Dead or not dead? what do you think?

[okidoki 13]

Hi All,

I need some advice and maybe some coding help.

in my HC3 environment I have about 10 Ecobright dimmers, in Fibaro they are multilevel switches. They act the same as the Wally from fibaro more or less. 

I have serious problems with these devices now and then: the HC3 reports that they are nog longer available. See the image.  A red cross appears and from the HC3 the device is nog longer accessable. The device itself functions if you use the button on the device to turn it on or dim the light.  This happens randomly, sometimes for 3 months everything is ok, and then suddenly 1 drops out. I include it again (power cycle, include etc) and then after a week months or weeks anoher device has the same issues. I have also other devices, fibaro dimmers, qubino dimmers, etc, and they are stable. , I reported earlier about this behaviour in this topic.

I'm in close contact with the dutch developers but up to now they don't know where to look. A firmware upgrade of the device did not help.

 

 

Please login or register to see this attachment.

 

Last week it happend again, it was no longer available.  I did some digging into the device. I read out the device info of the device (http://x.x.x.x/api/devices/295. The result is below:

You see the bold section:

 

{
  "id": 295,
  "name": "LichtHal",
  "roomID": 222,
  "view": [
    {
      "assetsPath": "/dynamic-plugins/com.fibaro.multilevelSwitch/assets",
      "jsPath": "/dynamic-plugins/com.fibaro.multilevelSwitch",
      "name": "com.fibaro.multilevelSwitch",
      "translatesPath": "/dynamic-plugins/com.fibaro.multilevelSwitch/i18n",
      "type": "ts"
    }
  ],
  "type": "com.fibaro.multilevelSwitch",
  "baseType": "com.fibaro.binarySwitch",
  "enabled": true,
  "visible": true,
  "isPlugin": false,
  "parentId": 294,
  "viewXml": false,
  "configXml": false,
  "interfaces": [
    "energy",
    "fibaroFirmwareUpdate",
    "levelChange",
    "light",
    "power",
    "zwave"
  ],
  "properties": {
    "parameters": [],
    "pollingTimeSec": 0,
    "zwaveCompany": "",
    "zwaveInfo": "3,6,4",
    "zwaveVersion": "1.12",
    "categories": [
      "lights"
    ],
    "configured": true,
    "dead": true,
    "deadReason": "power",
    "deviceControlType": 23,
    "deviceIcon": 15,
    "emailNotificationID": 0,
    "emailNotificationType": 0,
    "endPointId": 0,
    "energy": 0,
    "firmwareUpdate": {
      "info": "",
      "progress": 0,
      "status": "UpToDate",
      "updateVersion": "1.12"
    },
    "isLight": true,
    "log": "",
    "logTemp": "",
    "manufacturer": "",
    "markAsDead": true,
    "model": "",
    "nodeId": 37,
    "parametersTemplate": "0",
    "power": 0,
    "productInfo": "4,56,2,2,0,2,1,12",
    "pushNotificationID": 0,
    "pushNotificationType": 0,
    "saveLogs": true,
    "serialNumber": "h'6049569db3ea21a755da8f391fcb5698",
    "showEnergy": true,
    "smsNotificationID": 0,
    "smsNotificationType": 0,
    "state": false,
    "updateVersion": "",
    "useTemplate": true,
    "userDescription": "",
    "value": 0
  },
  "actions": {
    "abortUpdate": 1,
    "reconfigure": 0,
    "reset": 0,
    "retryUpdate": 1,
    "setValue": 1,
    "startLevelDecrease": 0,
    "startLevelIncrease": 0,
    "startUpdate": 1,
    "stopLevelChange": 0,
    "toggle": 0,
    "turnOff": 0,
    "turnOn": 0,
    "updateFirmware": 1
  },
  "created": 1621675895,
  "modified": 1621675895,
  "sortOrder": 162
}

 

 

I also got the list of a similar Ecobright device that still works and i did a diff on both files: Here is the diff view:

 

 

Please login or register to see this attachment.

 

 

What you see is that apart form the serial number et so, the only difference is the node id (i expect that) and the two fields "dead" and "deadreason"

In the device that works the value of 'death' is false and in the device that is dead the label is true.

 

So I figured, since i can still get the info from the device, that my device still exist in my zwave network. Could it be that by simply setting these back to 'false' the device becomes available again in the HC3?

What do you think? is this possible? Can the value of these parameters be changed? 

 

My next step is to try to write a piece of lua code to try this out. But i could use a litte help by that as i do find it hard to acces these complicated lists.

 

Do you think this could work?

 

thanks,

jacco

 

 

 

 

 

 

 

 

 

[tinman 984]

you need zniffer for that, and you need to check:

- if the ecodimm is sending something to your HC

- if ack comes back

- if ecodimm is sending anything 

 

 

[petergebruers 1,231]

10 hours ago, okidoki said:

So I figured, since i can still get the info from the device, that my device still exist in my zwave network.

Not necessarily, you are looking at cached data, but you surely are on the right track.

 

As @tinman said, it might be a little hard to know the cause with 100% certainty without a Zniffer

 

First, you have to know what "dead" means when we're talking about a mains powered device.

• The device gets this status when Fibaro HC sends data to the device, but it does not respond and X retries fail as well. When a device has this status, Fibaro HC does not try again (unless you change some settings or un-dead it first). This is because sending data to a really dead devices causes very long (up to 10 - 20 second) delays on HC, so it is worth avoiding that
• If the device sends data to the HC (eg status, temperature power), the HC will un-dead it
• If you have *polling* enabled, every X minutes the HC checks the device so it can go dead even though you have no active control or scene running

 

I would say:

• If a device has no power, it goes "dead", eg you unplug a Wall Plug. Or a Fibaro Dimmer 2 in 2-wire setup without bypass and low LED load might "drop out"
• If the device has a "firmware bug" that causes it to crash, it will go dead but power cycling will bring it back. It is rare, but I've seen reports.
• If the device has a "firmware bug" that causes it to forget its configuration, then it will have forgotten the network it belongs to, a power cycle does not help. Also rare. You have to exclude and include device. FGK-101 is battery powered but is easy to reset accidentally, that's an example that comes to mind. Or Neo Coolcam panel switches.
• If the device supports "remote reset" (which modern devices do!) then a bug, or a "rogue button press" can cause it to exclude itself. In that case, the device will no longer be on the HC, so technically it is not dead but gone

With Zniffer, you could capture the network data over a longer period of time, and see if it sends those "reset" commands or forgets its config.

 

If you want to do some "Monte Carlo" diagnosis... Try to exclude and include one of them in secure mode, and see if it changes anything. Secure mode makes the devices behave differently and immune to "rogue" packets on your network. Not recommended on all your devices because it increases latency (because HC3 only does S0 security, S2 will change that).

[okidoki 13]

Thank you @tinman & @petergebruers for your thoughts on this and the explantion. I have a feeling the firmware of the device crashes. However, what i think is srange is that the frequency of going dead varies over the dimmers. Some devices have nevenr gone dead, others already 5-8 times. 

 

I could use Zniffer to diagnose, however, i'm not in the position to modify anything about the device. Its up to the supplier to fix it. Because my HC3 and the other devices are very stable, i'm quite sure its the device instability that causes the problems. 

 

 

Today i received another firmware version, so i can upgrade an hope for the best.

 

 

[petergebruers 1,231]

5 minutes ago, okidoki said:

However, what i think is srange is that the frequency of going dead varies over the dimmers. Some devices have nevenr gone dead, others already 5-8 times. 

I wonder if it has anything to do with their power supply (in a broad sense, I mean both mains, the electronics or even the type of load). What theoretically could happen is that the voltage to the MCU sags, and if it then writes garbage to the NVRAM, then when it recovers the devices considers the memory invalid and acts as if it was reset (so no NodeID or HomeID and parameters set to default). Maybe have a look at different parameters that could explain different frequency of crashes. On a HCx it would look as if the device "just disappeared". Does this match your observations? I mean it still exists on your HC but part from that acts as if it was brand new and unused?

 

It is difficult to debug such a device... It can output "debug" statements to a serial port if they enable a "debug build" of the firmware, but I don't know how an end-user could safely monitor that, the port is mains referenced...

[petergebruers 1,231]

By the way... There is another device that is known to "disappear" on very, very rare occasions and that is the ZW5 version of Fibaro's Wall Plug. It has been mentioned by a handful of users only, but it is equally... enigmatic.

 

 

Edited May 25, 2021 by petergebruers

[okidoki 13]

Interesting. I see similar behaviour!

btw, i also have issues with my 2 wall plugs, i also have the fw02 version....