Insteon PowerLinc issues

[KHarms]

The Insteon PowerLincs seem to suffer – in varying degrees – with spontaneous disconnects from the USB Bus. They can be working perfectly one minute and then disconnect from the USB Bus. When this happens Windows sends out the USB Disconnect message to all applications. Sometimes this problem manifests as a “general hardware error” being returned from read and write operations being done on the USB interface. Neither is good news.

What happens is that HCA assumes this to be a power outage event and attempts to restore connection to the PowerLinc. This sometimes – but in many cases – does not work. The only solution is to manually power cycle the PowerLinc.

Why does the PowerLinc disconnect?

We don’t know and can only offer the advice that SmartHome tech support supplies: Make sure that there are no power saving settings on your computer that can power down the USB sub-system. This may be in the Windows Power Management features and it may be in the Bios settings. Check both. Second, SmartHome tech support recommends that you use a powered USB Hub for connection of the PowerLinc.

SmartHome tech support has said this:


"There are not many suggestions that we can offer in these cases. We have only seen that some computers do not work well with the USB interface. FTDI Chip, the maker of the interface stated that it was an incompatibility between windows powersave and the chip’s instruction set. Typically replacing the unit does not resolve the issue."


and


"Yes, unfortunately, it typically has to do with an incompatibility with certain chipsets in computers. We had a customer that had 2 laptops with the exact same OS, but the PLM worked on one, and not with the other. And it came down to the two computers had an Intel chipset and the other had an Nvidia."


We have fought this problem for months and we have tried all the software solutions that we can think to compensate for the problem and are left with the uncomfortable fact that this is indeed a hardware problem. The recourse of the software developer is always to blame the hardware. We really try and not do that but this time we have no choice.

Try the SmartHome suggestions - we know of several cases where use of a powered hub fully resolved the problem - and if that fails you may want to consider one of the serial interface connected PowerLincs.

[barry]

I have had this problem with my 2412U Insteon modem for months. It is good to know a powered USB hub might help. I have tried 3 different PCs (an XP Dell laptop, a win7 x64 HP Desktop, and a no-brand XP desktop with absolutely nothing installed other than HCA), and all experience the same problem, so it's hard to believe it is a chipset issue unless there is only one particular chipset that works correctly. From what I've experienced, after running for more than 24 hours the 2412U eventually stops responding to the software. I have tried both HouseLinc2 and HCA 9.0 and both have this problem. The problem can be resolved by either unplugging and re-plugging the USB cable and restarting HCA, or by rebooting the PC.

My solution was to schedule a reboot at 3am, and that avoided the problem 99% of the time. It seems like there is a "buffer" or something similar that overflows and locks up the USB if left running more than 24 hours. If reset (or rebooted) at 24 hour intervals, it does not seem to happen. After all of this, I'm now wondering if it is the combination of X10 and Insteon devices that I have that are upsetting the 2412 somehow.

I hope this helps someone else having these issues.

[Rob-NoVA]

I just got an Insteon 2413U and suffered the disconnects almost immediately. The machine running HCA is a Windows 7 box and based on info from this thread, I went into Power Options->Advanced Settings and disabled "USB selective suspend" for both on battery and plugged in. So far (keeping fingers crossed), I've had no disconnects for 24 hours, where previously, it was happening in less than 8 hours.

[Rob-NoVA]

Spoke too soon... :( Got home from work and the interface was down again. had to unplug/replug the USB cable and connect via the HCA->Hardware->Setup menu.

At the time of the outage? I saw this error in the log. 4,Error,09/28/2010 18:37:00,1,,,,SendTask: error 31 from Windows GetOverlappedResult. Service for interface 0 is terminated

What does that mean?

[kjbbasic]

Which version of the ftdi driver are running?

Latest is 2.08.02 http://www.ftdichip.com/Drivers/VCP.htm

Perhaps I've been fortunate but I have not had this issue (2412U) . I'm running HCA on a windows 7 laptop, which has been running for weeks with no issues. Powered USB hub is in use....tho wasn't always the case.

Does your setup allow for suspend/sleep or hibernate? Look at advanced settings->Sleep and verify all settings. Mine are set at either off or never since I want the laptop to run 24x7 ...only the monitor is allowed to turn off.

Also for the USB Root Hubs in device manager, under the properties/ power management tab for the hub I make sure the box that allows the computer to turn off this device is unchecked.

[Rob-NoVA]

Thanks for the info, kjbbasic. I am running the latest FTDI driver and power management is disabled for all USB Root Hubs. I hadn't checked that setting before, but it was already disabled. I moved the 2413U to a powered hub today. We'll see what that does. I did need to reboot the computer today, so I messed up my testing. Silly security patches... ;-)

[kjbbasic]

Any success with the powered hub?

[KHarms]

This is odd because the 2413 *generally* had fewer connect issues that the 2412. I posted about that someplace.

Try connecting it to it using the 2413U pick and then try the 2413S pick.

The 2413S pick only uses the serial port created by the driver.

The 2413U uses the serial port initially and then switches over to talk directly to the FTDI driver.

[Rob-NoVA]

Sorry for the late reply all. I've been away. Since moving to the powered hub, I've had zero issues. It's connected as a 2413S in HCA, FWIW.

[geek_mike]

I seem to be a day or two behind Rob trying much the same things with my 2413U.
I have tried a lot of combinations but everything has the problem if I have HCA talking to the 2413U. Typically I had been getting about 8-10 hours between crashes. Today with the updated 2.08.02 FTDI driver I did get close to 24 hours before the crash. This is a sample size of one but that is the longest I have seen this run between lockups. In my test, plugging and unplugging the USB has not cleared the problem, nor has power cycling the module. Rebooting the computer has been the only recovery method that has been working.
It had not occurred to me to try adding a powered USB hub. So I have just done that a few minutes ago. I am still using the 2413U driver interface in HCA. I have not yet shifted over to the serial version in HCA with this configuration. Although I did have it set to Serial in one of my earliest configurations and had the problem then. The hub that I am trying is a Belkin one that was sitting in my spare parts pile and was working fine when I last used it about a year ago. I am due to spend some time on travel later this week so it may not be till the weekend before I can report any results.

[geek_mike]

After running all night in the new configuration with the power hub it had not locked up but there were large number of "Power restored at ..." error messages in the HCA log. So I am giving up on that and now trying the same hardware configuration this time with the serial HCA driver selected.

[KHarms]

Please re-read the 1st email in this whole thread. It offers the best advice that SmartHome has offered on this.

Guys: We really wish that there were not PowerLinc issues. We tried to do all that we could in handling the disconnects when they happen but this is outside of a software "fix".

After all the real "fix" would be for the PowerLinc not to just disconnect on its own at random times.

[jhendric1]

I reciently changed from a 2414u to a 2413u and started having "power failure" messages appear as well. Never had them before and now I was having them regularly even though I know for a fact the power did not go out. My PC's power scheme was set to "Energy Star" so I changed it to "Home/Office Desk" and haven't had the issue since. Hibernate is disabled as well. I do not run a hub.

BTW, the reason I converted to a 2413u from the 2414u was because of it's dual band capability of powerline and rf. Response time from my Triggerlinc's(rf only) dropped from around 3 painfully long seconds to less that 1/2 second. I was using a W800RF32 with DS10's due to the fast response time but the reliablity stunk with missed "ON" messages everyday. With the 2413u/Triggerlinc combo I now have the speed I need with the reliablity I need and one place I use this is on the door into the garage. Opening the door and waiting for the light to come on in the dark gets old quick.

[KHarms]

What can the developers say. We don't like this situation at all. But We can't seem to find any more suggestions other than to use a serial version if you are being plagued by this.

[geek_mike]

I have done a lot more testing of several configurations. There seems to be 3 separate problems in using the 2413U.

1) Don't use the USB option in HCA to talk to it. I find that once it gets an error returned (such as no reply from the remote device) it reports that error for every transaction after that. Apparently it is not cleared somewhere, HCA or the driver I can’t tell where. If instead you use the serial setting in HCA (yes I know this is a USB device) but it still works and it does not show this error state. For me that meant that instead of running less than a day before it was down, it then runs for many days (I have had 2 weeks in one stretch) without a problem.

2) Installing a powered USB hub between the PC and the 2413U does fix a problem for many configurations. Modern USB ports in PCs come in a number of chip configuration. The standard places a limit on the power that can be drawn on these connections. That limit is typically ignored in the implementation of the powered USB hubs. Note that there are now laptops that often have one port that is designed to provide more power for charging cell phones (BlackBerry and other things). The problem you hit is that apparently when the 2413U is having a long conversation and talking to other Insteon devices and at the same time gets a command from the PC, it appears the 2413U exceeds its internal power capacity temporarily and logs a power failure if it can't draw more from the USB connection to keep up. I see this by sending the commands well-spaced out (one minute apart) and all works fine when it is directly attached. However, if I send the commands to many devices and have them send replies back as fast as HCA will run (typical of the way I set up the sunset device turn on) then it often gets a power fail fault. This seems to happen about 1 in 2 times for my sunset series without the powered hub. It does it on all the ports in my desktop CP both the ones on the mother board and the ones on an add-in USB card from a different vendor. Adding a powered hub and I never see the problem again. Eventually, I want to try either my Dell laptop which has a high power USB port, or an upgraded one on the desktop that has the extra power rating. I suspect that will eliminate the powered hub in my configuration.

3) The third problem is what makes the 2nd one so bad. Once you get the power fail signal. It does not clear even though the power is clearly back. It only clears after the OS has rebooted. Restarting HCA does not fix it. HCA keeps right on reporting a power fail. Power cycling the 2413U does not fix it but rebooting the system does. Interestingly, I also own a serial Insteon controller and it effectively had the reverse problem. It would lock up and power cycling the controller not the OS cleared the problem. For a while I kept that one fixed by putting the controller on a mechanical timer that turned it off and on twice a day (at times I know there would be no commands) to assure it was always reset regularly. It was my efforts to fix this that got me to switch to the 2413U.

I just checked the log viewer and my configuration has now been running for 8 days without a problem. I had previously rebooted to patch the OS.
This current long run will probably end on Tuesday as that is the day Microsoft will be releasing many OS patches from what I hear. Thus, I will have to reboot for those installs.

I hope this information is useful to someone.

PS: if you figure out how the clear the stream of power fail messages in HCA without an OS reboot, I would like to hear about it.

[KHarms]

The Power outage messages come from HCA getting a disconnect message from Windows.

The USB PowerLinc, as you know, come with a driver that creates a virtual serial port. When you open it using the 2413U model then it talks to the device driver directly - not going through the Windows serial APIs. When you open it with 2413S then it does call the Windows serial APIs and those in turn call the device driver.