Designing a reverb driver on drawing table - what sounds good?

[Mick Bailey]

If I get your numbers right you have a high pass filter with a cut off frequency of 720Hz. Seems high, but if it sounds good I guess.

That's what I ended up with after scoping the driver coil to see how well it was being driven at low frequencies. In practice you have to remember that this is the filter corner frequency rather than a hard cut-off. Whilst the figures suggest little reverb effect below this - especially on the low E string - the constant-current drive and impedance of the driver coil means that in practice the reverb sounds good at low frequencies.


Another point I'd make is adding a cap across the recovery coil teminals, or making this switchable off a rotary switch to select from a number of values. This has a drastic effect on the frequency peaking and can result is some really good effects. Some of the older 'Cascade' units have a cap soldered inside the unit to improve the output. It's easy to try out. I use this method to tune reverb circuits that are a little weak or lacking in dimension.

[MarkusBass]

I'm still not sure of how hard I should drive the driver stage.

Have you checked the links I provided before? It is very well explained just in the first link that was provided: Spring Reverb

Mark

[Chuck H]

I'm still not sure of how hard I should drive the driver stage...

I read in one of the links that over-saturating the input won't add any significant distortion...

I think you're interpreting those stipulations incorrectly. If you saturate the driver stage you will indeed be reverberating a distorted signal and the resulting reverb will be distorted WRT the dry signal.

Saturating the INPUT TRANSDUCER IN THE PAN is what the second statement is referring to. For the classic spring reverb sound you probably actually want to be close to saturation of the input coil. This is not the same thing as saturating the driver stage.

I guess I can focus on the pick-up end of the circuit, and the frequency response.

Somehow the entire discussion has gotten lost. You asked specifically about driving the pan. Our advice has been about driving the pan. Including the frequency response discussed. Applying the frequency response advice given to the recovery stage is almost guaranteed to sound bad. I thought the circuit application of the advice was obvious for several reasons. It was targeted for exactly what you asked about and covered in the links provided.

Trying to SPICE a spring reverb circuit will probably prove a lot like trying to lick your elbow. All the parts are there to observe but it's still impossible. For starts I think it's highly unlikely there is any sort of accurate pan model. Even to the purpose of replicating the dynamic EQ of the transducers. Especially confounded by the notion of saturating the input transducer.

EQ what's going into the driver stage with your LF and HF roll off. Doing it here reduces the likelihood of distorting the driver, which is bad. You want the recovery stage to be as linear as possible for starters. If you find the need to adjust what's going into the driver stage or coming out of the recovery stage you should do it by bench measurement for actual voltage and current levels and by ear for desirable frequency response. not by SPICE.

[überfuzz]

Trying to SPICE a spring reverb circuit will probably prove a lot like trying to lick your elbow.

Chuck H, you obviously never tried to simulate circuits in a spice software. You'd be surprise... Ah well, your comment was in good spirit. :-)

I have checked all the suggested links, or most of the stuff. It's evident that I've missed some thing you haven't. Please explain how I should go about calculating a proper signal level for my pan (this would be the stage on the driver side) without a measurement of the suitable current level..?

About the frequency responses of both stages: all suggestions seems legit. I'm well aware of the, lets call it funky, impedance of the reverb pan as a whole and what's causing it. However, I haven't really gotten to the set up of my cut off frequencies, that's why I ask for suggestions or advices regarding that. In addition to this, I'm well aware of the fact that my simulations of the reverb pan might be slightly off, but not as wonky as that elbow. ;-)

Cheers and thanks!

[Chuck H]

Please explain how I should go about calculating a proper signal level for my pan (this would be the stage on the driver side) without a measurement of the suitable current level..?

Well... The amount of drive the pan requires is published and covered in the links. This doesn't take into account the location of the reverb in the circuit, which could be pre clipping stages as in vintage amps or post clipping as in modern amps. This is actually an important consideration when determining how to drive the driver and balancing gain structures. If that helps at all. In a vintage type design, which is the most common implementation of a spring type reverb, there is commonly a point of diminishing returns in the level driving the driver where there may be some clipping when the amp is turned up loud. This is necessary to have adequate signal to the driver at lower volumes for a good reverb tone. That is, with the reverb placed in the middle of the preamp you can't approach saturating the input transducer of the pan with the amp at moderate volume UNLESS there is too much drive to the driver at higher levels UNLESS you include some sort of level control, like the erroneously labeled "dwell" control, to adjust this parameter. Most amps and reverb systems don't do this because the details of proper circuit operation become too complex for average users. So instead a compromise level must be chosen. Chosen is the key word here because the compromise is at the discretion of the designer and isn't a hard parameter. With a reverb tucked in the middle of a preamp, and therefor at the mercy of IT'S levels, there will be a point where there is too little drive for good tone, too much drive resulting in distortion or both. That compromise is up to you.

Now, while I've done some SPICE modeling for some types of circuits with good results I don't think I would even bother for this circuit because the compromises must be chosen by ear on the bench for best results. This does require some things be in place before hand. Like knowing your driver design will produce adequate current to drive the pan and designing in some ability to adjust the signal level to the driver stage for EQ and output. But I don't need a SPICE model for those things either. But I'm not interested (and I'm sure you aren't either) in arguing the relative merits of using SPICE models for this sort of thing any longer. I hope you find some of these considerations helpful regardless of how you approach the design aspect.

[J M Fahey]

Quote Originally Posted by Chuck H View Post
Trying to SPICE a spring reverb circuit will probably prove a lot like trying to lick your elbow.

Chuck H, you obviously never tried to simulate circuits in a spice software. You'd be surprise... Ah well, your comment was in good spirit. :-)

Can you simulate a reverb spring in spice software?

I guess that would take an accurate spice model of a reverb spring, including not only its inductance (that's easy) but its myriad resonances, which strongly affect impedance and frequency response, and also the delays and reflections involved.

I very much doubt anybody has such a model, in which case simulating it remains as easy as licking your own elbow.

[g1]

I was also under the impression that none of the tank "models" were considered any good.

[überfuzz]

g1 - At the moment I'm under the impression that it's harder to get proper impedance measurements, than it is to set up a model of the impedance in a tank.

The information in the links DON'T provide any info of my tanks signal level (current). It seems I have a tank that's off the charts, so to speak. It's a type 9 tank with impedances G/B. The data charts only goes to F... Bummer!

[Chuck H]

The complications in modeling a reverb pan compound drastically when you consider the ideal circumstance of approaching saturation of the input transducer coil, making the dynamic impedance a dynamic dynamic as well as all the phase anomalies and resulting overtones of the reflecting vibrations contrasting with decay time, what the EQ and effect of the reverb in all these conditions does when blending with the dry signal, blah, blah. Ok... So the question of the day becomes "How does it sound?" With so many dynamics even the best SPICE model, accurately implemented won't tell you much because there's just way too much going on. Making circuits adequate for the task is easy. Tuning it is done by ear. So what is the SPICE model for? Unless one just likes to play with SPICE models. I've done some of that. Trying to do it with a reverb circuit though... That would either frustrate me into an ulcer because of all the compromises or pop a vein in head from all the complexities. Still, it's a hell of a challenge and I suppose that can be valid on it's own if you're the sort.

[Chuck H]

From what I just found the "G" impedance transducer is listed as 8333 Ohm input impedance / 735 Ohm DC. If that seems logical. I don't know how to extrapolate a curve from this. I'll guess that the 8333 ohm input impedance is at 1K since that seems to be a standard test frequency for most reverb specs I've seen.

[überfuzz]

With so many dynamics even the best SPICE model, accurately implemented won't tell you much because there's just way too much going on.

Yes but this is a simulation.., i.e. there's no way I'll ever be able to count for all variables effecting the system.

The other day I came to remember a discussion I had over at the LTspice forum at yahoo. I was tinkering with a spring reverb model. There were about as many frustrated comments how blunt my model was over at the LTspice forum as there are comments here on how impossible it is to simulate. They insisted that I should simulate the decay, not just the impedance behavior. Whereas I found that if I got the impedance behavior somewhat similar to what we would expect to see in a real pan it gave me a simulation that was surprisingly close the the finalized reverb circuit. Eh, minus the decay. :-)

Thanks for the impedance info! Maybe I can extrapolate using the signal levels of the A to F pans? I mailed the company asking for the signal level, no reply...

[Mick Bailey]

That driver coil is a motor and as such has a complex back-emf component that interacts with the input signal in a non-linear manner. How you would analyze this in a spice model beats me.

My own observations with reverb circuits are that you need to experiment; I had a theoretical design in mind for a commercial unit and the finished product after debugging/optimization was some way off what I'd started out with. Choice of reverb tray has a fair influence on the sound, like choosing a speaker.

[MarkusBass]

Please note that the OP asked only about the required frequency response of spring driver and current that is needed to drive the spring. He didn't ask about full simulation of spring reverb circuit. This requires advanced knowledge of mathematics and DSP. The guys here: UA WebZine "Ask the Doctors!" April 06 | What Makes Spring Reverbs Sound "Springy"? they know how to do it.

Mark