4069 VCO - TERRIBLE tracking

[cyclic]

Hi,

I decided to use Rene Schmidt's 4069VCO as a project to teach myself EAGLE and make my own PCB for it. I also wanted to use it in VC-LFO territory so I ditched the DC-blocking capacitors and buffered both outputs with a TL072 instead.

Mission accomplished!

HOWEVER. My V/Oct tracking is nonexistent. Like, not even close to getting a single octave. Its still useful for weird noises, but useless for my keyboard.


I've switched out the trimpot and trim resistor, and I've switched the expo-converter transistors for some which are (theoretically) better matched than my first pair were.

I've attached my schematic in case there is something useful(ly wrong) there for my problem.

At this point I'm close to giving up and starting again with a TH555VCO (of which I already have a pretty good stripboard version)

Any help out there?

cheers

Lance

---

schematic.png
Description:
Filesize:	40.2 KB
Viewed:	847 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[wackelpeter]

maybe change the expo converter... if all components are okay and there's no failure within your build.


Just looked again throug the Electro Notes Archives and found an issue about expo converters... There is one that looks like it should fit for this application and i guess it's more stable... also there are more options to tweak it... what i had to do for example to get my TH4046 VCO's tracking...
it's in EN37 and i think figure D or E should fit in there...

P.S. maybe the brand of IC used for the VCO core affects the tracking or some unstable temperature at the NTC's... also both trannies should be thermally coupled...

[cyclic]

thanks.

yes, I tried two transistor versions for the expo converter, firstly 3904/3906s which I forgot to match, so then I checked and slotted in something else which I had in my box. unfortunately with a different pinout so had to do the annoying twisty leg thing... But they were 'matched' much better (as below)

both were equally bad at tracking.

I guess that what you are saying is the same line of thought I had re: accuracy of my expo conversion.

I guess the question arises about my matching. I've been using an AVR transistor tester, which tells me Hfe and "Uf", which I'm led to believe is actually meant to be "Vf" which is actually Vbe. (Certainly it gives my figures in the 0.7v range which seems right.)

http://www.eevblog.com/forum/beginners/understanding-my-transistor-checker-and-transistor-specs/

I thought that - Vbe - was what I was meant to match, ie base-emitter voltage. Is this right?
Do I need to also match Hfe, or something else entirely?

cheers

Lance

[wackelpeter]

Here's a great guide to transistor matching including a small easy circuit by Ian Fritz:
http://ijfritz.byethost4.com/MiscProj/trans_mat.htm

Just klick on the doc link and you'll get the pdf.

Well you have to match first a few NPN and then some PNP and choose those with the closest value...

But i would suggest to try out that other expo converter from EN37... you can build that on breadboard and hook it uo to your VCO after disconnecting the original converter...
For the 2K Tempco you can simply use a 2K Resistor first or a NTC or some temperature detectors from the KTY series... The Latter don't have exactly 2K a bit less so use a small series resistor... what's important is, that both transistors are glued together and the Tempco also has thermal contact contact...

The mentioned converter has a few more parts but you can play with some of the resistor values to get an better result...

In my 555VCO's i used unmatched transistors from the same batch and got the Tempco thing with one of these KTY-sensors and they track well over 4 octaves... which obviously could be improved by using a Tempco and matched transistors...

[cyclic]

Thanks,
I looked at the EN37.
I reckon I'll breadboard a simple converter again to test just so I don't have to keep desoldering everything, but If I can't get it going I'm going to call it done and live with it as is right now.

Is it possible that putting the input attenuator on one of the CVs is skewing everything, as even when nothings plugged in to it it is still effectively an extra 210k path to ground from the first transistor base? (10kNTC, 100k input and 100k pot)

I did find that the input attenuator for the sync was affecting the frequency quite significantly even with nothing plugged in for a similar reason (ie an extra cap and resisitor to ground in the main oscillators pathway) so I've removed that one already, and maybe I need to do the same for the CV input...

I don;t want to redo the pcb design for an extra opamp in the better converter right now.

It just seems like it would be a better use of my time (later!) to layout my own TH555VCO which will likely have better tracking, as well as actually being more useful as an LFO since i often want tri or sine for that.
Since this was at least partly just for my own learning of EAGLE, I've accomplished that and at least now I feel generally more confidant enough to have a go at more complex circuits next time.

[wackelpeter]

when i was looking through the archives of EM i found this today:
http://electro-music.com/wiki/pmwiki.php?n=Schematics.ScottGravenhorstsVersionOfVCO4069ByReneSchmitz

Scott Gravenhorsts take on Rene Schmitz' 4069VCO with an different expo converter.

[cyclic]

Yes I've seen that, buts linear.

Im sure the original must work fine, I just need to look again.

[JovianPyx]

[JL]

How close are you matching the transistors? They should be matched within a few mV. A while back I built a similar low-budget VCO using 40106 and an NPN/PNP expo converter (basically this). Tracks fine within 4-5 octaves. A couple differences you might notice in the expo circuit for the 40106 VCO: the resistor from the positive rail to the emitter of the PNP/base if the NPN is 100K, while on the 4069 VCO it is 1M. Also there is a 330R resistor between the PNP collector and the negative rail, while on the 4069 VCO the PNP collector is tied directly to the negative rail. Might be worth changing those parts around, in addition to double checking your transistor matching

[cyclic]

Yes, ironically I had also breDboarded that 40106 vco including buffered the out and sorted the pwm. And then for some reason I redid the whole thing around the 4069 instead and I can't really think of why that was now!
I hadn't checked the tracking on it tho so it doesn't help me now either...
I've had a few other Jobs take priority but I might get another look this weekend.

Thanks for all the feedback everyone

Lance

[cloudberry]

[piedwagtail]

A quick look at this and especially the last post:

the mixing of those CV inputs into the pnp is a likely a nightmare of impedances. Currents are probably leaking back down other inputs.

I'd suggest test breadboarding an inverting mixer opamp with all the inputs dealing with the impedance issues then a second inverting opamp to invert and buffer. Follow with the NTC join back at trimmer voltage divider / pnp junction.

Looking at Scott's treatment, he's using two virtual earth buffers to keep the circuit clean of unwanted interactions.
The Pitch CV of unknown source impedance is driving into the high impedance of pin 6 to buffer to the relatively low impedance of the tune 10k pot. Then he mixes it all together with the offset and raw fm into the high impedance of pin 2. The opamp law, that it strives to maintain the same voltage on both input pins, is applied here to keep pin 2 like pin 3 at ground. So pin 1 swings as a voltage around to compensate for any difference by driving out a current. In pin 6's case , pin 7 develops a voltage across the 10k feedback resistor by driving current through it to ground. Back at the case of pin 2, this drives a current through the transistors.

R

[cloudberry]

+1 Piedwagtail's response. I defer to his greater knowledge, and my thoughts were to do a proper active voltage summer too. (-I think Schmitz's objective was to design an oscillator with a low parts count).

Still sounds blimmin' marvelous though.

[steved]

I am still fumbling, tried many times to get this VCO tracking properly for me.

I put this together but moved the connection to T1's base to between the 1.5k resistor and v/oct pot and now my circuit tracks great with my old Moog satellite keyboard..

Not sure if this is specific to the -4.5 to +4.5 keyboard output, but this helped me-

*i used a 1K pot not 10K in the attachment.*

---

4069vcotracking.jpg
Description:
Filesize:	58.59 KB
Viewed:	382 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[piedwagtail]

[Sebo]

Did you check this version?
http://www.aleph.co.jp/~takeda/radio/farmVCO4069E.html

[mercury8]

has anyone ever got this circuit to track accurately with the original expo convertor and NOT the linear one. if I could get this to track 3 octaves id be so happy.

[cloudberry]

Well, yes. See above.
In the top schematic, replacing R9 with a 1k trimmer gives you a high frequency trim. Rene Schmitz suggested this somewhere, I forget where, sorry.

[cyclic]

So, 3 years down the track and one of the nuts fell off the back of my saw out.

I had given up on using this for anything except weird noise or as a vclfo.

None of my extra features worked, couldn't even get pwm to work either.

So I pulled it out, aaaaand- I removed the additional cv input and attenuator.

I was right. The extra input made a nightmare of the input summing.

I replaced the trim R with a 1k and I now have very acceptable tracking.
It's not great, but that extra pot WAS CLEARLY the problem.

Just so anyone else who reads this knows the final result.