I released PocketDigi 1.0.11 at http://pocketdigi.sourceforge.net. There are following new features implemented: RSID, PSK63F, PSK125, sampling rate conversion, tx delay/tx tail/ats3 preamble/postamble settings, Handheld PC 2000 improvements, Waterfall FFT optimization and basic Smartphone support.
RSID is the famous Reed Solomon ID of Patrick Lindecker F6CTE. Patrick added this feature into his MultiPSK software year ago. Until now only MultiPSK (and since last week also Cesco's FDMDV?) implemented this feature. Instead of CW ID or Image ID, a sequence of 15 MFSK tones is sent at the beginning of every TX to identify mode and center of frequency of the following data stream. RSID has very high redundancy, so if the RSID is decoded with zero or one error, there is extremely low probability of false detection. The feature is very handy to automatically detect mode and submode of rare modes and therefore encourages experimentation. Also it is easy to tune MFSK16 mode as no tuning is needed at all. I found RSID very useful with ATS-3b transceiver, at it has only narrow not very flat CW filter. Even if the whole digital signal does not fit into the filter, if the mode has high enough redundancy and is strong enough, it will be still decoded if tuned properly. But without seeing the whole spectrum of the signal on waterfall, it is very difficult to tune it. With RSID receive on, one only needs to get the RSID signal into receiver passband to make sure the signal will be tuned correctly.
PSK63F is a nice mode invented by Nino Porcino IZ8BLY. It is a crossbreed between MFSK16 and PSK63. It uses convolution code and varicode of MFSK16, but for hardware layer it uses binary phase shifting at 63.125Bd. PSK63F tries to fight ionospheric flutter by faster phase modulation and forward error correction. It is well known that BPSK31 does not survive polar flutter. BPSK63 is better, but one loses 3dB. For DX work, PSK63F could gain back that 3dB and still show lower sensitivity for polar flutter.
PSK125 is nothing more than 2x faster, 2x wider and 3dB less sensitive BPSK than PSK63.
I finally added sample rate conversion. It took me a bit longer, because it is done in fixed point arithmetic again to make it computationally feasible on Windows CE devices. I have good experience with my iPaq 3630 and Jornada 720, they show very little clock error. But my new laptop shows as high as 4% error on TX and 2% on RX. Even PSK31 did not work at that level of clock error.
With the release of ATS-3b kit by Steve KD1JV, I received reports about having trouble to get the communication interface working. ATS-3b is controlled by Manchester modulation generated by sound card, which is translated to digital levels by a simple slicer. Some sound cards generate ugly transient response at the start and end of sound. I added four new parameters to control the shape of transient response: TX delay, TX tail, ATS3 preamble and ATS3 postamble. The first two are silence intervals. The next two are intervals of Manchester idle tone. The idle tone is shaped by raised cosine function now to minimize clicks, if it is longer than 50msec. Also Jornada 720 generates some ugly click about 200msec after sound output opening. I was not able to find out the cause, but setting TX delay to 250ms either avoids the click or at least shifts the modulation after the click.
Until now, I relied on testing of Handheld PC 2000 build by Helge Tefts. Last month I bought Jornada 720 on e-bay for whooping $40+shipping. I did some user interface improvements on that platform, made some missing features available like context menus on RX/TX windows etc. Jornada 720 has a two LiIon cell accu. I bet ATS-3b may be powered from the handheld's accu.
I replaced complex FFT by a real FFT for waterfall calculation, which lowers current consumption. Now all the DSP calculation uses less current than user interface updates + FEC and certainly less than backlight on my iPaq 3630. This is quite interesting. It shows that for HAM radio digital modems and PDA sized display, DSP CPU will be probably counter-productive as DSP is usually clumsy for general purpose computing. To improve current consumption even more, I would probably need to replace some stock user interface components from Microsoft like RX window by a custom one or get back to a custom design with very small monochrome display and simple user interface.
Ondra OK1CDJ bought himself a Smartphone HTC S710. The device does not have a touch screen and did not show menu in PocketDigi-1.0.10. Smartphone only accepts menu with two main level items. I modified PocketDigi to detect Smartphone and modify menu accordingly. There were some more changes to make it work on Smartphone. Here is a screenshot from Ondra running PocketDigi on Smartphone and FT817.
In my opinion the non touch screen devices are more suitable for portable hamming as they are more rugged. Also Smartphones are designed for less energy consumption, often using smaller display, slower CPU, having integrated thumb keyboard and handsfree connector.
Ondra's Smartphone runs on Windows Mobile 6.0. He uses a tool to emulate mouse cursor to access tuning on the waterfall and to switch between RX/TX windows. I plan to improve usability to be able to tune by the D-pad. If you have a Windows Smartphone with WM5 and newer (no touch screen), please give it a try. I need feedback as I do not own any. I configured PayPal donations at pocketdigi.sourceforge.net. Please consider donating small amount. From the donations I will buy a Windows Smartphone to develop on.
If you have any problems with the software, if you have suggestions for improvement, please do not hesitate to send me an e-mail.
At the end of this e-mail is a list of changes since 1.0.10.
73, Vojtech OK1IAK
2007-01-01 Vojtech Bubnik <bubnikv@...>
- Added configuration of TX delay, TX tail, ATS3 preamble and ATS3 postamble
- Channels menu split to Channels and Modem to fit on 240 row screens.
- Handheld PC 2000 (Jornada 720 and others): Dialog boxes have OK and X
buttons on title bar now.
- Basic support for Windows Mobile 5.0 Smartphones with at least 240x240
You will still need a mouse emulator utility to control waterfall and switch
between RX/TX windows.
2007-12-22 Vojtech Bubnik <bubnikv@...>
- Added RSID support. Thanks to Patrick Lindecker F6CTE for Delphi source and
- Default value of TX prefetch buffers set to 3 on behalf of user report.
- Sample rate correction finalized. Automatic sample rate detection based on
user expereince with MultiPSK.
- Added menu accelerators. Very useful when using device with internal or
Alt-T: tools, Alt-C: channels, Alt-M: Modems, Alt-A: mAcros
- Added F1-F12 function key accelerators to execute macros and Ctrl-F1-F12 to
- Handheld PC 2000 (Jornada 720 and others): Dialog windows are now centered
on the screen.
- Handheld PC 2000 (Jornada 720 and others): Bugfix - menu is displayed again.
- Handheld PC 2000 (Jornada 720 and others): Context menu on RX/TX windows
opens on Alt-Tap or double tap.
- Battery consumption improvement: Waterfall calculation optimized.
2007-11-18 Vojtech Bubnik <bubnikv@...>
- Added PSK63F and PSK125 modes
- Mode dependent macros
- Sample rate correction library ported