FET help

[Carey Fisher]

I've designed a circuit with a shunt connected JFET as a volume control.
It's basically the circuit in Win Hill's Art of Electronics, p140, Figure
3.33 top. The problem I'm having is that I have two channels using separate
JFETS but using the same control voltage. And, due to the variation between
FETs, the same control voltage gives different values of attenuation between
the two channels. Anybody got any ideas how I can compensate for the
FET-to-FET variations? I'm using MMBF4393 JFETs with Vgs(off) spec'd @ -0.5
to -3.0. My actual control voltage is from 0 to -2.5 V.
Thanks in advance for any ideas.
Carey Fisher
careyfisher@ncsradio.com

 

[John S. Dyson]

In article <Ysdsc.37536$Wg6.33759@bignews5.bellsouth.net>,
"Carey Fisher" <dontsendme@no.crap> writes:

I've designed a circuit with a shunt connected JFET as a volume control.
It's basically the circuit in Win Hill's Art of Electronics, p140, Figure
3.33 top. The problem I'm having is that I have two channels using separate
JFETS but using the same control voltage. And, due to the variation between
FETs, the same control voltage gives different values of attenuation between
the two channels. Anybody got any ideas how I can compensate for the
FET-to-FET variations? I'm using MMBF4393 JFETs with Vgs(off) spec'd @ -0.5
to -3.0. My actual control voltage is from 0 to -2.5 V.
Thanks in advance for any ideas.

The matching problem that you are dealing with is one of the major

disadvantages of using JFETS in volume control applications.

Ignoring the obvious selection of components, and adding bias and
scaling the control voltage for the individual components,
another significant issue:

Matching problems are extreme for the 'high impedance' state. So,
you'll have slightly better performance (when the components
are matched and/or other considerations are made) if you use
the low impedance range of the fets (e.g. RDS to 10*RDS or so.)

For gain control applications and maximum dynamic range, assuming
you have enough voltage in the control signal, higher Vp components
tend to be more linear than low Vp components at high signal
levels (esp after using the 50% negative feedback mechanism.)

I have played with FET AGC stuff for over 30yrs, and you CAN make
it work well, but nowadays I would only use them in audio for
limiting the signal before A/D.

John

 

[Genome]

"Carey Fisher" <dontsendme@no.crap> wrote in message
news:Ysdsc.37536$Wg6.33759@bignews5.bellsouth.net...
| I've designed a circuit with a shunt connected JFET as a volume
control.
| It's basically the circuit in Win Hill's Art of Electronics, p140,
Figure
| 3.33 top. The problem I'm having is that I have two channels using
separate
| JFETS but using the same control voltage. And, due to the variation
between
| FETs, the same control voltage gives different values of attenuation
between
| the two channels. Anybody got any ideas how I can compensate for the
| FET-to-FET variations? I'm using MMBF4393 JFETs with Vgs(off) spec'd
@ -0.5
| to -3.0. My actual control voltage is from 0 to -2.5 V.
| Thanks in advance for any ideas.
| Carey Fisher
| careyfisher@ncsradio.com
|
|
|

AC couple the signals to your fet agc stage then inject a DC bias
current (10uA?) into the fet. Use 'DC' feedback to set the voltage
across the fet from this DC current according to your control input.
This should fix the RDS of the fet.

Hope that makes sense

DNA

 

[Genome]

"Genome" <Genome@nothere.com> wrote in message
news:TAisc.18$ag2.0@newsfe1-gui.server.ntli.net...
|
| "Carey Fisher" <dontsendme@no.crap> wrote in message
| news:Ysdsc.37536$Wg6.33759@bignews5.bellsouth.net...
| | I've designed a circuit with a shunt connected JFET as a volume
| control.
| | It's basically the circuit in Win Hill's Art of Electronics, p140,
| Figure
| | 3.33 top. The problem I'm having is that I have two channels using
| separate
| | JFETS but using the same control voltage. And, due to the variation
| between
| | FETs, the same control voltage gives different values of attenuation
| between
| | the two channels. Anybody got any ideas how I can compensate for
the
| | FET-to-FET variations? I'm using MMBF4393 JFETs with Vgs(off)
spec'd
| @ -0.5
| | to -3.0. My actual control voltage is from 0 to -2.5 V.
| | Thanks in advance for any ideas.
| | Carey Fisher
| | careyfisher@ncsradio.com
| |
| |
| |
|
| AC couple the signals to your fet agc stage then inject a DC bias
| current (10uA?) into the fet. Use 'DC' feedback to set the voltage
| across the fet from this DC current according to your control input.
| This should fix the RDS of the fet.
|
| Hope that makes sense
|
| DNA
|
|

Arse..... that's bollocks..... it goes and sticks the fet in its
transconductance region. Feel free to call me an idiot.

DNA

 

[Carey Fisher]

"John S. Dyson" <toor@iquest.net> wrote in message
news:c8rqc3$5a4$1@news.iquest.net...

In article <Ysdsc.37536$Wg6.33759@bignews5.bellsouth.net>,
"Carey Fisher" <dontsendme@no.crap> writes:
I've designed a circuit with a shunt connected JFET as a volume control.
It's basically the circuit in Win Hill's Art of Electronics, p140,
Figure
3.33 top. The problem I'm having is that I have two channels using
separate
JFETS but using the same control voltage. And, due to the variation
between
FETs, the same control voltage gives different values of attenuation
between
the two channels. Anybody got any ideas how I can compensate for the
FET-to-FET variations? I'm using MMBF4393 JFETs with Vgs(off) spec'd
@ -0.5
to -3.0. My actual control voltage is from 0 to -2.5 V.
Thanks in advance for any ideas.

The matching problem that you are dealing with is one of the major
disadvantages of using JFETS in volume control applications.

Ignoring the obvious selection of components, and adding bias and
scaling the control voltage for the individual components,
another significant issue:

Matching problems are extreme for the 'high impedance' state. So,
you'll have slightly better performance (when the components
are matched and/or other considerations are made) if you use
the low impedance range of the fets (e.g. RDS to 10*RDS or so.)

For gain control applications and maximum dynamic range, assuming
you have enough voltage in the control signal, higher Vp components
tend to be more linear than low Vp components at high signal
levels (esp after using the 50% negative feedback mechanism.)

I have played with FET AGC stuff for over 30yrs, and you CAN make
it work well, but nowadays I would only use them in audio for
limiting the signal before A/D.

John

Thanks John. What would you normally use nowadays for a voltage controlled

attenuator in audio applications?
Carey

 

[Winfield Hill]

Carey Fisher wrote...

I've designed a circuit with a shunt connected JFET as a volume control.
It's basically the circuit in Win Hill's Art of Electronics, p140,
Figure 3.33 top. The problem I'm having is that I have two channels
using separate JFETS but using the same control voltage. And, due to
the variation between FETs, the same control voltage gives different
values of attenuation between the two channels. Anybody got any ideas
how I can compensate for the FET-to-FET variations? I'm using MMBF4393
JFETs with Vgs(off) spec'd @ -0.5 to -3.0. My actual control voltage
is from 0 to -2.5 V. Thanks in advance for any ideas.

Hi Carey. The JFET gain-control circuits and discussion on pages
138 to 141 are only interesting in that the technique works at all,
and can be made reasonably linear (using the two-resistor trick)
for the case of small drain-voltage excursions, say +/-200mV max.
When making a single gain-control circuit it may be necessary to
select or screen the JFET before use, which makes the technique
poorly suited for production. Even after screening any thought of
making calibrated voltage-controlled attenuators is best set aside.
As should be any thought of making a matched pair of attenuators.

Should you persist, the formulas on page 138 - 139 imply you may
have some success if you add a fixed voltage (which may be positive
or negative, depending on the JFET Vg0s) to one of the JFET gates.
This can be most easily done with a pair of opamp inverting amps.

The best use for the JFET attenuator circuit is in servo-controlled
gain applications, such as the amplitude control for a Wien-bridge
oscillator. Here the gain must be some unknown but specific value
near 3.0 (or 2.0 for the two-integrator style) in order to stabilize
the oscillator's amplitude and insure low distortion. A feedback
amplitude-control circuit creates the JFET gate voltage. With some
care this technique allows one to make a simple oscillator with
distortion in the low parts-per-million range, which is very nice.
We'll be presenting such a circuit in the 3rd edition of our book.

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[Tam/WB2TT]

"Carey Fisher" <dontsendme@no.crap> wrote in message
news:Y4msc.20408$3W1.9308@bignews6.bellsouth.net...

"John S. Dyson" <toor@iquest.net> wrote in message
news:c8rqc3$5a4$1@news.iquest.net...
In article <Ysdsc.37536$Wg6.33759@bignews5.bellsouth.net>,
"Carey Fisher" <dontsendme@no.crap> writes:
I've designed a circuit with a shunt connected JFET as a volume
control.
It's basically the circuit in Win Hill's Art of Electronics, p140,
Figure
3.33 top. The problem I'm having is that I have two channels using
separate
JFETS but using the same control voltage. And, due to the variation
between
FETs, the same control voltage gives different values of attenuation
between
the two channels. Anybody got any ideas how I can compensate for the
FET-to-FET variations? I'm using MMBF4393 JFETs with Vgs(off) spec'd
@ -0.5
to -3.0. My actual control voltage is from 0 to -2.5 V.
Thanks in advance for any ideas.

The matching problem that you are dealing with is one of the major
disadvantages of using JFETS in volume control applications.

Ignoring the obvious selection of components, and adding bias and
scaling the control voltage for the individual components,
another significant issue:

Matching problems are extreme for the 'high impedance' state. So,
you'll have slightly better performance (when the components
are matched and/or other considerations are made) if you use
the low impedance range of the fets (e.g. RDS to 10*RDS or so.)

For gain control applications and maximum dynamic range, assuming
you have enough voltage in the control signal, higher Vp components
tend to be more linear than low Vp components at high signal
levels (esp after using the 50% negative feedback mechanism.)

I have played with FET AGC stuff for over 30yrs, and you CAN make
it work well, but nowadays I would only use them in audio for
limiting the signal before A/D.

John

Thanks John. What would you normally use nowadays for a voltage
controlled
attenuator in audio applications?
Carey

I believe HiFi amplifiers use a servo driven rotary potentiometer.

Tam

 

[John S. Dyson]

In article <Y4msc.20408$3W1.9308@bignews6.bellsouth.net>,
"Carey Fisher" <dontsendme@no.crap> writes:

"John S. Dyson" <toor@iquest.net> wrote in message
news:c8rqc3$5a4$1@news.iquest.net...
In article <Ysdsc.37536$Wg6.33759@bignews5.bellsouth.net>,
"Carey Fisher" <dontsendme@no.crap> writes:
I've designed a circuit with a shunt connected JFET as a volume control.
It's basically the circuit in Win Hill's Art of Electronics, p140,
Figure
3.33 top. The problem I'm having is that I have two channels using
separate
JFETS but using the same control voltage. And, due to the variation
between
FETs, the same control voltage gives different values of attenuation
between
the two channels. Anybody got any ideas how I can compensate for the
FET-to-FET variations? I'm using MMBF4393 JFETs with Vgs(off) spec'd
@ -0.5
to -3.0. My actual control voltage is from 0 to -2.5 V.
Thanks in advance for any ideas.

The matching problem that you are dealing with is one of the major
disadvantages of using JFETS in volume control applications.

Ignoring the obvious selection of components, and adding bias and
scaling the control voltage for the individual components,
another significant issue:

Matching problems are extreme for the 'high impedance' state. So,
you'll have slightly better performance (when the components
are matched and/or other considerations are made) if you use
the low impedance range of the fets (e.g. RDS to 10*RDS or so.)

For gain control applications and maximum dynamic range, assuming
you have enough voltage in the control signal, higher Vp components
tend to be more linear than low Vp components at high signal
levels (esp after using the 50% negative feedback mechanism.)

I have played with FET AGC stuff for over 30yrs, and you CAN make
it work well, but nowadays I would only use them in audio for
limiting the signal before A/D.

John

Thanks John. What would you normally use nowadays for a voltage controlled
attenuator in audio applications?

Frankly, nowadays (probably for the last 10yrs), it is probably

irresponsible to suggest the FET design to a new designer (without
LOTS of disclaimer caveats.)

I'd suggest looking for the SSM-type devices from Analog Devices
or the THATCORP devices.

The FET design can be made to work, but I'd strongly suggest that
the Gilbert type devices (or similar) as typified by the That
or SSM devices are usually better.

John

 

[martin griffith]

On Sun, 23 May 2004 23:00:59 -0400, "Carey Fisher"
<dontsendme@no.crap> wrote:

I've designed a circuit with a shunt connected JFET as a volume control.
It's basically the circuit in Win Hill's Art of Electronics, p140, Figure
3.33 top. The problem I'm having is that I have two channels using separate
JFETS but using the same control voltage. And, due to the variation between
FETs, the same control voltage gives different values of attenuation between
the two channels. Anybody got any ideas how I can compensate for the
FET-to-FET variations? I'm using MMBF4393 JFETs with Vgs(off) spec'd @ -0.5
to -3.0. My actual control voltage is from 0 to -2.5 V.
Thanks in advance for any ideas.
Carey Fisher
careyfisher@ncsradio.com


Hi Carey,

I don't have a copy of Win's epic with me, but

Fets are a nightmare to get to track accurately in stereo. I used to
(or attempt to) calibrate fets circuits in Urei and Audio Design types
of compressor/limiters years ago in recording studios, its quite
difficult to get them accurate within 0.3dB over a 20 db range, let
alone as a volume control, with a range of say 50dB. ( ok I know you
dont need that accuracy, usually, at low levels)

Fets were used for controlling audio in the 70's, because they were
the only option. Fets took over from CdS photocells and lamps as gain
control elements, another can of worms. Another option from the 70's
was the CA3080? an OTA, another nasty way of controlling audio,
although I never did play with them on the bench. These designs (IMHO)
should not be in modern analogue design books, except to show how it
was done in the good old days. The old NatSemi books seemed to be full
of naff audio circuits using fets

Fets are suboptimal as wide range audio attenuators these days, think
VCA's, much better specs, higher signal capabilities and generally
flexible architecture, as they usually sit in the inverting VE part of
the opamp, effectively as a current attenuator, and you can optimise
for either noise or distortion.

You could go the TI PGA2310 route, but that means having to write
software, try and get the GNU C complier running, keep the digits
away from the audio pcb tracks etc, while with a VCA all you need is a
10K linear pot, and a descent dual opamp,NE5534 etc on a lump of strip
board, and a few R's and C's, and 20 minutes with a soldering iron.
BulletProof.
(I have left out the Dallas type digipot, cos they are headroom
limited by their rather sad supply rails, and generally linear law not
dB/log laws)

Look at the Thatcorp range of VCA products,my favorites, or look at
some of the schematics on the Rane web site, YES, a company that
publishes the schematics of it proaudio products. I've picked up some
nice tricks from Rane ccts.( I dont like their "standard" balanced
input stage, but it works). Rane usually use the SSM range of VCA's,
but they always seemed a bit pricey, in the UK, and always bundled
with RMS and log converters etc, which I rarely needed.



martin



"When all else fails, digitize everything, use fiber optic cable and enter a
whole new realm of problems."

"We won’t use the words Microsoft and reliability in the same sentence."

<Found on the Rane tech pages>

 

[Winfield Hill]

martin griffith wrote...

Look at the Thatcorp range of VCA products,my favorites, or look at
some of the schematics on the Rane web site, YES, a company that
publishes the schematics of it proaudio products. I've picked up some
nice tricks from Rane ccts.( I dont like their "standard" balanced
input stage, but it works). Rane usually use the SSM range of VCA's,
but they always seemed a bit pricey, in the UK, and always bundled
with RMS and log converters etc, which I rarely needed.

Hey, Martin, do you miss your old studio-engineering days?

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)

 

[martin griffith]

On 25 May 2004 05:46:55 -0700, Winfield Hill
<Winfield_member@newsguy.com> wrote:

martin griffith wrote...

Look at the Thatcorp range of VCA products,my favorites, or look at
some of the schematics on the Rane web site, YES, a company that
publishes the schematics of it proaudio products. I've picked up some
nice tricks from Rane ccts.( I dont like their "standard" balanced
input stage, but it works). Rane usually use the SSM range of VCA's,
but they always seemed a bit pricey, in the UK, and always bundled
with RMS and log converters etc, which I rarely needed.

Hey, Martin, do you miss your old studio-engineering days?

Thanks,
- Win

(email: use hill_at_rowland-dot-org for now)
certainly do!,

but living 30 seconds from the beach is hard to beat
wouldnt mind a bit of work now and again, I have heard that they do
have a small TV station here, but I havent got around to finding where
it is, anyway its probably staffed with highly qualified media
students with no with no practical experience!
martin



"When all else fails, digitize everything, use fiber optic cable and enter a
whole new realm of problems."

"We won’t use the words Microsoft and reliability in the same sentence."

<Found on the Rane tech pages>

 

[Ban]

Carey Fisher wrote:

"John S. Dyson" <toor@iquest.net> wrote in message
news:c8rqc3$5a4$1@news.iquest.net...
In article <Ysdsc.37536$Wg6.33759@bignews5.bellsouth.net>,
"Carey Fisher" <dontsendme@no.crap> writes:
I've designed a circuit with a shunt connected JFET as a volume
control. It's basically the circuit in Win Hill's Art of
Electronics, p140, Figure
3.33 top. The problem I'm having is that I have two channels using
separate JFETS but using the same control voltage. And, due to the
variation between FETs, the same control voltage gives different
values of attenuation between the two channels. Anybody got any
ideas how I can compensate for the FET-to-FET variations? I'm
using MMBF4393 JFETs with Vgs(off) spec'd
@ -0.5
to -3.0. My actual control voltage is from 0 to -2.5 V.
Thanks in advance for any ideas.

Thanks John. What would you normally use nowadays for a voltage
controlled attenuator in audio applications?
Carey

If you want a dB-linear attenuation and up to 20dB gain and also +/-15V
operation and want very low costs I would recommend the SSM2164 quad VCA.
With a bit of extra circuitry you can linearize them.
Better still, but expensive and with only digital control PGA2311 or Cirrus
CS3310
Also Wolfson makes a VCA with similar characteristics.
The Panasonic motor-driven pots are obsolete now, ALPS still makes some
(very expensive).
FETs do not make good attenuators, high THD, low level only, high drift.
--
ciao Ban
Bordighera, Italy

 

[Jim Meyer]

"Carey Fisher" <dontsendme@no.crap> wrote in message news:<Ysdsc.37536$Wg6.33759@bignews5.bellsouth.net>...

I've designed a circuit with a shunt connected JFET as a volume control.
It's basically the circuit in Win Hill's Art of Electronics, p140, Figure
3.33 top. The problem I'm having is that I have two channels using separate
JFETS but using the same control voltage. And, due to the variation between
FETs, the same control voltage gives different values of attenuation between
the two channels. Anybody got any ideas how I can compensate for the
FET-to-FET variations? I'm using MMBF4393 JFETs with Vgs(off) spec'd @ -0.5
to -3.0. My actual control voltage is from 0 to -2.5 V.
Thanks in advance for any ideas.
Carey Fisher
careyfisher@ncsradio.com

You can use FETs and get then to track very well if you use the
FETs as switches.

Put a FET in series with the audio signal and switch it
completely on and off with a PWM format gate signal with a basic
repetition frequency much greater than the higest audio frequency.
Put a low pass filter after the switch and you're finished. The
output will follow the PWM ratio almost exactly with very little
difference between two different incarnations of the same circuit fed
with the same PWM signal.

Jim

 

[Max Hauser]

To Martin Griffith's wise observations could I add a few more? (I have some
practical products and patents and publications in the area of analog gain
control in recent decades.)

FETs: Two separate basic methods will linearize the electrically variable
channel resistance in a FET. Both were re-invented repeatedly in different
applications from the early 1960s or earlier and at least one of them is
alive in industrial practice, in the specialized world of electronically
tunable monolithic analog filters (continuous-time). The long history of
these FET techniques is among the topics in a 1996 overview and bibliography
that I and some collaborators put together [1], listed at the end.

It might be obvious, but for a matched two-channel version of one of these
FET circuits, monolithic multiple MOS or junction FETS might be useful.
They tend inherently to be well matched. (In the 1970s I could have
suggested FET part numbers but don't know what's sold today.) Updated
suggestions below.

Good bipolar-transistor mechanisms for continuous gain control (including
Dave Blackmer's -- db's -- circuit in the dbx gain-control products, origins
of THAT Corp. if I recall) come not from nonlinear, temperature-sensitive
effects like gm of emitter-coupled pairs (simple "OTA" products) but from
more accurate (not approximate) large-signal variable-gain mechanisms of
which BJT pairs offer basically two. These closely relate to the so-called
"translinear" circuits that give fundamentally accurate algebraic
large-signal relations among currents. (They and their history are surveyed
also in the book chapter [1], with adjacent in-depth chapter on translinear
amplifiers by Barrie Gilbert, who largely introduced that class of circuits
in 1968, and named it later.) As a broad rule, translinear gain control
tends to be a little noisy compared to FET methods, but can be more linear
in the signal path.

The early history of electronic gain control contains elegant ideas, such as
single-component automatic gain control (AGC) from a well-placed
self-heating thermistor (Kornev demonstrated 40dB AGC this way in the
Russian journal _Pribori i Systemy Upravleniya_ in 1967) [1].

Today much gain control is done with switched resistors, which can be both
more linear than the various continuous gain-control methods and if done
right, also can be near fundamental minima for noise. For example, Linear
Technology has tiny digitally controlled programmable-gain amplifiers (PGAs)
consisting of MOS op amps with switched input and/or feedback resistors,
gain range to 40dB. The digital control is direct and parallel (no
microprocessor required) though it is discrete and digital, in steps (such
as 1-2-4-8 or 1-2-5-10 V/V formats). Like other products mentioned already,
these were designed for more uses than specifically audio but may be useful
also for audio. For example, the LTC6910 (single) or LTC6911 (two-channel)
families, current info respectively at

http://www.linear-tech.com/prod/datasheet.html?datasheet=908
http://www.linear-tech.com/prod/datasheet.html?datasheet=1079

(There is also a "Buy Now!" button for small to moderate quantities.)


Overview and bibliography mentioned above are in

[1] Max W. Hauser, Eric A. M. Klumperink, Robert G. Meyer, and William D.
Mack. "Variable-gain, variable-transconductance, and multiplication
techniques: A survey." Chapter 14 in Sansen, van de Plassche, and Huijsing,
editors, _Analog Circuit Design,_ Kluwer Academic Publishers, 1996, ISBN
0792397762. (I don't know if this paper can be posted because of Kluwer's
copyright, but it exists on paper as well as in the book.) Immediately,
more info on the Kluwer book can be found at the current Web page

http://www.wkap.nl/prod/b/0-7923-9776-2?a=1


Hope this helps -- Max Hauser

 

[Carey Fisher]

"Jim Meyer" <jmeyer@nektonresearch.com> wrote in message
news:21ede509.0405251143.5ee4f5c0@posting.google.com...

"Carey Fisher" <dontsendme@no.crap> wrote in message
news:<Ysdsc.37536$Wg6.33759@bignews5.bellsouth.net>...
I've designed a circuit with a shunt connected JFET as a volume control.
It's basically the circuit in Win Hill's Art of Electronics, p140,
Figure
3.33 top. The problem I'm having is that I have two channels using
separate
JFETS but using the same control voltage. And, due to the variation
between
FETs, the same control voltage gives different values of attenuation
between
the two channels. Anybody got any ideas how I can compensate for the
FET-to-FET variations? I'm using MMBF4393 JFETs with Vgs(off) spec'd
@ -0.5
to -3.0. My actual control voltage is from 0 to -2.5 V.
Thanks in advance for any ideas.
Carey Fisher
careyfisher@ncsradio.com

You can use FETs and get then to track very well if you use the
FETs as switches.

Put a FET in series with the audio signal and switch it
completely on and off with a PWM format gate signal with a basic
repetition frequency much greater than the higest audio frequency.
Put a low pass filter after the switch and you're finished. The
output will follow the PWM ratio almost exactly with very little
difference between two different incarnations of the same circuit fed
with the same PWM signal.

Jim

Thanks to all for the ideas. I've managed to use pots to offset the control
voltage between the two FETs and achieve acceptable, but not great, balance
between the two channels. I've had to stick with the FET approach because
the PCBds are already laid out and ready for production. Adding a couple
SMT pots was relatively easy. My next design will either use VCAs or
digital pots with rotary encoders for the volume controls. Thanks guys...
I've learned a lot!!
Carey Fisher

 

[Tam/WB2TT]

"Max Hauser" <maxREMOVE@THIStdl.com> wrote in message
news:10b7urhqfao60fc@corp.supernews.com...

..................................
It might be obvious, but for a matched two-channel version of one of these
FET circuits, monolithic multiple MOS or junction FETS might be useful.
They tend inherently to be well matched. (In the 1970s I could have
suggested FET part numbers but don't know what's sold today.) Updated
suggestions below.
.................................................
Max,

Strange you should mention the 1970s. Back then I needed an AGC gain element
for broadband use to about 500 KHz. Built one with a JFET, but discarded it
for reasons I can't remember. Ended up using two diodes as the shunt
element. The diodes were in series to the control current, and oppositely
poled in parallel to the signal. This caused some cancellation of distortion
products. Also had tried a Raysistor. Anyhow, it worked quite well, and
stayed in production for over 10 years.

Tam

 

[Tim Wescott]

Carey Fisher wrote:

I've designed a circuit with a shunt connected JFET as a volume control.
It's basically the circuit in Win Hill's Art of Electronics, p140, Figure
3.33 top. The problem I'm having is that I have two channels using separate
JFETS but using the same control voltage. And, due to the variation between
FETs, the same control voltage gives different values of attenuation between
the two channels. Anybody got any ideas how I can compensate for the
FET-to-FET variations? I'm using MMBF4393 JFETs with Vgs(off) spec'd @ -0.5
to -3.0. My actual control voltage is from 0 to -2.5 V.
Thanks in advance for any ideas.
Carey Fisher
careyfisher@ncsradio.com

I have heard of using selenium photo resistive cells (along with light
sources) for controlling gain -- but I don't know if they'd match any
better than JFETs.

--

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com