Very, Very basic guide to Q65

NOTE: Performance estimates are related to 1296 EME use. All comments based on WSJTX 2.4.0 -rc4 and are subject to change as program changes are made.

Summary

For those who don't like to read a lot:
  • Q65 supersedes QRA64 (QRA64 is no longer offered as an option)
  • Q65 has a visible sync trace for all but the weakest signals
  • Q65 signals can be decoded when no sync trace is visible
  • Q65 copes very well with the wide Doppler spreading usually found on the higher Microwave bands (10GHz and up)
  • Q65 always uses AP information when it is available. AP cannot be turned off
  • Q65 does not use Deep Search. It needs no data base of stations and grides
  • Q65 has autosequencing available
  • Q65 used single tones (not dual tones) for shorthand messages, and these do not automatically decode. They are typically NOT used except under special circumstances.
  • Q65 (60s period) is now also supported by MAP65
  • Q65 outperforms JT65 under all conditions for 1296 EME
  • Q65 has 15, 30, 60, 120 and 300 second periods (but 15s is not available for EME)
  • Q65 has averaging. Make sure averaging is cleared when required. Set it to clear after decode.
  • Q65 (like other modes) benefits from stable, on frequency, signals
  • Q65 may benefit from 1Hz frequency steps, though 10Hz steps are normally fine
  • Q65 for 1296 EME typically uses a Tx Audio setting (sync tone) at 1500Hz
  • Q65 is normal used with COFM Doppler correction.
  • My suggestion for 1296 EME is to use Q65-30B unless one or both stations is/are QRP.

Q65 is a new digital mode which replaces QRA64 in WSJTX 2.4.0. If you want all the details and operating instruction the best reference you will find are the documents authored by Joe Taylor (K1JT) and other members of the development team. This guide is just intended to be an overview based on a couple of months of testing the use of Q65 on 1296 EME.

At the moment I believe the only official documentation on Q65 can be found here - Q65 Quick Start Guide. Q65 is also briefly covered in the WSJTX 2.6.1 User Guide.

Low tech overview

Q65 is a digital mode which has some features of QRA64 and some features of JT65. It has a sync tone like JT65, but only puts about 25% of the Tx energy in in the sync tone rather then the 50% used by JT65. This means more of the energy can go into the message tones, but the sync tone will be slightly less visible on the waterfall. However the sync tone can be usually be seen on all but the weakest signals. QRA64 used a different sync scheme with no visible sync tone.

Q65 makes extensive use of AP (A Priori) information. AP information is information you already know or which becomes known as the QSO progresses. When you start out, you always know your own callsign. Quite often you know both the callsign and the grid of the DX station you are trying to work, or those become known during the first part of the QSO exchange. Q65 uses this information to help it decode signals. This is exactly the same as what you (or at least many of us) do when working CW. I may copy muyown call in one transmission. For a random string it might take 4 or 5 transmissions (= averaging) It's a lot easier to copy my own callsign that a random 5 character alphanumeric string. I can probably copy my own callsign at a 3dB lower signal level than I can copy a random string (that's AP). The same goes for a Dx callsign. If I know what I'm listening for I can copy it at lower signal strength than I could if I had no idea what it was. The same again goes for reports. I can copy an "OOO" report more easily then an RST report, and even with RST reports I know that I'm typically looking for something like 559 or 459 or 55N. I'm not going to be looking for 372 or 299, so I have some idea whether or not I'm copying something that makes sense. If someone is sending 372, it may take a few repeats before I'm sure of that.

AP does the same sort of thing, but in a much better way and with much higher confidence. AP is designed in a way that the chance of it getting something wrong is vanishingly small. If you want the details on that, I'm sure there are reference papers and I'll list them at the end of this page.

The end of line decodes codes in Q65 are different from those in other modes and at first appear somewhat crypic, not always showing the value you might expect. Very generaly speaking, here is what they usually mean:

  • Q0 - This means a decode has been obtained without the use of any AP information. Usually this would be a plain text message
  • Q1 - This usually means that AP has decoded the message as a CQ call, with the DX Call and Grid unknown to AP
  • Q2 - This usually means that AP has used your own call as AP knowledge, but has assumed nothing about the DX Call or Grid
  • Q3 - This usually means that AP has used AP knowledge of your Call and the DX Call in the decode. It then looks for a grid, a report, 73 etc.
  • Q32 or similar means that a Q3 decode has been obtained after averaging 2 transmissions.

Most of the time you will see Q3 decodes during a QSO. Sometimes you may see a decode code that doesn't instantly make intuitive sense, but don't worry about it. The way AP works can sometimes generate an decode code that can be only be fully explained by those familiar with the internal details!

False decodes are very, very, very unlikely to occur. There are multiple checks and error correcting codes involved on the decoding which brings the chance of a false decode down to very close to zero. In several months of testing involving thousands of off-air and simulated signals, I only saw one false decode, and that was in an impossibly weak signal (requiring around 30 averages of a 300E transmission to get a decode). Even then the error was very easily identified and would not have been mistaken for a real decode.

Q65 differs from JT65 and QRA64 in that whenever it can use AP it does. You can't turn it off. It can't use AP on a plain text message because there is no AP information. So if you send "Tnx Joe 73", it can't (and doesn't) use AP. It looks at AP first to see if it can be used, but when it fails, it does to an unassisted, direct decoding of the message. AP is used because it enables weaker signals to be decoded.

Q65 has averaging available and an option to clear the average once a decode has been obtained. It's important to be sure than when averaging is used, it is first cleared before any new signal averaging is attempted. Averaging averages every undecoded signal since the last decode (if clear average on decode is set) or last manual clearing of the average. That means if you sit listening to noise with averaging on, then you find a signal and start listening to it without first clearing the average, the average accumulator will be full of noise and you will never decode the signal unless it's strong enough to decode in a single period without averaging.

Frequency accuracy, stability and step size

Since you can decode Q65 signals at a level when the sync tone is no longer visible, for really weak signal work you need to accurately know your frequency. You don't need to know it with 1Hz accuracy, but you probably do need to be within about 25Hz of where you should be. You might get away with being 100Hz off frequency if the Dx station is using a wide Ftol setting, though for best results Ftol should probably be kept to 100Hz or lower.

Q65 has no specific built in function to track a drifting signal (it does not have any form of AFC built in. which I think JT65 does). The internal algorithm can cope with a small amount of drift, but the more a signal drifts, the less sensitive the decoding may be.

Frequency step size can affect decode sensitivity is a similar way to frequency drift. Smaller is better. Much depends on the signal spreading, the submode in use and the rate of change of Doppler shift. For 1296 operation my testing suggests that 10Hz steps can be anything from as good as 1Hz steps in the best case, and as much as 1dB worse in the worst case. Not a huge difference for most contacts, but it could be critical on very weak signals. Note that the inability to change frequency during Tx operation (which is the case for quite a few rigs) also makes a difference in some modes. CFOM, for example strictly requires that both rigs change frequency during the Tx period.

If one station has 1Hz steps and can change frequency in Tx mode and the other station lacks either of those abilities, the optimum solution is that the first station uses "Full Doppler to DX Grid" Doppler control, while the other station has Tx and Rx on the sked frequency.

Which Mode? Which submode? Which Audio Tx Frequency?

Q65 allows you to chose mode length (30s, 60s, 120s, 300s) and submode (A, B, C, D, E etc.). For 1296 EME operation if you just switch from JT65C (with AP, DS and Averaging) to Q65-60C (with averaging) you won't see much difference. Q65-60C will do slightly better on the very weakest signals, but 95% of the time most stations aren't looking at the very weakest signals, so the choice of mode won't matter much.

What would make a difference is the use of Q65-30B. You will lose maybe 2dB in sensitivity, but (unless you are running QRP), would still probably be able to easily work 90% of the stations you could have using JT65C (+Ap +DS +averaging), but a QSO might take 2.5 minutes rather the 5 minutes. Very useful in a contest or a DXpedition pileup. Q65-120C/D will let you work stations who are several dB weaker than you could have worked with JT65 while still having a reasonable QSO time of 10 minutes or so. And when all else fails, there's Q65-300 which will allow you to achieve contacts with even weaker stations, provided your PA is happy with key down operation for 5 minutes and the prospect of a QSO lasting 30 minutes isn't a problem. DL3WDG has heard my 1296 EME signal (240W and a 3m dish) on a 14 turn helical antenna using Q65-300E (and no averaging was needed on some transmissions). I've also worked him on 1296EME with a full 2-way QSO when he was running 100W to a 77cm (30") dish, using Q65-60B. So Q65 operation requires some operator skill to decide on which mode may be the most efficient for difficult QSOs. I don't think there's any good reason not to use Q65-30B as the "default" mode for most contacts on 1296 EME. A default Tx frequency of 1500Hz also seems like a reasonable idea. When signals are weak and you don't see a visible trace on the waterfall, your RF frequencies not only have to the the same, your Tx Audio frequency also has to be the same. 1500Hz seems like a good choice.

Shorthand Messages

Note: A TX audio setting of 1000Hz rather than the more commonly used 1500Hz will make the use of Shorthand messages less confusing. See Q65 SH modes

Q65 does have a provision for shorthand messages, but they are more like the ones used in QRA64 then the ones used in JT65. Given the ability of Q65 to decode even very weak signals, there will probably be less incentive to use the shorthand messages than there was with JT65. There are 4 tones (assuming you have 1000Hz as your Tx and Rx audio frequency settings.

1000Hz - Tune
1250Hz - OK to send
1500Hz - RRR
1750Hz - 73

The 1500Hz tone (RRR0 and 1750Hz tone (73) operate just like the two-tone shorthand signals in JT65, except it's up to you to read them off the waterfall display and interpret them. Whether they will be visible when Q65 can no longer obtain a decode is a matter to be determined. It may depend on how well optimized your waterfall is and how good you are seeing single tone signals with Doppler spreading. The 1000Hz and 1250Hz tones are for "setting up" a contact if necessary. The first station sends a 1000Hz tone. When the second station has found that tone and coordinated frequency, they send a 1250Hz tone. After that the QSO begins. This was used with QRA64 because QRA64 has no sync tone trace and so could be a little difficult to see at low signal levels. Since Q65 does have a sync tone, the use of the 1000Hz and 1250Hz tones is unlikely to be needed very often.

Summary

Q65 from an operator point of view, is much like JT65 (the dominant mode on 1296 EME right now). However it offers more flexibility with the choice of 30, 60, 120 or 300s periods and Q65-60C is more sensitive then JT65C (even when deep search, which is not available with Q65 is used with JT65C. Unless you are dependent on MAP65 for your EME operation (as some stations may be on 144MHz - and even 432Mhz) there is really no good technical reason not to move to Q65.

  • Q65 supersedes QRA64 (QRA64 is no longer offered as an option)
  • Q65 has a visible sync trace for all but the weakest signals
  • Q65 signals can be decoded when no sync trace is visible
  • Q65 copes very well with the wide Doppler spreading usually found on the higher Microwave bands (10GHz and up)
  • Q65 always uses AP information when it is available. AP cannot be turned off
  • Q65 does not use Deep Search. It needs no data base of stations and grides
  • Q65 has autosequencing available
  • Q65 used single tones (not dual tones) for shorthand messages, and these do not automatically decode. They are typically NOT used except under special circumstances.
  • Q65 (60s period) is now also supported by MAP65
  • Q65 outperforms JT65 under all conditions for 1296 EME
  • Q65 has 15, 30, 60, 120 and 300 second periods (but 15s is not available for EME)
  • Q65 has averaging. Make sure averaging is cleared when required. Set it to clear after decode.
  • Q65 (like other modes) benefits from stable, on frequency, signals
  • Q65 may benefit from 1Hz frequency steps, though 10Hz steps are normally fine
  • Q65 for 1296 EME typically uses a Tx Audio setting (sync tone) at 1500Hz
  • Q65 is normal used with COFM Doppler correction.
  • My suggestion for 1296 EME is to use Q65-30B unless one or both stations is/are QRP.