[HBR] HBR2K -- Chapter 14 -- Large Signal Performance, Part 6

[Mike Feher]

Walt -

I apologize, but I somehow missed your response to my question. No doubt
about it, you are learning a lot as you go along and helping the rest of us
as well. There are, however, from my perspective, a few flaws in your
calculations and I will attempt to see if I can clarify them. With regards
to the noise floor that determines SNR, of course what you get at the
antenna terminals is the best SNR you will have. Every stage after that will
degrade it. I am sure you have by now run across the formulas that determine
the NF degradation through each stage as a function of individual stage NF
and gain. Obviously, the most offending stage is typically the mixer, as it
has the least gain. The front end does help this out substantially, but, in
the process it also generates noise, so, in reality the ultimate noise floor
cannot be achieved. What is the ultimate noise floor one may ask. Well under
well matched antenna and receiver terminations and at room temperature of
290 degrees Kelvin, it is -173 dBm/Hz. Or, simply kT, where k is Boltzman's
constant and T is the temperature. Often you see it represented in total
power, as kTB, where B is the bandwidth of the receiving system prior to
going to the detector. The receiver has so much gain in the early stages
that the latter ones barely contribute to noise at all, and especially, the
audio amp can be ignored. The real important thing as I mentioned before is
what hits the detector, which is your decision making circuit. So, B is
typically defined by the IF bandwidth, however, it is not as simple as that.
For example if you did not have a selective front end, well then you would
also encounter "image noise" and the noise floor to start with could be 3 dB
higher than anticipated as the image noise now also passes through the
receiver. Ideally, your noise bandwidth would be determined by your IF
bandwidth, if your IF filter had a perfect transition region from the
passband to the stopband of the filter. In reality, that also is not
achievable, so while your "X" dB bandwidth may be 2.4 kHz, the actual noise
bandwidth is somewhat wider and can be easily figured out by integrating the
area of the filter's response. OK, I am going on again so let's get to the
point. Let's assume that your IF is equivalent to 3 KHz prior to allowing
the signals to get to the detector. So, what is kTB for 3KHz? Well it is
simply the -173 dBm/Hz plus 10 times the log of 3000, or, about -138 dBm.
You claim that your receiver's noise floor is -107 dBm, or, 31 dB lower than
ideal, utilizing your filter. This implies that your receiver's NF (noise
figure) is 31 dB. Does that make sense? I would be surprised if it was over
15 dB. So, to me, right now things just do not add up. Hope every one is
still awake. 73 - Mike



Mike B. Feher, N4FS
89 Arnold Blvd.
Howell, NJ, 07731
732-901-9193



----- Original Message -----
From: <waltah@earthlink.net>
To: <hbr@mailman.qth.net>
Sent: Saturday, May 10, 2003 12:33 PM
Subject: Re: [HBR] HBR2K -- Chapter 14 -- Large Signal Performance, Part 6


> Mike Feher wrote:
> > Hate to be the "Devil's Advocate" here, And, posssibly you answered this
> > in a previous post. In either case, could you please remind of the
> > bandwidth that you are using to measure your noise floor?
>
> A lot of the reason for posting all that stuff is to get other thoughts
> and viewpoints.   If I knew what I was doing, I'd have finished the
> project six months ago.    Since this is in large measure a learning
> experience, please advocate!
>
> I measure the NF in the time-honored ham way -- with an AC
> voltmeter on the audio output; actually, across the primary of the
> output transformer.   I set the audio gain for a convenient reading on
> noise alone -- say 1 volt -- then apply the signal and adjust it for
triple
> the reading -- 3 volts in this case.   That would be roughly a 10 db
> ratio of (signal+noise)/noise.
>
> So the measurement bandwidth is set mainly by the audio bandwidth
> -- probably 5 kcs or so -- and the bandwidth of the filter in use,
> usually 2.4kcs.
>
> Such a casual approach can cause a few db error when comparing to
> measurements made on other sets and would show the HBR2K
> unfavorably if the other set measurement was made with a narrower
> bandwidth;  the effect would be 3 db/doubling, right?  Measuring at
> 4kcs would make the NF 6 db higher (worse) than measuring at 1
> kcs bandwidth.
>
> Of course my noise isn't uniformly distributed over the 4kcs -- there's
> more in the 2.4kcs corresponding to voice intelligence because most
> noise (now) actually comes from the mixers and must get through
> the filter. Only the 3180kcs IF noise (generated after that filter) is
> broadband enough to cover the audio section bandpass in a uniform
> fashion.
>
> (In very high quality receivers it is common to include a second
> selective filter at the end of the IF chain right ahead of the detector to
> take out the out-of-bandwidth IF generated noise but the added
> complication didn't seem worthwhile for this project.)
>
> I think the method is good enough for measuring my progress.   And
> the absolute answer it gives squares with my subjective impression --
> this receiver is 'getting there' but it is still too noisy.   I expect
that
> when a really good set is adjusted to comfortably receive signals on
> even a fairly quiet band (say, 20 when things are good -- does
> anyone remember the last time that happened?) if you then
> disconnect the antenna, the set goes essentially silent -- you might
> not know it was on.   HBR2K at its current state gets considerably
> quieter, but no one would yet call it 'silent.'
>
> Walt Hutchens
> KJ4KV
>
>
>
>
> ************************************
> Visit the HBR Receiver Web Site with over 100 pictures of receivers and
> construction notes...... via http://www.qsl.net/k5bcq/
>
> Retrieve reflector archived data via
http://mailman.qth.net/mailman/listinfo/hbr
>
>