As announced some time before, I traveled to Munich Monday evening and returned yesterday (Tuesday) evening. These 29 hours were split in 5 hours of driving, 18 hours with Chris Henkel (also known as "o8o8" or "bytesatwork" , who owns an EPro which served as a guinea pig) and his wonderful family (by the way: Chris is a fantastic cook!), and 6 hours of driving back home.
What was the goal of this trip, out of amical reasons?
I brought with me a small circuit board comprising several sub-circuit variants and a rotary switch which allowed to compare directly the effect of these variants on the EPro's volume response. It was an extended version of the much simpler switch board that I installed last March during the Without Touch 3.0 festival in his EPro. The latter had only two switch positions which allowed to switch between the original circuit (R67 = 3.3k) and the Moog modification (the known 3 resistor and 2 diodes network). At that time we had neither the time nor the occasion to do extended measurements and tests out of realizing that the Moog mod worked and smoothed the volume response. There was still a small question mark left since I used 1N4148 diodes (more common in Europe) instead of the 1N914 diodes as Bob Moog had suggested. Both types have identical technical data so that there should be no problem. But Randy George told us the same evening that he had tested both diode types in his EPro (sequentially, not in a direct A-B comparison) and found that the 1N914 would give better results. I could not believe that but planned already to build a test board which would allow to compare both diode type variants in order to see if there was really a difference, or if the "felt" difference resulted only in diode and resistor production tolerance. Some days later this issue was (naturally) discussed on Levnet and Peter Pringle told that he wasn't still satisfied with the modified volume response of his EPro.
Before I tell - finally - what happened yesterday, I'll quickly repeat my theory of operation. It has been based mainly on lots of theory, some guess and a bit of luck, since I do not own an EPro myself and complete schematics are not available:
R67 is part of a circuit which does volume cv processing. This circuit is located between the discriminator stage after the volume oscillator and the internal VCA of the EPro. It seems that the voltage of the volume cv output is not affected by any modification of this circuit. The original circuit (with constant R67) has a linear response. That means that the internal volume cv is strictly proportional to the volume discriminator output. The snappy volume response is an effect due to unlinearity either in the VCA or the volume discriminator circuit. In order to reduce this effect, Bob Moog divided the response curve into 3 zones with different slopes, using the two diodes not only for automatic switching but also for a soft transition between these zones, so that one does neither hear nor feel the bends in the curve. My idea was now to use four or even five zones with still less conductance, hoping that this would give the desired results.
So the prototype which I took to Munich had 5 sub-circuits which can be selected by a rotary switch:
1) The original 3.3k resistance
2) The Bob Moog 3 zones modification, built with 1N4148 diodes
3) The Bob Moog 3 zones modification, built with 1N914 diodes
4) The Thierry Frenkel 4 zones modification
5) The Thierry Frenkel 5 zones modification
The first thing I did after connecting my prototype was to check some EPro internal voltages in order to see if some of my theories were right and if the following tests would make sense. Then I soldered rapidly an "adapter" with a 1/4" jack and plug and about 4" of wire between. This adapter, connected between the audio output of the EPro and the amplifier (for audible feedback), allowed me to connect my pocket oscilloscope. During all tests we
What was the goal of this trip, out of amical reasons?
I brought with me a small circuit board comprising several sub-circuit variants and a rotary switch which allowed to compare directly the effect of these variants on the EPro's volume response. It was an extended version of the much simpler switch board that I installed last March during the Without Touch 3.0 festival in his EPro. The latter had only two switch positions which allowed to switch between the original circuit (R67 = 3.3k) and the Moog modification (the known 3 resistor and 2 diodes network). At that time we had neither the time nor the occasion to do extended measurements and tests out of realizing that the Moog mod worked and smoothed the volume response. There was still a small question mark left since I used 1N4148 diodes (more common in Europe) instead of the 1N914 diodes as Bob Moog had suggested. Both types have identical technical data so that there should be no problem. But Randy George told us the same evening that he had tested both diode types in his EPro (sequentially, not in a direct A-B comparison) and found that the 1N914 would give better results. I could not believe that but planned already to build a test board which would allow to compare both diode type variants in order to see if there was really a difference, or if the "felt" difference resulted only in diode and resistor production tolerance. Some days later this issue was (naturally) discussed on Levnet and Peter Pringle told that he wasn't still satisfied with the modified volume response of his EPro.
Before I tell - finally - what happened yesterday, I'll quickly repeat my theory of operation. It has been based mainly on lots of theory, some guess and a bit of luck, since I do not own an EPro myself and complete schematics are not available:
R67 is part of a circuit which does volume cv processing. This circuit is located between the discriminator stage after the volume oscillator and the internal VCA of the EPro. It seems that the voltage of the volume cv output is not affected by any modification of this circuit. The original circuit (with constant R67) has a linear response. That means that the internal volume cv is strictly proportional to the volume discriminator output. The snappy volume response is an effect due to unlinearity either in the VCA or the volume discriminator circuit. In order to reduce this effect, Bob Moog divided the response curve into 3 zones with different slopes, using the two diodes not only for automatic switching but also for a soft transition between these zones, so that one does neither hear nor feel the bends in the curve. My idea was now to use four or even five zones with still less conductance, hoping that this would give the desired results.
So the prototype which I took to Munich had 5 sub-circuits which can be selected by a rotary switch:
1) The original 3.3k resistance
2) The Bob Moog 3 zones modification, built with 1N4148 diodes
3) The Bob Moog 3 zones modification, built with 1N914 diodes
4) The Thierry Frenkel 4 zones modification
5) The Thierry Frenkel 5 zones modification
The first thing I did after connecting my prototype was to check some EPro internal voltages in order to see if some of my theories were right and if the following tests would make sense. Then I soldered rapidly an "adapter" with a 1/4" jack and plug and about 4" of wire between. This adapter, connected between the audio output of the EPro and the amplifier (for audible feedback), allowed me to connect my pocket oscilloscope. During all tests we
