The Vera is outside? Is it sheltered in some way, like in a box?
Yes plastic UV stabilized sealed box , with vents , not in direct sun. With cable entry glands and drip curves on the cable conduit
What two devices are at the end of what cable?
The two controlled loads Switch A and Switch B are at the end of a very long cable (30 meters long). I mentioned this as it may explain the observed behavior of load affecting control reliability
Vera(0) -> SwitchA(6) -> Outlet(11) -> SwitchB(23)
Ok thanks this was very helpful and I've worked out that what confused me is the difference between Neighbors and Routes
The Route list is what intermediate devices to go through to get to the device.
The Neighbors list is what devices are next to this device.
In Neighbors ID 1 is the Vera - which will then cause a Route of 0 to be used
In the routes 0 Means Go DIRECT and so will never be used with a second hop.
With UI5 the manual route is directly supported but I had to click Update Neighbors to get this to work
What does this mean? How is the load exaggerating the problem? Load should have no impact on signal quality.
I stand by the statement that load exaggerates the problem. Which is with no load - control works within 1 or 2 re-tries , with a load control never works.
It may be a voltage drop due to the long lead causing the Zwave chip problems maybe its excessive radio transmission from the long lead. I don't know the causal effect - I am only reporting the observed behavior.
Your reliability issue is a signal quality issue. If the routing doesn't help, I would suggest relocating the Vera.
After reviewing all the Zwave routes Ive determined that the cause is that the Zwave protocol has determined the neighbors incorrectly - I have drawn myself a diagram and what I have found is that the Vera ONLY routed via the 5 closest devices to it. ie it did not try and use a device halfway, even though some of the devices it chose because they were close to the Vera were actually further away from the final device and did not have a LOS to the final device.
Unfortunately with the 5 devices in the shed right next to the Vera this caused the Zwave network to determine a route which was essentially useless. All chosen routers have equally bad signal to the final device. What I needed to happen was for the Vera to use the device (Switch A device ID 18 ) that was halfway to the final device.
I’ve fixed it manually by adding in the intermediate position device into the manual routing and everything is good now.
FYI I’ve uploaded a diagrammatic layout of the Zwave devices
Device ID 11 is the problem device
Device ID 17 is a battery device and does not take part in routing - but I’ve included it for completeness
Device ID 1 is the Vera
Neighbors (Before Manual intervention) - Taken from the Vera Neighbors attribute of the four devices
device 11 - 1,6,7,8,17
device 18 - 1,6,7,8,14,
device 19 - 1,6,7,8,14,
device 17 - Not viewable (I’d guess because its battery powered)
So this is so wrong
The Neighbors of 19 does not include 11 and 19 and 11 are within 300 mm of each other
The Neighbors of 18 does not include 11 and 18 to 11 is 3 meters distance and 11 is the closest device to 18
– This is the mistake that Zwave has made that is causing the problem. It has missed a high quality neighbor and so is not forming a mesh to route via this device.
I placed 18 where it is so as to provide a mesh connection from the Vera to device ID 19, as Zwave is advertised as mesh - but it appears that in this case the Zwave protocol did not form a good multi-point mesh.