So I wanted to generate an audio wave using the SERIAL port of the Commodore 64. Since the lines ride at 5 V DC, but can alternate between 0V (signal on) and 5V (signal off), pulses should be possible. I wondered if I could modify the waves further, by changing the rise time. But that was not possible. 

Using ATN line to generate pulse waves

See picture from my oscilloscope above. When the C64 is idle, the ATN line stays at 5V. If we want the attention of devices connected to the serial port, we do this by dropping the line to 0V. In my code I made that the default situation. So "do nothing' in my case means keep the ATN line at 0V. Then, when wanting a pulse-wave, "let go" and the line will rise to 5V DC, but at the cost of a peak, as you can see. The pulse-width is 50%, as you can see, so at 50% of the period, I drop the line back to 0. Again at the cost of a peak, before it goes back to 0V. In this instance above, the peak-to-peak voltage is a total of 8.6. 

The rise time (or drop time) is very fast. I am unable to modify it with software. It is within microseconds. Even if I set it to go to 5V and follow immediately with a call to drop it to 0V, the peak is there. 

I had a plan to feed that back to EXT-IN on the SID, to passively mix with the other three channels, but that is impossible with this Vpp level. The recommendation is for any EXT-IN signal coming in the SID, to ride at 6 DC, but only at 3 Vpp. 

With 8.6 Vpp I am nowhere near that max limit, so let's not do it!

Example of music with SERIAL port ATN pulse wave

Nevertheless, it is possible to just hook up the pulse-wave to an audio set and have that extra channel. As a proof of concept I simply modified my PULSE-FX player and have the player not use gate on/off to generate the pulse wave on a channel, but use the SERIAL port ATN line low/high for the wave. 

Check the MP3 of the result (serial port pulse wave at left channel):