The MIG II Preamplifier
This is an add-on for a Leslie 760 or similar speaker. It was designed and marketed by Keyboard Partner and is meant to make a transistor Leslie sound like a valve Leslie. It is no longer available because of the availability of the Russian miniature pentode valve. I have such a valve so I made my own MIG II preamplifier using the circuit diagram supplied by Keyboard Partner as a starting point. All credit goes to Keyboard Partner for an excellent design and I hope I have not caused offence by reproducing the design for my own use.
I have made a few changes, particularly to the gain structure. This is no reflection upon the original design but a consequence of the actual parts used in my implementation. The Keyboard Partner Website does identify that the using alternate manufacturers IC's will affect the outcome. Indeed, no type number is given for U3 in Keyboard Partners' Website. I had initially thought this may have been a 4049 Hex Buffer, but it was not until I attempted to install one that I realised it had 16 pins and the design called for a 14 pin device. I used a 4069 type instead.
The circuit employs a Russian RF pentode valve, the 1SH18b in its first stage. This valve was designed for use in the MIG fighter plane. Sadly it is not easy to find now (2010) but 5-years ago they were readily available as surplus items. I can only imagine how much they would cost if they were made today and bought new; perhaps £100 or maybe £200 each. The electrode spacing must be minute to work at such low voltages. In quiescent state and 30v supply, the voltages measured here are Anode: 15 v, screened grid: 25v cathode: 0.179v Heater:1.45v. The data I have found on the internet is fairly scant and I'm not sure how the cathode is configured. It is obviously directly heated and the voltages I give are for the two heater connections but how they equate to a cathode bias voltage, I do not know. Using the values in Keyboard Partners' design gives a 10v drop across the 33k ohm anode resistor and driving the stage hard results in the anode bottoming out at about 7 volts so the biasing is fine.
The second stage of amplification, albeit with a gain less than 1 in my version, utilises a CMOS logic device in linear mode. As I understand it, a CMOS device so used, produces a good deal of second and other even harmonic distortion as clipping is approached. It may seem an odd thing to do but when the aim is to produce distortion, anything goes. The third stage, the 4016, is a switching stage. I would expect it also adds further even harmonic distortion.
The combination of stages may be compared to those in the valve amplifier it emulates. The first stage is much the same as in a valve amplifier, producing similar distortions and the second stage is akin to a push-pull output stage, again with similar distortion characteristics.
I designed a printed circuit board using entirely free tools. Firstly entering the schematic into LT Spice IV and generating a netlist file and then importing that into Free PCB.I have used LT Spice before. I used it in the design of the Guitar Amplifier,shown elsewhere on this site. In that application, I used the Spice simulation features to develop the design. In this case I simply used it as a means to generate a netlist. The Simulation would have been very useful but the necessary data for the Russian valve and for a the CMOS devices, particularly the 4069 operating in linear mode was not available. I had not used Free PCB before and I am very impressed with its ease of use. Wonderful. Here is the bottom copper trace and the Top Silk screen. High resolution images may be obtained by clicking on the images.
I made the PCB using the PCB fab in a box system. I've used numerous methods for making PCB's in the past but this is probably the best I have achieved to date. The drilling leaves a lot to be desired. My main difficulty here is in seeing, despite using an illuminated magnifier. One day I shall make a CNC drilling rig. It is however, adequate.
I powered up the board with 30v and connected an input from my audio mixer. Tones were generated within a PC running Linux 64Studio. I viewed the output on an ageing CRT oscilloscope. I noticed the circuit misbehaved when driven into clipping. The positive half waveform did not clip normally but made an attempt at clipping but the peak appeared to break though at a level greater than the supply voltage on the CMOS circuits. Surely impossible? It turned out that the input positive excursion exceeded this supply voltage it broke through the 4016 bypass switch (U2b in Keyboard Partners' circuit diagram). Since the whole reason for the circuit is to clip, or to approach it, this behaviour is undesirable. For now, I have lifted the blocking capacitor C2, effectively disabling the bypass circuit.
This is a trace of the valve anode voltage when driven hard into clipping at 100 hz. Note the nicely rounded or soft clipping. The peak-to-peak voltage is about 16 v.
Here the 100hz drive at similar levels to above is augmented by a superimposed 5khz signal . Again the anode voltage is displayed. It shows that the higher frequency gets through even in the clipping region. Hard clipping would give flat tops and bottoms with little of the higher frequency getting through in those regions. This means that the characteristic of the sound source can be heard through the distortion. A highly desirable feature.
This is the anode signal set at 5v p-p. There is no discernible distortion. Driven at this level, the valve will contribute little to the distortion of the preamplifier as a whole. With component values as given in the the Keyboard Partner website, it just this sort of level that would drive the second into hard clipping. In my version, I have reduced the gain of the second stage from 1.77 to 0.319. The second stage clips hard at about 4.5 v p-p output. The drive necessary to achieve this is 4.5 * 1/0.319 = 14.1 volts.
The third stage, the switching stage receives a R7/(R7+R11) = 5/6 of the second stage output. This would appear to equate to a gain that allows the two stages to clip simultaneously. I did not change these components from design.
Here is a trace of the final output when driven by a 100hz + 5k hz signal hard into clipping. This looks as though it should sound OK.
My organ gives about 4.5 v ac output. I adjusted the input attenuator to allow a decent drive to the valve stage. I.e. I changed the grid leakage resistor to 150k ohm.
Installation and Trial
Here is the finished preamplifier installed in my Leslie 760: quite straightforward. I have not connected the switching inputs as yet. I couldn't wait to find out how it sounded.
I set up the 'main' level to give 3.5v ac on the anode, measured with a multimeter and with the organ swell fully depressed, a handful of stops fully out and a weight on the keyboard holding down about 4 keys. Then I set the bass and Treble gains to give a tolerably loud sound.
I have had a bit of a go and I like it. It will growl, it will scream and it will play cleanly. Surprisingly, it seems to have improved the coherence of the pedals. Before, they always seemed to sound somehow apart from the rest of the organ and now they fit. However, I am well aware of the power of the placebo and that the opinion of the pill maker, the pill here being a preamplifier, should be treated with the utmost suspicion.
What is lacking is an adequate organist, which I am not. It is just possible that I may improve with time and with practice and if I do, I may include a recording on this site.