The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Mixed-Signal Design >> AGC
https://designers-guide.org/forum/YaBB.pl?num=1176125337

Message started by aaron_do on Apr 9^th, 2007, 6:28am

Title: AGC
Post by aaron_do on Apr 9^th, 2007, 6:28am

Hi all,

If i have an AGC which uses a piecewise log-linear VGA, how do i prevent the vga from going back and forth between two states? i.e. signal is higher than Vref so decrease gain => signal is lower than Vref so increase gain etc...

thanks,
Aaron

Title: Re: AGC
Post by mg777 on Apr 9^th, 2007, 9:42am

Is it the A_v versus V_control that is quantized or is it the transfer characteristic V_out versus V_in that is piecewise linear? The two are different. As long as A_v versus V_control is monotonic it's OK.

M.G.Rajan

Title: Re: AGC
Post by ACWWong on Apr 9^th, 2007, 10:50am

aaron_do wrote on Apr 9^th, 2007, 6:28am:

you use hysteresis....

Title: Re: AGC
Post by aaron_do on Apr 9^th, 2007, 7:01pm

thanks for the replies...

so the AGC is actually switching states back and forth slowly? Any good references?

thanks,
Aaron

Title: Re: AGC
Post by ACWWong on Apr 10^th, 2007, 2:18am

when i say hysteresis, i mean like how any household thermostat works, you need to offset gain up and gain down decision making...

i'm not sure about references... i've always designed my own AGC strategies.

Title: Re: AGC
Post by mg777 on Apr 10^th, 2007, 10:12am

Quote:

so the AGC is actually switching states back and forth slowly?

Can your system tolerate a steady state error between the AGC set point & actual?

M.G.Rajan

Title: Re: AGC
Post by aaron_do on Apr 10^th, 2007, 7:00pm

Actually i've never designed AGC before and I was just having some doubts. I'm not worried about steady-state error for the moment...I intend to implement two AGC loops. A coarse one and a fine one. Right now i'm looking at the coarse one.

What i want to do is have a peak detector and filter, then have a comparator bank to get about 6-8 control signals. They can select the gain level. So you end up with 9 gain levels (one is always set). So i'm worried that if i have say 10 dB gain differences, the output amplitude will osciallate by about 10 dB. I've read some papers but none of them seem to talk about that.

ACWWong, i'm starting to understand what you mean now by hysteresis. The idea had occured to me but i thought it would be too complex (i'm going for low power) to implement. Is there a simple way to do this?

thanks,
Aaron

Title: Re: AGC
Post by aaron_do on Apr 10^th, 2007, 7:42pm

Hi ACWWong,

thanks for the tip on hysteresis. I think I figured out how to get it working...

BTW if i need linear in dB, and i want to use a comparator bank, does that mean i need to have a dB linear scale of resistors? i.e. series resistors from VDD to GND with exponentially increasing values. I guess the other way would be to have an exponential peak detector...

thanks,
Aaron

Title: Re: AGC
Post by ACWWong on Apr 11^th, 2007, 1:57am

my opinion is to use exponential detector every time... i think a good google/IEEE search for limitter/RSSI designs will give you some ideas....

Title: Re: AGC
Post by aaron_do on Apr 11^th, 2007, 11:55pm

OK...sorry i made a mistake...it should be a logarithmic detector. No wonder i couldn't find anything on exponential detectors.

Title: Re: AGC
Post by ACWWong on Apr 12^th, 2007, 3:50am

oh course i meant log as well ;)