Just a thought . . . Lost links, poor com issues etc

[Teken]

A few months ago I started a thread on the SH forum about what is expected and will happen if a device runs out of links.

 

Some of the replies were that if a device such as a Icon switch was to be maxed out in its link table that the said device would simply roll over and drop the oldest link etc.

 

Question: Would this not explain in some instances where people have reported that scenes, events, have stopped working?

 

Please bare with me and my line of thought. If someone decides to incorporate Icon devices which can only store 30 scenes compared to the 417 that a full swl can.

 

During the course of their system growth / additions to the complexity of the linking process this would either come very close, or max out the Icon link ability.

 

I state this because with out the aid of a ISY / HL2 the average person would not know where the capacity is, and if they had come to that peak link allowed.

 

Why hasn't anyone brought this issue up? How common is this scenario?

 

Again I ask because my experience thus far in several installs has come extremely close to the 30 max links, and with out the aid of the HL2 / ISY the user would assume all is well when in fact any other future add on's would result in things to the user as not working etc.

 

This explains why others who simply re-add a supposedly bad device etc back into the system that it works. But, in reality another device / scene has been bumped off which they are not aware of, yet!

 

My solution for the two installs was to break the news to the friends that they needed to either dumb it down as to what they wanted to control, or to fork out some more cash on swl that has larger data link tables.

 

Insight

[LeeG]

Just as the specs related to ramp rate settings are not accurate for some of the ICON switches, I just wrote (and read back) 33 link records into an ICON relay v1.2 HW, v28 FW. Don't know how much memory the ICON relay has but more than is advertised in the specs.

[Sub-Routine]

I pushed some Icons into service several years ago for devices that don't belong to more than 30 scenes and are not usually controlled locally. At the time there was a significant price difference.

 

I have replaced a few with SwitchLincs as our system grew.

 

The Icons are great for n-ways if you don't need the level indicators.

 

Up to you,

Rand

[Brian H]

The originals looked like modified Switchlincs and may have more features than advertised. I also have a few Icons that have features not in the specifications.

 

That said don't count on anything being there that is not in the specifications. Some day you may get an ugly surprise.

[IndyMike]

Hello Tekken,

 

Lee has probably already given the most direct evidence of what the Icon units are capable of (vs advertised). As Brian pointed out, however, this could change with the next model revision.

 

History -

I have many older Icon switches and plug in units. I actually prefer the single LED in the switches. The Icon line is the SH entry level to Insteon. They are intended to whet your appetite and entice you into the mainstream switches and modules. They are advertised as having lower current/wattage capability and fewer features.

 

For the older units (I don't have any new plug in Icon units), the hardware is identical to the more expensive Switchlincs. These units used the same relays and Triacs. This makes complete sense from a manufacturing standpoint - carrying multiple configurations increases the chances of parts shortages, manufacturing errors, and increases cost. The lower "rating" on the Icon units would therefor appear to be a marketing tool intended to move people toward the more expensive Switchlinc line. I have no problem with this, they are simply "de-rating" their Icon units more and effectively cutting their profit margin on them.

 

Keep in mind that the above is "History". It could, or may already have, change should Smarthome get enough Icon unit volume to make a different configuration economically feasible. In reality, I doubt this will occur.

 

Link Table Upsets

I can't honestly say that I have had troubles with link table losses in any switch or plug in. When I have encountered a problem, it was usually due to user error or a programming failure.

 

KPL's are another matter - My early KPL's (rev 1.3 and earlier) did have problems maintaining their tables. Smarthome replaced these units with Rev 1.65 and they've been humming along for years now. To be fair, I very likely had my "Insteon blessed" Boosterlinc installed in this time frame. It could very easily have contributed to programming errors and link table corruption.

 

If I were to guess at the likely causes of link table corruption, the list would be:

1) Interrupted/Incorrect programming - unfortunately, due to the scene nature of Insteon programming, this could affect multiple devices. It's one of the reasons I "bench program" my units prior to installing.

 

2) User error (I'm guilty of this one). It's actually rather easy to inadvertently perform a factory reset on a switch. I have a number of locations where the switch fits "snugly" into the faceplate. It's pretty easy to accidentally perform a factory reset when simply trying to airgap.

 

3) Power outages. My experiences with EEproms dates back to the 80's and 90's. At that time the devices were susceptible to "upsets" due to power on/off sequences. We employed both hardware and software protection to prevent this. I do not know how the technology has progressed since that time.

 

So, in reading over the above I can see that I've provided a lot of words, but have not addressed your question (typical).

 

While your theory of "wrap back" on Icon unit link tables is plausible, the unit hardware appears to support more than the 30 advertised links. Considering that KPL's appear to be more susceptible to link table loss (417 link capability), I would lean more toward upsets during the actual programming of the device.

 

IM

[Teken]

I think on the next install I will try and max out the icon units and see what happens.

 

Can anyone please confirm the information that the once the device has reached its maximum link capacity that older links will be bumped.

 

This is one Q I would like a confirmation on.

 

I have a friend who intends to deploy only icon devices so this will be a great test bed to confirm my idea / theory of what will happen.

 

Also, would it be fair to say that the current V5.0 + devices have either removed previous features via firmware. Also, have indeed restricted links to the max of 30.

 

A few of you have offered information from older units that have indeed done this. I believe moving forward SH has more than likely dumbed down these devices pretty much to spec.

 

Hence why one member asked about longer ramp rates etc yet he was unable to do so. The common denominator was that his icon device was too new and had the latest firmware release which more than likely killed off those extra features.

 

As always the feed-back from the senior members here have proven insightful, and positive.

 

Much thanks

 

Teken

[LeeG]

Confirmed. Newer ICON does wrap at link record 33.

 

ICON Relay 2876SB v5.3 HW v39 FW

 

Wraps link record number 33 over link record 1. This is a logical memory wrap point as link record 32 is the last 8 byte entry of a 256 byte memory segment. Link record 33 would be the first link record in the next memory segment.

[Teken]

LeeG,

 

Most excellent of you for confirming and taking the time to complete the test for me and any future member wanting to know this info.

 

Having said this: Do you understand what I am stating here as to how this could be one of the problems many people have perceived as being a com issue?

 

There has to be tens of thousands of people who have Icon devices incorporated into their Insteon system which do not have a ISY / HL2 and are maxing out the link table at some critical point and along the way days, weeks, months finally notice something does not work!

 

Imagine this is a very complicated scene with 40 plus devices, and all of the sudden one or more icons have rolled over the previous link which was associated to a huge scene / group . . .

 

The outcome would be total mayhem, and broken, corrupt link tables etc.

 

Insight . . .

[LeeG]

I verified that link records created programmatically in violation of the stated ICON device specs will wrap. That is a failure of the application that is creating the link records in violation of the stated specs. One could argue the device should catch that application failure. Unrelated to ICONs, Smarthome/SmartLabs has suggested a technique that exploits the memory wrap characteristic to determine PLM memory size.

 

In your last example you indicate no ISY/HL2 which I take to mean the links are created using the manual Set button method. I don't know what the ICON firmware will do if more links are attempted with the Set button method than the ICON device can support. That is a good question. It might simply reject more links than the device can support.

[Teken]

I will try to see during the next install with icons devices only as to what happens. As this client does not have any controller for this install so this will test my theory . . .

[Teken]

I will try to see during the next install with icons devices only as to what happens. As this client does not have any controller for this install so this will test my theory . . .

[IndyMike]

Lee,

 

Thank you as well for providing the Hard data on this. One piece of fact hand is worth countless "pontificates" (i.e. myself) on any forum.

 

One of these days I will try to track down the technique for writing links to these devices (the wrap test) that you provided earlier.

 

Thanks again,

IM

[LeeG]

Some additional test results using the new ICON Relay. My previous tests were done using the SD Set MSB/Peek/Poke method of reading/writing link records. It is this method that wrapped link record 33 at 0EF8 over top of link record 1 at 0FF8.

 

Today’s tests use the ED (Extended) ALDB 2F command for reading/writing link records. The positive results are that this command does not wrap back. It writes to the end of the supported link database area and stops writing new link records. The link records do not wrap back. The same thing happens when writing link records using the manual Set button method. Link records are written to the end of the supported link record area with additional new link records not written and not wrapping back over existing records.

 

The odd thing is that only 24 link records can be written. Neither the ALDB 2F command nor the Set button will write records past link record 24. Of course links attempted past that point are not functional. I linked KeypadLinc buttons using the Set button for link records 23 and 24. These link records were written with the Set button and the links are functional. Pressing the respective KPL buttons control the ICON Relay. All attempts to link another KPL button which would be expected to be written in link record 25 do not occur and of course the KPL button is not functional as far as controlling the ICON Relay.

[Brian H]

Thanks Lee for the added tests and data.

Only 24 links. Sounds like a glitch that may need addressing by Smartlabs.

[Teken]

Lee G,

 

I can't thank you enough for all the testing and leg work. Has no one ever asked this simple Q?

 

If they have I have never once read about anyone testing the theory and seeing what actually would happen.

 

This is why I asked this Q here because I knew someone would know and give me the straight dope!