Water heater eating X-10 signal

| Dan, | | What would you think about trying some sort of high pass filtering | from either leg of the 2 pole breaker feeding the heater to ground?

You can get that in a UL-listed package as, e.g., the Leviton 6285 (though that isn't quite how they intend it to be used). I would be inclined to try the 0.1uF capacitor across the switch first. If it works that would be strong evidence for noise. Since you have the scope you might want to take a look. While the scope won't decode the X10 bits it might show you a steady noise source. Beware that if you have the original ACT Scope-Test (as opposed to the "2" version) it may be too selective to be useful.

Trying the CR234 might be a good idea as long as it isn't too old. (Older ones didn't handle extended codes.)

Dan Lanciani ddl@danlan.*com

Thanks. I'll give it a try.

I'll look that up. Thanks.

Appendix C, FIGURE C-3 (within dashed line rectangle)...

Thanks for these links, I'll definitely check them out.

Unfortunately, my scope is a rather old, 70's era, dual channel analog model - no storage capabilities at all.

I've been able to do a little more testing, and I'm beginning to believe that my "stable" system wasn't so stable after all; that it was really just barely working, and the addition of the water heater circuit was simply a straw that broke the camel's back, so to speak.

I dug out an extra WS467 from my X-10 parts box. I used a small table lamp for testing its operation, right at the panel. Using clip leads, I wired: from ground to one prong of the lamp, from the other prong of lamp to the black lead of the WS467, from the blue lead of the WS467 to what I will call a "test probe". I also wired the ESM1 from a ground to the blue lead of the WS467, in an effort to "see" what it was seeing.

I set the WS467 to the same housecode as my pole lamp, which is C4. Then I fired up my HomeSeer script to continually send C4 ON, sleep 5 seconds, send C4 OFF, sleep 5 seconds, send C4 ON, etc.

I first connected the test probe to the load side of the breaker feeding the lighting circuit in question. Immediately I saw new behavior on the ESM1 display: seemingly on each command, the LED bars indicated four (!) distinct signal amplitude peaks. That is, 5 bars would light, then drop to 3 or 4, then back up to five, repeating for a total of 4 peaks, then silence (which I assume is the sleep call in the script). This happened consistently at the breaker connection. I didn't think to measure at the time but I would say from memory that the total duration for the 4 pulses was around 2 seconds. It really looked like something was echoing the control signal 3 times, except that the green X-10 indicator never came on, and the WS467 never responded.

This breaker is installed about halfway down the height of the panel. At the bottom of the panel, there is an unused, exposed stake tab on that same phase. I moved my test probe to that tab. I saw the same repeating 4-peak pulses, except now, on every 3rd pulse, the ESM1's X-10 indicator came on. The WS467 began to intermittently respond to the commands (some it missed, some it responded to, about 50/50). I then flipped off the water heater's breaker. Nothing changed on the ESM1, but the WS467 "success rate" went up: most of the time it responded, but it still missed a command here and there.)

Unfortunately I ran out of time at this point, but the next test I wanted to try was to leave the ESM1 connected, and switch out the WS467, to see if its addition was causing this new, "4-peak" behavior.

I'll post my results (I feel I owe you guys at this point %^)

Thanks.

From your other post, it sounds like you're getting corrupted signals and the only way to see them is with a storage scope. You can get inexpensive USB Digital Sampling Oscilloscopes.

The 2nd & 3rd one are the same - HVW designed it and sold it under their own name until Parallax bought it.

Or you can use your scope to design an RC circuit like the one in AN236 making sure the maximum amplitude is in the 1-5V range and then use a soundcard to record it as a .WAV file which you can then view in a Wave editor. I've shown how to do this with low voltage (5V) signals from IR or RF receivers here...

You might want to add clamping diodes on the input to the soundcard. I'd try to generate a schematic but a power glitch this AM destroyed my router. My drawing program is on a different machine from the only one I can connect to the Internet so I'm limping along on sneakernet until I can find a replacement router. (D-Link DI-707P)

| I dug out an extra WS467 from my X-10 parts box. I used a small table | lamp for testing its operation, right at the panel. Using clip leads, | I wired: from ground to one prong of the lamp, from the other prong of | lamp to the black lead of the WS467, from the blue lead of the WS467 | to what I will call a "test probe". I also wired the ESM1 from a | ground to the blue lead of the WS467, in an effort to "see" what it | was seeing. | | I set the WS467 to the same housecode as my pole lamp, which is C4. | Then I fired up my HomeSeer script to continually send C4 ON, sleep 5 | seconds, send C4 OFF, sleep 5 seconds, send C4 ON, etc. | | I first connected the test probe to the load side of the breaker | feeding the lighting circuit in question. Immediately I saw new | behavior on the ESM1 display: seemingly on each command, the LED bars | indicated four (!) distinct signal amplitude peaks. That is, 5 bars | would light, then drop to 3 or 4, then back up to five, repeating for | a total of 4 peaks, then silence (which I assume is the sleep call in | the script).

Are you sure the sequence wasn't low then high? That would be normal behavior with a repeater, assuming the ESM1 is registering both the original and repeated versions. So you see C4, C4 repeated, ON, and then ON repeated (where repeated is in addition to the second copy from the transmitter). A Leviton test set would show only two because it measures only the "ON" after it is selected by its (fixed) address.

| This happened consistently at the breaker connection. I | didn't think to measure at the time but I would say from memory that | the total duration for the 4 pulses was around 2 seconds. It really | looked like something was echoing the control signal 3 times, except | that the green X-10 indicator never came on, and the WS467 never | responded.

I would try the 0.1uF capacitor across the WS467. It is possible that you have a noise source not related to the water heater but the addition of the heater circuit changes things enough to make it significant. Keep in mind that the WS467 does not have any AGC so it is the absolute level of the noise that matters (even if it is tiny compared to the good signal). Another possible experiment would be to use a lamp or appliance module rather than the WS467; the former are not as sensitive.

Dan Lanciani ddl@danlan.*com

Dan,

Thanks for the suggestions - all excellent. I'll do some more testing.

If I no longer post replies you'll know I finally fried myself sticking wires into my mains panel.

Off to the parts house. Thanks.

While there, grab a 25VAC wall transformer if you're interested in recording lengthy sequences with a soundcard. It will provide isolation while getting the line voltage down to a level that's safe for humanoids before devising a circuit to make the 60Kz and 120kHz levels safe for the soundcard.

Alternately, you can open up your ESM1 and use it as the source for input to a soundcard. John Galvin reverse engineered a schematic for the ESM1 shortly after it was introduced.

This is so obvious that I should have thought of it years ago as the need for lengthy PLC captures arises frequently. (In my defense, I usually use my DSO 'scope.) Recording them as .WAV files allows for editing and analysis in a Wave editor (e.g. Goldwave) with the length only limited by disk space. Tapping into the ESM1 really simplifies things. You can get 60Hz on PIC pin

13 and the signal on PIC pin 18, it's isolated and all low voltage. If I were ELK, I'd add a 1/8" stereo jack and a few resistors and double the price. ;)

So what exactly would I be looking at on the screen if I opened up my ESM1 and instead of ruining it (just a comment on my soldering skills or lack thereof), I miraculously was able to solder two wires to the right PIC pins and to ground (I assume) and hook them up to a miniature (or is it submini?) stereo jack. Would that output feed directly into the AUX input of a sound card, hopefully one on a laptop?

-- Bobby G.

Give me a few days and I'll try to put together a web page with "eight-by-ten colour glossy pictures with the circles and arrows". I've been really struggling lately with my spinal cord injury (I may soon need a power chair - what did your dad's cost?) and need to repair/replace the network switch that failed yesterday before I try to tackle this. I cannot access my network HDD (which is where I store pics) until that's done. Tasks which used to take an hour or two now take days or weeks.

While the details may change once I get into the nitty-gritty, I will probably recommend just soldering a jack to the ESM1. I have some 3.5mm M/F stereo extension cables that I got from All Electronics or one of the other surplus dealers. I'm sure Ratshack sells something similar. I'll cut one asunder and solder the female end to the ESM1. There's no room to mount a bulkhead type jack internally but the lead can pass through the case and I can then design a small adapter board with voltage dividers and/or clamping diodes to prevent damage to the soundcard. The male end of the cable can then connect the external adapter to the soundcard input. In this way, the ESM1 is unaltered (except for the connector dongle) when the external circuitry is unplugged.

You need to use "line-in" although an "aux" input is probably the same (I've just never seen one labelled "aux".) and I would suggest, out of an abundance of caution, that a separate sound card (PCI or USB) would be safer than the built-in soundcard emulation of most laptops (many of which do not have "line-in" but only "mic" which will not work for this.

Take your time. We purchased a few different chairs from the "fell off a truck" place. It took me about 4 months of looking actively because we needed to buy within 200 miles (until we bought the electric ramp van, that is). Street prices go from $400 up. The best deal I got was from an ISOLDIT that curiously made me sign a receipt that says "unit moves smoothly." It had sat in their store for a few weeks so I was really curious, but it did move, so I thought I could use it for spare parts at $135.

Of course, when I got home, it began to stutter badly. Fortunately, someone had installed brand new batteries (worth the purchase price all by themselves!) and the cause was a battery wire that wasn't screwed down tightly using lock washers. The slightly used, heavy duty chair would have sold for $2000+ new. Buying them new puts you in the $1600 range for the cheapest crap Medicare will pay for and way, way up to Recarro leather seats and full power lift and recline. Then there's the $30K I-Bit that can climb stairs but not fit under tables!

Sadly, there are lots and lots of chairs on Ebay that have has very little use. You can figure out why. Local pawnshops and wheelchair societies also have used ones. Most important advice is to sit in the one you intend to buy. We have four because three of them turned out to be very difficult for Dad to sit in for long periods of time because of his ruptured discs. One of my future projects is to mount a nice car seat from the junkyard to one of the units with a bad chair.

Interesting - the laptop option would make it more portable and I am pretty sure the one I have in mind has a LINE IN and if it fries I would be out the $20 I paid for it at - you guessed it - "fell off a truck" Bay! (Don't know why I called it AUX!)

Thanks!

-- Bobby G.

I did a quickie feasibility test using my scope card and just clipping my scope leads to R9 (Pin13) & C3 (Pin18). There's a screenshot at...

I forgot to include the scale on the 60Hz trace. It's 1v/DIV so this pin can probably connect directly to the soundcard. The PLC signal will need to be attenuated - the 39K & 10K resistors I show here...

should work.

I'm not an engineer so I'd appreciate any suggestions that Dan & Jeff might have.

As luck would have it, I was able to capture some noise which I see periodically. It shows up as 1 bar (my ESM1 is pre-ELK) but does not affect any of my X-10 devices (although it might if I had a weaker signal).

Using this with the techniques & software I recommend in the IR/RF link should allow you to record lengthy PLC sequences. We don't need really high recording resolution. My scope sample rate was 20K/s and we could go even lower with the soundcard to minimize disk space.

Sorry - I think this obsoletes Bobby's Monterey. ;)

If you can't get there from here, try...

Does that pattern read 011101?

-- Bobby G.

Yes, I scrolled the display to a startcode before capturing the screen.

BTW, I think the noise is from my Dell 17" CRT monitor. I'm not sure whether it's EMI affecting the ESM1 or on the powerline but the ESM1 only shows it when the monitor is awake a displaying something.

I wrote some software a few years ago that automatically scans a .WAV file, cleaning up the noise, normalizing amplitudes, outputting a CCF code and displaying the waveform in an oscilloscope like plot. That way people could capture IR & RF codes and send me the .WAV file rather than having to send me the hardware so that I could quickly convert them to CCF. If ever I start feeling better I might be able to do something similar to automatically analyze X-10 PLC codes recorded this way.

It's not going into the junkbin just yet unless you can shrink all of this down to the size of a cigarette pack with a line cord. (-: But the ability to capture raw line data to disk with just a sound card should be extremely useful in debugging certain sets of problems, particularly something like overnight phantoms and noise related events. Every weapon in the arsenal helps. I'll bet it would help solve Mr. Land's problem or at least shed serious light on it.

With the plan I had to distribute a number of wall warts around the house feeding a rotary switch and then the ESM1, this would give a very good way of being able to inspect the powerline at several places around the house quite easily and in great detail AND with a recorded log. That could be a

*very* good tool for analyzing the effect of things like signal suckers on the entire household wiring network. Before this, with no way to record or compare the signals other than through the LED bar graph display, it wasn't really workable. Now it is. Good work! With both the power feed and the I/O line on sockets, I should even be able to retain its function as a portable device.

-- Bobby G.