Insteon communication questions

[Vyrolan]

1) Which "direction" do the signals go on the powerline...if that makes any sense? Do they go out along the neutral or the hot or both?

 

2) Do dual-band devices always send out both RF and powerline communications? Do they do it for every message repeated? It seems they must since two dual-bands can repeat messages across phases for two other powerline-only modules.

[oberkc]

I understand that the signals are transmitted at the zero crossing point on the AC wave, which suggests that they are transmitted on the hot side of the powerline. I have also understood that they are always repeated via RF by those devices pocessing such a capability, but I am not confident of this enough to declare it fact.

[Brian H]

The combination Transmit and Receive tuned transformer. Is connected to the Neutral and Line AC Input connection. With a capacitor on the Line side blocking the 120 volt AC from frying the little sucker.

[Vyrolan]

I have also understood that they are always repeated via RF by those devices pocessing such a capability, but I am not confident of this enough to declare it fact.

That's my understanding as well...it would seem that a dual-band device must transmite any signal (triggered locally or received via either porwerline or RF) on both RF and powerline...all the time. It must transmit received RF as Powerline since wireless devices can control powerline-only....it must transmit received Powerline as RF since they can bridge phases. But then with the 3 hop repeating, it seems like there must be a massive flooding...dual-band device #1 sends out both RF and powerline....both are received by 3 other dual-band devices B, C, and D....unless those 3 perfectly detect that the two are duplicates, they would repeat both...so 12 more messages sent (3 devices sending 2 responses to 2 received messages). Seems excessive to me...makes me really wonder about the flood of dual-band devices and if it won't make things worse instead of better.

[IndyMike]

Hello Vyrolan,

 

RF devices do not communicate "constantly". They transmit during a predefined "quiet" period.

 

Each powerlineline transmission is made up of 5 packets of information as shown below. The initial transmission can be followed by up to 3 additional "Hops". Each hop is separated by a one zero crossing quiet period. RF devices communicate during this period. They supposedly do so "synchronously" so that their signals "add" rather than interfere.

 

Clear as mud?

 

[Vyrolan]

RF devices do not communicate "constantly". They transmit during a predefined "quiet" period.

 

Each powerlineline transmission is made up of 5 packets of information as shown below. The initial transmission can be followed by up to 3 additional "Hops". Each hop is separated by a one zero crossing quiet period. RF devices communicate during this period. They supposedly do so "synchronously" so that their signals "add" rather than interfere.

 

Clear as mud?

That actually makes perfect sense as far as interleaving powerline and RF...thank you so much for the detailed answer.

 

Moving along...

Are the RF transmissions also repeated 3 times from the initial sender like the powerline are?

 

Are the "Hops" like we see in Level 3 Event Viewer indicative of which send attempt by initiator it is? I actually assumed the "Hops Left" was like networking TTL...so like my message from A could be repeated by B to C to D to E, but E wouldn't repeat since that was already 3 hops... I thought the initiator of a message would make 3 attempts to send the message, and that message would only be repeated a maximum of 3 times. Here's an example:

• 
  [*:3ar6tixe] A sends message with 3 "Hops Left"
  [*:3ar6tixe] B receives A's message and repeats it with 2 Hops Left
  [*:3ar6tixe] C receives B's message and repeats it with 1 Hops Left
  [*:3ar6tixe] D receives C's message and repeats it with 0 Hops Left
  [*:3ar6tixe] E receives D's message but does not repeat

Then that whole process happens up to 3 times (three send attempts by A). Then D may receive directly from A and so the jumps are bigger than just 1 device. It seems like it has to do something like this TTL concept otherwise messages would repeat forever, no? If A sends a message received by both B and C, when B and C repeat, they will hear each other's repeats and repeat again...does it understand those are duplicates and not do it? The TTL solves that by just limiting the repeats...so I assumed that's how it worked...is that mistaken?

[Brian H]

This white paper may give you some information.

http://www.insteon.net/pdf/insteondetails.pdf

It has not been updated for the later stuff and still shows the old RF frequency of 904MHz not the present 915MHz.

[IndyMike]

That actually makes perfect sense as far as interleaving powerline and RF...thank you so much for the detailed answer.

 

Moving along...

Are the RF transmissions also repeated 3 times from the initial sender like the powerline are?

 

Are the "Hops" like we see in Level 3 Event Viewer indicative of which send attempt by initiator it is? I actually assumed the "Hops Left" was like networking TTL...so like my message from A could be repeated by B to C to D to E, but E wouldn't repeat since that was already 3 hops... I thought the initiator of a message would make 3 attempts to send the message, and that message would only be repeated a maximum of 3 times. Here's an example:

• 
  [*:rvutj1b5] A sends message with 3 "Hops Left"
  [*:rvutj1b5] B receives A's message and repeats it with 2 Hops Left
  [*:rvutj1b5] C receives B's message and repeats it with 1 Hops Left
  [*:rvutj1b5] D receives C's message and repeats it with 0 Hops Left
  [*:rvutj1b5] E receives D's message but does not repeat

Then that whole process happens up to 3 times (three send attempts by A). Then D may receive directly from A and so the jumps are bigger than just 1 device. It seems like it has to do something like this TTL concept otherwise messages would repeat forever, no? If A sends a message received by both B and C, when B and C repeat, they will hear each other's repeats and repeat again...does it understand those are duplicates and not do it? The TTL solves that by just limiting the repeats...so I assumed that's how it worked...is that mistaken?

 

This is actually pretty close. Have a look at pages 19-27 of the document that Brian provided. It gives some nice examples.

 

RF is repeated at each "quite period" in the plot I provided. Like the powerline transmission, the Hop count is decremented for each repeat.

 

Your list is an example of a "single thread" communication. In practice, many devices will be involved with each transmission. As Brian indicated, the paper has not been updated for awhile. There have been improvements to the protocol to involve more responders.

 

As an example, the white paper indicates that the original transmitter will transmit only once, and that listeners will repeat only once. Newer revision units, however, will transmit and repeat at each hop.

 

• 
  [*:rvutj1b5] A sends message with 3 "Hops Left"
  [*:rvutj1b5] All devices in range repeat with 2 Hops Left
  [*:rvutj1b5] All devices in range repeat with 1 Hops Left
  [*:rvutj1b5] All devices in range repeat with 0 Hops Left

[Vyrolan]

This white paper may give you some information.

http://www.insteon.net/pdf/insteondetails.pdf

It has not been updated for the later stuff and still shows the old RF frequency of 904MHz not the present 915MHz.

Thanks for the link!

 

This is actually pretty close. Have a look at pages 19-27 of the document that Brian provided. It gives some nice examples.

 

Your list is an example of a "single thread" communication. In practice, many devices will be involved with each transmission.

• 
  [*:39qmvu9w] A sends message with 3 "Hops Left"
  [*:39qmvu9w] All devices in range repeat with 2 Hops Left
  [*:39qmvu9w] All devices in range repeat with 1 Hops Left
  [*:39qmvu9w] All devices in range repeat with 0 Hops Left

Cool....that's kind of what I expected...I kept my example simple to try to keep it clear in case I was way off. Do they do any dupe checking? If switch A receives the same message with the same number of hops left from multiple other devices, does it realize they are all the same or does it just repeat them all?

 

Thanks for the info guys...I'm always one to want to really understand what's going on.

[IndyMike]

Cool....that's kind of what I expected...I kept my example simple to try to keep it clear in case I was way off. Do they do any dupe checking? If switch A receives the same message with the same number of hops left from multiple other devices, does it realize they are all the same or does it just repeat them all?

 

All devices respond during the same time window. The devices are synchronized by the 60Hz zero crossing of the powerline. From the image below, all devices that hear the initial transmit will follow by transmitting during the Hop1 time window. Multiple transmissions occurring at the same instant will increase the amplitude of the Hop1 signal.

 

The waveform shown below involves only 2 devices on an Isolated circuit so you cannot see amplitude "addition" during the Hop transmissions. However, the timing will be the same regardless of the number of listeners/repeaters. This timing is for a transmission only. It does not include a response transmission (acknowledge) from a device. The timing would apply to the ISY scene commands. Multiply the time frame by two for command/response transmissions.

 

[Vyrolan]

Cool....that's kind of what I expected...I kept my example simple to try to keep it clear in case I was way off. Do they do any dupe checking? If switch A receives the same message with the same number of hops left from multiple other devices, does it realize they are all the same or does it just repeat them all?

 

All devices respond during the same time window. The devices are synchronized by the 60Hz zero crossing of the powerline. From the image below, all devices that hear the initial transmit will follow by transmitting during the Hop1 time window. Multiple transmissions occurring at the same instant will increase the amplitude of the Hop1 signal.

 

The waveform shown below involves only 2 devices on an Isolated circuit so you cannot see amplitude "addition" during the Hop transmissions. However, the timing will be the same regardless of the number of listeners/repeaters. This timing is for a transmission only. It does not include a response transmission (acknowledge) from a device. The timing would apply to the ISY scene commands. Multiply the time frame by two for command/response transmissions.

Yea after posting that question I was reading through the white paper and discovered the answer. That makes sense how both the powerline and RF can be independent and also reinforcing of each other and themselves. I also noticed that devices won't send a new messages until the first message is completely done...but it seems the time windows are sufficiently small...in your waveform, the lifetime of a message is ~200ms, so even if a signal got queued up, it would not produce a significant delay. How does it handle collissions with regard to that? I didn't notice anything about that in the white paper but I'm not through it all yet either. If something sends a signal, essentially nothing else can send until all the hops of that signal are complete... Suppose during that time, two other devices both get clicked...when the first signal's final hop is complete, both will try to send their message and there will be a collision. In any normal-sized installation, I don't think there's much chance for tons of switches being clicked in the same ~100ms, but I wonder if you could generate a bunch if there would be a noticeable (1s or more) delay in the last one to go through.

[IndyMike]

Yea after posting that question I was reading through the white paper and discovered the answer. That makes sense how both the powerline and RF can be independent and also reinforcing of each other and themselves. I also noticed that devices won't send a new messages until the first message is completely done...but it seems the time windows are sufficiently small...in your waveform, the lifetime of a message is ~200ms, so even if a signal got queued up, it would not produce a significant delay. How does it handle collissions with regard to that? I didn't notice anything about that in the white paper but I'm not through it all yet either. If something sends a signal, essentially nothing else can send until all the hops of that signal are complete... Suppose during that time, two other devices both get clicked...when the first signal's final hop is complete, both will try to send their message and there will be a collision. In any normal-sized installation, I don't think there's much chance for tons of switches being clicked in the same ~100ms, but I wonder if you could generate a bunch if there would be a noticeable (1s or more) delay in the last one to go through.

 

Vyrolan,

 

I'm not sure that the white paper mentions collision avoidance. The Insteon modules DO have collisions avoidance (verified by a SmartLabs Engineer back in the day).

 

In the early days, if you pressed two switches "nearly" simultaneously, the second would "flash" that it was unable to communicate to linked devices. I just tried this with an older KPL, and a dual band SWL - no flashing on the SWL?? Things may have been "improved" again.

[Vyrolan]

I'm not sure that the white paper mentions collision avoidance. The Insteon modules DO have collisions avoidance (verified by a SmartLabs Engineer back in the day).

Yea it kinda hand waves the whole topic...it mentions how they'll wait for a previous message to complete before sending the next one, but nothing about when they collide. There's any number of ways they could resolve it that would only take a few cycles (~100ms). It's more just curiosity that could you nail a decent number of switches in a short enough time such that all the collisions do get resolved but the last message processed is actually delayed by a noticeable amount of time (in my head defined as 1s or more).

 

In the early days, if you pressed two switches "nearly" simultaneously, the second would "flash" that it was unable to communicate to linked devices. I just tried this with an older KPL, and a dual band SWL - no flashing on the SWL?? Things may have been "improved" again.

What was the result? Did both things actually get sent? Or did it just not work and not flash? =p

[IndyMike]

In the early days, if you pressed two switches "nearly" simultaneously, the second would "flash" that it was unable to communicate to linked devices. I just tried this with an older KPL, and a dual band SWL - no flashing on the SWL?? Things may have been "improved" again.

What was the result? Did both things actually get sent? Or did it just not work and not flash? =p

 

Both activated and communicated with the ISY. To be fair, both of these are "responder only". They do not control scenes and therefor do not request cleanup messages from anything but the PLM (pretty short communication time).

 

A scene controller, linked to many devices, might create problem for the second switch pressed.