Variable frequency digital oscillator

[CarlPugh]

I am redesigning a circuit that uses a variable frequency digital oscillator.
Originally the circuit was designed by HP.
The circuit uses a 74S132 (quadruple 2-input positive-NAND schmitt triggers) and a 74LS00 (quadruple 2-input positive-NAND gates)
The circuit operates at a center frequency of 10 MHZ and changes +/- 1.0 MHZ when it is modulated.

Where should I go to get information on variable frequency digital oscillators? (WWW, books, magazine articles or ???)

 

[nbucska]

Why are you redesigning it ? Do you have the desired specs?
How do you want to modulate it? (dig/analog?)

The 10+/- 1 MHz should be easy -- but you could get the same with one single transistor.



<nbucska@pcperipherals DOT com> subj: eng-tips
read FAQ240-1032

 

[CarlPugh]

The company that manufactures this equipment has discontinued it's manufacture due to some other obsolete parts. (Not digital ic's) It's my task to design a replacement.

The user feel that the present design is satisfactory and does not wish to make drastic changes in the design.

The specs are skimpy. What is desired is that the present equipment be replaced with my design and everything works the same. It is in a commercial application and the volume is small.

The modulating signal is a DC to 1 meghertz analog signal.

 

[BrianG]

It would help if you could provide full details of the circuit application and which of the parts you have to redesign.

It is difficult to visualise what kind of digital circuit can produce a frequency modulated output of +/- 1MHz from a "modulating signal of DC to 1MHz", especially with the components you have identified so far.

It certainly won't be a linear f.m. modulator, and my guess is that it sounds like some sort of FSK (frequency shift keying) modulator with a +/- 1MHz shift, and that the modulating signal can have a maximum 1MHz "keying" rate.

 

[VEBill]

You could do something simple with diodes as switches to insert capacitors into the oscillator circuit. This technique is used all the time in ham radio transceivers.

If you need extreme crystal-locked accuracy, then you could use a DDS.

 

[CarlPugh]

BrianG

You commented "It is difficult to visualise what kind of digital circuit can produce a frequency modulated output of +/- 1MHz from a "modulating signal of DC to 1MHz", especially with the components you have identified so far."

Yes it is, but that is what it does.

The transmitter is a voltage to frequency convertor and the receiver is a frequency to voltage convertor with optical coupling between the transmitter and receiver.
It's a linear system, DC + AC in and DC + AC out.

The carrier is a 10 meghertz frequency modulated square wave.

All circuits except the voltage to frequency convertor are frequently used.

Does anyone have any information on sources of informationI on variable frequency digital oscillators.

Thanks for any help.
Carl

 

[VEBill]

"variable frequency digital oscillators"

What does the word 'digital' mean to you in this context? Is it because it was built with digital chips? Or because the output is a squarewave? Try Googling without the word 'digital'. VFOs (without the digital) are as old as the hills.

Is this circuit used to convey a baseband signal (bandwidth DC to 1 MHz) over a non-linear (On/Off) optical medium (perhaps an LED-based system) by FM'ing the signal onto the 10 MHz RF square wave with a deviation of 1 MHz? This sort of application and the chips used are normally called Voltage Controlled Oscillators (VCOs). Google 'VCO' and so forth...

I've run audio through an optocoupler; I simply biased the LED and the Phototransistor into their linear regions and kept the amplitude within reasonably linear limits. Worked fine.

 

[nbucska]

Both the 132 and 00 are still available -- why do you have to redesign it? If you need to change the specs, what are they ? A 10 MHz VCO with 10% modulation should be easy
unless there is some unusual requirement.

It seems to me the hardest part of this job is to find
out what the job is supposed to be...


<nbucska@pcperipherals DOT com> subj: eng-tips
read FAQ240-1032

 

[CarlPugh]

VE1BLL
This system uses an on/off signal to send information, and it also uses digital chips. It can be called a variable frequency digital oscillator.
Yes to your second paragraph. Yes I have Googled.
Direct modulation of a LEd is a poor approach for accuracy. The LED degrades over time and if the fiber optic cable is bent sharply, the output amplitude decreases.

nbucska
From my second posting of 13 Oct 05 "The company that manufactures this equipment has discontinued it's manufacture due to some other obsolete parts. (Not digital ic's)"
I could just copy the old design, but it would be more honorable to redesign it. And even if someone else's design is copied, it is still advisable to understand how the circuit works.
The user wants something that works the same way that the equipment that he has does. That's the spec.

MY ORIGIONAL REQUEST WAS FOR
"Where should I go to get information on variable frequency digital oscillators? (WWW, books, magazine articles or ???)"

Thanks for your help
Carl

 

[VisiGoth]

Carl,
You have asked for:
"The modulating signal is a DC to 1 meghertz analog signal."
and have stated:
"The transmitter is a voltage to frequency convertor..."

Those are incompatible statements. The first statement descibes a frequency to frequency converter and is possible with the devfices you mentioned. But it would have been based on a clock for the 10 MHz, which you did not mention.

The second statement would be compatible with a ring oscillator, which does not need a clock source (it would be the clock source) and this second statement would then be a voltage to frequency converte if the input signal was zero to 10 Volts perhaps. But that does not explain the +/- 1 MHz deviation unless zero volts in was used for -1 MHz.

Could you provide the voltage to frequency curve you desire (MHz per Volt and offset) and the precision you desire (especially a linearity). The phase noise must be very poor and the absolute frequency accuracy would be very poor by other standards.

You can google under ring oscillators.

Some ring OSC modulate the supply voltage and depend on the transconductance elements to charge the capaciatnce slower to create the change in frequency. That is a poor control affect for the parts you have.

It is difficult to guess which circuit you have and thus we might be guessing in the wrong direction for circuits you might have.

Could you tell us what pin the control voltage goes to in your circuit. And what the cotrol to frequency deviation is.

http://heim.ifi.uio.no/~bassen/FM-adc/digital-adc.html

 

[VEBill]

"...an on/off signal to send information..."

You mean that the input signal is purely digital (not analog) and the resultant RF is either 9 or 11 MHz? Some of this isn't consistent.

Geesh, if it is just digital, then here is all you need.

1) Make a 9MHz square wave.
2) Make an 11MHz square wave.
3) Use gates to switch between them.

If you're worried about non-synchronous switching and resultant runt pulses, then you'll need to add a few complications.

 

[CarlPugh]

http://www.discovercircuits.com/PDF-Files/4584vco.pds

is an example of a variable frequency digital oscillator.

What I am looking for is theory of operation of variable frequency oscillators that use NAND gates in combination with schmitt triggers.

Surely there must be one expert on this subject at eng-tips.

Thanks for your help.
Carl

 

[VEBill]

1) It sez right on the drawing: "VCO".

2) That one appears to be a kludge in that he's just screwing around with the PS voltage going to one of the chips. Not sure how much 'theory' is involved in that sort of approach.

 

[nbucska]

Carl:
Please read this FAQ.

<nbucska@pcperipherals DOT com> subj: eng-tips
read FAQ240-1032

 

[jimkirk]

You might also look for "voltage to frequency converters".

http://www.national.com/an/AN/AN-240.pdf

will get you to National's application note of the LM331 current to frequency converter with several different designs. I don't think they work up to 10MHz, but there's some good background information there.

Also check out Maxim's MAX083 (I think that's the number). A next generation of the old Intersil 8038 VCO chip. The Maxim part works to well over 10 MHz with good linearity.

http://www.analog.com/en/prod/0,,760_794_AD652,00.html

for Analog Devices' voltage to frequency converter. Again, maybe not the frequency range of interest, but they usually have plenty of background information that may help you.

http://www.edn.com/contents/images/090105di.pdf

for a design by Jim Williams that operates from 1 hertz to 100 MHz. Again, plenty of design info. (And any design with a 160dB range is just cool.)

I doubt a design based solely on gates and schmitt triggers will be terribly linear or stable with temperature. Is that a concern?

For the theory, VisiGoth provided the basic theory.

To make a voltage controlled oscillator based solely on gates and/or Schmitt triggers, a la your discoverycircuit example, make a ring oscillator or RC relaxation oscillator. The frequency is (primarily) dependent on gate delay for the ring oscillator, and R, C, and amount of hysteresis of the Schmitt trigger.

That makes an oscillator. To make it voltage controlled, note that the propogation delay of a gate, or the amount of hysteresis of a Schmitt trigger is dependent on the supply voltage.

As an example, the typical propoagtion delay of a 74HC02 goes from 25nS with a 2V supply, 9nS with a 4.5V supply, and 7 nS with a 6V supply. Put an odd number of them into a ring and you can calculate the total propoagtion around the ring which gives you the frequency. Since the prop delay varies with frequency, so does the frequency. Low frequecy with large prop delay, high frequecny with low prop delay. Also note that these are *typical* delays. the maximium over part to part variation and temperature can range up to 115nS at 2V and -40 to 85 degrees C.

To have some better control, insert a well controlled fixed delay in the ring, then by varying the (hopefully) relatively small gate delay you might improve on the linearity. For the range you're talking about, you want about 91nS to 111nS.

This, with some specs on prop dalay should tell you what logic families may be suited for your task.