Discussions on Low Level EME IV
These are Contributions to a Discussion on Low Level EME from the Moon-Net.
From: Bob, N4BP.
To: Wes Stewart
>
Subject: Re: Hearing CW
On Tue, 9 Jul 1996, Wes Stewart wrote:
> When the test ended, I was told that I had tested better than I did the
> first time, back when I was a kid. Now it is possible (probable) that the
> testing equipment is different and that the method is different too, but I'm
> convinced that my familarity with the test signal was significant. I'm not
> suggesting that this is science, but perhaps interesting.
>
That was fascinating to me, Wes, as I went through almost the identical
experience about two years ago at age 54. I've had a constant ringing in
my ears for years and frequent headaches. I finally went to a
neurologist who sent me to an ear/nose/throat specialist. Did the same
booth with the tones and the button to push when I heard them. When I
was done, he told me I could probably hear grass growing. Funny thing
is, when I'm in a crowd of people, it is impossible for me to understand
someone right in front of me who is talking to me. Guess my ears are
also trained for CW...
Bob Patten, N4BP
n4bp@shadow.net
From: Cliff Buttschardt
One item that might be of use to all that work CW and have a problem with
precise sending might be the use of CCW which stands for Coherent CW.
This is cw that is precisely timed at 12 wpm. The formation of
characters occurs in a computer using the internal crystal. It might be
interesting to also use that very conceipt on receive for what has been
reported to be at least ten db increase in signal to noise ratio. This
of course probably will be masked by libration, but the precise sending
remains valuable. Check VE2IQ's software at: VE2IQ
On Tue, 9 Jul 1996, Frank Lumney wrote:
> Hello to all
>
> I am a new subscriber to moon-net and have read with interest all your
> comments on various bands and rcvrs etc, But my biggest problem is with
> sloppy sending. Principally with spacing. Many operators tend to repeat
> their call without leaving a word break or space. I copy all the components
> of the call but cant tell where it begins or ends. I was on 1296 last
> weekend and there was a station on there yl3ag/a. Now I copyed all the
> components looking at my scribbles but that is a bastard call and I had no
> idea how to compose it.
>
> I operate 1296 eme and welcome requests for contacts any time of the month.
> If you make changes and would like to check your station on another station
> don't hesitate to call. It's a long dry spell between activity week ends and
> I would like to get more benefit from my expenses and efforts. I am in the
> unique position of being retired and widowed and live alone so my time is my
> own and am doing all the hamming that we all longed to do but coulden't
> because of other commitments.
>
> Frank w2uhi
>
From Leif Asbrink, SM5BSZ
Date: 1996-07-18, Time: 11.29.00
Greetings to all,
The low level eme discussion, available at
http://www.nitehawk.com/rasmit/wk_eme10.html is now more than 50
pages. Looking through this large text, I can not resist adding a few
comments:
First: What is a QSO (and why do we like to make a QSO)?
As I understand it most EME'ers want to work more initials all the
time. More grids, fields, DXCC countries or US states is also OK.
Although the QSO in itself is very exciting to us, the information
transmitted/received is really uninteresting, which becomes very clear
if you try to explain what we do to someone who never heard about
amateur radio before.The real fun about EME is the technical
challenge. Personally I do not mind "computer gizmo QSO:s" - To me
they would be equally valid as a normal CW or SSB contact.
To me a real QSO has to be random. I think working on skeds is a fine
way of exploring the limits of the communication mode. I frequently
use skeds for IONO scatter on 144MHz, but I do not look for skeds to
try to improve my standings on EME. But I do use the VHFSCHED.SKD info
to know where listen!
Why not a CQ or two when the sked is over ?? I would love to find a
comment in VHFSCHED.SKD "calling CQ after sked" !!!
Of course, when I occasionally have a sked QSO, I regard it as
perfectly valid. To me any QSO is valid as long as the operator can
decide that he has received the information constituting the QSO while
the QSO is still going on. If the decision is based on a text on the
computer screen, or if it is based on something the operator thinks he
can hear in the noise does not matter, even if the operator used a
tape recorder to bring the cw speed down, or just to listen once more.
It is the ability to react on something unexpected that requires the
operator to be present. After all he is responsible for his own call
sign, and he has to decide what to do if the other station deviates
from the standard procedure and for example asks a simple question
like: "3V8BB QRV?" This requirement, room for the unexpected, is
really the hart of communication, and it has to be an essential part
of any new computerised communication mode.
The interesting thing about computerised QSO's is that it is a
technical challenge. I doubt very much that I will live long enough to
see computers beat humans in contests or working DX epeditions. Surely
the hardware is already here - the cost for the processing power
needed is already a small fraction of the cost for the rest of the EME
station.Below I try to explain why I think computer QSO's are so
difficult - understanding the problems helps in making QSO's the
conventional way...
I think the low level EME discussion has shown that there is a general
consensus in that lowering the speed well below what is in use today
will allow lower power levels on the transmit side. The fundamental
question is: how many dB can we gain using the optimum coding if we
limit the total QSO time to say 15 minutes? I think the UNKN422.WAV is
a superb reference - to me it represents the S/N and QSB
characteristics for a signal that allows a random QSO in 10 or 15
minutes with normal CW on 144MHz (if the operator is good). The key
problem is to identify the call sign of the other guy, and to verify
that he has got my own call sign right. This means that we need
something like 12 characters or say 80 bits of information.
To detect the presence of the signal of UNKN422 with FFTDSP by AF9Y,
an integration time of something like 5 seconds is needed, and then
the probability of decoding the right information, 0 or 1, is far from
100%. Even when adding 3dB from doubling the average power by sending
continuously it is questionable if 5 seconds is long enough to allow
a reasonably high probability of decoding the right information, but
if we double the number of bits for some smart error correction, I
think communication would be near 100% provided that faraday rotation
is properly handeled. The QSO time would then be 160*5*2sec, or nearly
half an hour if we use a single frequency.
If we want a real improvement, one bit has to be longer than the
characteristic time of the QSB. Again,looking at UNKN422 I guess that
an integration time of 20 seconds would allow 6dB lower transmit
power, again 3 of them from continuous carreer rather than morse code.
If we use a single frequency, the time will be unacceptable - at
least for contesting. If we use several frequencies for coding
- transmitting only one at a time - the total time may be acceptable,
but then it will be very difficult to handle the problem when several
stations call simultaneously after a CQ.
Unfortunately the situation can not be improved by better modulation
methods if the bandwidth is already choosen for optimum S/N. There is
no way of building a reference phase that will be correct for long
enough periods of time because we can not predict the phase of the EME
signal over times longer than one over the characteristic time
constant, which is the same time constant that sets the optimum
bandwidth.
Using frequency diversity is a possibility to avoid the QSB minima.
In fact a spread spectrum technology might change the EME channel
characteristics in a way that would allow the use of much narrower
bandwidths. That is, when both tx and rx are programmed to jump
between frequencies in a predetermined way, both QSB maxima and minima
should vanish. The amplitude should as a consequence be more or
less constant, and then "I can feel it in my bones" the phase should
also stay more or less constant.If we then use a bandwidth of 0.1Hz we
should have gained about 20dB above today's CW, I believe that the
problem is that we will have to spread the energy over a quite large
bandwidth in order get rid of the QSB. On 441Mhz 70% of the power is
reflected from a surface only 11% of the moon radius, and at 144MHz,
the reflection is even more point like. To understand the QSB, we have
to think of the moon as a large number of mirrors spread out in a
volume of something like 300km wide and 6km deep. Maybe the RMS
deviation in phase between the different reflection points could
correspond to 500 wavelengths or so. This implicates a bandwidth of
300kHz or so at 144MHz. Although I think it is technically straight
forward, I do not think frequency diversity will be accepted. Jumping
between several frequencies with relatively high power, staying only a
few milliseconds on each, will cause problems to other users of the
band unless frequency plans are changed and a lot more of the spectrum
is allocated to EME.
So again: What we already have is close to optimum, and LOW POWER EME
IS NOW!!
What we can do is to improve a little bit. Optimum filters give an
advantage of a few dB in my experience, but maybe good operators can
detect equally well without them. These filters do not require DSP's -
it is just one technology among others. Another small improvement lies
in use of the phase. If the keying speed is significantly faster than
the QSB time, like on 144MHz, the phase is continuous between dots and
dashes, so it is possible to follow the phase - and to extrapolate the
phase into regions where the signal is weaker. The advantage is 6dB
above a simple amplitude detector, but compared to the human brain the
advantage is much smaller.
The only significant improvement that remains is to use the computer
to average over several equal sequences. This is not very easy as long
as the precise timing is unknown, but certainly possible. Most
operators repeat their call signs several times with constant cw
speed, and the computer can be trained to use that. Again, the
improvement over what we do today will probably not be so impressive -
we already use our brains to do something similar!!
Personally I like the technical challenge, and I have spent quite some
time in designing my EME station. I would not hesitate to use
computers if that would improve, but despite a lot of programming
effort I still have the best results using my own ears for the final
decision taking.
73's Leif
From: Clifford Buttschardt
Subject: Re: Compatible CW Schemes
Comments: To: Bill DeCarle ,
jhdavis@aol.com, Brian Beezley
As Brian has indicted, a form of Coherent CW would be benificial. The
software has been available for some time for 6, 12 and 24 wpm! The only
requirement is that the coding be precisely three to one as usually is the
case with CW. A dot one unit, a dash three and spaces three. These
programs are available from Bill DeCarle, VE2IQ in his COHERENT series and
from the LOWFER page as PCW (precision CW), which by the way contains
a look up table in which the 12 wpm only, dots and dashes are displayed!
Cliff Buttschardt K7RR
On Fri, 19 Jul 1996, Brian Beezley wrote:
> >The only significant improvement that remains is to use the computer
> >to average over several equal sequences. This is not very easy as long
> >as the precise timing is unknown, but certainly possible. Most
> >operators repeat their call signs several times with constant cw
> >speed, and the computer can be trained to use that. Again, the
> >improvement over what we do today will probably not be so impressive -
> >we already use our brains to do something similar!!
>
>
> One improvement that easily could be implemented today would be always to
> send synchronous CW. That is, the clock that determines CW-element timing
> never is allowed to stop. A synchronous receive filter might then extract
> the clock from the signal and use it to process the CW in various ways at
> optimal times. A synchronous CW clock might be coupled with one or two
> standard EME sending speeds to eliminate speed-estimation errors. The nice
> thing about this scheme is that it offers potential improvement when using
> fancy filters but is completely compatible with unassisted listening.
>
> The ear/brain already predicts when code elements are likely to occur, so
> filters that automatically synchronize to the sending clock may not offer a
> whole lot of improvement. But regular, synchronized code emission surely
> would aid unassisted listening as well.
>
> The synchronous-clock scheme could be embellished by transmiting Morse-code
> FSK with the second carrier far enough away that it doesn't disturb ordinary
> listening. The second carrier could be used in various ways. For example,
> you could detect each signal with synchronous matched filters, subtract the
> detector outputs, and use the resulting waveform to mildly modulate the gain
> of a receive channel tuned to the tone transmitting ordinary CW. The gain
> hint ought to provide some improvement in copy. Once again, the
> transmission scheme remains compatible with unassisted listening.
>
>
>
> 73--Brian, K6STI
> k6sti@n2.net
>
For comments, typo's and changes: Rein Smit, W6/PA0ZN
Last uptated 07/19/96.
Top Page.