Oscillator Opinions

R

Richard Dungan

Guest
All,

I am presently looking at the design of a capacitive sensor, for a
"contactless" push-button. The circuit shown at the URL below is used.
When a grounded object (such as a finger!) is brought close to the
sensor plate, it oscillates, and this is detected:

http://www.radix-design.co.uk/sens1.gif

As it stands, it seems to be wildly random in its sensitivity.
I think that some of this stems from a poor choice of type (BC846B) for
the two transistors. There seems to be dependency upon the manufacturer
of the transistors, which IMO supports this view.

I would be interested in people's opinions, both regarding the general
topology and the choice of components.

All thoughts appreciated. Please decode the mangled email address below,
if you wish to email.

Richard

--------------Richard Dungan--------------
Radix Electronic Design Ltd, Orpington, UK
sedATradixDASHdesignDOTcoDOTuk
------------------------------------------
 
Richard Dungan wrote:
All,

I am presently looking at the design of a capacitive sensor, for a
"contactless" push-button. The circuit shown at the URL below is used.
When a grounded object (such as a finger!) is brought close to the
sensor plate, it oscillates, and this is detected:

http://www.radix-design.co.uk/sens1.gif

As it stands, it seems to be wildly random in its sensitivity.
I think that some of this stems from a poor choice of type (BC846B)
for the two transistors. There seems to be dependency upon the
manufacturer of the transistors, which IMO supports this view.
Click to expand...
Are the "faulty" units too sensitive or too insensitive?

Could you increase the value of C1 to reduce the sensitivity of the circuit
so that more hand capacity is required (5pF instead of 3pF according to my
LTSpice simulation for C1=100p). Increasing C1 would reduce the sensitivty
to parasitic capacitances in the transistors and other strays.

Presumably, you have very short circuit track lengths since this circuit is
so sensitive to stray capacitance. Is the circuit mounted directly behind
the button? Is it screened from adjacent circuits / buttons?

Alternatively, if capacitance isn't the problem, could it be the spread of
gain? Could you try a device with more gain e.g. 847C?
 
In article <obfqn0tlqi4skehr7ru1k3fr2ta2b8dppk@4ax.com>,
Richard Dungan <postmaster@[127.0.0.1]> wrote:
All,

I am presently looking at the design of a capacitive sensor, for a
"contactless" push-button. The circuit shown at the URL below is used.
When a grounded object (such as a finger!) is brought close to the
sensor plate, it oscillates, and this is detected:

http://www.radix-design.co.uk/sens1.gif
Click to expand...
It looks to me like the circuit will depend a fair amount on the
transistor characteristics. If I had to make a circuit like this, I'd try
to base it on an op-amp. I think the total package count would be lower
too.

--
--
kensmith@rahul.net forging knowledge
 
On Tue, 26 Oct 2004 02:42:53 +0000, Ken Smith wrote:

In article <obfqn0tlqi4skehr7ru1k3fr2ta2b8dppk@4ax.com>,
Richard Dungan <postmaster@[127.0.0.1]> wrote:
All,

I am presently looking at the design of a capacitive sensor, for a
"contactless" push-button. The circuit shown at the URL below is used.
When a grounded object (such as a finger!) is brought close to the
sensor plate, it oscillates, and this is detected:

http://www.radix-design.co.uk/sens1.gif

It looks to me like the circuit will depend a fair amount on the
transistor characteristics. If I had to make a circuit like this, I'd try
to base it on an op-amp. I think the total package count would be lower
too.

FWIW, I think the circuit is crap. I'd build a couple of oscillators
Click to expand...
that could handle soomebody walking by, and figure out how to design
a circuit that stays predictable with anything at all against the
electrode.

Like two MVBs in the tens to hundreds of KHZ, beat together, and a high-
pass filter.

Implementation, of course, is left as an exercise for the reader.
[...]
Of course, proofing this, I want to kick myself in the knee.
What a pompous ass.

But unless you have really, really tight size or price restrictions,
it could be done much better, unless your intent is an exercise in
persnickety circuit behavior! ;-)

Cheers!
Rich
 
All,

Thanks to everyone who has posted so far. I would like to emphasise that
the circuit design is not of my making. I would not have done it this
way.

As I previously indicated, I doubt that the BC648B is the right part for
the job (if there is a right part).

Some versions of the original design used a preset pot to trim the value
of R1 to adjust for transistor variations (supposedly hfe, but I think
the situation is more complex than this). This is horrible. The whole
design appears to take no account of variations in characteristics with
temperature.

I'd still be curious to know if anyone can recognise any known
oscillator topology.

Thanks again for all of the comments,

Richard

--------------Richard Dungan--------------
Radix Electronic Design Ltd, Orpington, UK
ngsATradixDASHdesignDOTcoDOTuk
------------------------------------------
 
Richard Dungan <postmaster@[127.0.0.1]> wrote...
I'd still be curious to know if anyone can recognise any known
oscillator topology.
Click to expand...
As shown, that's not an oscillator. Rather it's simply a
poorly-biased one-stage amplifier with a "sense" electrode
tied to its input, that may show disturbances at times.


--
Thanks,
- Win

(email: use hill_at_rowland-dotties-org for now)
 
Richard Dungan wrote:
I'd still be curious to know if anyone can recognise any known
oscillator topology.
Click to expand...
It's an emitter-coupled multivibrator e.g.

http://www.4qdtec.com/mvbz/ec1.gif

http://www.4qdtec.com/mvibs.html
 
Andrew,

"Andrew Holme" <andrew@nospam.com> wrote:

It's an emitter-coupled multivibrator e.g.

http://www.4qdtec.com/mvbz/ec1.gif

http://www.4qdtec.com/mvibs.html
Click to expand...
With respect, I don't agree. I see that the circuit bears some
similarities, but I think that that is where it ends.

My view has been much the same as Win's, for some time: it is a feedback
amplifier with marginal stability which is pushed over the edge by the
capacitance at the sensor. I assume that the instability is brought
about by the additional lag in the feedback network, caused by the
sensor capacitance. However, I stand open to correction on this.

Richard

--------------Richard Dungan--------------
Radix Electronic Design Ltd, Orpington, UK
ngsATradixDASHdesignDOTcoDOTuk
------------------------------------------
 
Richard Dungan wrote:
"Andrew Holme" <andrew@nospam.com> wrote:
It's an emitter-coupled multivibrator e.g.
http://www.4qdtec.com/mvbz/ec1.gif
http://www.4qdtec.com/mvibs.html
With respect, I don't agree. I see that the circuit bears some
similarities, but I think that that is where it ends.
Click to expand...
Analysing the circuit in LTSpice, the waveforms look multivibrator-like i.e.
square(ish) output voltage, transistors alternately switching on and off,
vbe voltage ramps....
 
"Andrew Holme" <andrew@nospam.com> wrote:

I'm not using the same transistor.
Click to expand...
OK

Using the LTSpice default (generic) NPN, C1=22p, C2=3p, I get a 430 KHz
square wave with 70% duty cycle. Oscillation commences 80ms after power-up.
With C1=100p, C2=5p, I get 195 KHz.

With a 2N3904, C1=22p, C2=3p, oscillation commences almost striaght away but
I get a complicated waveform. Not sure if the components are harmonically
related or not!

It seems OK with 2N3904, C1=100p and C2=5p. I get a 70 KHz square wave
after about 65mS.
Click to expand...
All interesting stuff. It raises various questions, which relate as much
to the use of Spice, as they do to the real-world circuit. Certainly, I
had varying start-up times, and as a matter of convenience, and possibly
improved reality, I have a piecewise current source giving it a small
kick, quite early in the simulation.

I have not tried the 3904, I am using a BC846 model. This is more or
less a commodity part and is made by several different manufacturers. I
find this particularly scary for such a fussy circuit, since I would
expect process differences to be evident.

Richard

--------------Richard Dungan--------------
Radix Electronic Design Ltd, Orpington, UK
ngsATradixDASHdesignDOTcoDOTuk
------------------------------------------
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top