[?] Audio VCO design

David

Don't use the 4046, unles you are prepared to calibrate each one. I found

regards Graham H

--- View in courier: . VCC . | . +-----------|---------+ . | |8 | . [Rt] +----+----+ | .VC>--------+ | |_ Vcc | | . | +---7-O|D VC| | . [LED]--> [LDR] | _| | . | +----6-|TH R|O-4-+ . | | |__ | . | +---2-O|TR OUT|-3----->OUT . [R] | | GND | . | [Ct] +----+----+ . | | 7555 |1 .GND>-------+---------+-----------+

or:

. +--------[Ct]--------+ . | | . | | \\ HC04 | \\ | .Vc>---[R]--+ +---| >O--+--O| >--+-->OUT . | | | / | | / . [LED]---> [LDR] | . | | | .GND>-------+ +---[R]----+

The LED/LDR combo or an incandescent/LDR combo is available as a single device called a "Vactrol":

JF

The old metal gate 4046 has a much wider VCO range than the HC version, which craps out at low control voltages, so that's good. The frequency tolerance is around +-50%, however, so you might wind up with 75 Hz to 6 kHz or 25 Hz to 2 kHz, which might matter to you.

I'd either nudge the frequency specs up a bit to accommodate that tolerance, or if I wanted something more accurate, use a BJT diff pair to provide a charging current to a cap and a little comparator with well-defined hysteresis to reset it. That way you get a nearly exponential tuning curve, which is a good match to human hearing, and a much more dependable frequency range. If you're feeling posh, you could use an LM13700 and a few Rs and Cs for the whole job.

Cheers

Phil Hobbs

If you want a nicer sound than a square wave, you can generate a multiple of the desired frequency and run it through some sort of simple sine shaper.

John

I'll probably get pilloried for this, but it sounds like the app cries out for a 555 or 7555. :-)

You might have to scale the tuning voltage a bit.

Good Luck! Rich

The thresholds in a '4046 (CMOS) are variable unit-to-unit, some better precision can be had with a linear design; I'm thinking of XR2206, LM331, and similar chips. The basic scheme is to use the voltage input to program tracking current source + sink elements, onto an integrator-connected amplifier into a comparator with suitable hysteresis. Analog switches, diode switches, current mirrors, or transconductance amplifiers are the key components.

That 80:1 tuning range is pretty nasty. Do you need that in one swell foop, or can you use some form of ranging.? ...Jim Thompson

The LM331 can do that, is nicely linear but you'd have to add the proper control voltage offset so you get the range you want:

However, check whether supply voltage and power consumption fit your requirements. Also, only available in DIP package.

A really slick and precise solution is a micro controller with AD converter in there. The MSP430F2003 would be a good candidate and for the basic function you'd need almost no external parts other than a decoupling capacitor. Feed you control voltage to its built-in AD converter, digitally offset and scale its output to whatever you need and let the output set the overflow register of the internal timer. The timer output gets piped to a port pin and, voila, there is your output frequency. Even without any external crystal and just using the internal digitally controlled oscillator (DCO) the precision would be in the single digit percent. TI has super-cheap development kits for that series.

If you don't mind a linear 50mV-4V for 50Hz-4kHz, it's pretty easy to do this with a couple of opamps (or an op-amp and a comparator) and a transistor (BJT, MOSFET or analog switch). You set one amplifier up as a +/- integrator, output goes to a comparator with hysteresis which controls the +/- switching. You get a triangle wave and a square wave output.

Best regards, Spehro Pefhany

On a sunny day (Fri, 20 Nov 2009 18:13:51 -0500) it happened Spehro Pefhany wrote in :

OK, my turn... X MHz xtal, X + a bit MHz VCO (varicap), dual gate MOSFET mixer, lowpass. Can make a very wide range. But I would use the 4046 too.

The curse of a quasi photographic memory, I can remember nearly every web page I've ever been to via context clues:

Here is a recent memory of where a vco was using a few transistors. I'm sure Its only good at room temp, but the price is right:

"John Fields" David Chapman

( snip ASCII art monstrosity )

Very likely, the RoHS directive will prohibit the use of a Vactrol.

Cos the LDRs ( ie CdS cells) inside are made with evil and forbidden Cadmium.

Various folk have been trying to get the insane bureaucrats in Brussels to grant a general exemption for CdS cells on the grounds there is no similar replacement device available - but with no success.

BTW

I know about the one Marshall Amplification got for professional audio, but it expires in a few weeks.

... Phil

I liked this thing,

The nice thing about this circuit is its wire range. 50Hz to 4kHz is a very small part of the total range.

Tim

Many thanks to all those who have taken the time to contribute to this thread.

I'm overwhelmed by the number of different and interesting suggestions that have been offered. As I said in my original posting, I'm not looking for an ultra-linear or accurately repeatable solution so the simpler the better for this application. The cross-coupled multivibrator arrangement looks very simple and effective so I'll probably be trying that to start with and then checking out the other design suggestions.

Thanks once again for all of your suggestions. MUCH appreciated.

- Dave

In article , David Chapman writes

Just read my patent about linearized control....

(In PDF form on my website) ...Jim Thompson

I would use a microcontroller, as Joerg suggested, because then you'll need just one chip and one capacitor and you're done. I've tried it with a PIC:

The lite edition of the HI-TECH C compiler has even floating point support, so it is very easy to implement it. You can get the PIC for less than $ 1 for 10 parts:

Works from 2V to 5V. And you'll need a programmer. I have this one:

Don't use the PICkit3, until they have fixed the firmware :-)

The MPLAB IDE and the compilers (the non-optimizing versions, which works good for most projects) are free.

What an elegant solution! Many thanks Frank.

Your proposed way of solving the problem really highlights the difference between my ancient way of thinking which usually involves discrete components and some digital or analogue ICs, and the modern 'micro fits all' approach.

I do feel old :-( ....but then I am, so many thanks for bringing me up to date. I'll certainly give the PIC approach a go now.

Thanks again - Dave

My pleasure!

It is never too late to learn new things, e.g. I'm 36 and trying to learn piano playing :-)

BTW: the PIC is not the best solution if you want high precision, because of the missing capture compare timer, which would be possible with some AVR microcontrollers or better PICs. The best would be a hardware PWM module, which are integrated in bigger microcontrollers, but you wrote that it is not important, so my very simple solution might work for you.

In other words, a 555. ;-)

Cheers! Rich