You have something important missing from your circuit. That is power supply bypass capacitors. You should place a 10 uf cap from +9 volts to ground and another 10 uF cap from -9 volts to ground. Then on each 8038 you should also put a 0.1 uf ceramic cap from the +9v to ground and -9v to ground. The 0.1 uF caps must be installed as close as possible to the 8038 power pins with short leads going to ground.

All of these caps are needed. If you don't install them the 8038 will not generate clean waveforms. In addition, with more than one 8038, the oscillators will have a tendency to lock to each other trying to run at the same frequency.

RichardO:

The bipolar power supply that I'm using is this Craig Anderton design :


Are the 10 uf caps in this diagram what you're talking about, or should I use additional 10 uf electrolytic caps?

 

Are the 10 uf caps in this diagram what you're talking about, or should I use additional 10 uf electrolytic caps?

If the leads from the power supply to the 8038's is more than a few inches then you should put additional 10 uF caps near the 8038's. If you are not sure, always go with more caps rather than less.

 

Wires and circuit board traces cause nano-henries. At high frequencies, like a square wave generator, a capacitor is only good for a few inches.

 

If the leads from the power supply to the 8038's is more than a few inches then you should put additional 10 uF caps near the 8038's. If you are not sure, always go with more caps rather than less.

Makes me wanna just buy an organ instead of make one

 

Makes me wanna just buy an organ instead of make one

Yebbut...you have the bugs worked out of this one.

 

Wired it up last night, per latest schematic, square wave output only for simplicity's sake. No sound. Need to troubleshoot today and tonight.

 

In addition, with more than one 8038, the oscillators will have a tendency to lock to each other trying to run at the same frequency.

Why would they do this?

 

Hey, it works after all (I found a disconnected resistor at one of the 8038s). The output was way too hot with the 10K at the op amp. After I reconnected the resistor and flipped the power switch, a very loud shriek made my hair stand on end before I realized that I was finally hearing two notes (my goal). I may put a pot there in the future for volume control. For now I'm using a 640 ohm resistor instead of 10k. The tone is very clean and the volume is now manageable.
So now I'm going to go ahead with building the organ with thirteen 8038s. Thanks to all of you for your help with this!

 

The output was way too hot with the 10K at the op amp. For now I'm using a 640 ohm resistor instead of 10k.

Now you know why everybody said 100k was wrong.

Have fun.

 

Why would they do this?

Without proper power supply bypassing, some of the oscillator's output signal will appear on the power supply leads. Even the best oscillator circuit has some variation in frequency as a function of the power supply voltage. If the frequencies of the oscillators are close together then they will both tend to "pull" to the same frequency. Even if the oscillators do not frequency lock to each other, their outputs will be frequency modulated by the noise on the power supply.

Note that the frequency locking can be subtle. For instance if one oscillator is running near an integer multiple of the other they can lock with the frequency of one at the integer multiple frequency of the other. Oddly, in an musical note circuit like the one being discussed here, if two oscillators, are adjusted with one close to twice the frequency of the other they will lock at exactly one octave apart.

One further comment. Frequency locking can also be caused by a ground loop where one oscillator draws its operating (and output) current through the ground of the other oscillator. Careful ground routing, such as a star ground or a ground plane on a PCB, are solutions to this problem.

 

@RichardO
Thank you for the explanation, if I've understood correctly we are talking harmonics here, the best analogy I can think of is the one I learned at school (a hundred years ago.) Is that of a piece of string being shaken up and down at different rates at each end, it creates a standing wave eventually, which is presumably the frequency lock you are speaking of.
Steve.