markens Posted March 13, 2009 Posted March 13, 2009 Hi, I've been bringing up my new ISY99i system over the past couple of weeks, and battling a communication reliability problem. I finally isolated it to a single computer. But, the details seem strange to me: With this computer off, everything in my network is 100% reliable (both X10 and Insteon), including X10 scene triggers and more complicated Insteon scenes controlled by the ISY. When this computer is turned on, the ISY becomes unreliable. Scenes don't work reliably, the ISY event monitor does not see any X10 events (even though a lamp module on the same circuit works 100% of the time). I thought using a filterlinc (1626-10) would help. But it doesn't. Plugging the computer to the filtered plug of the filterlinc makes no difference in ISY behavior. This is the part I find strange. The normal configuration is for this and another computer to be plugged into a UPS, which is itself plugged into a fliterlinc plugged into the wall. The other computer has no effect on ISY reliability in this configuration. I tried plugging the problem computer into its own filterlinc, bypassing the UPS; no change. The problem computer is fairly old. Could that make a difference? For now I'm just leaving it off, and everything works fine. Not a good long term solution though. More details on my network, to be complete: SignaLinc (2406H) installed near panel for phase coupling. Relatively small house and Insteon network (~15 devices). PLM on different circuit (same phase) and about 20' from panel. No access points (which seems ok since everything works 100% reliably with problem computer off). I have an X10 TesterLinc (4819) which shows perfectly normal X10 behavior when plugged in to the same circuit as the problem computer. Even when the ISY does not see X10 events, the TesterLinc shows the events with good "quality" (120). There is zero 120kHz activity shown when idle. I don't have any way to measure X10/Insteon signal voltages on the power line. My A/V setup (on the same circuit as the problem computer, but further from the panel) is connected to a filterlinc. There does not seem to be any other noise (or signal sink) problems. ISY-99i, v2.7.0. PLM: v1.95, fw 7A. Any ideas about why this one computer is a problem, and what I might do to correct it? Thanks much! BTW, kudos for the top-notch support provided in this forum. I have learned so much reading previous items! --Mark
MikeB Posted March 13, 2009 Posted March 13, 2009 I assume the PLM is on the same circuit as the PC/UPS? In close proximity to eachother? If it's close, and it's possible/practical, you might consider moving it further away from the PC/UPS to see if it still affects the signal to/from the ISY. Otherwise I'd consider replacing the PC's power supply, which is probably the source of the noise. While I find FilterLincs to work well when I've needed them, I imagine it might not be possible for them to solve all powerline issues caused by a device. I have heard of old/bad PC power supplies causing lots of powerline havoc. Before you go any further, test well with that PC off/disconnected to ensure that's truly the source of your problems.
markens Posted March 13, 2009 Author Posted March 13, 2009 I assume the PLM is on the same circuit as the PC/UPS? In close proximity to eachother? If it's close, and it's possible/practical, you might consider moving it further away from the PC/UPS to see if it still affects the signal to/from the ISY. The PLM was on the same circuit, but is now on another circuit. No difference in behavior placed further away. Otherwise I'd consider replacing the PC's power supply, which is probably the source of the noise. While I find FilterLincs to work well when I've needed them, I imagine it might not be possible for them to solve all powerline issues caused by a device. I have heard of old/bad PC power supplies causing lots of powerline havoc. This sounds like a very likely explanation for my case. The PC in question is over 10 years old, running 24 hours a day for the entire time. I will most likely retire it and replace with much newer hardware. Before you go any further, test well with that PC off/disconnected to ensure that's truly the source of your problems. Indeed I will. So far, everything looks good. But I'll be testing for a few more days before I make long term decisions. Thanks, --Mark
rowland Posted March 14, 2009 Posted March 14, 2009 Sounds like the problem I had with my PC. If so, then a new power supply seems to be the only fix. But, it's also possible that your problem could different and present the same symptoms I was seeing. If so your problem could be fixed with a Leviton 6288 filter. IndyMike is the resident expert on this.
markens Posted March 14, 2009 Author Posted March 14, 2009 But, it's also possible that your problem could different and present the same symptoms I was seeing. If so your problem could be fixed with a Leviton 6288 filter. IndyMike is the resident expert on this. I tried to find specs on the Leviton 6288, but found nothing more than generic description. How does this filter differ from the SH filterlinc 1626? Thanks.
rowland Posted March 14, 2009 Posted March 14, 2009 The 1626 is a Bandstop whereas the Leviton 6288 is a lowpass. IndyMike really is the guy to address this. But while we're waiting for him you could check out http://www.forum.universal-devices.com/viewtopic.php?p=15660#15660
IndyMike Posted March 14, 2009 Posted March 14, 2009 I tried to find specs on the Leviton 6288, but found nothing more than generic description. How does this filter differ from the SH filterlinc 1626? Hello Mark, As Rowland indicated the filterlinc 1626 is a bandstop filter that is effective in a narrow region around the X10 120Khz center frequency. The Leviton 6288 is a lowpass filter with a breakpoint around 13Khz. It simply has a wider bandwidth. Since your Testerlinc is showing a "quality" of 120 you should have adequate X10 signal. The quality doesn't tell you signal level, but rather the number of valid X10 "counts". 120 would be considered very good. I find it interesting that your PC is causing problems even when plugged into the UPS. Your UPS likely has a fair amount of filtering incorporated itself and your PC is somehow making is past and back onto the line. X10 and Insteon signaling work in noisy environments because the transmissions occur in a "quiet" region near the AC powerline Zero crossing. Most devices that cause noise are switching near the AC powerline peaks, far removed from the Insteon/X10 transmissions. Another feature implemented in some X10 devices is Automatic Gain Control. This allows receivers to adjust their input sensitivity down to ignore noise on the powerline. Leviton added a further refinement to this called "gated AGC". This essentially sampled the powerline in the AC zero crossing region and adjusted the input sensitivity accordingly (sensitivity optimized to the zero crossing). With the above in mind, here's a couple of possible causes: 1) Your X10 module does not incorporate AGC (you didn't mention the model). There is noise on the line, but the module can pull the X10 out of the hash and operate reliably. BTW - the Testerlinc isn't the best at detecting noise on the line. 2) The noise is causing your PLM/Insteon device to adjust their AGC downward (reduced sensitivity) thereby causing communication problems. I've been told (by Smarthome engineering) that Insteon devices do employ AGC. Unfortunately I have no details on the implementation. I'm not sure if this is Gated AGC. So, after that long winded explanation, where does that leave us? You've already isolated the culprit so most of the work is done. If you are planning on replacing the PC anyway, that would likely be the route to take. If you're into a little experimentation, you could try the Leviton 6288. I've had good luck with this unit on a problematic Toshiba Laptop. Happy automating, IM
IndyMike Posted March 14, 2009 Posted March 14, 2009 Sounds like the problem I had with my PC. If so, then a new power supply seems to be the only fix. But, it's also possible that your problem could different and present the same symptoms I was seeing. If so your problem could be fixed with a Leviton 6288 filter. IndyMike is the resident expert on this. Rowland, Thank you for the high praise. I assure you that I am simply a student standing on the shoulders of those who have come before me (Ido Bartana and Uncle Phil Kingery to name a couple).
MikeB Posted March 14, 2009 Posted March 14, 2009 Thanks for the detailed info IndyMike - very interesting stuff.
markens Posted March 15, 2009 Author Posted March 15, 2009 IndyMike, Thanks for all the good info! It really fills in some knowledge gaps for me, especially how AGC affects things in the presence of noise. I think looking at the waveforms of my powerline with the bad computer on would be very interesting. What tools do people use to do this? (I figured the Testerlinc is not good at showing more complex noise problems.) Identifying the problem computer has erased a huge question mark for me. Correcting it will be straightforward at this point. --Mark
Brian H Posted March 15, 2009 Posted March 15, 2009 I have an X10Pro XPCP phase coupler that I use to look at power line signals. It has two transformer coupled tuned circuits in it and isolates the power line from the scope. One Line and Neutral set; I have a line cord on. The other set is connected to a pair of binding posts. I added a 10 Ohm resistor across the post to roughly simulate a power line. Mounted in a single gang electrical box with a wall plate cover. Works OK. ACT made a scope to power line tester, but you had to be careful about line and neutral being correct as reversing it could make the whole scope 120 above ground. I have seen home construction plans for a similar line to scope tester.
IndyMike Posted March 15, 2009 Posted March 15, 2009 Mark, Analyzing noise spectrum is definitely next level. I use a USB scope along with the ACT Scope Test2 adapter that Brian mentioned. The problem with using the adapter for noise measurement is that it is "tuned" to 120Khz. Frequencies outside it's tuned range will be attenuated. For this reason you can't use it for absolute measurements. I can be handy for comparison purposes. The first plot below is noise injected by a 15 Watt Sylvania standard CFL. The slow (low frequency) signal is the 120V powerline. The high frequency content is the CFL switching noise (~ 70Khz +/- 150 mv peak). Note how the noise "dies off" in the region of the zero crossing). The following plot is of a Neptune dimmable CFL. Although the P-P levels are similar, the noise is "crowding" the zero crossing. There is also noise content in the 140 Khz range which is uncomfortably close to the Insteon 130Khz. I'm currently using these bulbs without any problems. Quite frankly, I'm surprised things function as well as they do.
markens Posted March 16, 2009 Author Posted March 16, 2009 Brian and Mike, Great info, thanks. I had no idea about USB scope probes, but it makes sense! --Mark
Brian H Posted March 16, 2009 Posted March 16, 2009 I have a USB scope that I use occasionally. Basically with mine. You get what looks like a scope screen on your PC Monitor and have dial on the display you can set. Has a small external box that the probes connect to and a USB cable on the other end to connect to the computer.
Recommended Posts
Archived
This topic is now archived and is closed to further replies.