Missing events

[MindBender]

It's well over a year since I installed my Xanura modules. It took me some time to gather the courage to actually start writing software for the system, due to initial disappointments in the quality of the Xanura products (http://www.domoticaforum.eu/topic.asp?TOPIC_ID=66).

But now I've got my low power solid state running I decided to give it a try. Existing Linux drivers for the CM11 are either broken or don't support extended X10 codes, so I had to write my own driver. I don't mind, because that way I'm sure I've got a good driver, but these things are very typical for Linux.

While I was testing my driver, I discovered a very nasty phenomenon. If I switch two dimmers (DAIX10) shortly after another, the switch event of the second dimmer doesn’t get noticed! And ‘shortly’ in this case is everything quicker than 3 seconds! This is totally unacceptable: There’s no way I’m going to adapt the way I operate my switch to the limit of the system, let alone explain this to my girl fiend: “No honey, don’t operate more that one switch every three seconds or the silly computer will get upset.” Due to my earlier disappointments I was about to blame it on Xanura, but fortunately I remembered some limitation in the CM11 which may be causing this problem:
<i>The CM11 interface has a 10 byte output buffer. An extended message is 4 bytes long, two per event are sent by Xanura modules, one byte holds the length of the buffer and another holds a bitmask marking function bytes. So pressing the button of one dimmer fills up the CM11 buffer completely untill it's read by the computer, making his interface unsuitable to use it with Xanura modules.</i>

Now I’m using a custom firmware in my CM11 (http://www.domoticaforum.eu/topic.asp?TOPIC_ID=294), which also may be the cause of this problem. Can any body with an original CM11 please try to reproduce this problem? And can anybody with a better interface, like the CTX35, please verify if this problem doesn’t occur with such an interface? I’ll probably buy a XTB-II (http://www.domoticaforum.eu/topic.asp?TOPIC_ID=69), but that one uses an even older protocol, that of the TW523 so I’ll probably have to write my own firmware for that one too.

[Niknik]

Hi,

I'm very interested in this, as I'm beginning my home automation project right now (even though I have been watching x10 closely for some years now).
I was about to order a CM11, maybe I'll wait some more time to see if I should get another interface!

For now I'm coding some c# to log my rfxcom power use events (just received it yesterday). Using mono, I expect it to run on Win and Linux as well.

As far as your problem goes, my thoughts are that I'll probably be using "scene" events, where pressing a button/event would cause the lamps to dim/brighten to the preset levels...
Of course, this is no "solution" to the problem you describe.

Using a TW523, you'd have to take care of all the "interpretation" of the bits coming in through the power line. If you're willing to consider that, wouldn't it be simpler to get the data from the CM11 faster, so it never has a chance to fill its internal buffer? Or, if it's possible to change its firmware, change it so it holds more buffer space?

(I'm not sure about the "low level" timing details, but as long as you could empty the buffer from the CM11 as soon as possible, maybe it would keep the buffer from overflowing and losing data)

[Digit]

This is what the manual of my ATC TI-213 says:

<b>FROM HOST</b>
The TI213-RS232 has a buffer that uses 19 locations to store data from the host to be transmitted onto the power line. Only one location will be taken if the data is the same and no gap indications are transmitted. For example, sending ADIMADIMADIMADIMADIM will only use 1 location. The TI213 will accept up to a 124 byte data packet and can accept 31 dim/bright commands at one time if necessary.
<b>
STORING FRAMES RECEIVED FROM THE POWER LINE</b>
The TI213-RS232 has a buffer that uses 64 locations to store frames received from the power line. If the same frame is received more than once without a gap between each frame, only one location will be used. This is how the TI213-RS232 keeps from filling its buffer too quickly when receiving dim/bright frames. Note that the TI213-RS232 will stop storing identical frames after receiving 32 identical frames with no gap between them.

I haven't got anything transmitting over the powerline installed yet, so can't test very easily.

Regards,
Robert.

[Snelvuur]

If i read my ctx35 out, it sometimes stores alot of information if i dont read it out with a tool. So i should be ok what i do have however is the sending of rf to powerline. I have one of those marmitek rf => x10 senders, when i use the remote and press button 1 2 and 3 in rapid succession i only get 2 lights on. I think when my own tooling is done (rfxcom with ctx35 on it) wont have those issues and its a problem of the marmitek rf->x10 device.

However, i always had issues with the cm11, always some funny situations. The ctx35 has not let me down for a long time.

[MindBender]

Polling is no longer of this age. It's slowing down incomming information and it's not nessecary: The RTS/CTS handshaking lines can be used to detect a listing 'ear'. And even though RS-232 isn't a bottleneck here, I'd like to avoid that as well.

Ideally I'd build my own interface, using a Ethernet enabled zilog eZ80 controller: They're cheap and I've still got a couple of them laying around. Ethernet is fast, connections can be detected and there are enough recources available to synchronize to NTP, time-stamp incomming traffic and buffer days worth of events. And there still would be room enough left for a small web-server. I'm thinking about using the XTB-II hardware as an electronic interface becuase it's loud enough to suit my needs ([ur]http://www.domoticaforum.eu/topic.asp?TOPIC_ID=69[/url]).

But building such an interface would side-track me from my domotica project. And to get X10 collision avoidance well implemented isn't as easy as it seems. Though this would be the ultimate line interface so if other people are interested, we could combine efforts here...

[Han]

A normal interface will be fast enough to empty the buffer in time. Though I have several transmiting devices on the power line I have never noticed a problem with overflow. I now use a CM11 compatible USB (DIY) unit which is much faster in emptying the buffer.
An extended message uses 5 bytes. This is an error in the CM11 documentation. (The CM11A is translating the 5 bytes correctly).

Han Alink

[Snelvuur]

Well i'am still building around the ctx35, so far it looks good. My house is far from build so once its done i can see if it gives any issues.

@mindbender: i still have my cm11 here, if you need it for testing purpose be my guest.