I know that a few people here use datamodes for SOTA including JT65 and JT9.
In a recent update of WSJT-X (1.8) there is a new mode FT8. I tried it out at home on 20m and 6m and it works great on both. The QSO structure of FT8 is the same as JT65/9 qsos but the main difference is that the transmit/rcv segement is only 15 seconds opposed to 1 minute for JT65/JT9. A total qso time of 1 minute versus 4 minutes.
This results in decodes to about -20db rather than -26db but the gain is in shorter transmit/receive times which can overcome propagation issues such as qsb. Also if portable the amperage per qso would be lower giving obvious benefits for battery requirements.
Now for a question. Due to the fact I dont have any old PC’s lying around, I can’t gauge what the decode time is like on the processor, or to be more precise, what the decode time would be on some small portable intel device. On my i7 PC there is not much of a noticble difference in decode time, but considering its less than a second on JT9 its very hard to measure.
If anyone is interested or willing to try out FT8 there is lots of activity on 7.074, 14.074 and when conditions allow 50.313 on something slow. This would enable us to see if it was a suitable mode for SOTA using old/cheap equipment.
Lots of us now with fingers crossed for an Android app for FT8. Could demand a much more powerful Android phone though than the one I have currently. I’ve no interest in taking any device (laptop, tablet etc) additional to a mobile phone, out on activations.
Just in from a quick local activation of The Cloud G/SP-015.
I downloaded WSJT-x-1.8-rc1 and ran it back to back on a pair on low spec Windows tablets.
Hipstreet W7in Z3725G BayTrail Atom CPU, 1GB Ram, 16GB Flash, old Win8/Bing + £1 USB audio dongle.
Linx 1010B Z3735F BayTrail Atom CPU, 2GB Ram, 32GB Flash, Win10Home 32bit r1607 + onboard sound.
The Win8/Bing tablet onboard sound drivers do not support 48kHz sampling which WSJT needs, hence the USB sound card. I don’t recall 48kHz on the on board sound card being supported in the past so at some point the drivers for W10 on the Linx1010B have been updated and they now do support 48kHz using the on board sound.
I setup both machines running alternate periods on FT8 and added a cheap headset to the USB sound card and lthe internal mic/speakers on the Linx.
Both machines happily decoded each other’s messages at the end of each period.
These low spec Atom machines run this and other WSJT modes with no problems. I think these may be the slowest x86 family chips on sale now.
I picked up an 11.6" Acer Chromebook a few years back for about $120. After the Celeron processor and 2GB RAM started getting a little slow for browsing the internet with a bunch of tabs open, I got a better Chromebook and my old one became an Ubunutu (Linux) based system with all of my Amateur Radio programs. Runs fldigi and WSTJ-X just fine, and I’ve used it on a couple activations. Might be a little larger and heavier than a tablet (a whopping 2.75 pounds), but it has a real keyboard and the screen is protected when closed without the need for a case. Even has 2 real USB ports, so no adapters like you’d need for a tablet or smartphone.
It really comes down to your budget, weight/size restrictions, and technical abilities, but most modern systems should be able to run WSJT-X just fine. After all, most “cutting edge” Amateur Radio technologies are a decade or two behind the rest of the world.
I think it’s a little more complex than just porting the source code across. We are doing some Android development at work where the source code was available. In the end it was easiest to simply rewrite the software from scratch.
That being said, the careful matching of the WSJT suite of modes to the rather specific channel characteristics is novel. Obviously that has been done for decades in the professional sphere but not for these requirements as there is little commercial interest in doing what we do. That was the main reason that we got “shortwave” at all back in the day - no commercial interest.
I’ve been running tests in the last few days on Linx 8 (Atom Z3735 quad-core @ 1.83GHz, 1GB RAM, 1280x800 8" display, 377 grams, running Windows 10) + Yaesu FT-817nd. I’ve modded the FT-817 to trigger VOX from the data input on the back so it only needs a simple audio cable from the audio jack to the data port, the whole package is very portable. The tablet’s CPU power seems to be enough, it manages to decode before the next cycle starts.
The audio chipset on the Linx 8 is locked at the 44kHz/16bit setting but WSJT-X 1.8.0 RC1 seems to be running fine (decoding 3-4 US stations per cycle last night). I will look for a newer audio driver tonight anyway and even try the PC, just to make sure I’m not missing decodes because of this. Andy, do you remember where the 48kHz requirement is mentioned ?
I got the requirement from the WSJT docs. This quote is from the online docs at Joe Taylor’s K1JT web page (but it is for 1WSJT-X 1.6)
"WSJT-X expects your sound card to do its raw sampling at 48000 Hz. To ensure that this will be so when running under recent versions of Windows, open the system’s Sound control panel and select in turn the Recording and Playback tabs. Click on Properties, then Advanced, and select 16 bit, 48000 Hz (DVD Quality). "
I know that when I tried JT9 and JT65 on my 7in Windows 8 tablet, I could not get any decode using the on board sound. Using a USB dongle set to 16bit/48kHz solved that problem as 48kHz was not supported by the on board sound.
As I said, I think the drivers in Win10 for the Linx1010 have been updated as it now offers 44100 and 48000 rates and I’d put money on the fact it never used to do so! The Linx8 is the same hardware as the Linx1010 apart from the screen IIRC so it could be just a driver issue. It would be good to know if there is a better driver but the fact it works is nice to know. There were plenty of Linx8 on eBay when I looked at good prices.
Sound card on the Linx 8 is 48kHz/16bit now with the latest drivers, so no external USB card needed. The CPU handles decoding comfortably, it only spikes to 50-60% for less than 2 seconds. I am a bit worried that sometimes I see considerably more traces on the spectrum than decodes … not much I can do about it I guess.
I am amazed that you say this, Razvan. In the first week on FT8 (22 July - 29 July) I have made 450 QSOs, including more than 40 on the 60m band. Are there perhaps stations replying to your CQ calls that you can’t decode?
I usually keep an eye on the waterfall, just to know if I need to prepare for a reply (that’s a JT65 habit). Not much to see there.
But i do have quite a high RX noise level on my backyard vertical, so there might be replies completely buried in the noise. I’ll take the setup out for a field test this week to G/CE-005, see if it’s better.