Help understanding theory behind voltage controlled oscillator

[vmallet]

I have a Voltage Controlled Oscillator (VCO) in my hands that's part of a modular synthesizer (audio) and I am really trying to understand how it works. After hours of reading on op-amps, comparators and other related circuits I understand a lot of aspects of the design but I am struggling with the most critical part, the actual oscillator.

To make things a bit spicier I am working from schematics I drew from studying the PCB closely without the luxury of being able to remove components. Mistakes are likely even though I tried to double check things.

The schematic is attached below.

Questions:

• The U106/U104 pair seems to be the heart of the beat; what makes them oscillate? Of course Q101 is here to throw a wrench in my understanding of the picture.
• I am really thrown off by the U102A / U102B NPN pair (Analog Devices SSM2210, see datasheet below). I've cross-checked their connections a few times and re-checked the datasheet, I really don't understand what role they play and how they can control anything.
• U101 and its networks seem to be a summing amplifier designed to collect the various tuning inputs to adjust the frequency. But how does its output do anything, being tied to the base of U102B?
• Speaking of U101, I have trouble reading the value of its feedback resistor R107. "orange orange red black red". In one direction it's 332 ohms at 2%; in the other direction it's 202K at "orange"% which apparently "orange" is not a thing for a tolerance spec. And it's probably not 332 ohms because a measure in-circuit is about 2K and it couldn't be higher than its actual value, so that doesn't work. Mystery!

The PCB is powered by +12V (VCC), -12V (VEE) and GND.

The two input signals "SYNC_IN" and "UNUSED_1" are optional inputs on the PCB and they don't need to be connected for the circuit to oscillate so I am ignoring them for now.

The two potentiometers RV1 (FINE) and RV2 (COARSE) allow for fine / coarse adjustment of the frequency. It can go from below 1 Hz to well above 10-15 KHz.

The SAW_SIGNAL output routes through one more unity-gain LF351 amplifier (with Rf = Rin = 22K) before reaching the output jack that can be used throughout the rest of the synthesizer; I have omitted this part on the schematic.

Datasheets:

• NS LF311 Voltage Comparator: https://www.jameco.com/Jameco/Products/ProdDS/837812.pdf
• ST LF351 JFET Op amp: https://www.st.com/resource/en/datasheet/lf351.pdf
• Intersil CA3130 op amp: https://www.renesas.com/us/en/www/doc/datasheet/ca3130-a.pdf
• Intersil CA3140 op amp: https://www.renesas.com/us/en/www/doc/datasheet/ca3140-a.pdf
• AD SSM2210 Audio Dual matched NPN: https://www.analog.com/media/en/technical-documentation/obsolete-data-sheets/SSM2210.pdf
• Fairchild PN4391 NPN Switch: https://www.alldatasheet.com/datasheet-pdf/pdf/54318/FAIRCHILD/PN4391.html

[ledtester]

This little blurb on VCO might help:

https://www.circuitstoday.com/voltage-controlled-oscillator

The main takeaway being that you are probably dealing with a relaxation type VCO (presuming that the SAW_SIGNAL is a sawtooth wave)

I also found this:

https://electronics.stackexchange.com/questions/233607/sawtooth-vco-and-current-source-for-modular-synthesizer

Reading through the comments gives a lot of insight into how it works.

So it appears that you have a relaxation type VCO and the timing capacitor is C102. Q101 is used to discharge the cap.

In any case, it seems to be a pretty standard design.

[ledtester]

Some more sleuthing...

the stackexchange page mentioned:

I have been working on a current source and sawtooth core VCO based on the Electronotes/Aaron Lanterman/Ian Fritz designs/ideas.

Googling around I found:
- https://electro-music.com/wiki/pmwiki.php?n=Schematics.Schematics
- https://electro-music.com/wiki/pmwiki.php?n=Schematics.SawtoothVCOByIanFritz

which led to:
- http://home.comcast.net/~ijfritz/sy_cir2.htm

That page isn't active anymore, but it is available via the Wayback Machine, e.g.:

https://web.archive.org/web/20071015113517/http://home.comcast.net/~ijfritz/sy_cir2.htm

You might find more info by Googling "Terry Michaels Electronotes v.62".

[vmallet]

This little blurb on VCO might help:

https://www.circuitstoday.com/voltage-controlled-oscillator

[...]

Some more sleuthing...
[...]
which led to:
- http://home.comcast.net/~ijfritz/sy_cir2.htm
[...]

Wow this is epic, rabbit hole of good information around that topic. Thanks a lot, it is going to give me plenty to read to try to make sense of this.

And now I don't feel too proud because I had searched around first before posting here but didn't get anywhere. Well, actually I had bumped into this one:
https://schmitzbits.de/expo_tutorial/index.html
and turns out this guy has some similar VCO designs and links to Ian Fritz' VCO for inspiration but I hadn't read all the way there yet.

[David Hess]

I think the bipolar pair in combination with U103 is suppose to generate an exponential output current for a linear input voltage (voltage in to the power of fixed X) so that the control voltage goes by octaves instead of linearly.

The oscillator itself uses C102 and U104 to integrate the current producing a positive ramp output.  Comparator U106 resets the voltage across C102 using Q101 when the ramp reaches its positive peak determined by R92 and R94.

"UNUSED_1" has to go somewhere for the circuit to work.  Did you miss something?

A different design which accomplishes the same thing and mentions its usefulness for audio synthesis can be found on page 15 of Linear Technology application note 14.

[vmallet]

Thanks for the explanation. I'm starting to get the gist of how it all works but the details related to the bipolar pair and U103 still escape me; more reading to do.

"UNUSED_1" has to go somewhere for the circuit to work.  Did you miss something?

Yes I did and I _finally_ found it! I've been looking over and over because in all the comparable VCO designs with a bipolar pair the collector of the first NPN (U102B in our case) is weakly tied to VCC and I couldn't see it here. Turns out there was a sneaky trace exactly right under one of the resistors tied to that network and going through a via and I missed it. Anyways we now have R87 going from U102B to VCC.

I'm attaching the revised schematic.

I renamed "UNUSED_1" to "FM_LINEAR", it's an input that lets us tweak the VCO's frequency and without anything plugged into its jack it's going nowhere.

A different design which accomplishes the same thing and mentions its usefulness for audio synthesis can be found on page 15 of Linear Technology application note 14.

Thanks I'll take a look.

[vmallet]

Oh and I think I figured out the R107 resistor and its weird colors: it's probably a 1K (or so) "Tempco resistor" to do the temperature compensation of the exponential scale:
https://store.synthrotek.com/Tempco-Akaneohm-1k-Resistor_p_354.html

(from digging after reading Ian Fritz' explanation of his design, pointed out by ledtester earlier in the thread)

[David Hess]

Oh and I think I figured out the R107 resistor and its weird colors: it's probably a 1K (or so) "Tempco resistor" to do the temperature compensation of the exponential scale:
https://store.synthrotek.com/Tempco-Akaneohm-1k-Resistor_p_354.html

Ok, that makes sense.  The differential pair is the standard configuration for generating the exponent function but it requires two forms of temperature compensation.  The input transistor compensates for the temperature dependent Vbe and the resistor compensates for the change in transconductance with temperature.

The example circuit I mentioned from Linear Technology handles the temperature compensation in a different way; it uses the extra transistors in a transistor array for heating and temperature sensing so that the transistor used for the exponent function operates at a constant temperature and no temperature compensation is required.