Tweet
Sometimes I get tired of the circuit design and analysis work I have to do as part of my job. When that happens, I usually go off and do circuit design and analysis for fun. And you thought that bus drivers were strange in their choices of vacations... 
Today's little mental holiday led me off to reverb pans, that topic having been current here a lot. Reverb pans are a lot like pencils; there are many kinds and sizes, most people kind of know how they work, and manage to use them without a lot of thought. There are more modern updates and changes that do much the same thing, but there is a core of folks that prefer the feel and results of the old fashion Eberhard Faber #2 to any mechanical pencil or pen.
The Beltone BTDR-2 is by most accounts an OK-ish replacement for the usual reverb tank if you design the circuitry to go around it and tinker till it sounds right. But it's not a direct replacement. My speculation was that I could design something with it that was a modestly useful direct replacement.
In my mind, that means that the thing is the same size or smaller than the usual reverb tanks, mounts in a similar way, and connects in and out with RCA phono jacks like the originals. I allowed myself a departure in that I thought it was OK if it had a DC power jack added, as I didn't think I could make it practical with batteries.
The idea is that this hypothetical lump could be used like the reverb pan in an amp; the original pulled out, a DC power wire and plug concocted, then the new thing plugged in with the old wires as a retrofit.
That sounds simple enough, but it took me a fair amount of thinking to figure out how I think it would work. The problem is untying all the knots that were put in the original circuits to drive the cuss-ed spring tanks in the first place.
The BTDR-2 is easy enough to use. It needs 5Vdc at something under 100ma to run. When it gets that, it takes in audio at its input at something about like +3dbm or so and gives you two outputs that are delayed some, with repflections and such, at the output at roughly the same as the input level. These can be used separately or mixed to mono.
Real reverb pans come in a wide range of input impedances and as such need a wide range of input drive signals. They also produce miniscule outputs, on the order of 6-10mVrms. This needs recovery amplification on the order of the amount that moving coil hifi cartridges do. Worse, the better drivers for reverb pans drive them HARD to get past the 60-100db signal loss in the tanks from input to output. And the inputs have their nominal impedance only at the nominal 1kHz measurement frequency. They're actually inductors between 10mH and 250mH with varying DC resistances. Getting anything even crudely like a smooth frequency response requires driving the input with a waveform that makes a constant current per frequency happen in the inductive inputs, and that means that the good reverb drivers have outputs that ramp up in frequency at a rate of one time constant, doubling in voltage every octave. Fortunately, all the realistic ones give up on driving the coils much above 5-7kHz, or the driving voltage would get really out of hand.
And that's the real problem with subbing in a BTDR-2. You have to un-wind that input drive voltage from doubling each octave, AND let the drive circuits think they're driving a real coil, AND get the drive voltage to the BTDR-2 set up to be about level with frequency from low guitar up to maybe 7kHz, as well as something like line level or above. The fact that reverb drive circuits may either drive the input with a current-source signal, letting the voltage rise to keep the current per frequency level, or with a pre-equalized voltage source that rises with frequency but ignores how much current flows, assuming that the current will sort itself out. Some of them assume a floating, non-grounded input on the tank too. I didn't think I could do anything about that without a fancier power supply than I wanted to require, so I left off doing that.
The approach I took was to make the driver happy by putting a coil on the board for the driver to drive. The advent of switching power supplies has made small, radial-package inductors available in the market. You can get inductors in the Bourns family (and probably others) up to 120mH with a series resistance of less than 100 ohms for about $2. A pair of these lets you fake the input impedance of a "1475 ohm" Accutronics tank if you put them in series. By picking other inductors from the same line, you can fake the inductance and resistance of the lower-impedance reverb tanks. The drive circuit will then not know it's not driving an unusually stable reverb pan input.
But that can have as much as +/-30V peaks on the inductors. So I designed an R-C network to take the inductor voltage as an input, and reduce the voltage linearly with rising frequency by using a cap to shunt voltage to ground. This unwinds the rising voltage drive to a linear-with-frequency signal. It has the advantage of lowering the really high drive voltage at high frequency. The signal still probably isn't at the right level, but signal level changes is what opamps are for, right? So I allowed myself one dual opamp. The front end opamp takes the frequency-leveled drive signal and converts it to the right level for the BTDR-2 by having a variable gain. Just to be sure the opamp doesn't kill the BTDR-s, there are some clipping diodes to keep the level into it from getting out of hand. The second section of the opamp buffers the BTDR output and drives a resistor/pot network to cut the signal back down so the "recovery" circuit in the amp will get only its measly 10mV and won't overload.
Power supplies was an issue. the BTDR needs 5V, the opamps need at least that, and it's poor economy to have to build a whole new power supply for this thing. About the best I could do was to put in a series of three voltage regulators. A 7824 will take in up to 40V safely. This drives a 7815, which then drives a 7805. That seems downright silly, until you note that 100ma at 40V is 4W, and things could get hot. This sequence of regulators spreads the 4W out so the regulators are all eating about the same power, and this gets each of them under the 2W that a TO-220 can get rid of in free air. So this thing can be driven by DC supplies from about 9V up to 40V by selective population of regulators and jumpers. So it's probably going to be at home in any amp with a power supply between 9V and 40V positive.
That doesn't help much with tube amps, unfortunately. Another 100ma out of a 500V B+ is going to generate another 50W of heat even if I had a circuit to do it and get rid of the heat. Not elegant. A switching step down is also not a good idea inside a tube amp. About all I can come up with here is a small toroid-based power supply running from the 120Vac or a wall wart separate from the amp.
That part is still TBD.
The circuit board is about 2.5 "by 5.5" at present, with the DC power, input and output phono jacks on it.
Of course there is plenty of room inside a cannibalized reverb tank.
Sigh. Back to the day job.
Today's little mental holiday led me off to reverb pans, that topic having been current here a lot. Reverb pans are a lot like pencils; there are many kinds and sizes, most people kind of know how they work, and manage to use them without a lot of thought. There are more modern updates and changes that do much the same thing, but there is a core of folks that prefer the feel and results of the old fashion Eberhard Faber #2 to any mechanical pencil or pen.
The Beltone BTDR-2 is by most accounts an OK-ish replacement for the usual reverb tank if you design the circuitry to go around it and tinker till it sounds right. But it's not a direct replacement. My speculation was that I could design something with it that was a modestly useful direct replacement.
In my mind, that means that the thing is the same size or smaller than the usual reverb tanks, mounts in a similar way, and connects in and out with RCA phono jacks like the originals. I allowed myself a departure in that I thought it was OK if it had a DC power jack added, as I didn't think I could make it practical with batteries.
The idea is that this hypothetical lump could be used like the reverb pan in an amp; the original pulled out, a DC power wire and plug concocted, then the new thing plugged in with the old wires as a retrofit. That sounds simple enough, but it took me a fair amount of thinking to figure out how I think it would work. The problem is untying all the knots that were put in the original circuits to drive the cuss-ed spring tanks in the first place.
The BTDR-2 is easy enough to use. It needs 5Vdc at something under 100ma to run. When it gets that, it takes in audio at its input at something about like +3dbm or so and gives you two outputs that are delayed some, with repflections and such, at the output at roughly the same as the input level. These can be used separately or mixed to mono.
Real reverb pans come in a wide range of input impedances and as such need a wide range of input drive signals. They also produce miniscule outputs, on the order of 6-10mVrms. This needs recovery amplification on the order of the amount that moving coil hifi cartridges do. Worse, the better drivers for reverb pans drive them HARD to get past the 60-100db signal loss in the tanks from input to output. And the inputs have their nominal impedance only at the nominal 1kHz measurement frequency. They're actually inductors between 10mH and 250mH with varying DC resistances. Getting anything even crudely like a smooth frequency response requires driving the input with a waveform that makes a constant current per frequency happen in the inductive inputs, and that means that the good reverb drivers have outputs that ramp up in frequency at a rate of one time constant, doubling in voltage every octave. Fortunately, all the realistic ones give up on driving the coils much above 5-7kHz, or the driving voltage would get really out of hand.
And that's the real problem with subbing in a BTDR-2. You have to un-wind that input drive voltage from doubling each octave, AND let the drive circuits think they're driving a real coil, AND get the drive voltage to the BTDR-2 set up to be about level with frequency from low guitar up to maybe 7kHz, as well as something like line level or above. The fact that reverb drive circuits may either drive the input with a current-source signal, letting the voltage rise to keep the current per frequency level, or with a pre-equalized voltage source that rises with frequency but ignores how much current flows, assuming that the current will sort itself out. Some of them assume a floating, non-grounded input on the tank too. I didn't think I could do anything about that without a fancier power supply than I wanted to require, so I left off doing that.
The approach I took was to make the driver happy by putting a coil on the board for the driver to drive. The advent of switching power supplies has made small, radial-package inductors available in the market. You can get inductors in the Bourns family (and probably others) up to 120mH with a series resistance of less than 100 ohms for about $2. A pair of these lets you fake the input impedance of a "1475 ohm" Accutronics tank if you put them in series. By picking other inductors from the same line, you can fake the inductance and resistance of the lower-impedance reverb tanks. The drive circuit will then not know it's not driving an unusually stable reverb pan input.
But that can have as much as +/-30V peaks on the inductors. So I designed an R-C network to take the inductor voltage as an input, and reduce the voltage linearly with rising frequency by using a cap to shunt voltage to ground. This unwinds the rising voltage drive to a linear-with-frequency signal. It has the advantage of lowering the really high drive voltage at high frequency. The signal still probably isn't at the right level, but signal level changes is what opamps are for, right? So I allowed myself one dual opamp. The front end opamp takes the frequency-leveled drive signal and converts it to the right level for the BTDR-2 by having a variable gain. Just to be sure the opamp doesn't kill the BTDR-s, there are some clipping diodes to keep the level into it from getting out of hand. The second section of the opamp buffers the BTDR output and drives a resistor/pot network to cut the signal back down so the "recovery" circuit in the amp will get only its measly 10mV and won't overload.
Power supplies was an issue. the BTDR needs 5V, the opamps need at least that, and it's poor economy to have to build a whole new power supply for this thing. About the best I could do was to put in a series of three voltage regulators. A 7824 will take in up to 40V safely. This drives a 7815, which then drives a 7805. That seems downright silly, until you note that 100ma at 40V is 4W, and things could get hot. This sequence of regulators spreads the 4W out so the regulators are all eating about the same power, and this gets each of them under the 2W that a TO-220 can get rid of in free air. So this thing can be driven by DC supplies from about 9V up to 40V by selective population of regulators and jumpers. So it's probably going to be at home in any amp with a power supply between 9V and 40V positive.
That doesn't help much with tube amps, unfortunately. Another 100ma out of a 500V B+ is going to generate another 50W of heat even if I had a circuit to do it and get rid of the heat. Not elegant. A switching step down is also not a good idea inside a tube amp. About all I can come up with here is a small toroid-based power supply running from the 120Vac or a wall wart separate from the amp.
That part is still TBD.
The circuit board is about 2.5 "by 5.5" at present, with the DC power, input and output phono jacks on it.
Of course there is plenty of room inside a cannibalized reverb tank.
Sigh. Back to the day job.
Comment