[sdiy] VCO that goes SHARP in highest octaves??

[Phil Macphail]

Thanks for sharing the results with us all. I originally fitted the resistor for the same reasons Harry mentions when using an LS358 in the expo circuit, where the transistors are closely (thermally) coupled. When I used an SMD transistor pair in an LFO I found that square-wave FM suffered from some thermal effects - the switch to a lower frequency would result in a fast initial change in frequency followed by a slower drift to the final frequency. The series resistor reduced this effect so I have always tried to minimise the power dissipation in any situation where transistor matching is required. I have also used cascode transistors as an alternative to the resistor with similar results,


Phil.

> On 19. Jul 2022, at 01:17, Neil Harper <metadata at gmx.com> wrote:
> 
> 
> 
> On 7/14/22 19:02, Ian Fritz wrote:
>> It would be interesting to compare the two separately, if you are so inclined.
>> 
>> Ian
> 
> I had a chance today to do exactly this!
> 
> (cents)	C4	C5	C6	C7	C8
> BEFORE	0	0	0	+3	+11
> IAN	0	0	-1	-3	-5
> PHIL	0	-1	-1	0	+2
> 
> each one required a slight adjustment to the scaling calibration.
> 
> Ian's solution (1/2 integrator cap, 1/2 reference voltage) returned the
> VCO to the expected operation matching the TH version. I'm used to using
> the HF-TRIM to null out this amount of error.
> 
> Phil's solution (4.7k at 137000's Iabc input) flattened the dip as well.
> The results here might not be perfect, because I lost the original
> integrator cap after modifying for Ian's solution and could only replace
> it with a generic/noname 1000pF cap with no claims of being NPO/COG. So
> maybe that could account for the +2 cent rise at C8?
> 
> Anyways, sorry for the bad science. (0805 components and a speckled
> epoxy floor do not mix)
> 
> I gave Phill's solution another try on another oscillator:
> 
> (cents)	C4	C5	C6	C7	C8
> BEFORE	0	0	0	+2	+8
> PHIL	0	0	-1	-1	-2
> 
> 
> This time, with the original NPO/COG cap in place, there was no rise at
> C8 - although the initial rise was less.
> 
> Anyways, I'm really happy with both solutions - thank you Ian Fritz and
> Phil Macphail for identifying the problem and coming up with solutions.
> 
> 
> --
> /// Neil Harper
> /// Every Wave is New Until it Breaks