I found the cause of the 1V/octave scaling problem I mentioned in my previous post. The problem was that I substituted a CA3086 transistor array for the CA3046. This is listed as a substitution on Dave Brown’s Mouser BOM. I probably chose the CA3086 because that’s what Bridechamber had in stock at the time, or something.
At first glance the two parts are viable substitutes for audio applications. Even looking at the Intersil web pages for both parts does not reveal the difference. The difference is in the HFE current gain parameter; the CA3086 has a gain of 100, while the CA3046 has a gain of 110. The Intersil web pages actually have a typo; HFE is stated as 100 for both. It wasn’t until I did a parameter-by-parameter comparison of the PDF data sheets for the parts that I found the difference.
So, the current gain is lower in my circuit, and hence the current coming from the CV summing amp isn’t sufficient to increase the oscillator frequency by 1V/oct. To correct this I reduced the value of the 51K resistor that follows the scaling trimmer pot. For my test I used a couple of clip leads to put a big resistor in parallel. With the resistance lowered I was able to get the oscillator to track at 1V/oct with no trouble. I have some 47K resistors on hand; I’ll try replacing the 51K resistors with those.
UPDATE: the above analysis is incorrect. The CA3086 is a perfectly viable substitute for the CA3046 in this context. The final entry on my calibration problem is here.
