Using Gilbert Cell for Volume Control

[kyrgeo]

I have never seen it done before, but I don't see a reason why it shouldn't be possible.

The Gilbert cell is a fully balanced four-quadrant multiplier circuit, that is extensively used in RF telecom as a mixer.

There are two inputs, and the output is simply the multiplication of the two input signals, times the gain constant. In this case, one of the inputs would be your balanced or single-ended input signal. The other input would be a DC voltage generated with the volume pot. For zero DC input voltage of the volume control, you get maximum attenuation.

Using this circuit, the audio signal will not pass through the volume pot. Another advantage is that the input impedance of the amp is much better defined, since impedance variations due to the volume control are eliminated. Inverting configurations are now easy to handle, since there is no more variation in loop gain as the volume is changed.

The only problem I see is noise. Any AC noise on the volume control input will be mixed with the input signal, giving all sorts of intermodulation products. However, this should be relatively easy to filter out (now that I think of it, a large capacitor after the volume pot would also act as a "soft start" circuit, and the volume changes would be smooooth), and also using large emitter resistors in this input to keep the gain low should help.

Any opinions on which input should be used for the audio signal and which for the volume control?

Comments are welcome!

Kyriakos Georgiadis

 

[Charles Hansen]

Hello -

I believe this is the standard way that a VCA (voltage-controlled amplifier) is implemented. For instance, look at:

http://www.analog.com/UploadedFiles/Data_Sheets/15183786ssm2164.pdf

the data sheet for an SSM2164. On page 7 there is a schematic that is very close to what you've drawn.

Hope this helps,
Charles Hansen

 

[jcarr]

kyrgeo:

I also considered doing something like a Gilbert multiplier attenuator, but haven't tried to pursue it in earnest, mainly because time is finite and I already have too many other priorities. The following thread includes some weblinks to other patents and reading material that you may find interesting.

http://www.diyaudio.com/forums/showthread.php?s=&threadid=8884

hth, jonathan carr

 

[Bricolo]

can this design be used for multi channel?
I mean, tying the bases of the transoistors connected to the pot all together (1 pot for as many channels you want)

 

[dhaen]

4 quadrant multipliers

Sorry to dampen things,

Before digital, this was the standard way to mix and fade video. I will warn you that making several channels track correctly is a difficult job, and requires periodic tweaking.

Cheers,

 

[Dan Banquer]

Vca?

Hi K.G.
I don't remember my VCA stuff that well but let's have a go anyway. It looks like V1 is your signal input and V2 is your DC control voltage. Like anything else in engineering there is no free lunch. The VCA will give lower noise than the standard ladder attenuator, but the distortion at greater than 10db of attenuation has been found audible by many. DC feedthrough may also be a problem.
For further research I would suggest going to the THAT Corp website and take a look at their equivalent schematics.
BTW, I was thinking of using a VCA for a headphone amp.
Dan Banquer

 

[Terry Demol]

I have never seen it done before, but I don't see a reason why it shouldn't be possible.

The Gilbert cell is a fully balanced four-quadrant multiplier circuit, that is extensively used in RF telecom as a mixer.

There are two inputs, and the output is simply the multiplication of the two input signals, times the gain constant. In this case, one of the inputs would be your balanced or single-ended input signal. The other input would be a DC voltage generated with the volume pot. For zero DC input voltage of the volume control, you get maximum attenuation.

Using this circuit, the audio signal will not pass through the volume pot. Another advantage is that the input impedance of the amp is much better defined, since impedance variations due to the volume control are eliminated. Inverting configurations are now easy to handle, since there is no more variation in loop gain as the volume is changed.

The only problem I see is noise. Any AC noise on the volume
control input will be mixed with the input signal, giving all sorts of
intermodulation products. However, this should be relatively easy
to filter out (now that I think of it, a large capacitor after the volume pot would also act as a "soft start" circuit, and the volume changes would be smooooth), and also using large emitter resistors in this input to keep the gain low should help.

Any opinions on which input should be used for the audio signal and which for the volume control?

Comments are welcome!

Kyriakos Georgiadis

You ave just created a VCA, and it will work as a vol control.
For more info on this type of circuit searc for MAT04 and
download data sheet, there is a good example.

VCA's are used widely in audio and infact every channel has one
in consoles like SSL which LOTS of our music is mixed through.
They are used mainly for compressors (which are also used
on 99% of recordings) and fader automation systems.

Unfortunately they have a reputation of being sonically pretty
coloured and have a bad name because of this, however the
harmonics are reportedly pretty benign (low order mainly
second). Bottom line is this circuit won't come close to a high
quality pot or stepped attenuator if transparency is the goal.


Cheers,

Terry

 

[dkimble]

VCA

FYI

Here are links to VCA ICs manufactured by National Semiconductor:

LM3080 Operational Transconductance Amplifier [Discontinued]:

http://www.national.com/pf/LM/LM3080.html#Datasheet

LM13700 Dual Operational Transconductance Amplifier with Linearizing Diodes and Buffers:

http://www.national.com/pf/LM/LM13700.html



Dave Kimble