ASM-1 VCO clocked and framed!

[Magnus Danielson]

Hi!

I have long wondered just how far one can talk a VCO sweep from
homebuilt stuff. Last night I finally got time to make a little more
methodicall search at work with the ASM-1 VCO. I knew it could sweep
widely, but at home I dont have equipment worth noting... 

Anyway, searching from the middle to the upper and lower ends it
stopped at 102 kHz and 89 mHz. That is a range of over 20 octaves.

The VCO will draft away in the high range due to a raising floor, so
it will not reach the bottom at the reset. Since this raise will it
take less time to reach the 5V level that creates the frequency will
be slightly larger than intended. When looking at the waveform at this
frequency one also see an rather large ripple just as the waveform
reach the bottom level. One migth first think that this is a high
frequency error, but view this part in lower frequencies it becomes
apperent that the frequency has little to do with this additional
noise. A close look at the transition area and also monitoring the
output of the LM311 one see that this ripple is caused as a result of
the LM311 turns off. This ripple also travels to the powerlines, since
it came up on my monitoring of the CV-summing op-amp.

The VCO stopped at 89 mHz (That is 0.089 Hz or over 11 second
period!), when I looked at it I discovered that this was the result of
the summing op-amp, it could not deliver more than 14.17 V (which then
is divided by a resistor net to a 57th part).

The output of the summing op-amp covered a range of -8 V to 14 V.
One should keep in mind that the CV input gets slightly scaled by the
CV input summing cursuit.

At the "rest-point" frequency, that is, totally unmodulated, will the
VCO oscillate at 880 Hz or there about. This is consistent with my
little math excersie which gives:

	     3
	f = ---
	 0  CR
	      1

Given that

	R = 1.5 Mohm
	C = 2.2 nF

which gives 909 Hz, some 3% off..

Actually, those 3% is in the right direction :)

I have been thinking of an alternative reset cursuit that would feed
current into the current summing point rather than shorting the cap.
This would have the benefit that the operating voltage for the fet or
bipolar would be constant as opposed to the jfet shorting the cap.

The falltime of my VCO is 732 ns and constant over the modulation
range. This is the constant error of cycle time that has to be
accounted for in the high ranges. Fortunatly is the pitch judgement
bad in the 20 kHz and above range :)
Also, the floor raise will make a worse error.

I will put the hardcopy images available on the web sometime soon...

When you see them, keep in mind that there are several measurement
errors, so you can't really accept all the digits in the numbers etc.
They where intended to get the basic feel of what is going on rather
than good numbers and nice curves. However, it is nice to fool around
with a good sampling scope, so I did have some fun :)

Comments and suggestions are wellcome.

Cheers,
Magnus