Something New from Doc Sketchy

[indigoid]

for your next trick, make it generate an optimised (fewest different parts) BOM using only E12 values

[Dr. Sketch-n-Etch]

What does that mean for Frequency Shifter owners of light year's past?

Nothing. I used a thing on the internet called QuadNet to design the optimal PDN for my previous design, and my new code gives exactly the same values.

The issue was that a) QuadNet is a pain, and b) it wouldn't allow me to design down into the LFO range.

[Dr. Sketch-n-Etch]

for your next trick, make it generate an optimised (fewest different parts) BOM using only E12 values

Actually, all kidding aside, I'm doing exactly that, but with E96 values. It won't be an "optimized" BOM, though -- I'll have to specify the C values I want to use, and then it will automatically generate the 1% resistor pairs to achieve the RC values. I know how to do it -- I just haven't done it yet.

[Dr. Sketch-n-Etch]

UPDATE: I wrote the little code for the resistors, and now I can use the PDN90RC function to calculate all the RC time constants for the phase displacement network, set the capacitor values, and then use my new E96RES function to find the nearest lower value of resistance from the E96 series. By subtracting that result from the original value and using the function again on the difference, I can find the value of the second resistor to put in series with the first to give RC within about 0.02% of the desired value.

For example, one of the RC values is 0.06308266 s. So, if I choose a 1uF capacitor, the resistor I need is 63083 ohms. So, I use the function once:

E96RES(63083) = 61900

Then I use the function again with the difference:

E96RES(63083 - 61900) = 1180

So, the two resistors in series are 61900 + 1180 = 63080, which is within 3 ohms of the target of 63083 for a relative error of -0.004%. Using this method to generate the resistor pairs for every stage, and then calculating the new RC time constants gives a phase angle plot which is indistinguishable from the original one.

Pretty sweet, eh?

[KSS]

Very sweet!

[Dr. Sketch-n-Etch]

UPDATE: I just finished laying out the 4-channel morphing scanner design with auto-scaled CV range on a single PCB which is 4" x 6.4". It's pretty elegant. I'm making one for somebody already. If anybody else wants one, please let me know.