1.8.9 beta, usb connectivity problem

[werribee_au]

Upgraded from 1.8.8 to 1.8.9 on the 21/01/10

Have had a couple if times (24th & 27th) where comms has been lost with the station, the diag file says that the usb has been disconnected but it has not. The first time it happened I thought, well one of those things, but the second time I thought, Ummm look into this further.

I never had this problem with version 1.8.8 release or 1.8.8 beta. Thinking of reverting back to the 1.8.8 release but before I do I'd like to have a debug listing when the fault occurs with 1.8.9.

The times that this has happened the debug has not been active.

I have a scheduled reboot of the station notebook at 2am each morning. (I like to keep things fresh)

So, when I check the "Config|Debug Log" it logs ok until a restart and then it becomes unchecked.
Is there an ini setting I can turn on so the debug is enabled after a restart?

If so, maybe I can then make a more detailed post of the problem.

[steve]

Is there an ini setting I can turn on so the debug is enabled after a restart?

Set Logging=1 in the [Station] section. But most of the interesting stuff gets logged to the diags file anyway. Could you attach the file where it says the USB has been disconnected? And can you describe what actually happened - the display stopped updating? You say the USB wasn't disconnected, but the message doesn't necessarily mean that the plug has physically been uplugged, it means that Windows has lost contact with the device; there's not much that Cumulus can do in that case. The USB code in Cumulus didn't change in 1.8.9.

[werribee_au]

Attached is the diags sub directory before I did any editing.

I can't describe the problem as I was unaware until this morning (5am) that there was and issue that occured at 1739hrs on the 27th. the web site was still being updated and the log jan10log.txt file was still being appended to. So it was clearly a connection issue between the software/USB/station console, somewhere

I have added "Logging=1" to the [Station] section of the ini file hopefully it falls over again to see what it reports.

The only other option is my notebook is falling over and that I something I don't want. I'd then have to build up a box and buy a ups.

[steve]

What was it that alerted you to the problem? Your graphs had flat-lined or something?

You could try the 'restart if unplugged' option; then, if it is just some kind of USB software/driver glitch, it should recover OK on the restart.

[werribee_au]

What was it that alerted you to the problem? Your graphs had flat-lined or something?

Flat-lined, Yes

You could try the 'restart if unplugged' option; then, if it is just some kind of USB software/driver glitch, it should recover OK on the restart.

Ok, I'll give that a go and see what happens.

I've checked that box and will see what happens. It does seem strange that is has only reared it's head after the release to beta upgrade though. Been solid as a rock for months with 1.8.8

Thanks for the help Steve.

Keep ya all posted

[werribee_au]

I got an error this morning "Sensor lost contact" @ 07:20:04hrs

1928.8015 : 07:24:04 USB device is plugged in
1928.8031 : 07:24:04 request EW data block, written = 9, addr = 00B2E0
1928.8062 : 07:24:04 EW data line 05 01 2F FF 07 E0 37 E3 0F
1928.8062 : 07:24:04 EW data line 06 FF 7F F0 FF ED CF FB C7
1928.8062 : 07:24:04 Data: 01 2F FF 07 E0 37 E3 0F FF 7F F0 FF ED CF FB C7
1928.8062 : Lost sensor contact!!!
1928.8265 : Ignoring bad data: intemp = 204.699996948242
1928.8265 : 07:24:04 EW data line 07 F0 07 ED 3F E3 7F F1 17
1928.8265 : 07:24:04 EW data line 08 E2 C7 E0 07 F0 87 EE E7
1929.7984 : 07:24:14 USB device is plugged in

There is nothing in the diags file only the above excerpt from the debug log file.
More around this entry in the attachment.

Is that intemp value in deg C?

Does this mean that there was an error in the data packet? The usb connectivity appeared fine.
Maybe I should change the batteries in the transmitter they are over 6 months old.

I'll think wait a little longer to see if it happens more regularly before changing the batteries.

Everything else seems to be running ok.

[steve]

I got an error this morning "Sensor lost contact" @ 07:20:04hrs

Seems to happen quite often to some people. Mine loses contact all the time, and I don't need Cumulus to tell me - I'm frequently getting dashes on the display.

Is that intemp value in deg C?

Does this mean that there was an error in the data packet?

Yes. Of course, the 'lost contact' flags being set may also have been an error!

[werribee_au]

Hello Steve,

I had another flat-line which started at '6822.2672 : 20:59:28' in the debug.log file.
Looks like Cumulus it can see that the usb is plugged in, just no data from the console. The scheduled reboot at 2am resolved the problem. Changed the today.ini timestamp and edited out the flat data in the feb10log.txt file and everything downloaded fine from the data logger.

Can you pick anything up in the logs?

Is it worth going back to 1.8.8 to see if it happens again?

Thanks

[steve]

The only thing of note is this:

6821.2672 : 20:59:18 request EW data block, written = 0, addr = 000000

The 'written = 0' says that Windows failed to send the command to read the data (so there could never be a reply). This does at least mean that the problem is detectable - it's not just that the station isn't replying, it's that Windows can't talk to it. But detecting it is one thing, doing something about it is another. I could make Cumulus restart (as in the unplugged case) but I doubt whether it would help. Possibly worth trying...

If you want to try 1.8.8, that's fine by me, but none of the Fine Offset code has changed for a long time.

This could well be the same problem as in this thread, using 1.8.8: https://cumulus.hosiene.co.uk/viewtopic.php?f=4&t=2349

[Super-T]

As a matter of interest, are these usb comms problems only on XP systems?

[steve]

As a matter of interest, are these usb comms problems only on XP systems?

One appears to be on XP SP2 and the other on XP SP3.

[Super-T]

I still think that there is a problem with the Fine Offset in the way that it handles the USB connection but I have no way to prove that and Fine Offset do admit to the DirectX problem in XP.

[Charlie]

I still think that there is a problem with the Fine Offset in the way that it handles the USB connection but I have no way to prove that and Fine Offset do admit to the DirectX problem in XP.

Can you provide a reference for further reading about the DirecetX problem in XP? I'm still chasing issues that only began after adding a Fine Offset to my hobby list.

[Gina]

I wonder why a few people have problems with XP when most don't. I think it's more likely to be other software installed in XP rather than XP itself - or maybe some subtle difference with the hardware. I've run Cumulus on several machines but never seen this problem. Even with many years experience with both hardware and software, I still have to put some things down to MAGIC!

[Super-T]

There's a thread here somewhere for my directx problems. Basically I couldn't start games that were using Directx...something to do with DirectInput. Had to unplug the desktop unit while I started a game and then when the game was running, I could plug the unit back in.
https://cumulus.hosiene.co.uk/viewtopic.p ... it=directx
Fine offset acknowledge the problem and blame DirectX9 as the problem is not there in DirectX10.
Go figure.