More Power

Making Sine Waves with Oomph

For both the ESR and Octopus test adapters, it would be nice to have a cheap & easy to make sine (and perhaps square) wave source that could be built into the same box. There are plenty of schematics around for Wien bridge generators, unfortunately they tend to be for about 1kHz. Also, if you notice, common "jelly bean" op amps have a rather high output impedance, not what you want for an octopus.

The octopus I built was designed with about 90ma flowing through the voltage divider. Even with a generator directly connected without a voltage divider, you would want an output impedance of about 50 ohms maximum when using a 1000 ohm reference resistor. Having too low a source impedance will distort the waveform display.

There are of course, audio amps are made to deliver power into a low impedance. Most of them have too low a bandwidth for this purpose. Two exceptions are the LM386 and LM384 audio amps. With bandwidths of 300kHz and 450kHz respectively, these hold promise. The power rating of the 386 seems a little low for this purpose, but I have one handy.

This sine wave generator is based on the example shown in the 386 data sheet (search for "386").

The circuit was assembled on a solderless breadboard and was tested at both 1.5 kHz and 150 kHz.

Notes:

For 100 kHz operation, 1000pF caps(C3, C4) and 1592 Ohm timing resistors(R3, R4) can be used.
The range of adjustment of R2 that produced good sine waves was extremely narrow. Once the pot was adjusted, the sine wave generation was stable even if the timing capacitors or the load on the output was changed.
Finding a suitable lamp proved more challenging than expected. Most flashlight bulbs draw too much current to work in this circuit. Of the lamps carried at Radio Shack, two were found that would work. They were rated at 12V - 25ma and 6V - 25ma, most of the tests were run with the 12 volt bulb.
Based on rather limited information, the optimum operating range of a lamp would be about 8% volts and 25% current. The stability of the oscillator depends on the non-linearity of the bulb's resistance. The VI curve for the lamp can be seen here The few times this was mentioned in the literature, the filament was said to be dimly glowing or not glowing during operation. .
Operation at 150 kHz: Some crossover distortion was visible under all conditions. This could be minimized by using a voltage source of 5-9 volts. The maximum output was 2 V(peak to peak). Supply voltages above 9 volts resulted in high frequency parasitic oscillations
Operation at 1500 Hz: When using the original layout, the maximum output was 2 V(peak to peak. By adding a voltage divider at the input to the RC circuit, it was possible to increase the output voltage to 4 V(peak to peak), this trick didn't work at 150 kHz. To avoid clipping in the output, a minimum power supply of 7.4 volts was required at the increased output voltage. Refer to modification "A" on schematic. I didn't spend much time trying to optimize this modification, so it may to possible to increase the output voltage even more.
One could raise the output voltage more by using a second 386 as amp. Keep in mind that the maximum input voltage for a 386 is +/- .4 volts
Adding capacitor C1 increases the inherent gain of the 386 from 20 to 200. Since the circuit has to operate at a gain of three, perhaps six with modification "A", it's not clear why the engineers at National put this part in their schematic . Capacitor C2 is called a bypass cap, no value given, the datasheet doesn't explain what this does and the circuit seems to work ok without it. C5 and R5 are probably there to suppress high frequency parasitic oscillations. If used, C7 would typically be .1uF

Previous notes on this project

7/3/99
Initial testing on a solderless breadboard revealed the following:

The circuit in the National datasheet can be adapted to run at 100 kHz and is insensitive to variations in load. 100 pF timing caps/1000 ohm resistors were used to run at 150 kHz, just switching the caps to .1 uF allowed it to run at 1.5 kHz without further adjustment.
The 386 has a maximum input of +/- .4 volts, since a Wien bridge must operate at a gain of 3X, this limits the output to +/- 1.2 volts. On my initial tries, I could only get 2.0 volts (peak to peak). (The 384 is limited to +/- .5 volts input.)
The circuit was extremely sensitive to the values of the lamp & feedback resistor used. Based on very limited data, it looks like a lamp needs to be run at about 8% volts and 25% of rated current. The feedback resistor needs to be set so the amp produces a gain of three. Remember that the amp needs to produce the current through the lamp in addition to the output current.
The tests were run with a feedback resistor of 156 ohms and a 12 volt/24ma lamp. This lamp seemed marginal for operation. Will try a 6 volt/24ma lamp later.
The 386 lived up to it's reputation for unwanted high frequency oscillations, be sure to put a bypass cap on the power supply pins.

[Home Page]