Hello all,
Another Patrick here.
For an honest comparaison, logically the result might be normalized to a
standardized text throughput (based, for example, on BPSK31). For example, the
Olivia 32-1000 has a throughput of 24.4 wpm. MFSK16 about the same as PSK31 (a
little less than 50). This because, it is mechanical to improve a minimum S/N:
if you decrease the speed transmission by 2 you have a gain of 3 dB on the
minimum S/N (however in the reality it is only about 3 dB because the new
parameters are not at the exactly same quality of "optimality").
Note: a word is a group of 6 characters.
In fact, it is not totally honest, because characters can be 6, 7 or 8 bits
large and of course the more bits you have and better it is (with 8 bits you
can transmitted accented latin characters or non-latin characters, when you
can't in 7 bits and in 6 bits you cannot distinguish small or capital letters).
Now to normalize it is easy. For example, if the text throughput is half the
normalized text throughput, the determined minimum S/N is increased by 3 dB and
so on (10* log(ratio)).
73
Patrick (F6CTE)
----- Original Message -----
From: Patrick Novak
To: digitalradio@yahoogroups.com
Sent: Saturday, July 12, 2008 9:06 AM
Subject: Re: [digitalradio] Re: KV9U - Mode Sensitivity / Robustness
Yes... that would be the ONLY proper test... the statistics would tell us...
cheers
P
Tony wrote:
Patrick,
I for one would like to try different modes and compare them
under real conditions. There can be too many variables.
I agree, I think the simulator can tell a lot about mode performance, but it
has it's limitations.
It might be a good idea to test modes according to type. Seems fruitless to
compare a mode that's known to be extremely sensitive and robust against the
"average" chat mode.
Good point about switching frequency when testing; need to mind the wide
modes in the narrow band segments etc.
There was a lot more activity along these lines when sound-card modes were
first introduced. Wonder if anyone recalls digital beaconing?
Station S59DOR had one on 10 meters that would transmit the different
Hellschreiber modes.
Macros allowed the software (IZ8BLY's Hellschreiber) to switch modes
automatically. Have a screenshot of the beacon if anyone is interested.
Glad to see the enthusiasm Patrick! It's the stuff that keeps the hobby
alive.
Tony, K2MO
Kings Park, NY
----- Original Message -----
From: "Patrick Novak" <[EMAIL PROTECTED]>
To: <digitalradio@yahoogroups.com>; <[EMAIL PROTECTED]>
Sent: Friday, July 11, 2008 11:09 PM
Subject: Re: [digitalradio] Re: KV9U - Mode Sensitivity / Robustness
Hi Tony and it is good to see your posting to the group.
I for one would like to try different modes and compare them under real
conditions. There can be too many variables. I understand it is impossible
to do it exactly but how about this scenario:
We will start a contact and send a short file in one mode, then switch to
a different mode (close/same frequency/band) and send the same file
again.
Try several modes and then compare the number of errors in each mode.
If we have a large sample set we will be able to make an educated guess at
the mode which performs best.
There could be a group or individual who will keep the results table, or
each or OF us (experimenters) will keep own statistics.
What do you think? What everybody thinks?
regards es 73
Patrick
VK2PN
Tony wrote:
Rick,
>Has anyone compared Domino EX with the top modes that get
>through the worst conditions and, if so, what have you found?
I've compared a variety of sound card modes for sensitivity and
robustness using an HF path simulator (see below).
I first tested each mode for sensitivity by finding the lowest
signal-to-noise ratio that would still produce 100% decode. These were
direct-path simulations with no ionospheric disturbance.
For what it's worth, DominoEX 11 baud, had about the same sensitivity as
MT63 and Olivia 500/4. DominoEX 4 baud had about the same sensitivity as
MFSK16.
Olivia 500/16/8 were more sensitive than DominioEX 11 as well; Olivia
mode sensitivity would increases with an increase in the number of tones.
Things got interesting while testing for robustness with the ionospheric
simulator. The Domino EX modes fell short to the point where they would
not decode at all with certain disturbances regardless of SNR.
The DominoEX modes seemed sensitive to frequency spread found on the more
disturbed ionospheric paths, especially polar and low-latitude paths.
I think it's important to remember that these are simulations and on-air
testing may show different results. There are no static crash simulators
with Pathsim so it's hard to simulate things like mode recovery time etc.
If anyone is interested in path simulation, it's important to normalize
the audio before it's feed into the simulator. Normalizing keeps the audio
at the same level so one mode is not any louder or weaker than another.
Tony, K2MO
Kings Park, NY
______________________________________________________
Sensitivity Test ( lowest SNR that will produce 100% decode)
(DIRECT PATH - NO IONOSPHERIC DISTURBANCE)
RTTY 45................... -5db
Feld Hell.....................-6db*
DominoEX-11........ -8db
MT63 1K................. -8db
Olivia 500/4.............. -8db
Olivia 500/8..............-10db
PSK31................... -10db
Olivia 500/16 .......... -12db
MFSK16....................-14db
DominoEX-4 baud...-15db
* Simulated quiet band condx. Good contrast between text and white
background.
______________________________________________
Sensitivity Test SNR -12db
(DIRECT PATH - NO IONOSPHERIC DISTURBANCE)
MFSK16.....................Perfect Copy
Olivia 500/16..............Perfect Copy
DominoEX 4 baud....Perfect Copy
Oivia 500/8..................90%
Olivia 500/4.................90%
DominoEX 11 baud...90%
PSK31.........................90%
Feld Hell......................Faint Text *
MT63 1K.....................No Copy
RTTY 45 baud............No Copy
* Feld Hell somewhat readable at this SNR under extremely quiet condx.
_______________________________________________________
Sensitivity Test SNR -15db (faint signal in waterfall)
(DIRECT PATH - NO IONOSPHERIC DISTURBANCE)
DominoEX 4bd..........100% Copy
MFSK16.......................95% Copy
Olivia 1K/32.................90% Copy
Olivia 500/16...............90% Copy
Olivia 500/8................. <50%
THE QUICK BROWN FOX :UMPS OB$[T,3AZY DOG
T.`8 dIC `"N FSX JUd VCR g&?GZY 1PaOlj/== BROWZ+_
/|mu_S R pHB\7JY 9OG
Olivia 500/4...................No Copy
DominoEX 11bd...............50%
tTHE QUICK BDOWN FOXGUMPSLiVER IE LA ' Y BCI
THQUICK) ROWN FOX JU S OVCu THE LAZY DOG
TH QUo OWN FN0 ZUMLx OGR THE tZY DOa? L
PSK31...........................<50%
E QUIhK BRO FOs J.MPS OVEC THE s
-t( a OGTr �ITBRO N Ft uJU0PbtaVEXE LAZY t
OpHE uPIOK BROWN FtoMPS OT0R li0 LA rY Ctep
MT63..........................No Copy
Feld Hell.....................No Copy
RTTY 45....................No Copy
_______________________________________________________________
High-Latitude Path Simulation
(Moderate High-Latitude Disturbance, 3ms path delay, 10hz frequency
spread, SNR -8db)
(3ms path delay, 10hz frequency spread, SNR -8db)
MFSK16......................Perfect Copy
Olivia 500/16...............Perfect Copy
Olivia 500/8..................Perfect Copy
Olivia 500/4..................90%
Feld Hell........................90%*
DominoEX 4bd..........No Copy
DominoEX 11bd........No Copy
PSK31.........................No Copy
RTTY45........................No Copy
MT63............................No Copy
* Text to background contrast marginal but readable.
_______________________________________________________
Mid-Latitude Path Simulation
(Storm condition Mid-Latitude Disturbance, 2ms path delay, 1hz frequency
spread, SNR -8db)
MFSK16......................Perfect Copy
Olivia 500/16...............Perfect Copy
Olivia 500/8..................90%
Olivia 500/4..................90%
DominoEX 4bd............90%
MT63 1K.......................90%
Feld Hell........................<90%*
*Good text contrast, missing characters
DominoEX 11bd
THE QUIRK BROWN FOX JUMPTYOFttT3G LAZY DOG
THE QROWN FOiRJUMPtJeoER TLAZY DOG
THE QUICK NNOWOXlrUMPS OVER ti8AZY DOG
PSK31
THe EaoICK BROWN F Jto c OTeR THE LAZY DOG
THE Q4Icš BROWN MOX p HMPS OVIRHE LAZY DOG
Tr4 QUICK BROWNa o e o JUMPS r -ER THE LY
RTTY45
XMXSPYUIVK BRGWN FOKGDUMHR KZ4 ZILAZYHDOG
OVZHE QUFKH?492,6FF JINQ QOCSKQ0/6&)'6 BMXDQ
NQ QUICK YBRTZNLFMY JUVHS OVER PHE )-+6 DOG
______________________________________________________________
Low-Latitude Path Simulation
(Storm condition Low-Latitude Disturbance, 6ms path delay, 10hz frequency
spread, SNR -8db)
MFSK16...................Pefect Copy
Olivia 500/16............Perfect Copy
Olivia 500/8..............90%
Olivia 500/4..............90%
Feld Hell....................Poor Copy*
DominoEX 4bd .......No Copy
DominoEX 11bd......No Copy
MT63 1k....................No Copy
PSK31......................No Copy
RTTY45.....................No Copy
* Feld Hell letters were diffuse (spread) making it difficult to read.
--
~~~~~~~~~~~~~~~~~~~~~
Patrick VK2PN
QF56pe
Sydney Australia
------------------------------------
Announce your digital presence via our Interactive Sked Page at
http://www.obriensweb.com/sked
Check our other Yahoo Groups....
http://groups.yahoo.com/group/dxlist/
http://groups.yahoo.com/group/contesting
http://groups.yahoo.com/group/themixwgroup
Yahoo! Groups Links
--
~~~~~~~~~~~~~~~~~~~~~
Patrick VK2PN
QF56pe
Sydney Australia
