Goodbye FT8, Hello Olivia, The MAGIC Digital Mode For HF!

Sure.

I just wanted everyone to know that Joe's description is typical for matched filter comparisons: he's not contriving a new way to count SNR, for example.

73
Chip W1YW

 

Yest FT8 is very much "canned", but it is still a fun mode. I've used PSKx, JTx, and RTTY but haven't used Olivia yet. If I want to rag chew I'll use SSB or PSK.
When the band conditions are poor (which is often lately) weak signal digital modes are still very much alive. We had a CME hit yesterday. But last night I saw plenty of signals from Australia, New Zealand, Japan, South Korea and the Philippines (and plenty of USA) on 15m and 17m with a little wire dipole and less than 50 watts. I've made many JT65 contacts with as little as 1 watt. SSB? Dead as a hammer, especially above 14 mhz. I'd rather make a contact than listen to static or some old farts comparing their latest surgeries.....

 

Reference from QEX last year by both Steve K9AN and Joe K1JT. This paper describes their open source decoder specifically for JT65, but I think its safe to assume most of it applies to FT8 as well.

https://physics.princeton.edu/pulsar/k1jt/FrankeTaylor_QEX_2016.pdf

As to specific reasons for design choices,

"JT65 uses timed transmitting and receiving sequences one minute long. Messages are short and structured so as to streamline minimal exchanges between two amateur operators over potentially difficult radio paths. ... "It should be obvious that the JT65 protocol is intended for the basic purpose of completing legitimate, documented two-way contacts, but not for extended conversations. "

Regarding embedded "clock" / symbol data, page 16, top left, a concise list,

JT65 Message Processing, stages 1-12, especially starting at,

6. Receiving software at B: remove impulsive noise; detect synchronizing signal, measure its frequency and time offset. Etc etc etc ...

Regarding statements of SNR values,

Jump to Appendix A back on page 15.

Note the distinctions made between those used in conventional professional literature (Eb / No) to characterize decoder performance versus comparisons based on SNR for Amateur Radio applications, and the equations (15-16) where SNR2500 is written in terms of the Eb/No.

It would be interesting to see how Olivia or other digi modes would fare under similar analysis and simulation.

Meanwhile, for good or bad, the market has made its choice known.

73 de John - WØPV

 

There is no doubt about that.

If I remember correctly, this was the known-bit-pattern signal that was interleaved in the data, and used to do best-case fitting, correct? If so, it assumes that the system clock is the bit-clock source, and that its agreement with the received signal's bit edges was already within 60ms.

Indeed.

 

In other words, you really think you earn an award for letting your computer talk to another computer you don't even know is there? I think your computer should get the award, because on FT8, you don't do squat.

Hmm I think I heard someone talking to me so imma gonna log it as DX cause I can get a paper saying im cool.

 

OK guys, we all need to know that for ham radio operations in general, signal strength reports and SNR reports are pretty much a crap shoot. Maybe for certain commercial operations they are much better defined. For hams the reports are pretty much for our entertainment, amusement or amazement. They have been going on since before the days of RST reports. Take them all with at least a grain of salt or maybe even a tsp. of salt.

The amazing thing to me and what really counts is that some of the digital mode software can decode signals that you can't even see on the waterfall. The signals are low enough that when mixed with the noise around them they disappear and often can not even be heard if you are listening to the RX audio!

Where the 'clock' comes from is immaterial! REALLY! It is my contention that every means of communication gets a 'clock' from somewhere whether it is the latest and greatest digital mode, old CW or even the human voice.

Don't believe me? Try copying by hand (or in you head) someone sending very poor CW where the inter element or inter character spacing varies greatly or where the dot and dash element lengths are almost the same. You will find it very difficult to get good copy. Then try it with a computer. Copy will be even worse! Somewhere in our brain a 'clock' is running. It is measuring time passage. It is far more flexible and forgiving than a computer's clock but is running just the same.

"Decoding" human voice works the same way. Don't believe me? Try listening to speech that has been sped up or slowed down by a factor of four without changing the pitch of the voice. (There is software, etc. that will allow you to do that.) It is very difficult to catch everything being said. All the same syllables are there and they are all in the same order but the problem is our brain is use to processing speech ('data') at a certain rate or somewhere close to that rate. 'Rate' implies an element of timing is involved. How do you measure time? With a clock of course! So somewhere in our heads a clock is running and our brain has been trained to expect to hear words / syllables at some rate. If they come too slowly or too rapidly our brain has a tough time making sense out of what it is hearing. a.k.a. Data out of sync with clock!

Dave K3GAU

 

Beats the heck out of letting your computer talk to others computers and claiming contacts when all you did was turn on the radio and go take a nap.

 

Did a little experiment today with JT and FT. Tuned to the 20m channels, quite busy of course, many strong sigs audible above noise.

Set my PC clock manually, at random, to a value obviously in error of the sequence by at least 10s. Of course the decoders did not produce any received data. Duh.

Then I set the PC clock again, manually, just "by ear", listening for the end of a xmsn interval and counting the seconds displayed until then start of the next. This was done simply using the Win 7 clock "Change Data & Time Setting" box, setting the figure to a round number, like nn:nn:00, and clicking when on cue.

First with JT65. Its long "off" segment made it easy. Even on the first pass, which was a bit clumsy, received data decoded, messages across the channel flooded in.

Verified how close this manual clock set technique was to a standard using https://time.is

Even my first sloppy pass was only about 0.7s off. Subsequent retries produced improving results with practice. Now most are validated to within 0.2s. This could be repeated without any verification at all.

Repeated it all with FT8. Of course with a faster cycle and less off time that's a bit more challenging. But got similar results and success.

Also verified that the PC clock need not be set accurately to the correct time, only to the nearest border of a 60s or 15s segment, ie, on JT the clock could be set for 23:00:00Z when UTC was really 23:03:00, no decode problem. Same with FT on 15s intervals.

So, it seems with JT / FT tolerances there may not always be a need for the internet, GPS, WWV, etc etc. as long as the shared "channel" is occupied by at least one audible signal. (Note - that signal may not even have to be disciplined with a standard either, but then of course, other signals that were would not decode.)

But I suppose an operator performed manual receiver system clock calibration method, even using an "in-band" target signal (obviously not workable when only inaudible signals are on channel) will still disqualify these modes as being legitimate amateur radio digital communication for some eh?

73 de John - WØPV

 

I don't understand why everyone seems to be "down" on people using FT8 for awards. It's certainly no different than someone using SSB droning on and on AND on with "CQ Contest blah blah blah" for hours on end. The level of QSO is just as impersonal. I happen find SSB contesting activity absolutely mind-numbing, HOWEVER, if it's your cup of tea, I'm happy for you. Contest until you pass out as long as you are having fun.

Have you seen those guys using these things called paddles and bugs? Pfft! Real men use straight keys!

Look, FT8 QSO's require human interaction with the computer in order to occur. The virus between the keyboard and chair has to operate the software. It's simply another method of communication between individuals using the radio/computer and an accurate time source for making contacts. It happens to be a digital mode where only a short bit of information is exchanged, which seems to be immensely popular at the moment. Some people like to talk, some people enjoy CW, some guys enjoy sending cat pictures over SSTV and others like computers and the digital modes. They all have their unique uses. They are simply tools in the box.

I hate Gin, but I enjoy Vodka. I also prefer cycles over hertz. RELAX! Perspective time folks, this is only a hobby, but portions of this thread remind me of the operating system wars played out over irc networks in the mid-90's. (Windows folks vs OS/2 users)

Now, WSPR qualifies somewhat for what you described, computers and radios listening to each other but not "talking". To the best of my knowledge, no awards are given for WSPR.

 

Not for me. Your experimental results were very interesting, and I'm glad you took the time to do them.

 

Agreed that there are other good digital modes for poor cndx, small stations etc. But the nicest about JT65 is to put together efficiency and comfort since JT65 and FT8 are synchronous modes (reference based on the clock) so it allows you to make DX while watching a movie or talking to the wife...

 

Olivia is so 2004.

 

That is the best explanation I have heard yet.

 

Ham radio is so 1914.

 

Perhaps even a proper "keyboard" like this