Topband: Minimum discernible signal ?

Tom Rauch W8JI@contesting.com
Wed, 10 Jan 2001 05:36:15 -0500


Hi Tony,

> There is much talk about the ICOM 756 Pro's low noise floor and weak
> signal capabilities when compared to analog or AF DSP receivers. The shape
> factor of the DSP filters are on the order of 1.2:1 and can be narrowed
> down to 50hz ! The multiple AGC loops are said to do a good job and
> blocking dynamic range is in the 125~130db class. Third order IMD dynamic
> range is around 90 ~95db.  With that said, those who test this rig say
> that there is more to the story than what these numbers reflect. In terms
> of weak signal performance, the IF DSP in combination with an extreamly
> low noise floor is said to be a cut above analog or crystal filtered rigs.

That's fine Tony, but you have to consider the spacing of the test 
signals used in the ARRL tests. Virtually all receivers today have a 
roofing filter that follows the first mixer. This filter is generally 10 or 
more kHz wide, and most often has a somewhat poor shape factor.

It's a "cheap" filter, and a wide one, but it makes a significant 
difference in test results. Testing outside the BW of the roofing filter 
 takes all later stages, including mixers and amplifiers, out of the 
picture.  

While wide spacing IMD and blocking DR tests are useful if you 
have a ham "down-the-road" operating 30 kHz or more away, there 
should *always* be a test well within the roofing filter BW of 
receivers. As far as I know, the ARRL does not do this.

If the testing party does not test or publish narrow-spaced tests 
(spacing much less that the BW of the roofing filter) we have no 
idea at all how the receiver will work in normal CW or SSB 
conditions. Any blocking or IMD numbers, if taken at more than a 
few thousand Hz spacing, are almost meaningless in the 
conditions we typically face on amateur bands below 50 MHz.

For example, my FT1000D looks very good in ARRL tests, yet the 
receiver was almost useless in the very first CW contest I operated 
with it. I heard all kinds of bloops and bleeps that weren't really 
there, because of design flaw in the noise-blanker that allows the 
noise blanker to mix signals ****even when the noise blanker is 
NOT turned on*****!!!!!

I looked at that rig, and wondered why on earth I bought it. 
Fortunately the fix for that only took a few cents worth of parts and 
a few hours of my time, and now it is a decent receiver.

However, if the modified FT-1000 and the stock FT-1000D were 
compared in the ARRL test...they would look the same.

I look at the roofing filter BW, and the test signal spacing, and if 
test signals are NOT both placed within  the roofing filter BW of the 
receiver I ignore the results (unless they are bad, in which case I 
know closer results will be even worse).  
  
> Question - Forgeting about desense from QRM for a momet, if two rigs are
> said to have identical minimum discernable signal figures by ARRL Lab
> standards, how is it possible that a receiver with DSP IF's is said to
> hear better near the noise floor than conventional crystal filtered
> receivers? 

Maybe, but it would have little to do with the DSP system vs 
analog filters. Digging into noise requires all stages are quiet, and 
the bandwidth is as narrow as possible. A 50Hz filter will ALWAYS 
make the MDS look good, even if you add it after the receiver. 
Noise (white noise) level is directly related to the BW of the 
receiver.

The question is how narrow a BW can you really use in normal 
conditions, and how "smooth" is the noise. In practice, I find a 250 
Hz filter is about as narrow as I can ever use and not have major 
problems tuning the station. If there is "rough" noise, like QRN or 
power leaks, a narrow filter can make things worse. My ears do a 
much better job than any filter (including DSP) will in that case, 
and during the summer I often use a 1.5 kHz filter or wider on CW. 

Narrow filters ALWAYS tend to ring out the noise, and blurr it over 
the signal. That's true no matter what the type of filter is, although 
smooth flat filters that are properly designed can be better, for a 
given low attenuation BW, than a crummy poor design. 
   
I always thought that if two rigs show a MDS of -140dbm for
> instance, that they would both hear the same? Or is there more to IF DSP
 then the MDS figure represents?

It sounds to me like the 50 Hz filter "inflates" the MDS. If you tack 
a 50 Hz wide audio filter on a regular receiver, the MDS normally 
goes down.  
 
> Question 2 - Obviously atmospheric and man made noise limits weak signal
> performance on HF but, what difference can one expect on a quiet near
> noise free band ( 15 meters in mid winter ) from a receiver that's said to
> ave  -135dbm and one with -141dbm? Under these conditions, and in terms of
> weak signal performance, is 6dbm a substantial gain? Again, forgetting
> about desense from QRM for a moment.

This is a 160 meter reflector. My 160 meter noise floor, after pre-
amplifiers, on a quiet winter night when the band is open to Europe 
is about -117 dBm measured at 3.5 kHz bandwidth. With a 350 Hz 
bandwidth, the "off-the-air noise floor" of my system is -127 dBm. 
With a 35 Hz filter it is -137 dBm.

To give you an idea of signal levels, W4ZV (the loudest station I 
normally hear when beaming Europe) is normally -30 dBm here. 
DF2PY is normally (average night) -80 dBm. Of course the weakest 
Europeans are down at or below noise floor.

I need a blocking DR of more than 100 dB at CLOSE spacing if I'm 
going to work a weak signal within six kilohertz of Bill. If I don't 
have that DR, it's my problem. I give a "hoot" about the blocking 
DR at a wider test spacing that the entire DX area of the band!

Contests are even worse, as we have multiple -50 to -20 dBm 
signals just a few hundred Hz from noise-floor signals. IMD and 
blocking are important at test spacings of a few hundred Hz!

That's why my primary receivers are now Drake R4C's with 600 Hz 
roofing filters and 125/250 Hz eight-pole IF filters, diode mixers, 
and MMIC IF amplifiers. I have a blocking DR with 300 Hz spacing 
(both signals in the passband of the roofing filter) of almost 150 dB 
now. The system is now totally limited by the spurs of closeby 
transmitters, rather than internal defects. 
 
> I enjoy working CW-QRP and digital mode QRP ( PSK31) and if these claims
> about the IC756 Pro are true, then it should be a real winner for weak
> signal work. Since I'm not a contester, I'm not worried about the less
> than perfect blocking dynamic range of the Pro.

Contesters and DXers should be worried. And if digital modes get 
more crowded, you might be too. Manufacturers pay far too little 
attention to transmitter waveshape and bandwidth on CW, 
harmonic distortion and IMD in transmitters when the audio system 
is used, and close spaced performance in receivers. I'd pay twice 
the amount for good RF performance than I would bells and 
whistles and 9 thousand filters. Harmonic distortion in SSB 
transmitter audio and modulator systems will be particularly 
important with new digital modes piped into the mic plugs on 
radios, and it is NEVER measured. 

> Lab stated that they were able to pull out individual CW siganls in the
> middle of pile-ups using the 50hz filter setting. They also said that they

So what?. Pulling an S-9 signal out of a plie up is nothing. Pulling 
an S1 signal out of a bunch of 40 dB over signals is great. They 
didn't say, or you didn't quote, how weak the signal was. 

> were able to pull QSO's 200 and 300hz apart during the CQ World Wide 160
> Meter CW Contest. How bad can the dynamic range be?

Bad.

he fact stations spaced 250 Hz apart can be "pulled-out" is 
absolutely meaningless. I can do that with a FT101E or a *stock* 
R4C, and both are horrid receivers that I would never consider 
using. What I want to do, and what most of us want to do, is be 
able to hear a noise floor signal a few hundred Hz from strong 
stations when multiple strong signals are present.

It also would be nice if someone tapped an engineer on the 
houlder on occasion, and reminded him that key clicks are not a 
necessary part of sending CW. 

DSP filtered radios, without a narrow IF filter upstream of the DSP, 
simply won't do that. The last time I looked A/D conversion with 
enough speed and bits would cost a thousand bucks. That would 
buy a lot of crystal filters, which when properly designed are just as 
good in BW and don't suffer overload!

Hopefully someday manufacturers will someday look at what is 
important and improve operation on crowded bands.

73, Tom W8JI
w8ji@contesting.com


--
FAQ on WWW:               http://www.contesting.com/FAQ/topband
Submissions:              topband@contesting.com
Administrative requests:  topband-REQUEST@contesting.com
Problems:                 owner-topband@contesting.com