CGS48 VCO calibration HF compensation

From circuitbending to homebrew stompboxes & synths, keep the DIY spirit alive!

Moderators: Kent, Joe., luketeaford, lisa

Post Reply
tardishead
Wiggling with Experience
Posts: 332
Joined: Sat Aug 03, 2013 3:00 pm
Location: London

CGS48 VCO calibration HF compensation

Post by tardishead » Fri Oct 18, 2019 10:49 am

If I calibrate my 3 cgs48 vcos starting at G1 the first 3 octaves I can get right within 0.5 cent.
But when I go for the next 3 octaves higher the best vco goes about 3.5 cents flat and worst up to 10 cents flat.
Is that about as good as can be without a hf compensation trimmer in the design?
And why is one worse than another? I am using original lm394 transistors. Is it possible one could not be as super matched as the others?
Last edited by tardishead on Thu Nov 07, 2019 7:13 am, edited 1 time in total.

tardishead
Wiggling with Experience
Posts: 332
Joined: Sat Aug 03, 2013 3:00 pm
Location: London

Post by tardishead » Sat Oct 19, 2019 3:25 am

CGS YAHOO group seems to be defunct
Anybody have any experience with these

User avatar
wackelpeter
Veteran Wiggler
Posts: 532
Joined: Fri Apr 17, 2015 5:56 pm
Location: Germany, MD

Post by wackelpeter » Sat Oct 19, 2019 5:12 am

i wish i could help, as i build 4 of those ASM1 VCO's which this is based on but i used a expo converter design by Thomas Henry (including HF trim) with the addition of some Buchls-style FM... and this works well... over at least 6-7 octaves i have just a tiny amount of precision lost... say startat 50Hz i can reach 1600 and 3200Hz with just 5Hz up or down in this region, while on the lower end it's mostly spot on or starting at 400 maybe 1 or 2 and worst 3 Hz off... (if i can trust at least my frequency counter)

Maybe just breadbord another expo converter and replace this for some test if you can achieve better results, then you now at least the core of the VCO is okay and maybe can solve that problem otherwise...

Perhaps it's possible to insert some HF compensation in that expo converter too (i think so)... breadbord and test if it betters the results...

Well i have to add i used a pair of matched 2n3904 or 3906 in my build (would have to look it up which one is pnp or npn i can't remember such things) I tested them with the transistor matcher YATMA that i think elektrouwe posted some time ago here...

also i guess it's obvious that you have used styroflex or wima FKP, MKP or any other better suited caps as the timing cap and perhaps look what kind of CA3140 or LM311 you use (especially meaning did they came from a reliable source and not some ebay sellers in asia or elsewhere as there were some complaints about fakes or poor reproductions lately)...

also maybe here some useful information could be found:
http://electro-music.com/forum/forum-165.html

elektrouwe
Common Wiggler
Posts: 157
Joined: Mon Apr 08, 2013 4:37 am
Location: wild south of Germany

Re: CGS48 VCO calibration

Post by elektrouwe » Sat Oct 19, 2019 6:03 am

tardishead wrote:...
Is that about as good as can be without a hf compensation trimmer in the design? And why is one worse than another?
are you talking about the 680R in series with the integrator C ?
Replace it by a 1 k trimmer and tune again. Then leave the trimmer or replace it back with a fixed R having the found value.
This is a "Franco compensation" which compensates VCO linearity error coming from the reset time pulse width. This reset time is defined by the comparators (constant) feed forward RC timing but it's also a (nonlinear) function of the comparator delay.
"Rossum compensation"for the expo is something else. It compensates the internal base bulk resistor of your expo pair. I guess the LM394 has a low enough Rbb, so you don't need this kind of compensation.

In almost every discussion about VCO tuning transistor matching is mentioned.
Matching has nothing to do with tempered scale accuracy !
It's only about temperature stability !

tardishead
Wiggling with Experience
Posts: 332
Joined: Sat Aug 03, 2013 3:00 pm
Location: London

Post by tardishead » Sun Oct 20, 2019 2:28 am

So if I replace the 680R with a 1k pot I treat this like a hf compensation trimmer find the correct value and sub with a fixed resistor?

Does anyone else have any experience with this vco?

elektrouwe
Common Wiggler
Posts: 157
Joined: Mon Apr 08, 2013 4:37 am
Location: wild south of Germany

Post by elektrouwe » Sun Oct 20, 2019 4:32 am

tardishead wrote:So if I replace the 680R with a 1k pot I treat this like a hf compensation trimmer find the correct value and sub with a fixed resistor?
yes. Calculation is also possible, if you can measure integrator reset time with a scope:
R_franco = T_reset / C_integrator
example: reset pulse width =2us, C= 2n2 => R =2/2.2 kOhm =909 Ohm

tardishead
Wiggling with Experience
Posts: 332
Joined: Sat Aug 03, 2013 3:00 pm
Location: London

Post by tardishead » Sun Oct 20, 2019 1:28 pm

yes. Calculation is also possible, if you can measure integrator reset time with a scope:
Right I'm going to look into this. It seems strange that Ken did not write about this in the build docs
Last edited by tardishead on Sun Oct 20, 2019 2:53 pm, edited 1 time in total.

elektrouwe
Common Wiggler
Posts: 157
Joined: Mon Apr 08, 2013 4:37 am
Location: wild south of Germany

Post by elektrouwe » Sun Oct 20, 2019 2:37 pm

[edit] answer removed because question has been removed
did you try the trimmer meanwhile ?
Last edited by elektrouwe on Wed Oct 23, 2019 2:17 pm, edited 1 time in total.

tardishead
Wiggling with Experience
Posts: 332
Joined: Sat Aug 03, 2013 3:00 pm
Location: London

Post by tardishead » Sun Oct 20, 2019 4:09 pm

mmmmmm I don't think my scope is high res enough. I can just about find it but it looks like an impact transient or a dampened oscillation. Not really a discernible pulse. I can't discern any pulse width out of it.

tardishead
Wiggling with Experience
Posts: 332
Joined: Sat Aug 03, 2013 3:00 pm
Location: London

Post by tardishead » Wed Nov 06, 2019 6:56 pm

elektrouwe wrote:[edit] answer removed because question has been removed
did you try the trimmer meanwhile ?
Thanks elektrouwe
I put in a trimmer in one of my VCOs and got very good results.
I found the resistor needed was 950R
Now I can track reliably 6 octaves (maybe more..... haven't checked yet!)
This is very much worthwhile
Thanks again!

elektrouwe
Common Wiggler
Posts: 157
Joined: Mon Apr 08, 2013 4:37 am
Location: wild south of Germany

Post by elektrouwe » Thu Nov 07, 2019 2:29 am

tardishead wrote: I put in a trimmer in one of my VCOs and got very good results.
I found the resistor needed was 950R
cheers! good to hear you got it solved.
950R - makes sense, close to my guess for 2us reset time.

tardishead
Wiggling with Experience
Posts: 332
Joined: Sat Aug 03, 2013 3:00 pm
Location: London

Post by tardishead » Thu Nov 07, 2019 11:57 am

I really don't know why something was not mentioned about this in the build docs. A trimmer at this point would have been a great addition and would make it easy to build a precision VCO. I changed the title of this thread so hopefully if someone has a similar query they will find out what to do

Post Reply

Return to “Music Tech DIY”