Node's becoming unreachable

  • Fellow board members!

    I've got this "huge" mansion πŸ˜› which is problematic!
    Let me tell you what I have:
    1 RPI 3 Domoticz with Mysensors Gateway 1.4 directly attached to RPI Headers. Radio is the 1000 meter range version with alu foil connected to the ground of the antenna. This radio antenna is extended with a extention wifi cable to place the antenna higher.

    In every room there is a node made from a Arduino Nano and a good USB power supply (Apple ones). Most of them have normal radio with cap installed.
    Every node in my house has a static ID.
    There is 1 node on the attic which has a 1000m radio which is not insulated with alu foil (to do) but is reaching downstairs with the right angle of the antenna. This node is a repeater and a parent for an other node to control my airco in the same attic. Otherwise the radio of the airco will NOT reach downstairs.

    I've got a shed also with an 1000m radio which is insulated and also extended with a wifi extension cable to reach the gateway. This board is a Arduino Mega with a 4 relay board, parking sensors and door sensors. This node is a repeater and also a parent for the node in the kids playhouse outside.

    The node in the kids outdoor playhouse is a Nano with normal radio with cap.

    There are several other nodes but most of them are repeaters.

    My problem:
    Sometimes my gateway cannot send a successful switch command to my shed. We are talking 15 meters away from each other with insulated radio's and extended antenna's in a line of sight.

    Sometimes the connection / node's become numb like the one for my airco. Then I cannot send a normal command to my Airco node relaying through my attic radio. Then I have to reset the attic node and then the airco node to get it live.

    There are several questions that I'm struggling with...
    1 Is it a problem that almost every node is a repeater?
    2 Is it a known problem that connections "disappear" from the network and after a reset it's running well?
    3 What can I do the make the system more reliable?
    4 Should I roll back to the basics and create a USB Gateway on my Pi and place the radio elsewhere? Or go EthernetGateway to be sure I reach my shed?

    I'm willing to invest quite some time to get this system up and running but I'm quite irritated about the connection drops and resets I have to give.
    I think the system is well designed but we are facing radio problems at least and maybe some scripting errors from my side regarding the use of a repeater node or dedicated parent routing.

    If someone is willing to help I would be very happy!

    I will post a sketch how nodes are placed and configured later today.

    FYI I've scanned with the poor mans wifi scanner and changed all the channels to 111.

  • 0_1474880697218_Scheme MyS.png

  • I'm too new to all of this to be able to help, unfortunately, but your diagram got me really curious - what does the BBQ node do??

  • @mrwomble Measure the temp of the meat on the BBQ and the kettle itself. I get a push message from Domoticz when the centre of the meat hits a certain temperature regarding the switch I flipped for cow, chicken, fish and so on...

  • Admin


    What did you use for the meat probe? (Been thinking about the same here...)

  • @tbowmo The IKEA Fantast probe with MAX6675 breakout board.

  • @Sander-Stolk Now that is taking home automation to a whole new level. Nice! Adding that to the list of sensors I want to build. It's a long list and growing all the time... πŸ˜„

  • Admin

    Maybe your could start logging in the shed node for a while to see what is going on.

  • @Sander-Stolk I have exactly the same problem with my bird house node. It is about 20 m away from the repeater. Both are 1000 m radios and aluminium shielded. The repeater uses LOW setting. The birdhouse node used MIN for 3 weeks and worked fine, then the dropouts appeared, sometimes for a whole day there is no message coming through. The repeater is working fine, I tested it with @AWI's quality meter sketch. The birdhouse node is 10 times checked, it is ome of the "My Slim AA ..." nodes, very simple setup. I recently changed the radio setting to LOW, but same results, it works for a few hours then dropouts start.

    Unfortunately I cannot advise you, just tell you that I have similar problems. I have no idea what to do, I discussed in detail with @Oitzu but I am still having dropouts.

    I will need to do further testing if the repeater can reach the node, what is sure is that the node cannot reach the repeater.

  • Admin

    one thought, is there any changes in the humidity in the air, when the signals drops out?

    Could be that the radios are just on the edge of what they can do reception wise, and then if the weather turns bad, and the humidity steps up, it causes some of the troubles?

  • @tbowmo I will observe. Today it is dry and there are still dropouts. But then for 20 m? with a +PA+LNA? I would expect this thing to go 100 m... Maybe a bad soldering point that is affected by humidity?

  • Admin


    Well it was just a thing that just popped up in my head. πŸ™‚ Even though they are marketed as 1000 meter, it all depends on multiple things, like any walls between the nodes, type of wall material, is it wet? is it dry? etc.

  • @tbowmo No, it does not seem to be the humidity. Now in the evening that the temperature drops and the humidity rises I have quite regular readings.

    Anyway, it's strange. Nobody is at home -- Conditions did not change. It did not work last night and during today (readings once very two hours more or less), but since the late afternoon I get readings every 10 minutes (as it is supposed to be). So: What is better now then it was last night and evening and during today? The moon phase???

  • Admin


    must be someone pointing an RF earthing wand at a nuclear moon phase filter somewhere, that makes the photons turn upside down, and enter the antenna in reverse order..

    (In other words, it's kind of black magic with RF sometimes πŸ˜„ )

  • @hek I will start debugging today when I'm at home from this node.

  • Hardware Contributor

    @Sander-Stolk - Maybe you can try a metalic/aluminium case? I have seen much better performance of my network since i change GW and repeaters to a aluminium case.

  • Ok! So I found some issues but not fixed completely...

    What I found out is that my Shed node is sending through my attic-node. Why? I dont know but parent was 1 and distance 2.
    After a Clear EEPROM on the Shed-node and disconnected the Attic-node, re'-uploaded the original sketch and voila a few Find parents and Fails but then I saw 4-4-0-0 instead of 4-4-1-0.
    So that was one problem fixed. The attic-node was not isolated with foil so now most of the time the Shed-node is sending OK and not so many Fails.
    I've ordered a new type of radio with a shield and a external antenna as some may know from the comparison video of the NRF24's on youtube.

    Is there a way to make the parent (0) static to prevent node's from seeking and broadcasting?
    So that I can force my Shed-node to always contact the parent 0 instead of roaming towards the strongest link.
    I know that I can redirect node's to use a specific repeater node but not a parent node.

    So far this update. When the kids are not in sleeping I will check their node's to find out if there is something wrong.

    Keep you all posted!

  • To be sure that the radio is not interrupted or failing in any other way because the lack of power, I've connected the radio with a voltage regulator so 5v in and 3.3 volt out with CAP.
    It helped a little but not much hence the order of a new radio for the Shed which is isolated.
    This is the one to test:

  • @sundberg84 said:

    @Sander-Stolk - Maybe you can try a metalic/aluminium case? I have seen much better performance of my network since i change GW and repeaters to a aluminium case.

    @Sander-Stolk said:

    This is the one to test:

    I'm checking these out to see if I can fix the radio problem once and for all

  • @Sander-Stolk said:

    This is the one to test:

    Looking at the module I see that the shield is NOT touching the ground of the antenna. Is that correct? Or is it connected all together on the back side of the module?

    This is also what I thought to see in my own experiments, that it was better when the Al foil was not touching the gnd of the antenna.

    Still my shielding is imperfect. It does not work at MAX setting.

    Looking at the proposed module I see that really everything is covered by the shield. I will try to enlarge my foil wrapping, but not touching the gnd.

    Hmmmm, guess I will have to do more tests. There are various opinions here about toching the ground with the foil or not.

  • Hero Member

    I don't have a good experience with the shielded radio's. Also with these there are many variations. The shielding of the 3 (different) shielded nodes I have is grounded. Which is what it should be.
    Much has been written about it on this forum I.e

  • I think I have to replace the shed-node. This Arduino Mega just stops in its loops and last night it just hangs. No more fast blinking led.
    Will convert it to a Nano but have to figure out how because of the many childs.

  • Hardware Contributor

    @Sander-Stolk said:

    Is there a way to make the parent (0) static to prevent node's from seeking and broadcasting?

    #define MY_PARENT_NODE_ID 0 (For 2.0.0)

  • @AWI ok, I guess you are right. I will do some more tests with grounded foil.

    Actually I have some really nice and thick Al metal. I could use it to build some better permanent shielding of the module.

    Actually I would like to give my repeater a metal enclosure. But it is built in a way that inside its case the 220 V are converted to 5 V. So I prefer the plastic case. I am too afraid that by some stupid failure there are 220 V on the case and somebody touching it. Hm, so maybe I should rebuilt the repeater with an external 5 V supply and a metal case.

    I am totally sick of these NRFs that all perform differently. πŸ˜‰

  • @karl261 said:

    I am totally sick of these NRFs that all perform differently. πŸ˜‰

    Same here! So I radically solved my problems with connectivity by moving to rfm69 radios. It took some time and effort to make 'adaptor' pcbs to connect rfm's to my existing nodes, especially for 5v nodes, as rfm's are not 5v tolerant and require voltage level converter. I don't have comprehensive statistics for now (finished switching only a few days ago), but even first tests gave stable communication range far beyond that I ever saw with nrf's