the simple spring has relatively short first slap (35mSec ?) and similar short repeat rate.

To do something different with multiple springs try to connect them in series, so you get re-reflections from the mid point discontinuity, for more diffusion...  or not.  Maybe connect two different springs to a single driver angled apart... If their return reflections crosstalk all the better.

Or just code up a digital reverb...  ;D

You are messing with a very mature technology so i suspect some of these things have already been tried and rejected.

JR

 

[Tubetec]

Cheers John ,
Yeah I'll maybe take a closer look at some of the Akg units from way back and see if theres any inspiration to be found there ,
Thanks again for the reply ,

 

[Rob Flinn]

AKG spring reverbs are the best of the bunch if you want something that gets close to a plate.   

 

[Tubetec]

thanks for chimeing in Rob ,
Im thinking a good tube drive stage makes all the difference ,its easy to rewind the input coils on the tanks  for 8 ohms ,the handy thing is the amp unit is good to drive speakers or cans also ,6v6 is a bit flahulach in power terms ,but the more you drive into the tank the more you get out ,a bit of compression and a couple of % points THD just makes it better too....

 

[brujo]

I don't think that the driver amplifier is relevant, but the makeup amplifier certainly is and tubes are a good choice for that.
Perhaps with four tanks (two stereo sets) and some crossed feedback the diffusion can be augmented.
If you use for example a short decay 3 spring tank (more springs equals less resonances) and a medium decay 3-spring tank with independent driver amplifiers feed from the same input, but feedbacking some of the signal from the medium tank to the short one and vice-versa would add progressively more diffused and longer decay. As the tanks would have very different first taps, the resonances would be minimised and the sound richer.
You can even mute the input signal on one of the driver amplifiers and drive that tank only from the output signal from the other, but keep feeding back the signal to the first tank driver, in order to have a longer reverb with less density on the first seconds of decay, i think the AKG BX20 did something like this.

 

[Bo Deadly]

I don't think that the driver amplifier is relevant, but the makeup amplifier certainly is and tubes are a good choice for that.

Although I have no real experience with spring reverbs, this is actually the opposite of what I was thinking.

I think a tube might make a good reverb tank driver because a tube makes a good limiter because the onset of clipping occurs over a larger dB range (a tube can be < 1% distortion at low levels and then 5% THD well before hard clipping whereas a solid state circuit will go from < 1% THD to hard clipping in 2 dB). So if you select a transformer that steps down the signal just the right amount so that the tube clips at the threshold were the spring does not go bonkers it might be clearer sounding simply because you  squeeze more signal into the springs without them jangling (as much). Something with a 4 ohm secondary would drive two 8 ohm tanks in parallel.

But my *guess* would be that noise is really where improvements are to be had. Specifically, the first thing I would study is mains noise. Have you ever rotated a microphone and found that at some specific orientation noise is greatly minimized? My guess is the recovery transducers of spring reverb tanks would be similarly affected.

If you're going to use 2 tanks you might make mount them in a clam shell fashion so that both spring trays are isolated all the way around. It might be interesting to see if noise is reduced by orienting the recovery transducers in a humbucking fashion [1]. You would have to mix the two signals together to get cancellation so that would eliminate stereo however (unless you use 4 tanks).

Of course filtering before and after might improve things further but I would probably outboard that until it's established what filtering suits whatever arrangement.

As for the recovery amp, a high gain op amp circuit is probably going to give significantly better noise performance than a tube.

Although I suspect that all of the above would have relatively little effect compared to simply placing the tanks as far away as possible from sources of EMI and particularly mains AC. Meaning running long cables off into the corner of the room were there are no outlets (being used).

Oh, and proper grounding is absolutely critical but it always is.

[1] Although this might necessitate that the trays are re-rigged relative to the chassis. And bolting two trays together would give them more mass and thus increase shock absorption.

 

[JohnRoberts]

While perhaps not obvious it is possible to overdrive springs and this creates unpleasant audible artifacts... I'm not sure if it's the drive magnetics saturating, or a mechanical interference, just know that it happens.

JR

 

[brujo]

Although I have no real experience with spring reverbs, this is actually the opposite of what I was thinking.

I think a tube might make a good reverb tank driver because a tube makes a good limiter because the onset of clipping occurs over a larger dB range (a tube can be < 1% distortion at low levels and then 5% THD well before hard clipping whereas a solid state circuit will go from < 1% THD to hard clipping in 2 dB). So if you select a transformer that steps down the signal just the right amount so that the tube clips at the threshold were the spring does not go bonkers it might be clearer sounding simply because you  squeeze more signal into the springs without them jangling (as much). Something with a 4 ohm secondary would drive two 8 ohm tanks in parallel.

Sorry, but I think that's a highly inefficient way of making a limiter, you could use a lamp or fet limiter within a solid state amp instead, it's cheaper, easier and runs way cooler, the SSL listenback mic limiter is very easy to build and works wonders for this application. While experimenting with reverb drivers for a local tube amp brand that I design for, I noticed that the only real difference in drivers has to do with power transfer. Be it correctly matching the impedances and delivering the right amount of power to the reverb drive coil. Note that I said "right amount of power" and not "max power", there is a small range of drive that gives the best results, if you exceed that you get more "springiness" as in the outboard Fender reverb unit (great for guitars, vocals and keys, but hates transient heavy material) and a higher risk of overheating the coil. Obviously, a low drive level gets you near the noise floor, but the tank's inertia is your primary concern. The ideal driver amp rating is around 1 Watt, and by trimming its gain the decay quality varies greatly, it's a good point to experiment.

If you're going to use 2 tanks you might make mount them in a clam shell fashion so that both spring trays are isolated all the way around. It might be interesting to see if noise is reduced by orienting the recovery transducers in a humbucking fashion [1]. You would have to mix the two signals together to get cancellation so that would eliminate stereo however (unless you use 4 tanks).

I tried that, the noise doesn't get much lower, but disassembling the spring suspension that holds the aluminium frame where the coils and the springs seat, and mounting two tanks in a new metal shielding box greatly reduces noise. Be aware that the reverb recovery coils and the power supply should be in opposite corners of the chassis for the best results.
The best noise performance is archieved when a high impedance recovery amp is put very close to the recovery coil, but if that's not possible you can always connect the 'hot' terminal of the recovery coil to a balanced cable and leave the other terminal of the cable unconnected, then you can use a differential amp to substract the common mode noise in the cable. Works okay.

As for the recovery amp, a high gain op amp circuit is probably going to give significantly better noise performance than a tube.

And a discrete BJT design will outperform both, and probably be cheaper, experiment!

Crossed feedback within stereo channels works great also, as you probably never hear a discrete reverb tail from only one side of a room.

 

[Tubetec]

With standard reverb tanks as used in guitar amps ,your always fighting induced  hum  from the mains transformer .Carefull orientation can help reduce hum a good bit as can extra sheilding . Taking the outputs from the tank in a balanced fashion helps greatly with induced hum also,it also easily allows the phase of one tank to be flipped relative to the other . Another interesting thing Ive done in the past was to add reverb to an existing guitar amplifier ,I added a wirewound pot (200 ohm) in front of the input transducer . That allows me to connect the spring directly to the speaker out of the amp ,a spare input channel is then used to recover the output of the spring ,and mix it back with the direct signal ,this works great actually ,only issue is there is a tendency for feedback to occur if your drive levels to the tanks are set to high . another benefit of this method of driving is that its the speaker outs of the amp that your effecting ,not just a signal from earlier in the preamp stage .
I had a chance to look at the Akg Bx-20 schematic, nothing ground breaking really ,just two independant springs with their own op amp driver and recovery stages ,mono and stereo inputs are available also .Another interesting feature of the Paia hot springs unit is that both tanks are driven in phase ,but the outputs of the tanks are wired in series out of phase ,the designer claims this gives a smoother tail to the verb . The idea of feeding back a portion of the output of one spring back into the driver circuit also is worthy of experimentation . Anyway ill keep looking up vintage and obscure reverb schematics just in case ,other than that just experimenting with different ways of recombining the individual tank outputs is bound to give some scope for different tonalities .
Kindest regards once again  to all who contributed .

 

[JohnRoberts]

Another thing that may not be obvious about recycling old springs is that they are designed for different  physical orientations.  The weight of the spring physically distorts the centering of the transducer.

JR

PS: I never messed with guitar springs, but I was over all mixer engineering at Peavey and we used truck loads of spring reverbs in the old powered mixers before digital became cheaper.  Yes, transformer hum was an issue, among other things.

PPS: Interesting (perhaps) tidbit, when we (Peavey) started making stuff in China we had to ship over spring reverbs from the US becuase the Chinese made poop sucked too bad to use.

 

[Tubetec]

Thanks John ,
Yeah I was aware of the mounting orientation with tanks alright ,all the attributes of  a particular tank ,such as impedence mounting etc are in the part code .
I used lots of peavey gear over the years and I do remember the earlier xr series mixers with the springs ,many of the rental companies here loved the peaveys ,real workhorses ,hard even for rookies to kill,and when they did go wrong easy to fix .
Its a true measure of good design and engineering when a product line remains more or less unchanged for years and years .
I just wonder these days ,will modern gear with Smpsu's and switching amplifier topologies stand the test of time like the old boat anchor/back breakers used to do.

 

[clintrubber]

And a discrete BJT design will outperform both, and probably be cheaper, experiment!

Some units (like for instance the D&R Stereo Reverb) use indeed discrete BJT-based mic-pre alike tank-recovery amps: the common hybrid of BJT & opamp.

In case of the D&R, it's still a noisy unit   

...but its stereo reverb (2 long tanks) is gorgeous.

 

[Tubetec]

I intend to use some discrete preamps salvaged from a Yamaha Pm3000, inputs are approx 3 kohm  vs 2250ohm  output transducer on the tanks. Just driving the tanks with the aux send seems to work ok ,the input transducers are around 200 ohm at the moment and noise levels from the tank are well below backround from a sm57 for a vocal input. I can easily rewind the drive transducer for low impedence from a small single ended transformer coupled tube stage ,ten feet of 29 gauge magnet wire  gives a resistance of .8 ohm ,which is rougly the value I measure from an original 8 ohm accutronics transducer.

 

[zamproject]

I had a chance to look at the Akg Bx-20 schematic, nothing ground breaking really ,just two independant springs with their own op amp driver and recovery stages ,mono and stereo inputs are available also .

Hello

You have to look better
The BX20 design is certainly one of the most advanced spring reverb driver
The input signal is splitted (balance trim) and send to two coils at both end of the spring,
Signal from spring is also taken at both end, and summed (invert phase with balance trim)
You have a perfect slap less system !!!
This is for one side...

In total there is 8 coils and two lonnnnnnnng spring about 2m
Spring construction is also complex, with mix of different size/diameter/winding pitch. 

Best
Zam

 

[Tubetec]

Thanks Zam,
Yeah I read about the spring being fed and recovered simultaneously from both ends in the user manual .
I find it a little hard to visualise without actually seeing a close up of the transducers though.
Its a clever design though and probably deserves the title 'King of the Springs'.

 

[abbey road d enfer]

Hello

You have to look better
The BX20 design is certainly one of the most advanced spring reverb driver
The input signal is splitted (balance trim) and send to two coils at both end of the spring,
Signal from spring is also taken at both end, and summed (invert phase with balance trim)
You have a perfect slap less system !!!
This is for one side...

In total there is 8 coils and two lonnnnnnnng spring about 2m
Spring construction is also complex, with mix of different size/diameter/winding pitch. 

Best
Zam

Actually, the most important factor in the performance of the BX20 is the fact that NFB is applied to the drive coils; each drive coil is coupled with a NFB coil, so the mechanical movement is servo'ed, which avoids the nasty noises that JR mentioned earlier.

 

[abbey road d enfer]

I had a chance to look at the Akg Bx-20 schematic, nothing ground breaking really ,just two independant springs with their own op amp driver

I think you really should look harder...

 

[Tubetec]

Hmmm alot more going on than meets the eye then , kinda hard to decern the coupling of multiple  the transducers from the circuit diagram

 

[zamproject]

Actually, the most important factor in the performance of the BX20 is the fact that NFB is applied to the drive coils; each drive coil is coupled with a NFB coil, so the mechanical movement is servo'ed, which avoids the nasty noises that JR mentioned earlier.

Yes you are right, I miss that in my basic description.
The receiver coil have two function, NFB loop to the same end drive coil (mixed with input signal)
and send to the output stage (with invert phase mix beside second receiver, that perform the same as first spring end)
two coils per spring end, 4 coils per channel, two springs 8 coils in total  8)

At least that's how I understand it

I have mine open right now for check/repair/calibration, not sure I can fix it this time... coils are ok, but I have a relative low resistance short to ground at one coils, seem to be degradation of insulation inside the wire bridge in gold can that link "big" wire from amp card to micrometric coil wire...
Each time I open the tank I find the mechanical build just amazing...

Best
Zam