cannot seem to get the UPB noise level down around here to the point where it doesn't mimic real UPB Ack's on the powerline. With a low noise level you can get a pulse back from any unit you address letting you know that it heard the command or not. This opens up the possibility of adding features to the program to attempt to resend commands that were not acked to increase reliability. But an Ack pulse is a simple pulse, not a message that can be verified or checksummed. The UPB standard also allows you to ask for such a message, but though I've turned on the flag to request that in ever command I send out I dont seem to get anything back. Perhaps that just isn't implemented in the devices that I have or perhaps I just dont understand something about it. In any case I high noise level will make the PIM think that you've received the ack pulse even when the device has not responded.
Before going to bed last night I checked the noise graph and noticed an interesting drop out after a short power failure we had just after 7 yesterday. Only lasted about 5 seconds, but it make the noise completely disappear for an hour. This makes me wonder about the same simulate preset dim issue I blogged about here Debugging UPB Signals. At 3:30 am or so after laying in bed awake for a while I got up to have a look at the data again. Usually at night there are very few lights on so noise would be lower if it was related to a failing CF ballast but higher if it was related to something dimmed to 0. Notice the dip just before 1am. Checking the XTension logs showed that there was a hit on the porch motion sensor and the deck lights dimmed from 0 to 100 for a while. While the lights were on the noise was low and when it turned them off the noise returned to the higher levels. The thing that makes this particularly weird is that this porch light is just 2 regular 100 watt light bulbs in an outside fixture, no CF no ballast nothing odd or unusual at all. But testing it again manually verifies that when they are on or off or dimmed to any value above 5% there is no noise, but dim them to 0 and they clog the UPB line. These are on a regular X10 wall switch. UPB and X10 are supposed to be able to coexist, but there is something about dimming to 0 that makes UPB unhappy.
This morning now I've checked a bunch of other regular X10 switches that I dim to 0 instead of turning off, so that I can have a gentle startup with them rather than jumping to 100% like they do if you just turn them off and about half of them add to the UPB noise and half do not. I can change a couple to not dim to 0 without too much trouble, but some I will need to debug as I need that functionality. Or just replace with a UPB capable switch that can do ramp rates. The 2 master closet lights are setup this way so that during the night if you walk into them the lights come on dim so as not to disturb a sleeping spouse but come on full during the day. One of the 2 causes this problem, one does not. Replacing the light bulb in the offending one doesn't seem to help so next step will be to replace the switch, or add a filter or something.
Seems very strange indeed that regular bulbs on X10 switches are the culprit here as most of the other problems I've ever had with noise have been due to failing CF ballasts, but nothing like that going on in the house at the moment.