Maker Pro
Maker Pro

variable high resistance

W

Walter Harley

Jan 1, 1970
0
Something I was working on recently got me wondering:

Suppose I want to have an adjustable resistance, that will be linear (that
is, ohmic) for AC voltages less than 1Vp and audio frequencies. The
resistance needs to vary from about 50k to 5MEG, controlled by some other
(analog) voltage or current. The relationship between control signal and
resistance does not need to be particularly linear or repeatable from part
to part.

So far, easy. Now, the slightly-harder part: the power supply is a pair of
3V lithium cells, and to maximize battery life it would be desirable to draw
< 100uA or so.

If it weren't for the current spec, I could use an FET optocoupler like an
H11F3, I think (not sure if they're ohmic up to 1V at lower resistances, but
the datasheet seems to suggest that at these high resistances they'd be
okay). But to get one of those down to 50k takes >1mA through the LED.

With higher voltage rails I could just use a JFET. But with 3V rails,
wouldn't the Vgs, and the unit-to-unit variation in Vgs, bite me? With only
2V between Vss and max negative signal I'm not sure I can keep the channel
from turning on.

I suppose one idea would be to use a charge pump to generate a more-negative
voltage rail. Or are there JFETs that will work with these low voltages,
without needing to be individually selected for threshold voltage? Or maybe
a low-power MOSFET? Any other ideas?

Thanks for any suggestions!

-walter
 
Walter said:
Something I was working on recently got me wondering:

Suppose I want to have an adjustable resistance, that will be linear (that
is, ohmic) for AC voltages less than 1Vp and audio frequencies. The
resistance needs to vary from about 50k to 5MEG, controlled by some other
(analog) voltage or current. The relationship between control signal and
resistance does not need to be particularly linear or repeatable from part
to part.

So far, easy. Now, the slightly-harder part: the power supply is a pair of
3V lithium cells, and to maximize battery life it would be desirable to draw
< 100uA or so.

If it weren't for the current spec, I could use an FET optocoupler like an
H11F3, I think (not sure if they're ohmic up to 1V at lower resistances, but
the datasheet seems to suggest that at these high resistances they'd be
okay). But to get one of those down to 50k takes >1mA through the LED.

With higher voltage rails I could just use a JFET. But with 3V rails,
wouldn't the Vgs, and the unit-to-unit variation in Vgs, bite me? With only
2V between Vss and max negative signal I'm not sure I can keep the channel
from turning on.

I suppose one idea would be to use a charge pump to generate a more-negative
voltage rail. Or are there JFETs that will work with these low voltages,
without needing to be individually selected for threshold voltage? Or maybe
a low-power MOSFET? Any other ideas?

Thanks for any suggestions!

-walter

logic level fets

NT
 
Walter said:
Something I was working on recently got me wondering:

Suppose I want to have an adjustable resistance, that will be linear (that
is, ohmic) for AC voltages less than 1Vp and audio frequencies. The
resistance needs to vary from about 50k to 5MEG, controlled by some other
(analog) voltage or current. The relationship between control signal and
resistance does not need to be particularly linear or repeatable from part
to part.

For a job the other week, I'd graphed the LED current versus resistance
for a couple of cheap, Silonix, MSL 37V51 opto-resistors, (Farnell
part 316-8785).

(rough values)
LED current CdS resistance
0 >10M
8uA 5M
10uA 2M
50uA 100k
100uA 40k
500uA 8k
1ma 4k5
5ma 2k
10mA 1k

regards
john
 
F

Fred Bloggs

Jan 1, 1970
0
Something I was working on recently got me wondering:
Suppose I want to have an adjustable resistance, that will be linear (that
is, ohmic) for AC voltages less than 1Vp and audio frequencies. The
resistance needs to vary from about 50k to 5MEG, controlled by some other
(analog) voltage or current. The relationship between control signal and
resistance does not need to be particularly linear or repeatable from part
to part.

So far, easy. Now, the slightly-harder part: the power supply is a pair of
3V lithium cells, and to maximize battery life it would be desirable to draw
< 100uA or so.

If it weren't for the current spec, I could use an FET optocoupler like an
H11F3, I think (not sure if they're ohmic up to 1V at lower resistances, but
the datasheet seems to suggest that at these high resistances they'd be
okay). But to get one of those down to 50k takes >1mA through the LED.

With higher voltage rails I could just use a JFET. But with 3V rails,
wouldn't the Vgs, and the unit-to-unit variation in Vgs, bite me? With only
2V between Vss and max negative signal I'm not sure I can keep the channel
from turning on.

I suppose one idea would be to use a charge pump to generate a more-negative
voltage rail. Or are there JFETs that will work with these low voltages,
without needing to be individually selected for threshold voltage? Or maybe
a low-power MOSFET? Any other ideas?

There are JFETs with guaranteed pinch-off less than 1.0V available. Just
exactly what kind of circuit did you have in mind?- an attenuator?- a VGA?
 
W

Walter Harley

Jan 1, 1970
0
Fred Bloggs said:
[...]
There are JFETs with guaranteed pinch-off less than 1.0V available. Just
exactly what kind of circuit did you have in mind?- an attenuator?- a VGA?

That would work, then.

The circuit is as unsophisticated as you can imagine: it's just a variable
load for an electromagnetic pickup in an electric bass. The issue is that,
because of the high inductance and DC resistance of these pickups, their
frequency response and overall tone (including other aspects such as the
amount to which the drag of the magnetic field damps the strings' motion,
etc.) depends strongly on the load they see.

Electric bassists (and guitarists) like to able to affect their "tone" in
various ways via controls on the instrument. Most modern basses either feed
the pickups into a passive tone network, or into an opamp-based buffer which
then drives an active EQ section of some sort. Neither solution covers the
range of desirable tones; some basses have switches to turn the buffering on
or off, or some bassists simply use different instruments depending on what
sound they want.

Anyway, I'd been wondering whether making the loading continuously variable,
rather than just switched on or off, would be musically useful. But I
wanted to be able to get the max impedance higher than what a commonly
available pot would do - above 1M, the selection of pots gets small. Thus
the question.

For the actual project at hand, it looks like there's not much sonic
difference when the load goes higher than 1M, meaning I don't need an active
solution at all. So it was more just a matter of curiousity - "if I needed
this, how would I do it."

Much appreciated,
-walter
 
W

Walter Harley

Jan 1, 1970
0
[...]
For a job the other week, I'd graphed the LED current versus resistance
for a couple of cheap, Silonix, MSL 37V51 opto-resistors, (Farnell
part 316-8785).

(rough values)
LED current CdS resistance
0 >10M
8uA 5M
10uA 2M
50uA 100k
100uA 40k
500uA 8k
1ma 4k5
5ma 2k
10mA 1k

Thanks, John. That's less current than I thought they would need.

But what's the response time like? ISTR the CdS photocells take a long time
to recover to high resistances - I need something that will track pretty
close to its final value within 0.1 seconds or less, preferably more like
..05 seconds.
 
J

John Jardine.

Jan 1, 1970
0
[...]
But what's the response time like? ISTR the CdS photocells take a long time
to recover to high resistances - I need something that will track pretty
close to its final value within 0.1 seconds or less, preferably more like
.05 seconds.
Slow, yes indeed!.
Step from 1k to 1M takes 400ms.
regards
john
 
K

Ken Smith

Jan 1, 1970
0
Something I was working on recently got me wondering:

Suppose I want to have an adjustable resistance, that will be linear (that
is, ohmic) for AC voltages less than 1Vp and audio frequencies. The
resistance needs to vary from about 50k to 5MEG, controlled by some other
(analog) voltage or current. The relationship between control signal and
resistance does not need to be particularly linear or repeatable from part
to part.



How about a semi-digital method?


50K
------+--------/\/\/---------
! !
! A1 ----- A2 !
X --!>----! Ref !---!>-- Y
! !
N ==============! !
-----
DAC


A1 and A2 are good quality buffering amplifiers. The circuits between X
and Y are arranged so that it is non-inverting. When N = 0, the input
looks like a 50K. When N is full scale, the input impedance is very high.

If you want a knob the user turns, replace the DAC with a pot the user
adjusts.
 
W

Walter Harley

Jan 1, 1970
0
Ken Smith said:
[...]
How about a semi-digital method?


50K
------+--------/\/\/---------
! !
! A1 ----- A2 !
X --!>----! Ref !---!>-- Y
! !
N ==============! !
-----
DAC


A1 and A2 are good quality buffering amplifiers. The circuits between X
and Y are arranged so that it is non-inverting. When N = 0, the input
looks like a 50K. When N is full scale, the input impedance is very high.

If you want a knob the user turns, replace the DAC with a pot the user
adjusts.


That was the first thing I tried; but it turned into an oscillator (a 21Hz
square wave, as it happened). But I had some other stuff going on in the
circuit too, that undoubtedly aggravated the situation - if I need to get
this working maybe I'll eliminate the extras and give it another try.
Thanks!
 
J

Joseph2k

Jan 1, 1970
0
Walter said:
Fred Bloggs said:
[...]
There are JFETs with guaranteed pinch-off less than 1.0V available. Just
exactly what kind of circuit did you have in mind?- an attenuator?- a
VGA?

That would work, then.

The circuit is as unsophisticated as you can imagine: it's just a variable
load for an electromagnetic pickup in an electric bass. The issue is
that, because of the high inductance and DC resistance of these pickups,
their frequency response and overall tone (including other aspects such as
the amount to which the drag of the magnetic field damps the strings'
motion, etc.) depends strongly on the load they see.

Electric bassists (and guitarists) like to able to affect their "tone" in
various ways via controls on the instrument. Most modern basses either
feed the pickups into a passive tone network, or into an opamp-based
buffer which
then drives an active EQ section of some sort. Neither solution covers
the range of desirable tones; some basses have switches to turn the
buffering on or off, or some bassists simply use different instruments
depending on what sound they want.

Anyway, I'd been wondering whether making the loading continuously
variable,
rather than just switched on or off, would be musically useful. But I
wanted to be able to get the max impedance higher than what a commonly
available pot would do - above 1M, the selection of pots gets small. Thus
the question.

For the actual project at hand, it looks like there's not much sonic
difference when the load goes higher than 1M, meaning I don't need an
active
solution at all. So it was more just a matter of curiousity - "if I
needed this, how would I do it."

Much appreciated,
-walter
I would consider what is called a transconductance amplifier. I do not know
how well they do above about 50 kOhms equivalent though. Generalized
immitance converters may be able to do this also. I see that you have some
reason for not blindly settling on conductive plastic potentiometors.
 
Top