QBI buttons QRI Lights and a strange problem

[smoothquark]

I would be grateful for some help - this has been bugging me for the last few days! House is still being refurbished and re-wired. All the electrics terminate in the garage (hub), with multiple QRI DIN modules controlling the lights throughout the house. Couple of weeks ago, I ran cortex to commission the QRI modules and left a single QBI module in the "hub" to control the lights as a temporary measure, has the decorating has not been done yet. - eg. Button 1 -> toggle Hallway lights and toggle Study lights. Set-up reflex and turned off the laptop. The reflex programmed buttons work fine (and still do), except that the LEDs on the QBI don't work - not a major problem.
Now, I have set-up my rackmount mini-itx system and thought I would get Cortex running - this is where I am now having problems - the QBI buttons and LEDs work fine and as expected when controlled via Cortex behaviour. However, when the buttons are physically pressed the results are completely random - sometime only one light comes on and at times all do - the light icons on Cortex are always right but the QRI relays seem to do something completely different. I just can't seem to figure it out. If Cortex is turned off, reflex works normally and I can get the whole thing to behave by going back to Cortex behaviour. Just physical button presses are completely haywire!
Don't know if I am doing something stupid - all the light relays are set to manual. Tried clearing reflex to see if there was a conflict - no help...

[Gumby]

Hi,

I can't say I have ever seen anything like this. It sounds like there are reflex packets being sent from the QBI buttons directly to the QRI when Cortex is running, with random behaviour due to race conditions between the Cortex response and the direct packet. There is an "advanced" control somewhere (can't remember where) which enables Cortex and Reflex to run in parallel - you should check that this is NOT enabled.

But since you say you have tried clearing the reflex program then this shouldn't be the problem, unless the clearing did not complete for some reason.

[qbatqbat]

I had a similar problem with a QBI at the beginning of the year after using the auto reflex programming option. Whenever I pressed a button with cortex running to toggle a light it would toggle twice and so leave the light back in it's previous state (or odd variations). Monitoring the network and comparing what happened when non QBI buttons were pressed for the same action it did look like both reflex and cortex were trying to handle the same event for the QBI.

I had trouble getting rid of this behaviour using the gui but using the node profile utility I saw there was still reflex packets set that would not clear but managed to type the correct thing in the PCIF command code box to clear it (sorry, I don't remember what!)

At the time I put it down to the module firmware being new and that cortex would probably be shortly upgraded to sort it so I left it like that (the QBI not having reflex to fall back on, but all the other modules having it without any problems).

For info, version for the QBI in question is:
Master version: 0F
Year: 09
Month: 01
Day: 20
Flave fw ver: 00
Variant: 1
Hardware ID: 020

Nick.

[smoothquark]

Thanks very much. I have been looking at the profiler - I think you are right - there is some sort of reflex still active. I am clearing the memory at present to see if I can get it to work....

[smoothquark]

Managed to clear the memory and started afresh. First programmed Button 1 - worked as expected. Then created an auto reflex - also SEEMED to work as expected. Then allocated Button 2 - 4 and NO further auto reflex. Now Buttons 2 - 4 work fine with Cortex running and do not work when Cortex is not running (correct behaviour). Button 1 on the other hand has gone haywire as before. So, by a process of deduction, it has to be the auto reflex that has caused the problem. As Gumby suggested, both Cortex and reflex are active - need to find a way to disable this. Perhaps this was the command that qbatqbat used via the PCIF command box?

[Gumby]

I don't know what qbatqbat used, but you could try clearing the reflex for the QBI and then manually creating the required reflex vectors - for toggling a QRI it's not that hard, see help file and instructions for starter kit for examples.

[Karam]

Hmm, I'll check this out in case there is some sort of bug, but meantime perhaps one thing you might do is send me a node profile analysis dump for that QBI, taken whilst Cortex is running the network.

[smoothquark]

Thanks Karam. Do you ever sleep? Shall email the profile to you shortly. I wonder if this problem may be due to the fact that I do not have a watchdog timer set? That might explain the observation in post #5 above, where the QBI button 1 seemed to work OK for a little while before going haywire. I guess what I need is someway of ensuring that reflex is disabled whilst Cortex is running.

[Karam]

This does look like a bug which we will fix. If anyone has similar issues in the meantime please contact me for a workaround.

[smoothquark]

Karam, you are a star! Problem solved.