Designer’s Guide Community

aamar

Community Member

Offline

Posts: 57
Germany

Log-domain converter
Mar 09^th, 2009, 9:07am

Dear Colleagues,

I have a linear in dB variable gain amplifier and I have a detector detecting the peak-Valley of the envelop of an AM signal. I want to implement a limiter circuit using the two blocks. So I detect the output of the amplifier and feed it back to the VGA.
I knew that I have to transfer the detector output in log domain before using it to control the exp-VGA inorder to have a linear relationship between the input and output of the amplifier.
My question now is how to convert the detector's voltage signal in log voltage signal using MOSFETs?. Can the translinear circuits make it?, but then I have to convert the detector's output voltage to current (OTA is needed)? Can anyone tell me if it is the correct solution or are there better solution?.

Thanks and best regards,

aamar

Back to top

IP Logged

buddypoor

Community Fellow

Offline

Posts: 529
Bremen, Germany

Re: Log-domain converter
Reply #1 - Mar 9^th, 2009, 9:37am

aamar wrote on Mar 9^th, 2009, 9:07am:

Dear Colleagues,

I have a linear in dB variable gain amplifier and I have a detector detecting the peak-Valley of the envelop of an AM signal. I want to implement a limiter circuit using the two blocks. So I detect the output of the amplifier and feed it back to the VGA.
I knew that I have to transfer the detector output in log domain before using it to control the exp-VGA inorder to have a linear relationship between the input and output of the amplifier.
My question now is how to convert the detector's voltage signal in log voltage signal using MOSFETs?. Can the translinear circuits make it?, but then I have to convert the detector's output voltage to current (OTA is needed)? Can anyone tell me if it is the correct solution or are there better solution?.

Thanks and best regards,
aamar

1.) First question: What do you mean with "linear relationship between the input and the output of the amplifier" ? Control input or signal input ?

2.) Im not sure if it wise resp. neccessary to convert the detctor output to the log domain. Why ? The exponential control characteristic of the amplifier requires a VOLTAGE (in volts) in order to exhibit the desired properties.

Back to top

LvW (buddypoor: In memory of the great late Buddy Rich)

IP Logged

aamar

Community Member

Offline

Posts: 57
Germany

Re: Log-domain converter
Reply #2 - Mar 9^th, 2009, 1:38pm

Thanks for reply,

Inorder to answer your questions assume the following relation

Gain=K*exp(Vc)

where K is a constant and Vc is the control voltage.

The relationship shows an exponential relationship between the Gain of the amplifier and the control voltage Vc. If I can transfer the Vc to log Vc then it will be as follows

Gain=K*exp(log Vc)=K2* Vc

So the linear relationship is between the Gain of the amplifier and the control voltage.

So the question is how to realize the (log Vc) using MOSFETs.

Hope it is now clear.

Thanks once again.

aamar

Back to top

IP Logged

buddypoor

Community Fellow

Offline

Posts: 529
Bremen, Germany

Re: Log-domain converter
Reply #3 - Mar 10^th, 2009, 1:04am

Yes, its clear - as far as my question 1.) is concerned.
That�s the normal behaviour of an AGC amplifier.
But you should know that there is a special reason for this exponenetial relationship (linear-in-dB).
And this touches my question Nr. 2) from above:

Why do you want to have this linear relationship between control voltage and gain ?
Because the exponential relationship has the advantage that the loop gain for signal fluctuations (and, thus, the stability properties of the closed loop) are INDEPENDENT on the input power itself !!
This can be proven.
Therefore my advice NOT to use the log domain for the control voltage.

Back to top

LvW (buddypoor: In memory of the great late Buddy Rich)

IP Logged

aamar

Community Member

Offline

Posts: 57
Germany

Re: Log-domain converter
Reply #4 - Mar 10^th, 2009, 1:58am

Thanks again for your reply,

But is it possible to regulate an AM signal to have it at the output of constant amplitude using an exp-VGA with a linear control voltage?,

I am trying it but I can't get it till now, and I don't have any proof that it is possible. (The best result I got is to reduce the modulation index e.g. from 0.5 to 0.2)

Can you tell me if it is suitable for such an application or not (limiting the amplitude of a signal or regulating it to have constant amplitude at the output)?

Because the exponential relationship has the advantage that the loop gain for signal fluctuations (and, thus, the stability properties of the closed loop) are INDEPENDENT on the input power itself !!
This can be proven.

Interesting, Do you know any reference regarding this topic?

Best regards,

aamar

Back to top

IP Logged

buddypoor Community Fellow Offline Posts: 529 Bremen, Germany	Re: Log-domain converter Reply #5 - Mar 10^th, 2009, 4:32am Well, I enclose a copy of a corresponding paper. Regards Edited: The attached paper: A Wide Range and High Speed Automatic Gain Control Eugenio J. Tacconit and Carlos F. Christiansen PAC 1993 was removed by the Forum administrator.
Back to top	« Last Edit: Mar 10^th, 2009, 8:51am by Forum Administrator » LvW (buddypoor: In memory of the great late Buddy Rich) IP Logged