« The Drug - John Dunn, Consultant, Ambertec, P.E., P.C. | Main | The Bluest Screen of Death - John Dunn, Consultant, Ambertec, P.E., P.C. »
The sketch below summarizes a basic model for signal to noise ratio (SNR) analysis:
Note that the SNR gets worse and worse with each successive stage of signal processing. It never improves at the output of a later stage over what the SNR of the input to that later stage may be.
Therefore, you want the first stage's gain G1 to be as high as design considerations will permit. Lots of gain at G2 and later simply won't make up for a poorly designed G1.
You can follow this conversation by subscribing to the comment feed for this post.
As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.
Having trouble reading this image? View an alternate.
Comments are moderated, and will not appear until the author has approved them.
Your Information
(Name and email address are required. Email address will not be displayed with the comment.)
Learn about our
free engineering
consulting referral
service at:
We have over forty
electrical, electronic,
mechanical and
software engineers
with expertise in more
than sixty-five categories
of technology and business.
All are members of the
IEEE and adhere to the
IEEE professional codes
of ethics.
No fee is charged for the
referral service.
Each member is an
independent consultant
and negotiates his/her
own consulting agreement.
Editor: Jerry Brown
Contributors:
Carl Schwab
Unless otherwise noted,
reprinting or republication
of anarticle on this blog is
authorized by crediting the
author and prominently
displaying the following
sentence at the beginning
or end of the article,
including the hyperlink to
"This report is republished
with permission of IEEE
Consultant's Network of
Long Island"
Yes, the gain of G1 must be high.. Feedback networks can also be employed to improve the gain (G1)which will help in improving the SNR..
Posted by: Account Deleted | October 06, 2011 at 10:25 AM
One potentially non-obvious point with the simplistic "make G1 high!" approach is that you have to *keep the gain from the input to ALL points in the signal path reasonably high*. I.e., if at any point in the signal path G1*G2*G3*...etc... heads back down to something approaching one, noise again adds directly as if it were present at the source.
This problem is typically seen in (arguably) poorly designed RF front ends, where someone had an overload problem and just added a big attenuator to try to fix it. In fact, the usual problem with designing good front ends is balancing gain distribution with overload protection -- based on presentations such as this one, some people come off with the mistaken idea that "as much gain as possible as soon as possible" is a good idea when, in actuality, if anything it's usually better to give up a little bit of system noise figure and lump the bulk of your gain towards the back end of a system in a "general purpose" receiver so that you have decent dynamic range/overload protection.
Posted by: Joel Koltner | October 06, 2011 at 12:10 PM
Bandlimiting is of similar importance, I have been privy to more than a handful of designs which had very sensible gain topologies but did not properly tailor the bandwidth at the gain stage.
Posted by: Frank Walker | October 06, 2011 at 12:59 PM
Thanks for your sharing this information.
http://www.maurymw.com/MW_RF/Noise_Parameter_Measurements.php
Posted by: Noise Figure | January 31, 2012 at 03:50 AM