[Articles] Significant flaw in Z-Wave chip discovered

[CT Droid]

It has come to light that a design flaw in the Sigma Z-Wave chip may affect your Z-Wave network performance, especially with larger networks. Mi Casa Verde, the guys behind the Vera 2 controller (stay tuned for the review!), which also supports Z-Wave, has discovered a pretty significant bug while troubleshooting existing Z-Wave installations.

Z-Wave is a mesh network, meaning every device has a 'routing table' which remembers how to get to the correct switch (think of it as a mini-internet). When you first pair a new Z-Wave device with a controller, they have to be very close, so the controller can learn about the device. Once this is done, you move the Z-Wave device to the location where you want it, and 'repair/heal' your network, which is supposed to force each Z-Wave device on the network to relearn how to communicate with other Z-Wave switches (and learn the best routes). This is where things go wrong. When you repair/heal the Z-Wave network, it is supposed to overwrite the existing routing table (since before issuing this command, it thinks it is directly connected to the controller), but this isn't happening. Instead it looks like the new routing info is being appended to the existing routing table, making it very difficult for Z-Wave devices to communicate, especially on larger networks.

Since Sigma is the only manufacturer of the Z-Wave chip, this will affect every Z-Wave component on the market. Mi Casa Verde's Vera 2 controller has a field upgradeable firmware option, so they are working hard on pushing out a new update which will resolve this issue. Not sure how other manufacturers will tackle this problem, but if you have a large Z-Wave network, and are experiencing problems, this might be why. I will update this post as more information becomes available.



View the full article

 

[electron]

This post has been updated, check the full link for more info.

 

[Tink]

Besides the update which clarifies that it is not a hardware (design or fabrication) flaw, but a firmware flaw, the information posted at the source site is not accurate.

The bug affects very TINY networks, and only those which were added in such a way that even with a small number of nodes, some of the nodes added were relegated to having to use routing to communicate with the primary controller. When (even with a small network) the nodes are added by forming concentric circles moving away from the controller, the bug is not experienced.

We also note that with HomeSeer, it is possible that our optimization procedure resolves the issue and is why our users have not seen this before.

I wanted to post this to alleviate anybody's concern over Z-Wave and to clarify some bad information that was provided.

Tink

 

[electron]

Thanks for the update Tink. This is a pretty confusing situation (and I updated the post to reflect this), so any additional info is always welcome.