From:
From: Mike, AF9Y
My station showed a non linear drift rate of aprox 0.02 Hz/sec
during the test. I doubt that many ham stations will be able
more stable. At the 2 Hz BW for FFTDSP, the signal would
remain in the bin for aprox 100 seconds. 100 seconds is the
period of integration for the 200x setting.
Conclusion:
-174 dBm will be the minimum detectable signal for a ham station
using a good preamp (0.5 dB NF) when running the FFTDSP program.
(For those of you familiar with the UNKN422 weak signal
challenge, as a comparison, the UNKN422 signal was aprox -156
dBm)
Here's my calculation for the MR Beacon level:
(Someone please correct me if I am wrong.)
MR 437.1 Beacon Tx level = 1.3 watt
= +31 dBm
Distance, Earth to Beacon = 1.0E7 Km
Path loss = 92.4 + 20 log (km) + 20 log(fghz)
= 92.4 + 20 log(1.0 e7) + 20 log (.437) =
92.4 + 140 - 7.2 = 225 dB
Signal level at Antenna =-194 dBm
Assumed Antenna Gain = 20 dBi
(include coax loss)
Signal level at Preamp =-174 dBm
I believe this test confirms the possibility of detecting the probe
with a 20 dBi antenna (including coax) and a 0.5 dB NF preamp.
Assuming 0.5 dB coax/combiner loss, four 436CP30 M2, 10ft antennas
would provide 21.5 dBic. I think this is about the minimum antenna
system required for a reasonable chance of detection. Also, I would
recommend mounting these antennas with non conductive masts and
combining the coaxes behind the reflectors. You can easily loose 1.5
dB if conductive masts are used.
Doppler Chirp correction would be required if the rate is greater than
.02 Hz/sec. If someone can confirm this need, I'll add it to the
program.
de Mike, AF9Y
http://www.webcom.com/af9y
Work: mwcook@itt.com
Home: mwcook@cris.com
From:
E-mail: James, G3RUH
From: Ian, G3SEK
Subject: Re: RX Experiments at -174 dBm
Mike Cook AF9Y wrote:
>As James points out, the antenna array temperature is also
>important. Here's what you can expect with different
>types of arrays:
>
> Commercial Dish with Cassegrain = 18 deg K
>
> Typical Amateur dish using dipole/horn feed = 65 deg K
>
> Opitmized 16 Yagi Stack = 85 deg K
>
> Wide spaced 16 Yagi Stack (poor stacking) >170 deg K
=======================================================================
At 430-something MHz, the yagi figures are still too pessimistic.
An optimized single yagi, such as the DJ9BV series, has the entire 'rear
panorama' suppressed by well over 20dB, and mostly more than 30dB.
Taking the 20dB figure, that means that when aiming at an object high in
the sky, noise pickup from ground at 290K will add only 2.9K to the
system noise temperature.
Unless the yagis are very badly stacked, the noise pickup from the rear
of the whole array should not be significantly different from one yagi.
At one time there were quite a lot of poorly stacked arrays of 16 noisy
yagis on 432, but thanks to natural selection there are very few of them
left today (yes, I could give callsigns :-)
DJ9BV has modeled several 4-yagi arrays at 432MHz. Including sky noise
of 15K plus noise generated by internal losses, rear pickup and sidelobe
pickup at low elevations, the best antennas came out in the 25-30K
region. Even the very worst didn't go above 50K. You should of course
add contributions from phasing/feedline losses and the RX noise
temperature, but with a good system the total from all sources needn't
be higher than the 85K quoted above.
With the 2Hz effective bandwidth, that looks like a noise floor of
-176dBm.
=======================================================================
>
>Does anyone know what temperature to expect for a 2 or 4
>bay, 10 ft helical array?
=======================================================================
Not I, but it's probably going to depend on the size of the backplane
and the stacking distance. Unfortunately I don't know of any equivalent
to the DL6WU rule for yagis, that would apply to helicals.
If the polarization of the Mars Relay beacon is circular, I'd agree with
James that a dish would be the most suitable, because a CP feed is so
easy to arrange. I'm extremely suspicious about the real-life CP gain of
multiple crossed yagi systems because the feed arrangements are so
messy.
An effective dish for 437MHz would have to be quite large because
spillover, feed blockage and other inefficiencies start to become very
important at smaller diameters. As a guide, a 12ft TVRO dish is very
effective for EME at 1296 and above, but useless at 432, so I'd suggest
15ft as an absolute minimum for receiving the MR. Otherwise you'd do
better with yagis or helicals - if you can make them work well.
73 from Ian G3SEK Editor, 'The VHF/UHF DX Book'
'In Practice' columnist for RadCom (RSGB)
Professionally:
IFW Technical Services Clear technical English - world-wide.
From: Tom Clark W3IWI
Subject: forward
From: Ray Soifer <71331.1337@CompuServe.COM>
To: "Tom Clark, W3IWI"
Cc: "Bill Tynan, W3XO" , "Phil Karn, KA9Q"
Subject: Article by SM5BSZ
Tom and Phil,
Since I don't think that either of you gets W2CRS' "VHF EME Report," I've typed
in the following article from the June 1996 issue, which arrived today (I have
no scanner and I'm not within range of a fax machine). Any errors or typos
should therefore be blamed on me, not on the author:
Optimum Use of EME Channel
by Leif, SM5BSZ
The use of morse code exactly the way we are used to is in my mind very close to
optimum use of the EME channel at 144 MHz. The suggestion by KA9Q and W3IWI,
presented in the April issue of the bulletin implicates very poor use of the
final amplifier. With single yagis and both ends, something like 500W is needed
for each tone, so the idea to transmit four tones simultaneously will require a
linear power amplifier capable of delivering peak power outputs in the order 8
kW. The four tones will all be in phase at regular intervals so the output
voltage needs to be four times what is needed for 500W which leads to 16 times
the power.
The best use of the transmitter is to transmit an unmodulated carrier. This
means that if we want to use a multi-tone scheme, the coding with N tones only
gives us N states if we do not count the state with no tone on.
AF9Y points out, in the same issue, that the characteristics of the EME channel
will determine the optimum waveform to use, and with the above in mind, this
means that the channel characteristics will determine the optimum duration for a
single data bit. On 144MHz the libration fading will typically modulate a
carrier with a few Hz, so therefore the receive channel needs a minimum
bandwidth of something like 5Hz.
There are two possible optimum modes of communication, one is to make the bits
so short that a significant amount of information can be transferred within one
QSB peak. The bandwidth then has to be in the order of 20Hz. This is in fact
the mode in use today, and switching from morse code to some other code would
not give any significant improvement.
Lowering the speed compared to what is common practice today will reduce
performance. Getting characters one by one will not give enough information -
the information is not only the characters themselves, but also the sequence
they come in. With a very slow speed CW one can use 5Hz bandwidth, which gives
a 6dB improvement in S/N, but the fading is much deeper than that, so although
one can much easier say that the signal is there, it is not possible to get any
useful information out of it.
Lowering the speed to much below normal CW, using 5Hz bandwidth and post
detector integration is a possibility to detect signals not using the QSB peaks,
but instead the average signal level. In my experience the average signal level
is about 10dB below the peaks (peak value within a one minute period). The
bandwidth reduction gives 6dB so we need another 4dB that have to come from post
detector filtering. To get that, a post detector time constant of about 1.5
seconds is needed. In a real circuit probably 5 or 10 seconds is needed since
in this mode information gets lost if one bit is lost. so the integration time
has to be long enough for the probability of the average signal dropping below
the threshold to be very small.
With the above in mind I would suggest two possible modes of improved EME
communication. One is to transmit a continuous carrier, jumping between say 128
different frequencies, staying at each one for say 5 seconds. The first and
last 5 seconds could be at frequency 0 and 127 respectively and each
transmission would have to be 2 minutes or more. Each frequency would be one
character, and decoding would be by means of something like FFTDSP by AF9Y, with
some simple routines added. The transmitter could be a SSB rig fed with a
single tone generated from a soundblaster. With a channel separation of 15Hz
the total signal would fit in the standard bandwidth of our rigs. I think this
way of making EME would mean the possibility to work EME using about 3dB lower
erp as compared with present methods. This is far below the 10-20 dB
improvement suggested by W3IWI so I guess he did not take QSB into account.
This communication mode, using the average signal level, will be very reliable
once the signal threshold is overcome. (The estimate of 3dB improvement is based
on the information that this is the level at which signals are detected with
FFTDSP compared to when a random QSO is possible - with a lot of time and
effort)
The other mode I propose I think would be much more elegant, and it is probably
slightly better. The real beauty of it is the compatibility with todays CW, and
the concept is simply to define "the standard EME time window." This means we
decide for example that the whole message has to be exactly 15 seconds. The
computer will calculate the appropriate speed and number of repetitions of
various parts within the message. Exactly the same sequence is then transmitted
during a full 5 minute period, synchronized to the 15 seconds window. With a
computer it is easy to calculate "sliding FFT," with say 256 points. This could
give a bandwidth of 20 Hz corresponding to two points in the FFT. Every second
we would get 40 such transforms, so in the whole 15 second window there would be
600 transforms. To store the power at each frequency and time would require a
matrix of 256*400 points which is no problem with a PC today. By just adding
point by point the 20 matrices belonging to the 20 windows within the whole 5
minute transmission period we would have made a post detector integration that
improves the S/N by 6.5dB, but this is the average S/N, so using the ears would
still probably be better. If we instead store in the matrices 0 if the
correspondiing power is below 2* the average noise power and 1 if the the
current value is above, the sum of 20 matrices would have a column of values
significantly larger than other columns if a weak signal is present. I am quite
sure that correct information can be extracted at levels well below those needed
today.
Both methods require good frequency stability. Of course I do not mean that
above suggested times are really optimum - I just want to point out the two
different types of modes that are the two possibilities that well use the EME
channel according to my understanding of the problem.
73s Leif / SM5BSZ
(W2RS note:
Leif is very active on 144 MHz EME but to my knowledge he does not currently
have an Internet e-mail address. He can be reached through the 2-meter EME Net,
weekends on 14.345 SSB. Leif is well known for his development of an innovative
polarization diversity system, using 45-degree increments and a voting device at
the receiver, which achieves close to optimal results in combating Faraday
rotation without the need for mechanical Z-axis rotation.
His calculations in the above article appear to assume random operation rather
than prearranged schedules. My own work has shown that, at 144 MHz, it is
possible to complete a prearranged schedule with a S/N ratio 6-8 dB lower than
that required for a random QSO; this results, I believe, from "cerebral signal
processing" as the receiving operator's brain makes use of the information it
has when trying to copy the incoming signal.
73, Ray)
From: Brian Beezley, K6STI
Subject: UNKN422.WAV
Mike, I wanted to tell you about my ongoing efforts to crack your
UNKN422.WAV EME callsign mystery. I'm motivated primarily by the technical
challenge; the $100 prize you offer will barely pay for the electricity I've
expended! I think this is a really neat challenge you've come up with, so
I'm copying this note to the EME reflector and to the TAPR HFSIG reflector
(where there's recently been some discussion of CW decoding) in hopes of
spurring others to try tackling the problem (perhaps one more time).
Earlier this week I seriously looked at the problem for the first time. I
spent the better part of two days on it. Listening to the file without
benefit of signal processing, I could just barely hear a tone pop out of the
noise for a second or two. (The first time I listened, I heard only noise.)
This was so unexpected that I downloaded the file again
(http://www.webcom.com/af9y) to make sure I had the right one!
I cobbled together a special version of DSP Blaster to process the .WAV file
instead of real-time, sound-card audio. I found neither its LMS noise
reduction nor 15-wpm matched filter helpful.
For quite some time I've wondered whether a matched filter really is optimal
for processing CW signals for detection by ear. A matched filter is optimal
for threshold detection of signals in noise under certain conditions, but I
wonder whether the conditions hold when preprocessing a signal for detection
by ear. I don't know what the ear/brain does to detect CW, but it's
certainly more complex than simple thresholding. For example, for many
years I've preferred to use an SSB-bandwidth filter to recover weak HF CW
signals rather than a 500-Hz CW filter. The additional noise somehow makes
signal detection seem easier. (I haven't run a test to verify that it
actually makes detection more reliable). I find listening through extremely
narrow filters somewhat disconcerting because they shape the
background-noise spectrum approximately to that of a sine wave at the filter
center frequency. So your ear must then distinguish between the desired
sine wave and weaker ones that bobble around in the background. Even though
I estimate that DSP Blaster's 15-wpm matched filter improves CW detection by
perhaps 1 dB for my ears, I sometimes prefer to detect CW in wideband noise
(even though listening for long periods that way can be fatiguing). (It's
interesting that most ops seem to vastly prefer narrowband CW filters.)
Since UNKN422 remained a mystery, I tried implementing a signal-processing
technique I had planned for DSP Blaster version 2. I've had my Pentium-100
for one year and this was the first time I ever ran up against its
processing-power limits. I had to recode the algorithm using
Pentium-specific, floating-point optimizations (pipelining). It immediately
yielded clearly audible dots and dashes throughout the .WAV file. Now I was
getting somewhere! Right away I discovered I had misestimated the sending
speed. I now was sure that recovering the callsign would be easy. I
visualized the "K6STI Cracks UNKN422 with Secret Algorithm" headlines. So I
listened carefully for some time, wrote down some possible calls, and
consulted your list of 1000 EME callsigns, among which is the mystery
callsign. Hmmmm...nothing matched very well. After more listening I found
that I could mentally transform any given sound segment into my current
best-guess callsign with very little wishful thinking! I was familiar with
this phenomenon but thought I'd outgrown it years ago. I've heard 160-meter
operators talk about it and I imagine it must be old-hat to you EME ops.
Twice I was quite sure of the callsign. I was ready to submit my winning
entry and thought I'd take one last listen. Then I discovered that I heard
completely different letters by altering playback speed and pitch! This was
quite a shock to someone who prides himself on his hearing acuity. Although
the dots and dashes are readily audible, the signal level is quite variable.
(I assume this is due to some kind of moon multipath.) Often the code
elements seem to have holes in them or to be smeared together. Sometimes
it's not clear whether it's one long dash or a dot and dash. I also
discovered that my callsign choices tended to depend critically on one or
two crucial letters. It was very easy to persuade myself to ignore other
sounds that really didn't fit but clearly weren't important because, after
all, surely this was the correct callsign! The fascinating thing about this
problem is that it involves both a technical and psychological challenge.
In frustration, I tried cascading a semi-matched filter with the
processor-intensive technique. It didn't help. I really didn't want to
guess at a callsign, so I kept listening over and over, tweaking algorithm
constants. More new callsigns. I got so involved with the problem that my
sleeping schedule became disrupted. When I reluctantly stopped listening to
take a much-needed nap, I fitfully dreamed about weak CW! I decided it was
time to return to normal life and finally gave up on the problem late one
night. Temporarily.
I have one more technical trick to pull. In fact, although it's not
implemented in the current version, this trick was the whole reason I
started the DSP Blaster project. I plan to implement this technique in
version 2 along with the processor-intensive algorithm. I think the
combination may provide enough S/N enhancement to yield a solution. It will
take some experimentation to get right and a fair amount of programming to
implement, so I expect it will be a little while in coming. Meanwhile,
maybe someone else will decide to try a concerted attack on the problem.
I think your UNKN422.WAV challenge would be a really neat term project for
university engineering students. It's a fascinating mystery, involves
signal-processing theory, practical engineering, programming techniques, and
human perception and psychology. It's really a lot of fun! You might try
to locate an engineering professor somewhere who is also a ham and lay it on
him.
73--Brian, K6STI
k6sti@n2.net
From: David, KT1X
Subject: Code vs. No code and VHF..inappropriate styles and debates
While reading the "debate" on the code and no code stuff, a couple of
thoughts came to my mind.
FIRST...Why is this being argued as a code vs. no-code issue on a VHF
Reflector? The VHF rules were changed years ago...the battleground is
really for easy access to HF previlages. The discussion over CW on VHF
should clearly be about effectivness (S/N, bandwidth, realities of
propagation, cost) alone. Isn't that clear!?? Issues such as its usefulness
to hams should dominate our discussion.
On HF, the arguments seem to be the following in favor of code:
It is historical..with out this OPERATING SKILL (NOTE I DID NOT SAY
TECHNICAL SKILL), one is not skilled in a mode used by daily by many HFers
for a variety of reasons.....personal, technical-economic etc. and yes, even
one that hasn't been mentioned,.....a bit of privacy.... since only good CW
ops are likely to eavesdrop on your narrow band converstation. CW is used
by many modern HF hams for a variety of reasons, including to escape the
brashness and BS of phone modes, with their "elbows out" get on the "soap
box" style of operating....not to mention the QRM of all types (this is the
same reason I seek out unoccupied bands and operate robust digital HF modes).
As, far as the effectiveness of CW goes, it is a COST EFFECTIVE weak
signal/low power technology, HF or VHF, unlike many of the more
advanced/but very interesting weak signal technologies which generally cost
an arm and a leg, if they are available at all to even the technically adept
ham.
Remember, much modern communications technology of interest to business is
almost antithetical to the ham spirit. Business folks want it to be
dependable, exploitable, dull, owned, centrally controlled, and other than
gadgets to "hook" or sell to the public, they want the technology to be
invisible to the average user.
Argueing whether or not CW is useful to commericals (I, like many hams was
one), or maintaining that a certain mode is/is not "outdated" based on what
the current use/training of communications techs demands is IMHO bogus. Ham
radio is not some sort of official training ground for telegraph ops or
computer/communications techs. It has traditionally been a multifaceted
hobby with a few basic enterance requirements to assure capable operating.
It seems to me that hams should do things because they are fun, educational,
interesting, work for our purposes, not to ape commerical communications,
except if it fits a ham oriented agenda. How exciting/cost effective would
it be to call up folks on the international telephone circuits just for a
brief chat, or to try out your latest telephone gadget?
It also seems to me that much of the "debate" on code/no-code is one with
some hidden agendas, not relating to the issue as presented. So far nobody
has mentioned them, although one might detect their presence in some of the
typical hostilities present in the usual code vs. no-code flap.
Many of these "arguements" are based on stereotypes and personality, and
generational culture clashes, as well as the need of certain individuals to
feel superior or to vent their frustrations. They are argued using crude
half-truths or veiled insults. It is a common human tendency, exacerbated by
the "cocky bullshit oriented American persona". "If I don't know it or use
it, what good is it!" This faulty attitude seems to be applied to justify
things whether the "it" is CW, digital, HF, packet etc. It is a fault
shared by "old farts" and "brash newbies" alike.
As a fairly active long-time ham whose had had a lot of experience and fun
on a variety of modes and bands, I am quite insulted to be characterized and
stereotyped by some soap box extremist from either of the two camps, and I
suspect that many others are, no matter what their interests, background and
experience, use of modes etc. In case you haven't noticed, these stereotypes
have been picked up by the media, and hams are either portrayed as geriatric
cases with straight keys, or socially challenged gadget geeks.
I personally find many of the styles of operating in ham radio today to be
obnoxious....on many bands and modes.....VHF/FM/HF/SSB etc. It's not the
mode or the technology, its the style of the ops, the wise-assed
know-it-alls, and those with narrow vision and a need to feel superior who
bug the be-jesus out of me. Of course, that doesn't surprise me, come to
think about it. In many ways I suspect ham radio is just reflecting the
general society, and it will become increasingly like it.
If we want ham radio to have a worthwhile future, this is where we must work
at improving things. If not, it will end up like the CB debacle..a lowest
common denominator sewer.
David
PS...I would be all for replacing the higher speed HF CW requirements with
one that was one reasonable speed, and then linking further operating
privilages to technical proficiency based on in-depth knowledge about things
like propagation, RFI, RF safety, design of filters, advanced modulation
techniques etc at a conceptual level other than memorizing terms and regs.
I know several long-time hams who were turned off by the lack of technical
challenge for even the old style Extra before the current round of "question
pool" exam system came in.
This challenge was what the original incentive licensing was supposed to do,
except that over time they dumbed the tech part down and over did the code.
At least if we reemphasized the tech part, we could argue that ham radio has
educational value. Problem is, don't know how this could be implemented
within the current "get rid of the FCC" mania...and I am sorry to say, it
wouldn't sell as many bells and whistles radios to an eager crowd of "gadget
consumers".
For comments, typo's and changes: Rein Smit, W6/PA0ZN
Top Page.