A couple simple changes fixed and enhanced the LM386 Wien Bridge Audio Oscillator but also discovered a big problem
From yesterday:
https://wb9kzy.blogspot.com/2025/06/lm386-wien-bridge-audio-oscillator.html
Several items from the punchlist have been solved / changed but one big problem was spotted.
1) the waveform looks "better" - accomplished by changing the feedback resistor in series with the diode pair from 10k ohms to 15k ohms. Using the open loop feedback resistor value of 5.7k ohms, the ratio of 5700 / 410 = 13.6 which represents a gain of about 14.6 for the maximum undistorted output. When the diodes were in the loop the resulting gain wasn't high enough to produce undistorted output ?
2) the waveform is now tuneable - I first tried a .22 uF (10x) cap in place of the .022 uF cap connected to the LM386 output along with a change of resistor value to 4.7k ohms from 51k ohms. This did not work resulting in either noise or a noisy square wave. Before giving up on the 10x cap idea I tried moving the .22 uF cap to between ground and pin 2 of the LM386. This worked ! Next step was to replace the 4.7k ohm fixed resistors with a dual 5k ohm pot. This also worked ! The waveform was now variable from less than 500 Hz to well over my audio limit above 12,000 Hz !
 |
| near the high end of tuning |
 |
| the low end of the tuning range |
One
final thing: added series 1k ohm resistors to both legs of the 5k ohm
pot to provide a lower frequency output but the results weren't good.
The lower frequency only went down to 370 Hz. But the
amplitude got weaker rapidly with higher frequency, not sure why.
Bottom line for now: 430 Hz is low enough for me.
3) I made a number of component changes, hopefully mostly minor, here is an updated schematic:
4) the big problem happened after I tried connected a speaker directly to the output. This caused the output to wilt, distort and change frequency. I had been using a small transistor radio type speaker connected to the output but through about 240 ohms of resistance (27 ohms + 200 ohms + 15 ohms). This was done to knock down the volume - it gets tiring listening to a loud single continuous tone. So I will have to abandon the speaker and connect an 8 ohm resistive speaker dummy load instead. Usually this circuit is run at a higher frequency, 1 kHz, which seems to help. I will try using larger output caps to see if that helps at the lower end. I may also try a different LM386 in case I have damaged this one in some way. And it appears that this application of the LM386 needs to be run at high volume in order to allow enough amplitude to actuate the diode feedback.
The National Semiconductor blurb in the Audio Handbook (see yesterday's blog post) just had a load of RL with no exact value specified - on other sample National Semi Audio Handbook circuits for the LM386, a speaker is shown so the lack of a speaker may be telling. Also, the video mentioned in the day before yesterday's blog specified a 330 ohm load:
https://wb9kzy.blogspot.com/2025/06/mistakes.html
So maybe the CPO idea is NG if the LM386 Wien can't hack a loud speaker - now low volumes through a resistor will be fine, especially with headphones. I'll keep working on it, especially getting it to work with a low impedance load. For a little while I thought I had most of it working :(
Best Regards,
Chuck, WB9KZY
http://wb9kzy.com/ham.htm
