Author |
Message |
metal_head_82
Joined: Dec 27, 2009 Posts: 34 Location: Germany
Audio files: 2
|
Posted: Mon Jan 02, 2012 5:50 am Post subject:
Expo converter trouble - need some help |
|
|
Hey everybody!
I tried toget my single opamp SAW VCO to 1V/Oct (thread here: http://electro-music.com/forum/topic-49901.html ).
I swapped the opamp with an inverting schmitt-trigger from a 40106 IC. Works just like the opamp and is already a single supply part (basically it's the same circuit - the opamp is used as schmitt trigger).
The idea was to keep this design as simple as possible. So I tried the expo converter from René Schmitz's 4069 VCO. It can be found here:
http://www.uni-bonn.de/~uzs159/vco4069.htm
I just put both transistors to ground (0V) because I don't have a negative rail in my circuit. I can get the converter to work over 3 octaves. After that the response of the VCO seems to flatten. So if I try to adjust the 1V/Oct trimmer further I can get that 4th octave working but this will mess up the setting for the other octaves. Perhaps the expo converter needs some "high frequency compensation".
Anyone any ideas? I'm stuck here...
Thanks in advance! _________________ "I don't care much about music. What I like is sounds." |
|
Back to top
|
|
|
umschmitt
Joined: Jun 29, 2011 Posts: 189 Location: brrlin
Audio files: 11
|
Posted: Fri Jan 06, 2012 7:11 pm Post subject:
|
|
|
Wow cool ! Three octaves are already nice for such a small number of components ! (Now you understand I'm not going to be of any help)
Actually I have a couple of questions : did you try the converter also with a regular op-amp ? And what did you do with that 22k resistor (initial CV ?) to negative rail in R. Schmitz design ? _________________ ::U::N::S::C::H::N::E::L::L:: |
|
Back to top
|
|
|
beep
Joined: May 05, 2013 Posts: 105 Location: Germany
|
Posted: Mon Oct 07, 2013 10:49 am Post subject:
|
|
|
PUSH! _________________ Is the future obsolete? |
|
Back to top
|
|
|
JingleJoe
Joined: Nov 10, 2011 Posts: 878 Location: Lancashire, England
Audio files: 14
|
Posted: Sun Oct 13, 2013 1:49 pm Post subject:
|
|
|
You may be reaching the limit of the VCO, use a smaller cap. This will also mean that lower CV is required to acheive the same frequencies you were hearing in your test. _________________ As a mad scientist I am ruled by the dictum of science: "I could be wrong about this but lets find out"
Green Dungeon Alchemist Laboratories |
|
Back to top
|
|
|
beep
Joined: May 05, 2013 Posts: 105 Location: Germany
|
Posted: Tue Oct 15, 2013 7:25 am Post subject:
|
|
|
that would just mean bigger CV input resistors, right? _________________ Is the future obsolete? |
|
Back to top
|
|
|
JingleJoe
Joined: Nov 10, 2011 Posts: 878 Location: Lancashire, England
Audio files: 14
|
Posted: Tue Oct 15, 2013 12:42 pm Post subject:
|
|
|
That would either make the range smaller or when combined with a smaller cap, decrease the availiable range. _________________ As a mad scientist I am ruled by the dictum of science: "I could be wrong about this but lets find out"
Green Dungeon Alchemist Laboratories |
|
Back to top
|
|
|
beep
Joined: May 05, 2013 Posts: 105 Location: Germany
|
Posted: Tue Oct 15, 2013 4:41 pm Post subject:
|
|
|
I tried the expo stuff today using the original single op amp saw vco circuit.
In the lower frequencies it works well but when entering the upper range,
the response gets weaker, just like metal_head described.
I changed the resistor then from 100nF to 10nF and noticed
that the oscillator is still at a high pitch even if the CV is at 0V.
I lowered the initial pitch then by replacing the 1M resistor at the
first transistor (the pnp) with a 10M resistor, but that fucks up the response.
looks like I need to have a negative voltage rail to get it working
properly, but I really want to get it working on a single supply. _________________ Is the future obsolete? |
|
Back to top
|
|
|
JingleJoe
Joined: Nov 10, 2011 Posts: 878 Location: Lancashire, England
Audio files: 14
|
Posted: Thu Oct 17, 2013 12:07 pm Post subject:
|
|
|
here's a crazy idea rather then use a dual rail power supply, make a 1 or 2 volt power supply (you can use an buffer on a voltage divider) and use it to bias up the emitter of T2 (going by this schematic http://www.schmitzbits.de/vco4069.htm you'd just connect the emitter of T2 to the output of the buffer at about 2V). _________________ As a mad scientist I am ruled by the dictum of science: "I could be wrong about this but lets find out"
Green Dungeon Alchemist Laboratories |
|
Back to top
|
|
|
beep
Joined: May 05, 2013 Posts: 105 Location: Germany
|
Posted: Tue Oct 22, 2013 1:30 pm Post subject:
|
|
|
*__* I did not watch the forum for some time and didn'T recognize your answer. Thank you, that sounds interesting. I'm gonna give it a try tomorrow, don't want to upset my girlfriend tonight
I'll tell the results immediately. Working on things together. I love this community _________________ Is the future obsolete? |
|
Back to top
|
|
|
beep
Joined: May 05, 2013 Posts: 105 Location: Germany
|
Posted: Thu Oct 24, 2013 2:35 am Post subject:
|
|
|
I tried the 2V emitter biasing thing, but it didn't work. It works with a voltage of around 0,2V. SCale is still messed up, but I used BC547B/5547B transistors, maybe it'll work better when I have the suggested ones (3904/06)
and I need a proper 1K trimmer. _________________ Is the future obsolete? |
|
Back to top
|
|
|
rene_schmitz
Joined: Oct 27, 2013 Posts: 16 Location: germany
|
Posted: Sun Oct 27, 2013 10:04 am Post subject:
|
|
|
Hi Beep and all,
I think the current that runs into the transistor buffer is the main problem. A FET would be a better choice here. And ultimately some form of reset time compensation is needed.
Cheers,
René _________________ My site: http://www.schmitzbits.de |
|
Back to top
|
|
|
beep
Joined: May 05, 2013 Posts: 105 Location: Germany
|
Posted: Mon Oct 28, 2013 8:22 am Post subject:
|
|
|
Hey Thanks for the answer
since I'm a noob, I don't have any experiences with a FET and how
it could be integrated in the circuit. I have a few BF245C.
Do I use them just like NPN? _________________ Is the future obsolete? |
|
Back to top
|
|
|
rene_schmitz
Joined: Oct 27, 2013 Posts: 16 Location: germany
|
Posted: Tue Oct 29, 2013 5:25 am Post subject:
|
|
|
A JFET source follower is superficially very similar to the BJT emitter follower. You might also need to change the resistor, 100k is probably too big. I'd start with 5-10k.
(Alternatively you can just use a FET opamp as a follower.)
Cheers,
René _________________ My site: http://www.schmitzbits.de |
|
Back to top
|
|
|
beep
Joined: May 05, 2013 Posts: 105 Location: Germany
|
Posted: Tue Oct 29, 2013 9:07 am Post subject:
|
|
|
I have no idea what to do...
according to the VCO4069 schematic,
do I have to replace T1 or T2 or both? _________________ Is the future obsolete? |
|
Back to top
|
|
|
rene_schmitz
Joined: Oct 27, 2013 Posts: 16 Location: germany
|
Posted: Wed Oct 30, 2013 3:57 am Post subject:
|
|
|
I had meant the emitter follower at the saw output in metal_heads schematic.
To clarify it, I've drawn this little schematic.
Its to be taken with a grain of salt, e.g. I've not drawn a trimmer for v/oct. And one probably needs to add a reset time compensation.
(Edit: thats now included in the drawing)
Cheers,
René _________________ My site: http://www.schmitzbits.de Last edited by rene_schmitz on Wed Dec 04, 2013 6:59 am; edited 1 time in total |
|
Back to top
|
|
|
beep
Joined: May 05, 2013 Posts: 105 Location: Germany
|
Posted: Wed Oct 30, 2013 4:41 am Post subject:
|
|
|
Hey, thank you for this schematic! )
I'm actually doing this with Opamps and I have an op amp buffer at the sawtooth output.
I'm sorry if I steal your time, but I don't know what a reset time compensation is and how to build it. _________________ Is the future obsolete? |
|
Back to top
|
|
|
elektrouwe
Joined: May 27, 2012 Posts: 143 Location: Germany
|
Posted: Wed Oct 30, 2013 6:22 am Post subject:
|
|
|
beep wrote: |
I'm sorry if I steal your time, but I don't know what a reset time compensation is and how to build it. |
let me answer for René, enough time stolen from him
when the sawtooth (Vc1) is ramping down it will reach the lower switching threshold of the 40106. The 40106 will then charge C1 as fast as possible until Vc1 reaches the upper threshold of the 40106 again. This action takes time - between some 100ns and some microsecs. Let's assume this delay time Tdly is 1us and the VCO frequency is 100Hz = 10000us.
The error caused by Tdly is 1us/10000us which can be ignored.
If the VCO is tuned to 20kHz=50kHz ,the error is 1us/50us =2% which could be a problem.
The solution would be to add a compensation voltage to Vc1, so that the 40106 always switches Tdly BEFORE Vc1 reaches the lower threshold !
In our 20kHz example the sawtooth ramp time should be 49us, @ 10kHz it should be 99us...
For CV=0 this compensation voltage is 0, for the highest CV this compensation voltage reaches the highest value: Vcomp is proportional to Fvco which is proportional to the sawtooth slope dVc1/dt which is fast for high charging currents and small capacitors:
dVc1/dt = I/C1
for a compensation dt =Tdly and dVC=Vcomp !
( draw a sketch and think about if you don't get it immediately
the we can rearrange the equation:
Tdly/C = Vcomp/I
because Vcomp/I is a resistor (Ohms law) and
both Tdly and C are constant, it is obvious that we can just put a series resistor between C1 and the 40106 input with the value
R = Tdly/C
to completly cancel out any Schmitt-Trigger delays.
this is the reason why many sawtooth VCOs even work with slow TL061-opamps as Schmitt-Triggers: just the right R in series with the integrating C and you are done !
I discovered this compensation scheme a couple of years ago and when I checked other VCO designs I saw that this was a really old story and has been done since decades, but it seems that it is still not very well known among DIY-VCO builders
PS: in high end VCOs you can often find another high frequency compensation. It is for compensation of the expo transistors bulk resistance. But that is another story... |
|
Back to top
|
|
|
rene_schmitz
Joined: Oct 27, 2013 Posts: 16 Location: germany
|
Posted: Thu Oct 31, 2013 6:07 am Post subject:
|
|
|
Thanks elektrouwe!
I just have two minor things to add. When you do reset time compensation in a schmitt trigger oscillator with a resistor in series with the cap (which is called a Franco Resistor, after Sergio Franco btw.), you will affect both thresholds, effectively just DC-shifting the signal instead of changing the amplitude. So you can to put a diode over the resistor to make it only affect one side.
(which explains why there's a 1n4148 across the 680 ohms resistor in the VCO4069 )
The emitter bulk resistance compensation was developed by Dave Rossum.
There's some reference in the Oakley One of Three VCO builders Guide:
http://www.oakleysound.com/vco5-bg.pdf
Cheers,
René _________________ My site: http://www.schmitzbits.de |
|
Back to top
|
|
|
beep
Joined: May 05, 2013 Posts: 105 Location: Germany
|
Posted: Thu Nov 28, 2013 3:03 am Post subject:
|
|
|
eeeer... the main problem was to get this work on single supply without -12V rail _________________ Is the future obsolete? |
|
Back to top
|
|
|
rene_schmitz
Joined: Oct 27, 2013 Posts: 16 Location: germany
|
Posted: Wed Dec 04, 2013 6:58 am Post subject:
|
|
|
Hi Beep and all,
I think the reason for the bad performance of the expo on the positive side is that at some point there is not enough current available to open the npn. So one should buffer the voltage first. For this I've drawn now a compensated emitter follower between the pnp and the npn. (A single supply opamp could perhaps also do, but that would be cheating. ) It looks good in simulations, I haven't built it however. When you add this buffering the input voltage range can be shifted towards the positive supply, and the negative supply is not needed anymore.
You'll find the image above in this thread updated. Also included are the HFT components now.
Cheers,
René _________________ My site: http://www.schmitzbits.de |
|
Back to top
|
|
|
beep
Joined: May 05, 2013 Posts: 105 Location: Germany
|
Posted: Wed Dec 04, 2013 7:55 am Post subject:
|
|
|
Thank you very much! You're great!! Will this circuit work the same when using an op-amp (single opamp saw vco) instead of the 40106? _________________ Is the future obsolete? |
|
Back to top
|
|
|
rene_schmitz
Joined: Oct 27, 2013 Posts: 16 Location: germany
|
Posted: Thu Dec 05, 2013 6:13 am Post subject:
|
|
|
HI Beep,
if that opamp/CCO also needs it's current pulled to GND then it should work.
Do you have a link?
Cheers,
René _________________ My site: http://www.schmitzbits.de |
|
Back to top
|
|
|
beep
Joined: May 05, 2013 Posts: 105 Location: Germany
|
Posted: Thu Dec 05, 2013 7:18 am Post subject:
|
|
|
it's the vco design from the guy who started this thread. _________________ Is the future obsolete? |
|
Back to top
|
|
|
rene_schmitz
Joined: Oct 27, 2013 Posts: 16 Location: germany
|
Posted: Fri Dec 06, 2013 12:09 pm Post subject:
|
|
|
This also sinks the current to GND, so it ought to work too. _________________ My site: http://www.schmitzbits.de |
|
Back to top
|
|
|
beep
Joined: May 05, 2013 Posts: 105 Location: Germany
|
Posted: Sat Dec 07, 2013 3:07 am Post subject:
|
|
|
Thank you very much for your advice and the schematics!
Can't wait to breadboard it. Already bought the 10K NTCs.
I know, dual supply is a much better solution for proper synth applications,
but I want to build something similar to the SH101,
just a playable synth, which can be powered from batteries.
Good for traveling. _________________ Is the future obsolete? |
|
Back to top
|
|
|
|