My journey with spring reverb drivers starts 3 years ago, when I salvaged a spring tank from a wrecked combo amplifier (together with the rest of the components). The tank wasn't in good condition either, since one of the wires holding a spring had broken. I had to make a new one from one strand of a copper wire, threaded through the magnet and the support and soldered into a loop, which worked surprisingly well. That's when I started working on the '19 version of this circuit with the help of a friend. That one quickly turned out into a build, which I then updated in the next years.
'19 Reverb
I want to show the '19 version without going too much in depth, just to show how far has it gone.
I hate a lot of things about it: from using LTspice as schematic editor, to the inverting input stage with 1Meg worth of Johnson noise right away. The main inspiration for this one were the circuits used in modern Fender and Marshall amplifiers, because I didn't even know how to pick some values for myself.
'20 Reverb
In 2020 I've updated the circuit. This already looks better, with a non-inverting buffer, a charge pump for bipolar supply, since I had found out that driving tanks with high enough input impedance requires more voltage; also this cleaned up the circuit nicely. The mixer at the end is less messy too. Another version after this one switched to variable gain on the recovery stage because of a misunderstanding I had about noise. I was really struggling with noise at high reverb levels because of the small voltages present on the output coil and the large gain required. Sadly, variable gain helps with headroom, which isn't really an issue when your input signal is a few mV, but not with noise.
Between 2020 and 2021 I came up with a version of the circuit I was finally happy with. I had discovered a couple of pages describing spring reverb in detail on ESP, which were an invaluable help. I don't try to hide the fact that I owe most of my spring reverb knowledge and the positive outcome of this long project to them.
This circuit is an adaptation of an ESP one as guitar pedal, which means working with different levels and supplies. I'd finally switched to the NE5532 as driver and recovery amplifier: this beautiful but power hungry op-amp works better than the TL072 both for the driver stage, since it's able to drive heavy loads with little distortion, and more importantly for the recovery stage, where its excellent noise specs meant a world of difference from before. With my old tank having a 150 ohm input coil, I found a single 5532 to be just enough, although paralleling two in a power booster configuration is an option. I then made an 8 ohm version for a friend with a simple discrete push-pull amplifier from ESP.
A fun note is that up to this point I managed to build all the updates on the same piece of veroboard, even making a TL074->072+5532 adapter by stacking two sockets.
Here's a demo of it:
'22 Reverb
And we're finally to this year! After looking back at the circuit, I've decided to refresh it a bit and go through some things I had until now left assumed, like the hefty low-passing on the recovery stage to keep down the hiss.
I'm proud of this version, probably the first that isn't just a patchwork of existing designs and has some thought put in every component. In practice, most of the changes aren't that big from the previous version, but there's a focus both on minimizing noise and achieving the best sound.
Let's start with the changes: the dry buffer is now in parallel with the reverb driver so that its noise isn't added to the input of the latter. The 5532 is a thirsty thing, but I've measured just about 150mV of voltage offset because of bias current with 1M, which isn't concerning. For the same reason I've changed the "dwell" control from input attenuator to gain (transconductance) control. I've also bumped up the maximum gain a bit: it's already higher than what you see ESP using because there the input signal is assumed to be 1Vrms, not guitar levels. Even with maximum gain I didn't notice clipping (not saying it's not there at all), so this should allow a more dramatic reverb than before, while still allowing something less bouncy or operation at line level with the gain down.
The tank I have access to is 150 ohm in 2250 out, so those are the values I worked with for the op amp drive version. The 2.2 ohm gain stopper for the 8 ohm tank is just a guess based on some calculations, you're free to change it if necessary.
Op amp drive? As with the last version, I'm featuring both the 5532 driving the coil or the discrete amplifier. The former works fine down to 150 ohm, in my opinion but isn't suited for lower impedances. The discrete driver works just as well if not with high impedances and is the only viable option for 8 ohm, so this time I went for this one also for my veroboard layout. There's also the option of paralleling two 5532 which should work well above 8 ohms and maybe even then, but I didn't go for it just because that would have left me with an odd number of op amps.
If you're wondering about the resistor in parallel with the coil, its value is close to the (calculated) reactance of the coil at 20kHz and limits gain above that. More suggested values on ESP. You can try messing with this value to change the timbre.
Reaching the output coil. Here, a low noise amplifier is a must, as I have learnt in my first attempt with a 072. The 5532 does an excellent job, and I've tried to help it by scaling down the impedance of its feedback network. As much as I've tried to keep noise down on the input, this is the critical part because of the gain involved.
The simplest but most exciting change comes from experimenting with C4 and R8 on the first schematic. These form an RLC low-pass filter with the output coil. You can think of it like a guitar tone control and similar considerations apply. It's a filter that can show some resonance around the cutoff frequency, and you can tune the frequency through the cap value, the peaking with the resistor value. Often the peak is between 2k and 4k. This is a very effective way of changing the sound of the reverb, and is also critical for hiss. In my struggle to keep noise at bay, I went pretty heavy handed with this and also the feedback capacitor on the recovery op-amp in the previous versions, while this time I managed to achieve a much more open sound by first approximating with a simple calculator and then by ear. With a 2250 ohm output coil the suggested values seem to accentuate the "drip" while keeping good noise performance.
The "reverb" control went from being a gain control to an attenuator. This because it allows me to keep the C8 filter constant (my personal preference) and because there's no chance of clipping the recovery stage. Even banging chords with the dwell at maximum the peak amplitude on my scope was 10-20 mV. Of course this changes with coil impedance and so should the gain of the stage. These tiny levels mean a lot of gain, so I don't think there's much more that can be done to avoid some hiss with the reverb at full (this will drown the dry signal completely anyway), but I find it more acceptable that with many pedals so it's not that bad really.
I'm giving you a few options for the mixing stage. The first one is suggested for pedal use, with or without the dry switch, since it keeps the dry at unity and the controls simple. The alternative mixing stage goes from fully dry to reverb only with a single knob and makes the dry switch unnecessary. In case it's not clear you don't need the reverb pot with this one, another advantage of the fixed gain recovery stage. This one is if you want reverb only while keeping only two controls or if you want more reverb than dry for some reason. Alternatively the dry switch allows you to cut the dry blend, which is useful if you want to use this for looping or effects loop in a mixer or similar. In case you're curious I chose dwell, reverb and dry switch.
Again, many thanks to Elliot Sound Products. The least I can do is share the article and the two projects on the site, which should be the starting point for anyone getting into spring reverb.
Hey there, great post. I've assembled the '22 reverb, but I'm having some trouble getting it to work properly. I'm not getting any sound from the reverb tank, and what does come out sounds a bit rumbly and distorted. I've checked for shorts, and it doesn't seem to be the issue. My suspicion is that I might not be using the correct values to drive the reverb tank. The tank I'm using has an 8 Ohm input and a 2250 Ohm output, with the input being insulated and the output grounded. I used the values in the veroboard layout, since it says in the description "8 Ohm in and 2250 Ohm out, otherwise tweak..." Just wanted to confirm if …