I want an electronic circuitry design for driving a piezo-electric
crystal having the following specs:
1> To make an oscillating frequency generator 20 Khz to 60 Khz
(variable).
2> Should be operated on a AA battery.
3> Should not be IC 556 based.
4> Should only use transistors.
5> Need block diagram, circuit diagram with component values.

---

Request for Question Clarification by sublime1-ga on 07 Jul 2003 08:34 PDT

ramchandran1...

Some clarification would be helpful, since your question title
refers to Variable Low Frequency oscillators, but you have also
stipulated a frequency from 20 Khz to 60 Khz, which would not
be considered to be 'low' frequencies (20 Khz is at the upper
range of human hearing). Did you perhaps mean to say 20 - 60 Hz?

Also, for the sake of others researching this, the terms you 
will often find are abbreviated as VFO for Variable Frequency
Oscillator, and LFO for Low Frequency Oscillator.

---

Clarification of Question by ramchandran1-ga on 07 Jul 2003 10:10 PDT

It is for 20 KHz to 60 KHz only (Ultrasonic) and it should sweep
between this range. It should be a VFO with a pulse rate of 80
pulses(approx)/ minute adn it should be controllable. That is, it
should sweep between 20 to 60 KHz 80 times per minute. A LED indicator
for this function is desirable.

---

Clarification of Question by ramchandran1-ga on 07 Jul 2003 10:11 PDT

It is for 20 KHz to 60 KHz only (Ultrasonic) and it should sweep
between this range. It should be a VFO with a pulse rate of 80
pulses(approx)/ minute and it should be controllable. That is, it
should sweep between 20 to 60 KHz 80 times per minute. A LED indicator
for this function is desirable.

---

Clarification of Question by ramchandran1-ga on 08 Jul 2003 01:36 PDT

regarding your comments: yes what you say is correct.alternately we
can use a simple  piezo crystal element {wafer} as a load. what we
need is sweep frequency generator(square wave)and this will be fed to
the element. we need not get steady frequency output from the crystal
and if sweep is not possible then it could be a dual frequency type
ie. when the l.e.d is on it could be low frequency. and in off
condition it could be high {60 KHz}

---

Clarification of Question by ramchandran1-ga on 10 Jul 2003 01:17 PDT

Why I insisted on a transistor level is only because of the field
effect factor. If I use an IC(556) it drains out the battery very
fast. I donot mind an op-amp provided it draws less current compared
to IC 556. It should only be a square wave and I need the harmonics
very much. It is to test effects of harmonics on humans at a
continuosly variable frequency. The purpose is to test it as a nerve
stimulator(non contact type- aero stimulator) and to study the effect
of electro magnetic field within a limited area. Since it will be an
analog circuit within which the frequency time span will be controlled
by a capacitor the linearity will be maintained to a great extent.
Secondly, since the response of the piezo element is pre determined by
the manufacturer there will be many peakings with reference to the
fundamental frequency and it may not respond properly for various
other frequencies which is ok with me. The fundamental frequency is
around 2.7 KHz the same will be driven on their harmonics ( starting
above 20 KHz). My problem is not with the circuit but the current
factor when operated on a AA pencil battery. If I use a transformer
the magnetic field interferes with my study and makes it go hay wire.
Probably you can consider rodent repeller units which use similar
frequencies. My load factor is approximately 300 ohms- which can be
limited with available current.

It should be remembered that piezo-electric crystals cannot be made to
"sweep" a frequency range.  The marvelous thing about quartz crystals
used as a frequency source is that they are as 'steady as a rock' on
one specific frequency.  A minor imperfection here is that the
temperature needs to be held constant or they could drift ever so
slightly off frequency.  This can be bothersome when the signal is put
through a frequency multiplier, which may multiply the drift by some
number like 12 or 16, thus rendering the signal perceptably off the
intended value.  The frequency of oscillation is dependent upon the
thickness of the wafer of crystal in the holder.
     You failed to say whether the output should be electromagnetic,
with waves capable of being received by some sort of VLF receiver
[very low frequency], or alternatively if you wanted the output to be
audio, in which case the waves would lend themselves rather nicely to
the job of being detected by the sensitive eardrum of a rat.

This seems way too hard to do at the transistor level.  Are you sure
you understood this homework problem correctly? :)

Are you doing this with discrete transistors?  Do you mean ONLY
transistors (like for an analog ASIC) or do you mean no ICs?  An
op-amp would seem to make the problem slightly simpler.

Does the output have to be sinusoidal?  A square wave would have
harmonics that show up in the output adding distortion to the signal. 
At 20kHz, there would be quite a big one at 60kHz, which the output is
going to have to be capable of following.

You might also want to think about the linearity of the sweep.  Driven
by an exponential, saw tooth, or triangle wave...

Incidently, this is low frequency as most of the resources online are
going to be giving circuits in the MHz.

Just my 2 cents...
Bradley

The linearity that I referred to was regarding the charging profile of
a capacitor related to the frequency sweep desired.  It seems that you
would require an even larger capacitance (already at 1.3Hz!) or higher
charging voltage to extract a relatively linear portion of the curve
or else you would have very fast frequency change in the beginning
followed by a very slow frequency change at the tail end.  I would
also expect that the temperature coefficient of a discrete transistor
circuit may be of some concern.

As an example, the 555 uses approximately 30 transistors to provide
its fuctionality.

While I generally dislike the tendency of people to throw a 555,
microcontroller, or exotic IC at all problems they come across, in
this case I am going to bring up the possibility of using a
microcontroller.  I know my analog design skills are woefully
inadequate to provide a transistor solution.

I would suggest that you look into a PIC microcontroller.  There are a
wide range of choices--looking up an 8-pin model, the power
requirements are less (possibly substantially) than a 555.  The power
requirements are reduced by lowering the operating frequency and/or
using the sleep functionality of the device.  This would give you a
reprogrammable, steady output that meets your frequency requirements. 
There may also be the possibility (if later deemed necessary) of
accounting for the non-linearities and frequency response of the
transducer, too.  The 8-pin PICs have 6 general purpose IOs, which
would allow both control inputs and an LED output (or two).  One
problem you'll come across is the 3V lower limit on the PIC's power
supply.  In other words, you may need two batteries.

Hopefully this has been of some help.  Perhaps Roadrunner or a
researcher will have more to comment.

Bradley