Dear LF friends
Thank you for the interest to my recent posting regarding
receiving on 137 kHz Band.
Unfortunately this escalated into a discussion on benefit
or drawback on DSP in a receiver. This was NOT my aim.
I would like to outline the principle problem on receiving
weak and very weak signals in close proximity to strong
signals in general.
The main point(s) of my posting are:
roofing filter bandwidth is critical to prevent
gain reduction by AGC being affected by strong signal
slipping in the passband of the roofing filter
Special (very narrowband) preselectors may be used for
receivers having broad roofing filters.
Relatively selective antennas such as tuned magnetic receiving
antennas may not help significantly.
Perhaps I did draw your attention too much onto my receiver.
I have only one receiver capable of tuning below 150 kHz
and this is (unfortunately) one using DSP. This attracted your
attention - too much to my understanding.
The design of the analog frontend with it's "broad" roofing filter
is the limiting factor in my receiver for the purpose of
receiving very weak signal in close proximity to strong
signals.
====================================================================
To Alan Melia:
The DSP block itself works perfect.
AM> Lack of dynamic range suggests not enough bits in the
AM> ADC but I might be wrong about that.
This is exactly the point. At the time where the receiver was
designed no moderate priced ADC were available which were
fast enough and had more than 16 bit of resolution. ADC with
24 bit resolution would have been the optimum for the purpose of
digitizing an IF of 456 kHz. The implemented 16 bit ADC in my
receiver restricts the theoretical dynamic range to 90 dB
(this was mentioned in my posting).
On the other hand, I don't be aware of any analog demodulator
which has such an dynamic range.
AM> One question I am not sure about....the concentration on
AM> linear phase roofing filter....isn't this only really
AM> necessary for digital signals??
"Really necessary for digital signals" is very much correct,
Yes. The aim of the designer was to not impair the phase
respond in the analog front end very much below the capabilities
of the DSP part. By using "low distortion" roofing filters
the DSP can make the best out of the signal.
AM> Though I must admit it probably gives good impulse
AM> response,
I can confirm this for AM and SSB modes. If I listen to a LF/MF
broadcast radio signal with real (not oversized) 9 kHz HF
bandwidth, I can tell whether the music played in the studio is
origining from an analog magnetic tape or from a CD. Demodulation
is realy "analythic" as compared to analog demodulators.
AM> and that seems important with the types of
AM> QRM and QRN on the 136 band.
Being phase linear helps to reproduce any modulation
with great precision. Impulsive QRN / QRM on the other hand
is another story. As you probably know a frequency (filter)
characteristic in the frequency domain is always related
to a response characteristic in teh time domain.
The narrower a filter and the steeper it's roll off,
the slower it's response in the time domain.
This point was discussed in length in this reflector last
year.
Personally I don't like much the narrow and steep filters
on my receiver to listen in CW mode. A lot of phantom audio
(not intermodulation products!!) is audible when weak CW
signals were to be detected aurally, during presence of
impulsive noise. If I can I stick with filters of 500
or 300 Hz, if sensitivity is no problem. CW signals come
much more contoured with these BW rather than with
200 or 50 Hz BW.
AM> I know what Toni means about design.....they don't make
AM> 'em like they used to Toni. Now if I could only hide this
AM> AOR inside an old 51-J case with all the knobs working on
AM> the functions directly.....
I understand what you and Toni mean. If you temporarily
operate my receiver, being used to the traditional
layout of front panel controls, then it is a real mess.
Now as I am used to it, I don't find it preventing me
to optimze the receivers parameters quickly. Direct
front controls ARE BETTER, no discussion, but I don't
like to pay for it too much. I put more emphasis into
internal quality (reproduction of signal) than on
"neatniness".
AM> I tried to e-mail you direct some time ago but the
AM> message bounced, I dont seem to have a working
AM> address for you.
Sorry for that. I do email in my QRL. So I have to use
what is available. I have no access to parameters related
to correct addressing.
I had similar problems myself. It helped for me to put
the addressees email address manually into address field
rather than using the automatic feature when selecting
"respond to a letter".
Obviously we have one "real" address and an alias address:
[email protected]
or
[email protected]
Try both addresses, filled MANUALLY into the address
field of your email software, at least one should work OK.
====================================================================
To Dave Brown:
DB> My initial reaction is that it seems the analog section of
DB> the rx was designed separately from the digital back end
DB> and no-one considered the overall receiver performance.
Definitely not. The designer team consisted of 2 persons
which used to cooperate successfuly for decades (one deceased
recently).
DB> Deriving the AGC as you described, ahead of the primary
DB> bandwidth determining filters in the digital IF, would
DB> appear to be the main issue.
This is what I tried to express.
DB> Can you manually control the gain in the analog section and
DB> turn off the AGC? Possibly not,
No way.
DB> Some form of coarse AGC in the analog section derived from a
DB> few steps between say middle and top of the A/D converter
DB> range would surely have helped solved the problem.
Actually there is one discrete step. When input levels exceed ?? dB,
a ?? dB attenuator is put into the front end. But this is not
the same as you think about.
DB> Interested in your figure of -102dBm noise level. Presume this
DB> was the external band noise?
Not necessarily. The close in DCF49 signal (-31 dBm) activates the
AGC and gain is reduced significantly. result: Noise figure of
receiver rises accordingly, further it reduces useful dynamic
range. If I tune to exactly 137.00 kHz and cannot hear any
"real" signal, there is already audio noise. Partially this
is local man made noise picked up by the antenna (I am living
in a suburban area), lot of electronic ballast whine audible
time by time. Main noise appears to be simply the receiver
noise level under gain setting condition affected by DCF49
falling into roofing filter's passband.
DB> And what bandwidth was it measured in?
The -102 dBm level was observed while receiving
DF8ZR. No real beat tone, but "modulated quasi white noise".
So I conclude this was the situation of "signal equals noise".
DB> receiver performance with noise blankers switched on.
I fully understand and agree with your concern. My receiver
doesn't have noise blankers. It uses noise reduction
(based on correlation perhaps) instead.
Best 73 de Gamal Soegiono
[email protected]
or
[email protected]
| 