Maker Pro
Maker Pro

low power vco?

A

albert

Jan 1, 1970
0
I need a low power vco. A 74HC13 or 4000 series logic oscillator would
be good, but I can't figure out how to make the frequency vary.

I need it to oscillate between 65 Kc and 70 Kc with a control voltage
in the 1 to 5 volt range, so it does not have to operate over a wide
range.

It has to be a 50 percent duty cycle however. It can have a sine wave
output (which is preferred), but a 0 to 4 volt ttl type square wave
output is ok.

I need someting simple, and I'm trying to avoid the low power 555
timers as they don't have 50 percent duty cycles and thee lowest power
variant of the 555 still draws 500 ua quiescent current.

I've googled and found good info on cmos logic based oscillators, but
no way to make them variable with a dc control voltage.

Any suggestions?

Thanks

A
 
J

John Fields

Jan 1, 1970
0
I need a low power vco. A 74HC13 or 4000 series logic oscillator would
be good, but I can't figure out how to make the frequency vary.

I need it to oscillate between 65 Kc and 70 Kc with a control voltage
in the 1 to 5 volt range, so it does not have to operate over a wide
range.

It has to be a 50 percent duty cycle however. It can have a sine wave
output (which is preferred), but a 0 to 4 volt ttl type square wave
output is ok.

I need someting simple, and I'm trying to avoid the low power 555
timers as they don't have 50 percent duty cycles and thee lowest power
variant of the 555 still draws 500 ua quiescent current.

I've googled and found good info on cmos logic based oscillators, but
no way to make them variable with a dc control voltage.

Any suggestions?

---
This'll get you the VCO, and you can use any micropower comparator to
get the supply current down, but I don't think it'll get you exactly 50%
out. For that you could take the output and run it through a dflop
divide-by-two.


VC>-----|+\
| >--+-->FOUT
+--|-/ |
| |
+--[R]---+
|
[C]
|
GND
 
A

albert

Jan 1, 1970
0
Thanks John, it almost looks TOO easy! I'm learning Spice now, and
this looks like a perfect use for it! Probably Maxim makes some nice
low power linear IC's too!

Keep Smilin'



On Sat, 24 Jan 2004 15:31:58 -0600, John Fields
 
M

Michael Black

Jan 1, 1970
0
albert said:
I need a low power vco. A 74HC13 or 4000 series logic oscillator would
be good, but I can't figure out how to make the frequency vary.

I need it to oscillate between 65 Kc and 70 Kc with a control voltage
in the 1 to 5 volt range, so it does not have to operate over a wide
range.

It has to be a 50 percent duty cycle however. It can have a sine wave
output (which is preferred), but a 0 to 4 volt ttl type square wave
output is ok.

I need someting simple, and I'm trying to avoid the low power 555
timers as they don't have 50 percent duty cycles and thee lowest power
variant of the 555 still draws 500 ua quiescent current.

I've googled and found good info on cmos logic based oscillators, but
no way to make them variable with a dc control voltage.

Any suggestions?

Thanks

A

You left out the most important thing, what frequency range are
you talking about? That inpart determines the solution.

Assuming, from your reference to the 4000 series, that this is audio
frequencies, the obvious solution is the VCO in the 4046 PLL.

Michael
 
J

John Popelish

Jan 1, 1970
0
albert said:
I need a low power vco. A 74HC13 or 4000 series logic oscillator would
be good, but I can't figure out how to make the frequency vary.

I need it to oscillate between 65 Kc and 70 Kc with a control voltage
in the 1 to 5 volt range, so it does not have to operate over a wide
range.

It has to be a 50 percent duty cycle however. It can have a sine wave
output (which is preferred), but a 0 to 4 volt ttl type square wave
output is ok.

I need someting simple, and I'm trying to avoid the low power 555
timers as they don't have 50 percent duty cycles and thee lowest power
variant of the 555 still draws 500 ua quiescent current.

I've googled and found good info on cmos logic based oscillators, but
no way to make them variable with a dc control voltage.

Any suggestions?

Thanks

A

Your application is essentially a data sheet example for the CD4046:
http://www.fairchildsemi.com/ds/CD/CD4046BC.pdf

Re the block diagram on page 2 and the design table on page 11, R2
sets the lower frequency limit and R1 controls the difference between
lowest and highest frequency. Be sure to strap all unused inputs to
ground rather than leaving them floating.
 
J

John Larkin

Jan 1, 1970
0
Thanks John, it almost looks TOO easy! I'm learning Spice now, and
this looks like a perfect use for it! Probably Maxim makes some nice
low power linear IC's too!

Keep Smilin'

It is too easy.

John
 
J

John Larkin

Jan 1, 1970
0
You left out the most important thing, what frequency range are
you talking about? That inpart determines the solution.

Assuming, from your reference to the 4000 series, that this is audio
frequencies, the obvious solution is the VCO in the 4046 PLL.

Michael



To convert Kc to KHz, multiply by 1.0000e0.


John
 
M

Michael Black

Jan 1, 1970
0
John said:
To convert Kc to KHz, multiply by 1.0000e0.


John
Oops, I missed that. But that just emphasizes why the obvious,
the 4046, was missed by the original poster.

Michael
 
J

John Larkin

Jan 1, 1970
0
With 'any micropower comparator'? Some, especially if they have
hysteresis, might oscillate, and some might just go closed-loop
linear, and some couldn't work that high. And, if it does oscillate,
how are you going to reliably map 1-5 volts to 130-140 KHz?

John
 
J

John Fields

Jan 1, 1970
0
With 'any micropower comparator'?

---
Well... probably not. How about "most any"?
---
Some, especially if they have hysteresis, might oscillate,
and some might just go closed-loop
linear, and some couldn't work that high.

---
They don't need hysteresis, in the commonly accepted sense, to work,
since voltage overshoot past the switching point and the time required
for the comparator to respond and reverse the output (voila,
oscillation!) makes it happen. For 140kHz that's only about 3.5µs
between edges, so it doesn't seem like it ought to be _that_
difficult...
---
And, if it does oscillate,
how are you going to reliably map 1-5 volts to 130-140 KHz?

---
1-5V into an opamp, set offset and span using the opamp, output of the
opamp into the VCO input?

Or maybe even just a simple voltage divider, it'll depend on what the
VCO's transfer function looks like, no?
 
J

John Larkin

Jan 1, 1970
0
---
Well... probably not. How about "most any"?
---


---
They don't need hysteresis, in the commonly accepted sense, to work,
since voltage overshoot past the switching point and the time required
for the comparator to respond and reverse the output (voila,
oscillation!) makes it happen. For 140kHz that's only about 3.5µs
between edges, so it doesn't seem like it ought to be _that_
difficult...
---


---
1-5V into an opamp, set offset and span using the opamp, output of the
opamp into the VCO input?

Or maybe even just a simple voltage divider, it'll depend on what the
VCO's transfer function looks like, no?


That oscillator has only three parts, but it's anything but simple.
Comparators have high-order lags (being, basicly uncompensated opamps)
so will usually oscillate with negative feedback. The frequency
depends on the internal lags and any additional external lag, so OK,
you can make this oscillate, although the frequency depends on the
internal delay of the comparator, which for CMOS parts will have a
serious positive TC.

If the comparator is reasonably symmetric, I'd expect the frequency to
be max when the + input was around Vcc/2, and decrease nonlinearly
above or below that.

With the right micropower comparator and some scaling of the control
voltage, this might work, but I'd be reluctant to commit a production
application to this.

This looks like a good place for a classic 2-transistor astable
multivibrator.

John
 
J

Jim Thompson

Jan 1, 1970
0
On Sat, 24 Jan 2004 17:03:15 -0600, John Fields
[snip]
They don't need hysteresis, in the commonly accepted sense, to work,
since voltage overshoot past the switching point and the time required
for the comparator to respond and reverse the output (voila,
oscillation!) makes it happen. For 140kHz that's only about 3.5µs
between edges, so it doesn't seem like it ought to be _that_
difficult...
---
And, if it does oscillate,
how are you going to reliably map 1-5 volts to 130-140 KHz?

---
1-5V into an opamp, set offset and span using the opamp, output of the
opamp into the VCO input?

Or maybe even just a simple voltage divider, it'll depend on what the
VCO's transfer function looks like, no?


That oscillator has only three parts, but it's anything but simple.
Comparators have high-order lags (being, basicly uncompensated opamps)
so will usually oscillate with negative feedback. The frequency
depends on the internal lags and any additional external lag, so OK,
you can make this oscillate, although the frequency depends on the
internal delay of the comparator, which for CMOS parts will have a
serious positive TC.

If the comparator is reasonably symmetric, I'd expect the frequency to
be max when the + input was around Vcc/2, and decrease nonlinearly
above or below that.

With the right micropower comparator and some scaling of the control
voltage, this might work, but I'd be reluctant to commit a production
application to this.

This looks like a good place for a classic 2-transistor astable
multivibrator.

John

Didn't we beat this configuration to death when it was posted by John
Fields before?

It's a crap circuit... totally unpredictable.

...Jim Thompson
 
J

John Fields

Jan 1, 1970
0
Didn't we beat this configuration to death when it was posted by John
Fields before?

---
A couple of years ago, as I recall. Replete with simulations by you,
Vin VS fout plots and screen shots of scope displays showing it working
by me, amidst much howling about that bumblebees can't fly.
---
It's a crap circuit... totally unpredictable.

---
_Totally_ unpredictable? It oscillates, doesn't it? Its output
frequency changes with voltage doesn't it?

Moreover, your simulator's output yielded something other than noise, so
my guess is that it, at least, could predict enough about it to come up
with an output.
 
J

Jim Thompson

Jan 1, 1970
0
---
A couple of years ago, as I recall. Replete with simulations by you,
Vin VS fout plots and screen shots of scope displays showing it working
by me, amidst much howling about that bumblebees can't fly.
---


---
_Totally_ unpredictable? It oscillates, doesn't it? Its output
frequency changes with voltage doesn't it?

Moreover, your simulator's output yielded something other than noise, so
my guess is that it, at least, could predict enough about it to come up
with an output.

Oscillating? Yes. At a predictable (by design) frequency? NO!

It's a crap circuit.

...Jim Thompson
 
J

John Fields

Jan 1, 1970
0
Oscillating? Yes. At a predictable (by design) frequency? NO!

It's a crap circuit.

---
Not to beat a dead horse, but since the unpredictability of the output
frequency is primarily due to the variations in the comparators' input
bias currents and response times, I'd argue that rather than being
_inherently_ a crap circuit, its crappiness is inherited from the
comparator.

But, I do agree with you in that even with, say, a 1% RC, just looking
at an LM393 with a bias current spread of 10:1 and a response time
spread of God knows what from unit to unit (and _only_ at Vcc = 5V and
Ta = 25°C) makes the whole circuit look like shit.

On top of that, since changes in Vin are going to force the comparator
to work at different point on the common mode slope, even if you could
predict a worst case working window for fout VS Vin it would probably be
huge.

Still, it _does_ oscillate, and its frequency _does_ change when you
change Vin. Must be good for something, huh? Perhaps a good example of
a can of worms?-)
 
J

John Larkin

Jan 1, 1970
0
Still, it _does_ oscillate, and its frequency _does_ change when you
change Vin. Must be good for something, huh? Perhaps a good example of
a can of worms?-)

I guess the op's problem is best met with a 4046, but that isn't as
much fun.

Propagation-delay-based oscillators and timing chains have been used
successfully. Several outfits, most notoriously LeCroy, have built ICs
that used strings of CMOS buffers as shift-register-like tapped delay
structures, and varied Vcc in a closed loop to tune the stage delays
to exactly what they needed. They *did* have good control over the fab
process to make this work.

Do modern on-chip DLLs (like in the newer Xilinx parts) work this way?

How about using two to four comparators in a ring oscillator? If you
use a current-programmable dual or quad comparator, you could use the
current programming input as the frequency setter. Or Vcc. This is
maybe more interesting as a delay chain than as as just an oscillator.

Still, the 4046 wins here.

John
 
J

Jim Thompson

Jan 1, 1970
0
On Sun, 25 Jan 2004 11:23:32 -0800, John Larkin

[snip]
I guess the op's problem is best met with a 4046, but that isn't as
much fun.

Propagation-delay-based oscillators and timing chains have been used
successfully. Several outfits, most notoriously LeCroy, have built ICs
that used strings of CMOS buffers as shift-register-like tapped delay
structures, and varied Vcc in a closed loop to tune the stage delays
to exactly what they needed. They *did* have good control over the fab
process to make this work.

Do modern on-chip DLLs (like in the newer Xilinx parts) work this way?

I don't know about Xilinx, but I *did* design such a scheme for "a
major player" several years ago. Coarse control was
switch-in/switch-out of inverters, fine control was via VCC
How about using two to four comparators in a ring oscillator? If you
use a current-programmable dual or quad comparator, you could use the
current programming input as the frequency setter. Or Vcc. This is
maybe more interesting as a delay chain than as as just an oscillator.

Still, the 4046 wins here.

John

Delay-based is bad... it produces a non-linear frequency versus
control-voltage curve. Some lurkers may remember my post where I
added a series R to the timing cap to get rid of the delay effect.

If there is interest I could start another "Design Exercise of the
Week" and show some VCO designs... I've only done a few dozen over the
years ;-)

...Jim Thompson
 
K

Keith R. Williams

Jan 1, 1970
0
I guess the op's problem is best met with a 4046, but that isn't as
much fun.

Propagation-delay-based oscillators and timing chains have been used
successfully. Several outfits, most notoriously LeCroy, have built ICs
that used strings of CMOS buffers as shift-register-like tapped delay
structures, and varied Vcc in a closed loop to tune the stage delays
to exactly what they needed. They *did* have good control over the fab
process to make this work.

Do modern on-chip DLLs (like in the newer Xilinx parts) work this way?

I don't believe Xilinx modulates the VCC, only the taps. Though
if you want the straight scoop, ask Peter Alfke (Xilinx big-wig).
He's a regular on comp.arch.fpga.

OTOH, I've done similar things with bipolar circuits. (modulating
the current source in ECL circuits)
How about using two to four comparators in a ring oscillator? If you
use a current-programmable dual or quad comparator, you could use the
current programming input as the frequency setter. Or Vcc. This is
maybe more interesting as a delay chain than as as just an oscillator.

At what frequency?
 
Top