Thanks this makes things connected and I start to understand.


Paul Keinanen wrote:
On 11 Oct 2006 13:33:26 -0700, wrote:

I'm dealing with a DSSS current signal at 70MHz. Before I start
designing anything, I am calculating the signal amplitude.

You did not specify the chip rate or base band bit rate and spreading
function. For GPS signals the chip rate is 1.023 MHz, and the signal
spectrum is about 1 MHz wide, so the front end selectivity should be
as close to this as possible.

For a terrestrial 70 MHz DSSS system, there are usually going to be
several independent DSSS transmitters sharing this 1 MHz bandwidth (or
whatever your chip rate is), some being close to the receiver and some
very distant.

If you want to receive the distant signal and use the whole 1 MHz
bandwidth all the way through the 140 dB gain, a weak distant
transmitter producing only 0.1 uV at your receiver antenna terminal
would be amplified to 1 V. However a local DSSS transmitter sharing
the same 70 MHz band with a different PN sequence could produce 1 mV
at your antenna terminal, which should also be amplified by 140 dB to
10 kV, which clearly is impossible ! The strong signals would saturate
the later amplifying stages and the weak signal would not get through.

A reasonable gain/bandwidth distribution for a 70 MHz system would be
to use a low loss band pass filter with a few MHz bandwidth at 70 MHz,
to take out any VHF band I TV signals and FM broadcast signals at 100
MHz. Use a low noise amplifier stage, followed by a filter with deep
sides which is matched to the bandwidth of the service (1 MHz for
GPS). Then perhaps one stage, possibly with AGC and then the
despreader, followed by a filter matching the bit rate (not chip rate)
of the signal, followed by multiple AGC amplification stages to
amplify the despread signal before the actual BPSK or whatever
detector.

In terrestrial systems at 70 MHz the noise temperature is well above
300 K, thus the antenna noise figure is much worse than 3 dB, so
trying to keep the receiver system noise figure below 3 dB might not
be too productive. It might even be better to suffer an additional 1
dB noise figure hit and use steep filters tailored to the chip rate in
front of the first LNA.

Paul OH3LWR

When the answer is not clear for our doubts, we argue about it for
better explanation, until the doubts are solved. I think this is a
healthy way to get answers.




Wes Stewart wrote:
On 11 Oct 2006 12:13:20 -0700, wrote:

I asked what I don't know. I am expecting a helpful answer.

The problem was, you got some prior answers and then wanted to argue
about it.

The frequency is 70MHz.

Then, assuming some type of filtering around that frequency that's
more or less where your system will likely oscillate.

One reason superhetrodyne receivers were developed to replace TRF
receivers was to develop system gain at different frequencies so as to
minimize the chance of oscillation. Another, of course was to achieve
better selectivity that with older technology was better accomplished
at lower frequencies.

Even a single conversion receiver usually develops its gain at three
different frequencies: r-f, i-f and post detection.

wrote:
Well, I may have to go back to the old question, because it's not
exactly a DSSS depending on cross correlation. What are the possible
bad effects when I keep connecting amplifiers? Shorter bandwidth?
Smaller IP3? There should be something I don't know.

Dynamic Range.

wrote in message
ps.com...
I am asking because it seems difficult, and maybe impossible. I've
never heard of this high gain before, but I dont' know why I cann't
stack amplifiers and I am looking for somebody who can tell me what
exactly the problem is: you see the NF is nice, and I can get the SNR
if the band is narrow enough.

The problem is that, if you have 140dB gain at one frequency, you need *over
140dB isolation* from your final output back to the original input to
prevent oscillations from occurring.

140dB isolation is very difficult for reasonable amounts of money!

I took a receiver design class at Rockwell-Collins when I was working for
them. The guidelines given at the class were no more than 70dB gain at any
single frequency. Any more than that, an instability can result. One of my
instructers in college told me that with the old school of design, allow at
least 1/2 inch of length for each 40dB of gain at a single frequency, and
even with those parameters, interstage shields would be necessary. One
technique that radio manufacturers is "egg crate" construction.............a
die-cast chassis with 1/16th inch walls separating each compartment that
contains a module is used.
The better RF generators also use this scheme. Examples of a receiver and RF
generator are the Racal 6790/GM and the Boonton 103D, respectively.
If you are still curious if your idea will work or not, I suggest that you
actually build the cascaded lineup and find out what happens. Sometimes, the
best way to learn about these things is to experience them yourself. I
remember the first cascaded amplifier lineup I built when I was going to
school. When I asked my instructor why the circuit was oscillating, he asked
me where the decoupling networks were. I asked him "decoupling what?". Oh,
that's what those 100 ohm / .01uF RC networks were that I used to see in
series with the collector/drain resistors. I used to jump out those 100 ohm
resistors, thinking that the extra collector/drain voltage would somehow
give the radio more gain!
It is always good to ask questions on the NG, but the best way to learn is
to actually build the circuit yourself. Nothing like the "school of hard
knocks".

Pete

wrote in message
ps.com...
When the answer is not clear for our doubts, we argue about it for
better explanation, until the doubts are solved. I think this is a
healthy way to get answers.




Wes Stewart wrote:
On 11 Oct 2006 12:13:20 -0700, wrote:

I asked what I don't know. I am expecting a helpful answer.

The problem was, you got some prior answers and then wanted to argue
about it.

The frequency is 70MHz.

Then, assuming some type of filtering around that frequency that's
more or less where your system will likely oscillate.

One reason superhetrodyne receivers were developed to replace TRF
receivers was to develop system gain at different frequencies so as to
minimize the chance of oscillation. Another, of course was to achieve
better selectivity that with older technology was better accomplished
at lower frequencies.

Even a single conversion receiver usually develops its gain at three
different frequencies: r-f, i-f and post detection.