For no particularly good reason, I’d like to build a variable-frequency power supply to directly drive a small Airpax two-winding, four-wire synchronous motor that drives a turntable. These are small clock-style motors that are made for either 50- or 60-Hz operation at 250 or 300 rpm. Often, they are operated with a capacitor, but a two-phase supply is probably the smoothest way to go. The value of the capacitor is selected for the frequency in use. If the frequency varies (to vary the motor speed), it will be wrong.
What I was thinking was to use a pair of LM1875-based supplies driving a small step-up transformer (these are nominal 110-V motors) supplied by an oscillator that would give me a precise quadrature pair of outputs, something along the lines of the Wien bridge HP bench generators with a variable capacitor. Yes, of course, more modern digital options exist. But I just thought I’d do it in the analog domain.
My goal is to arbitrarily vary the speed of the turntable from about 16 to about 90 rpm in two ranges, determined by the 33-rpm and 45-rpm pulleys on the motor that are used at the standard drive frequency. Could you point me to application notes that might be helpful in this instance?
I remember an oscillator circuit in AN-31 in our Linear Apps book. There is a “Low Frequency Sine Wave Generator with Quadrature Outputs.” This is easy to make, but I don’t think it’s that easy to tweak or trim over a range. The 0.02 and 0.01 and 0.01 caps could all be switched to set the range. Then you might need a couple of 22-MΩ pots ganged with a 10-MΩ pot. Needless to say, that is a lousy idea!
The “High Freq.” oscillator is the same idea. You will be working at moderate frequencies. You do not need a precision, low-distortion sine at either output. You need a pretty good sine. You might be able to gang a couple of LM13700s.
You might even be able to make a truncated triangle wave, and the motor would still be happy. The gain of an LM13700 may tend to drift with temperature, so your frequency might drift around. Maybe lousy. You could build one and try it out. If you’re just running it at home, with decent temperature control, it might be okay. But if you took it outside, it might drift lousy.
I know all about triangle waves: how to generate them symmetrically over broad ranges, convert them to sines, and make perfect quadrature matching. Wait a bit and I’ll send you a circuit, better than sines. It will make a good column, too.
At my first job, a project engineer asked me to find out why his project in production had 25% failures on one particular circuit. It was a 2N706 crystal oscillator driving a 2N706 buffer supplying a local-oscillator signal to a 7-dBm balanced mixer. The buffer output had to have certain RF voltage, and 25% were low. I fooled around before carting half the engineering department’s instruments out there and found the difference between radios that passed and those that didn’t: the 2N706 buffer had 50 mV more dc voltage emitter to base.
I could fix every radio by putting a 2N706 with the higher e-b voltage in the defective circuit. I reported this and was asked first how many this fixed. I replied, “10 in a row.” Then I was asked why my 50 mV dc made any difference at RF, and I answered, “That’s a good academic question and I don’t know the answer, but I fixed the problem and that’s what I was asked to do.” My question to you is to provide the academic answer of why a slightly higher dc VBE voltage affected its RF output.
Suddenly after 35 years, it has become a burning question to find out why the yield correlated. Ahem. And then I’ll have to ask you exactly what was the schematic of this oscillator. Swapping a transistor is one way to do it. But did you try adding in a little more bias current? I’ve done that.
I’m not an oscillator expert. I know the guy who was, Tom Mills, and he could answer that. But he had the bad luck to die a few years ago. Your bad luck, too. What frequency was this oscillator supposed to run at? How did you measure the VBE? Just the two terminals were measured? Or did you measure it while it ran as an oscillator? Whose 2N706s? What year? Obviously, you haven’t given me enough information.
I just ran into some good old bifilar-wound transformers for a 5-MHz oscillator that we used 40 years ago. We didn’t throw them all out. If it didn’t run right, we sometimes changed the transformers, and sometimes the resistor bias. I don’t recall that we ever changed the transistors. Good planar resistors are remarkably uniform.
To find some transistors with 50 mV of different VBE, it sounds like somebody at the transistor factory was sweeping a lot of dice into a bucket and bonding them all up. Even the bigger die would meet the JEDEC 2N706 spec, but it wouldn’t amplify right. If you opened up the cans, you probably would have found that the two different types of transistors were different geometries. The “bad” ones were probably bigger. That’s the best answer I can give you, since you ask 40 years late.
For more Bob Pease, go to http://electronicdesign.com/Departments/DepartmentID/6/6.html