Tuesday, 24 January 2017

TS590 tcxo, WSJT-X multiple instances and dx!

I have been trying out the TS-590 and I seem to have caught the dx bug, working all round Europe and into the US on 40 metres. This short wave radio thing is quite habit-forming.

I have fitted the tcxo (temperature compensated crystal oscillator) into the Kenwood TS-590.
The SO3 tcxo for the TS-590, along with the supplied adjustment tool and a £1 coin for scale.
It takes 100 of those pound coins to buy one.

For such a tiny thing, £100 seems rather a lot, especially as the rip-off eBay tcxo-alike costs about £10. And you even have to solder the Kenwood one onto the board yourself, while the £10 one comes already soldered onto its own replacement board.

I think that leaving the tcxo off a rig like a TS-5900 is a bad move. The cost of designing in a separate small board, screwing it in place and providing jumpers, then selling the txco separately all round the world, with costs of transport and the cost of storage plus financing at the dealers ... I guess that just putting it on the board to start with probably might cost under £20, and then 100% of rigs would have it fitted as standard.

I suppose not everybody uses WSPR and data modes as I do. Ah dear. Hard to comprehend, I know.

Anyway, you take off ten screws and the bottom case, unscrew the mini-board and unplug it, get the tcxo the right way round, solder it in six places, screw the mini-board back in, plug the board back in, take off two jumpers, and use the ten screws to get the case back in place. In manufacture this would just be one component amongst dozens inserted at the same time.

OK, rant over.

It works perfectly as far as I can see. They supply a little plastic dibber to adjust it for correct frequency (no doubt that bit of plastic makes up a large part of the £100). So far I have not adjusted it. I did a comparison with a couple of WSPR operators who are famous for being scrupulously accurate. On 40m I came with in 2hz, on 10m within 3hz. OK, I could fire up my GPS frequency standard, but the setting recommended by Kenwood is by ear. I doubt if my ancient ears can do better than the WSPR result. Eh? What was that? My ears do what?

Which goes to prove a couple of things. First, 10m was open! Yes, a burst of Es yesterday allowed me to do the WSPR tests on 10m. Second, 6m was not open, so that will have to wait. It would be better to test the 5th harmonic of the GPS standard against the TS-590 on 6m. I bet it will hardly matter. I have two requirements:-

1) the rig should not drift on WSPR to any great degree, and
2) it should start each time pretty well on the right frequency.

More tests needed eventually but so far both of those boxes appear to have been ticked.

Things are not so rosy on the WSJT-X multiple instances front. Changing the computer for a faster machine has made things just a little better. I can run two instances of any other mode just fine, but two instances of MSK144 is too much and the display starts to be compressed. Whether that has any effect on the decoding I have no idea (but I doubt it).

The faster machine helps, but does not solve things completely. I can even find some shortening on just one instance of MSK144, variable, but usually only about half a second.
This is with one instance, different computer, different rig and a fresh installation of software.

I have done a lot of comparisons. I will leave them all out of this blog. My conclusions are that my initial idea was right. I believe this problem arises because only MSK144 has real time decoding, which demands a lot from my processors. Changing the computer over made very little difference, as I was adding memory and faster handling at every point, but the processor speed is much the same.

My initial work-around of using MSHV for one of the instances still works. I can just stick with that arrangement.

There cannot be too much wrong with it. I worked some DX on 6m MSK144 tonight - GI4DOH, almost 250km away. Meteor scatter too; it took 45 minutes to complete while we waited for the pings. My first meteor scatter contact in IO74.

Do you know, there is something about amateur radio? I hope to turn the rig on and try a bit of 40m action around midnight. What will conditions be like? There is still that moment of expectation while I wait to hear who has heard my CQ. There is still the puzzlement of trying to work out if I have worked this place (... DXCC, square, state, oblast, island, station ...) before.

You can cloak it up in all sorts of modes and CAT defined frippery (and I do), but just working some interesting stuff is reward enough. For me, it always has been.

But you knew that already.

See you all on the bands.




  1. Hi Jim,
    I am surprised that a single instance is causing problems, never seen that here with laptop. But I have heard from others that they have had problems with older computers that they had hoped to dedicate to radio some running older non supported operating systems but air gapped from the internet. It know tha is not the situation at all in your case.

  2. For running multiple instance, define the link for each like :
    C:\WSJT\wsjtx\bin\wsjtx.exe --rig-name=FT991
    C:\WSJT\wsjtx\bin\wsjtx.exe --rig-name=IC7100
    73's de OV3T

  3. Thank you Thomas. What you say is what I do.
    It still does not work for MSK144 because I do not have enough processor capacity. I can run two instances of any other mode fine. Even one MSK and one of any other mode, but two instnaces of MSK is beyond the ability of any computer I have.
    The Fast Graph shows shortened signals, usually 14 seconds or down to 13 seconds even with Fast decode and low FTol.
    Maybe I need new a computer but these are multi core gaming specification computers all fairly new. They include an Intel i5 four core and 4, 6, and 8 core AMD processors.
    So, I have two computers with separate instances of WSJT on each. I have one instance of WSJT on one of them (IC-7300), and three on the other runing the IC-7100, FT-817 and TS-590. I can run all three rigs with WSJT on that computer with any mode except MSK144.
    I cannot run two instances of WSJT if they are using MSK144. Sure, I can set them up, but they do not decode the full receive segment.
    Even a single instance of MSK144 with deep decode and high FTol will shorten the trace to 12 seconds.

  4. i have the same problem with two instances of mshv with msk144 both. Tx delay 5sec when tx on one of them. But no problems in rx.