PMIX35 48 hour logging

Post by **l0rien** » Sun Jun 01, 2008 5:10 pm

I've been having problems with my install for some time now, but I could not put my finger on the actual problem.

With the PMIX35 I hoped to spot the problem. What I did: run a 48 hour test from my 'meterkast' and poll status requests every 60 seconds to either end of the network. In my case, two modules(E10 = ZAX12 and E06 = SAIX12). Recording noise, impedance, voltage levels and received returns from either modules. The measurement is from Friday 30th of May starting at 11:00, ending Sunday 1-06 around 13:00.

The resulting Excel file with the data and the graph below you can find here: A10 48 hours.xls

The graph with the data inserted below. A little explanation as to what you see:
* Voltages(in mV) are displayed on the primary axis(left)
* Impedance (times 100, ignore the legend x10) and Noise levels(in mV) are displayed on the secondary axis(right)

So the blue line "Voltage Out" shows the output voltage of the PMIX35 putting the status request on the line. It swings around the 5000(5 Volts). The Noise level is between 50 and 800 mV (right Axis). The Impedance starts at 3,5Ohm, moves up to 5,8Ohm etc. You see the returns of the modules as the triangles. Their position on the Y asis(left) indicates their return Voltages.

I'd expect with low noise I would get the best performance, which is at nighttimes. Looking at the data, that is not what I get. Now the million dollar question: What is my problem?

Why do I get such poor performance?

Any help would be greatly appreciated!

Post by **l0rien** » Tue Jun 03, 2008 3:34 pm

Anyone?

Post by **Snelvuur** » Tue Jun 03, 2008 5:42 pm

I cant really tell what the problem now is? You have poor performance, you just mean that the switches take a long time to process?

// Erik (binkey.nl)

Post by **MindBender** » Tue Jun 03, 2008 5:52 pm

<blockquote id="quote">quote:<hr height="1" noshade id="quote">Originally posted by l0rien
 Anyone?
<hr height="1" noshade id="quote"></blockquote id="quote">
I've looked at your data and I love your scientific aproach, but I don't have enough in-depth knowledge to help you here.
Though I have similar reliability problems and my PMIX35 is on it's way, so I'll let you know as soon as I can contribute to a solution.

Post by **Jfn** » Tue Jun 03, 2008 6:24 pm

What does the PMIX35 measure actually? Does it deal with capacitive/inductive ('imaginary') loads, or with real loads only?

The only thing I can think of is that somewhere on your wiring there is a piece of equipment that has a high capacitive or inductive load, which in turn introduces phase-shifts in the X10 signal or even attenuates it (Compare it to a waterhose full of holes, with the holes being loads that are attached to the wiring: Waterpressure at the end would be much higher if there were no holes):

What I would try should the problem remain, is put the PMIX35 and the lamp on the same outlet. Take a large extensioncord (at least 10 meters) and plug it into the same outlet as the PMIX35. Use the extensioncord to power the (micro-) module that is giving you troubles. My feeling tells me you should have much better signal readings.

Als het niet gerepareerd kan worden dan is het niet kapot!

Post by **l0rien** » Tue Jun 03, 2008 11:32 pm

<blockquote id="quote">quote:<hr height="1" noshade id="quote">Originally posted by snelvuur
 I cant really tell what the problem now is? You have poor performance, you just mean that the switches take a long time to process?

// Erik (binkey.nl)
<hr height="1" noshade id="quote"></blockquote id="quote">In the graph I have 48 hours of polling 2 devices for a status request every minute. Only the triangles in the graph indicate a response from a module. So I rarely get a response and I can't seem to figure out why exactly.<blockquote id="quote">quote:<hr height="1" noshade id="quote">Originally posted by MindBender
I've looked at your data and I love your scientific aproach, but I don't have enough in-depth knowledge to help you here.
Though I have similar reliability problems and my PMIX35 is on it's way, so I'll let you know as soon as I can contribute to a solution.
<hr height="1" noshade id="quote"></blockquote id="quote">I'm really interested on your findings with the PMIX35, please keep me posted!<blockquote id="quote">quote:<hr height="1" noshade id="quote">Originally posted by Jfn
What does the PMIX35 measure actually? Does it deal with capacitive/inductive ('imaginary') loads, or with real loads only?

The only thing I can think of is that somewhere on your wiring there is a piece of equipment that has a high capacitive or inductive load, which in turn introduces phase-shifts in the X10 signal or even attenuates it (Compare it to a waterhose full of holes, with the holes being loads that are attached to the wiring: Waterpressure at the end would be much higher if there were no holes):

What I would try should the problem remain, is put the PMIX35 and the lamp on the same outlet. Take a large extensioncord (at least 10 meters) and plug it into the same outlet as the PMIX35. Use the extensioncord to power the (micro-) module that is giving you troubles. My feeling tells me you should have much better signal readings.<hr height="1" noshade id="quote"></blockquote id="quote">Sorry I don't know what the PMIX35 measures exactly.
Love your analogy with the water pressure. The only thing is, let's say that I do have that kind of equipment somewhere in the house. At night time all of my equipment is turned off. The stuff that is not(router, switch, Tv on stby etc) is behind a filter (FM10). So I would assume that somewhere in the logging there would be obvious periods where there are no problems and the signals pass normally.

Putting the PMIX35 close to a module works fine, the only thing is, I want it to work from the central point (meterkast) not just next to a module

Post by **Snelvuur** » Wed Jun 04, 2008 3:14 pm

Maybe the switches have a build in dos defence, if its gets a status request to often, it wont reply at all? Just thinking out loud here.. try switching it to say every 5 min and see if its still so bad.

// Erik (binkey.nl)

Post by **l0rien** » Wed Jun 04, 2008 9:50 pm

In a previous test I polled every 30 seconds, so now I doubled it. Still, I will try a new test every 5 minutes and see what that does, thnx.

Post by **Digit** » Wed Jun 04, 2008 10:12 pm

I poll every 5 seconds. No problems. Well, i think even that is way to slow. But i don't have much X10/A10, only some outside lights and some less important inside. So i don't care much about that yet. If needed, i'll increase the polling rate.

But i would say you can forget about that influencing things.

Post by **Bwired** » Wed Jun 04, 2008 10:18 pm

I poll the Xanura CTX35 / PMIX35 every 400ms

Post by **l0rien** » Wed Jun 04, 2008 10:41 pm

So is this network 'performance' unusual? I can't imagine that everybody had a 100% functioning network to start with right?

Post by **Digit** » Wed Jun 04, 2008 10:46 pm

The moment my ACT TI213 moved to the fusebox, all problems were gone. I installed a filter for my TV, but not because i needed it, just as a precaution.

Post by **l0rien** » Thu Jun 05, 2008 5:59 pm

That's the problem, this measurement was done from the fusebox ('meterkast', srry)

The only way forward I see is to install an SFX40 and see what that does. The data does not support it will do anything significant, but it's worth a try and relatively cheap.