micro power square wave oscillator

[nukeymusic]

I'm looking for a schematic for a square wave oscillator which draws
around 10uA, powered by 2 to 5V and oscillates at around 100kHz with a
50% duty cycle. I tried out some schematics around a 4007 but none can
meet all the requirements.

any suggestions here?

regards,
nukeymusic

 

[David L. Jones]

I'm looking for a schematic for a square wave oscillator which draws
around 10uA, powered by 2 to 5V and oscillates at around 100kHz with a
50% duty cycle. I tried out some schematics around a 4007 but none can
meet all the requirements.

any suggestions here?

regards,
nukeymusic

An MSP430 micro might be able to do that, worth checking. Not 5V though.

Dave.

 

[[email protected]]

I'm looking for a schematic for a square wave oscillator which draws
around 10uA, powered by 2 to 5V and oscillates at around 100kHz with a
50% duty cycle. I tried out some schematics around a 4007 but none can
meet all the requirements.

A simple two transistor astable multivibrator should do the job. You'd
need low capacitance transistors - the SD214 is one part that might
just do the job

http://www.datasheetcatalog.org/datasheet/linearsystems/SD214.pdf

The BFR92 5GHz broad-band transistor would walk it, since it is an
appreciably lower capacitance part

http://www.nxp.com/acrobat_download/datasheets/BFR92_CNV_2.pdf

but you'd need to put some 22R of low inductance (surface mount)
resistor in series with each base to stop them oscillating at a GHz or
so on the stray inductances and capacitances around your layout. A
ferrite chip (non-wound) inductor might do the same job - they weren't
widely available back when I was playing with the BFR92 and it's PNP
complement.

Note the base of the BFR92 can't take more than 2V of reverse voltage
- on a 5V supply you have to be careful that your circuit doesn't
destroy the transistor when it turns it off. You should really build
and test the circuit in LTSpice before you risk blowing up real parts
(not that they are all that expensive - a dollar or so each).

 

[Jonathan Kirwan]

I'm looking for a schematic for a square wave oscillator which draws
around 10uA, powered by 2 to 5V and oscillates at around 100kHz with a
50% duty cycle. I tried out some schematics around a 4007 but none can
meet all the requirements.

any suggestions here?

I am thinking about a PUJT as a relaxation oscillator, perhaps coupled
to a CD4060, maybe? Have you considered that approach?

Jon

 

[Jonathan Kirwan]

<snip>
I was thinking of loading too..
Mmmmm....
1/100Khz = 10uS

Let's say a risetime and falltime of 1uS.
Say supply = 5V

By
I = C dv/dt
And assuming a 100% efficient circuit..

10uA = C 5V/1uS

C = 2pF !!

With those assumptions and if I did the math right, it suggests you
can't even drive a typical scope probe without going over the supply
current spec.

I suppose you are right. Dispels the PUJT relaxation thingy, I think,
as it dumps the charge each cycle and I think the 2n6028 has 100's of
pF, roughly. Not good.

Trapezoidal rise/fall, then? Or perhaps draw some charge at first
and then just move it back and forth in an LC + xtl tank with high Q,
replacing losses?

How do the watch folks do it? There must be some serious remodeling
of crystals for simulation work there (vastly more complex
understanding of them) in order to design something as efficient as
I've seen from them. Anyone know the details here?

Jon

 

[MooseFET]

I'm looking for a schematic for a square wave oscillator which draws
around 10uA, powered by 2 to 5V and oscillates at around 100kHz with a
50% duty cycle. I tried out some schematics around a 4007 but none can
meet all the requirements.

any suggestions here?

Its hard to beat the LTC1540 for low current. You will need really
high resistor values so don't breath on the PCB.

 

[[email protected]]

A simple two transistor astable multivibrator should do the job.

http://www.freepatentsonline.com/3705362.pdf

You'd need low capacitance transistors, so the BFR92 and its PNP
complement, the BFT92 might be useful

http://www.nxp.com/acrobat_download/datasheets/BFT92_CNV_2.pdf

Like the BFR92, the BFT92 will need a 22R (or bigger) base stopper,
and can't take more than 2V of reverse bias across the base-emitter
junction.

 

[Phil Allison]

"D from BC"

I was thinking of loading too..
Mmmmm....
1/100Khz = 10uS

Let's say a risetime and falltime of 1uS.
Say supply = 5V

By
I = C dv/dt
And assuming a 100% efficient circuit..

10uA = C 5V/1uS

C = 2pF !!

With those assumptions and if I did the math right, it suggests you
can't even drive a typical scope probe without going over the supply
current spec.

** Nonsense.

The load charging current only flows during the rise-time of the square
ave - which even in your hand-picked example is only some 10% of the time.

The average supply current is therefore ( for your example) only 1 uA.

Similar math applies for faster rise times making the AVERAGE current draw
essentially independent of the rise-time for fixed frequency and load C.


...... Phil

 

[Jonathan Kirwan]

Its hard to beat the LTC1540 for low current. You will need really
high resistor values so don't breath on the PCB.

You just reminded me of the HA7210, which just needs a crystal and a
cap to make a (mostly) square wave oscillator. The data sheet
suggests supply currents around 21uA @ 5V to 38uA @ 8V. Not 10uA,
though.

I see that the LTC1540 has low requirements when it just sits there
doing nothing, but would it do better than the HA7210 once you
surround it with four resistors and a cap, or so.

Jon

 

[Jonathan Kirwan]

21uA @ 5V to 38uA @ 8V

.... for 100kHz.

Jon

 

[[email protected]]

Since I'm posting via a telephone modem - my brother's house in Sydney
has every luxury except ADSL - posting a schematic would be time-
consuming. An LTSpice text file would work, but who needs it?

The patent I pointed to does include a number of schematics, and if
you need a schematic for conventional two-transistor astable
multivibrator, you can find one with google.

You don't seem to appreciate precisely how simple a two-transistor
astable actually is. I did once see someone build a two-transistor
multivibrator that blew up its base-emitter junctions, whence the
warning, but it isn't a complicated circuit.

 

[Frank Buss]

Since I'm posting via a telephone modem - my brother's house in Sydney
has every luxury except ADSL - posting a schematic would be time-
consuming.

Posting binaries in this newsgroup would be not a good idea. But why is it
time consuming? Lets assume a 56k modem. A JPEG image of about 200kB needs
less than a minute to upload, for which you can use one of the free image
hosting services and then posting a link to it.

 

[[email protected]]

Posting binaries in this newsgroup would be not a good idea. But why is it
time consuming? Lets assume a 56k modem. A JPEG image of about 200kB needs
less than a minute to upload, for which you can use one of the free image
hosting services and then posting a link to it.

Posting binaries to ths news-group is not an option - when I'm at home
I could post to abse, but I can't here.

I'd have to stick it on my web-site as a new page - which takes quite
a lot of fiddling around - if I thought that it was worth the effort.
Since astable multivibrators are almost trivially simple, I'd prefer
not to further clutter up my web-site for benefit of the terminally
dim.

 

[Frank Buss]

I'd have to stick it on my web-site as a new page - which takes quite
a lot of fiddling around - if I thought that it was worth the effort.
Since astable multivibrators are almost trivially simple, I'd prefer
not to further clutter up my web-site for benefit of the terminally
dim.

Not that I'm interested in an astable multivibrator schematic, but
uploading images is as easy as clicking on the upload button, e.g. with
this image service:

http://www.imageshack.us/

A breakout board for a display connector I designed and soldered some weeks
ago:

http://img329.imageshack.us/my.php?image=displaysteckerxb0.jpg

I routed this by hand. It was interesting to see how the Eagle autorouter
result looked like, with two sides and multiple vias

 

[MooseFET]

You just reminded me of the HA7210, which just needs a crystal and a
cap to make a (mostly) square wave oscillator.  The data sheet
suggests supply currents around 21uA @ 5V to 38uA @ 8V.  Not 10uA,
though.

I see that the LTC1540 has low requirements when it just sits there
doing nothing, but would it do better than the HA7210 once you
surround it with four resistors and a cap, or so.

The LTC1540 would be spending most of its time "doing nothing". Most
of its time it will spend near a rail.

The input current doesn't rise much when the difference voltage goes
up so the resistors can be many meg.

 

[[email protected]]

Post it yourself, if you are so keen to see it. The OP hasn't asked
for it, and I'm not motivated to do you any favours.

If you wanted to prove how clever you are. you could try to post a
circuit for a 50% duty cycle emitter-coupled astable multvibrator -
the smaller voltage swing at the active devices would reduce the
current consumption over the more familiar collector-coupled design,
and would make it easier to preserve a thin base-emtter junction, if
you could get it to work.

 

[legg]

100kHz crystal oscillator or a 200kHz ceramic resonator oscillator
followed by a FF if you need very close to 50% duty cycle.

Pity about the square wave requirement - that's where a lot of the
power goes, and a sinusoidal output can have fair dv/dt through the
logic transition levels.

The usual complications and expense show up in anything with inductive
energy storage..........

RL

 

[James Arthur]

Scale the impedances up from this...

http://analog-innovations.com/SED/CMOS-Osc-NoClip.pdf

For non-precision work I'd delete R4 and R5 rather than scaling them
up. Then connect right end of C1 directly to U2Ain2

...Jim Thompson

That drives U1A's input diodes...naughty naughty!

Any idea what the average crossover current on that
gadget is? The spikes are mA; the average could be
a bunch. A 74hcu04 might help, or one of those CMOS
transistor arrays things, CD4007 (?).

Here's a low-tech method:

Vcc +2v (regulated) / 8uA
-+-
|
+--------------+--------------------+
| | R4 |
R1 | | 100K
100K Q1 | R3 |
| 2n3904 |/ .--470K--+-------+------> 102KHz
+------+-----| C1 | | | 250mV p-p
R2 | | |>. 100nF | |/ | (semi-sinewave)
680K | | .--||--+------| |
| | | | Q2 |>. |
=== | | === 2n3904 | |
GND | | GND | |
| '--------------------+ |
| R5 | |
| 220K |
| | |
| === |
| Ct GND |
| 20pF |
'----------------||------------------'

I didn't use all the current, leaving some for a buffer.

With regular transistors you have to keep everyone from
saturating otherwise they're way too slow.
RF transistors would be easier--you could probably just
spin a bog-standard multivibrator and get better symmetry
to boot.

Yours is simpler and cuter, if the current's okay.

Best regards,
James Arthur

 

[Joerg]

I'd investigate the oscillator in the 4046. Beautiful symmetry, and it's
pretty low-power too. Hope you're not trying to drive much of a load
though.

www.ing.unirc.it/portale/didattica/files_docenti/006770027020060507195200.pdf

www.standardics.nxp.com/products/hef/datasheet/hef4046b.pdf

... etc.

I can't see the OP's post because he uses Google. He could check whether
he can make an oscillator out of one of these since they've got a nice
hysteresis:

http://ww1.microchip.com/downloads/en/DeviceDoc/21434h.pdf

Get the lowest voltage part available to minimize capacitive voltage swing.

 

[Joerg]

I suppose you are right. Dispels the PUJT relaxation thingy, I think,
as it dumps the charge each cycle and I think the 2n6028 has 100's of
pF, roughly. Not good.

Trapezoidal rise/fall, then? Or perhaps draw some charge at first
and then just move it back and forth in an LC + xtl tank with high Q,
replacing losses?

How do the watch folks do it? There must be some serious remodeling
of crystals for simulation work there (vastly more complex
understanding of them) in order to design something as efficient as
I've seen from them. Anyone know the details here?

First, low voltage. A fresh silver cell is just 1.55V. Then the clock
never really feeds anything except the input of a divider chain within
the chip. They'll probably do their darndest to keep its capacitance to
a minimum.

A square wave oscillator won't be a nice resonant architecture. So
you'll have to muscle capacitive charges around and it will consume more
power. It's like wanting to rapidly move the pendulum of a grandfather's
clock between its end points. This is why the OP might want to think
about whether it really has to be a square wave.