Post subject: Help! need some advice on cable compensators...
Posted: Thu Nov 24, 2005 10:01 am
Thu Nov 24, 2005 2:40 am
I am in a
quandry (Am trying to increase my vocabulary, and this is one of the
new words I learned). I need to make a cable compensator. I have a basic
idea of how the full device should work, but I am not sure how the cricuit
works exactly. The biggest problem (from what I have read) with the
compensator I need is that it needs a bandwidth of just over 1 Ghz.
If anyone can offer some advice, or direction to a website it would
be greatly appreciated!
Post subject: compensatorPosted: Thu Nov 24, 2005 11:27
When dealing with compensators, there are several questions you
need to answer:
1. Fixed compensation, (manually) adjustable compensation,
or automatic (self-tuning) compensation. These somewhat depend on whether
the conditions you're compensating for are going to change either significantly
2. Assuming it's gain flatness and constant group delay
you need, what is the range of correction needed? (Example: cable is
down 20 dB at 1 GHz)
3. How accurate does it need to be (1 dB? 0.1
dB? 1 nanosecond? 100 picoseconds?)
4. Active or passive?
Post subject: Posted:
Fri Nov 25, 2005 5:02 am
Joined: Thu Nov 24,
2005 2:40 am
hey mystery guest,
were actually very useful (read:I hadn't thought of all that!). The
compensator is meant to work on a very long coax line (about 60m if
I am not mistaken). Delay not really the issue, but gain flatness is
a BIG issue, I would like it to have as high an accuracy as possible
(It is on our radio telescope, so we are doing VERY low level noise
measurement). Also, the cable isn't really gonna change at all, so it
will need to be adjustable, but just once so that you can set it for
the specific cable and forget about it. Also, Passive I think is the
way to go!
I have the old unit here on my bench, and just tell
me if I am completely wrong in my assumption. It looks like a butterworth
filter, of which the Q is adjustable (with a little pot). I can't imagine
that it would work at a particularly high frequecy as they have used
old 5% carbon resistors... So what I would like to try is to replace
it with is an adjustable Q tuned circuit, and use the lower end of the
frequency responce slope as the compensation curve... Have I got the
wrong end of the stick, or am I on the right track?
Thanx a lot
for the help so far!
subject: EQPosted: Mon Nov 28, 2005 11:55 am
Do you really need
that much correction for the gain? The unit you have is one of the classic
configurations - a LPF with adjustable Q.
You might want to try
to measure the response you're trying to equalize, and then using optimization
to get the values for an equalizer. Tom Cuthbert's books discuss this
in detail. (www.trcpep.com).