Line Noise Interference Question

I've been reading here:

Where they say:

"As we apply a voltage to a solid state power supply, the current drawn is (approximately) zero until a critical ?firing voltage? is reached on the sinewave. At this firing voltage, the transistor (or other device) gates or allows current to be conducted. This current typically increases over time until the peak of the sinewave and decreases until the critical firing voltage is reached on the ?downward side? of the sinewave. The device then shuts off and current goes to zero. The same thing occurs on the negative side of the sinewave with a second negative pulse of current being drawn. The current drawn then is a series of positive and negative pulses, and not the sinewave drawn by linear systems. Some systems have different shaped waveforms such as square waves. These types of systems are often called non-linear systems. The power supplies which draw this type of current are called switched mode power supplies. Once these pulse currents are formed, we have a difficult time analyzing their effect. Power engineers are taught to analyze the effects of sinewaves on power systems. Analyzing the effects of these pulses is much more difficult."

Does the chopping of current near the zero crossing of switching power supplies affect X-10 or is it the filtering of such choppiness in the power supplies themselves that causes the problem? I have a lot of switching power supplies that cause problems, but a also a few that don't. I'd like to know what differentiates the two.

-- Bobby G.

| AFAIK, all of the diagnostic tools available have similar frequency | responses. If Dan Lanciani is right, its possible to have a noise source | that will not show up on them that will still jam X-10. He said he had a | ~200kHz noise source that interfered with X-10 but did not register on the | ESM1. I'm not sure whether he was using the ESM1 that I loaned him | (fullscale=10Vpp) or an ELK model (fullscale=5Vpp).

It was a brand new, out-of-the-box Elk model. Note that the original Leviton test unit did not register either.

| It might be that it | would register on the latter but not the former. It's hard to say because he | really didn't provide much detail about the actual frequency and amplitude.

The actual frequency was just over 200kHz, perhaps 205kHz. That's as close as I can estimate looking at a scope. The amplitude was quite low (maybe

50mV--I don't remember), affecting only two-wire wall switches in their off state.

To confirm the frequency response of the Elk meter at more reasonable amplitude I used an old remote control box which conveniently generates a steady 197kHz carrier at several volts. Again the Elk read nothing but all X10 operation on the line was inhibited. (I.e., this is an even better jammer than the

200khz signal generator I had used before. :)

Finally, I reprogrammed an RR501 (using my own firmware) to use 192kHz carrier for normal X10 transmission. Its transmissions were accepted by every normal X10 receiver I tried (including a CM11a).

As I've said, the cycle-counting technique used in many X10 receivers helps only with noise of lower frequencies and the analog filters don't have that sharp a cutoff. Test tools are doing you no favor by implementing a band filter narrower than that of the actual receivers.

Last time I posted about this you said you had ordered a signal generator so you could test the frequency responses for yourself. What did you find?

Dan Lanciani ddl@danlan.*com

You're over complicating things and looking at the wrong factors.

Switching power supplies are similar, in concept, to the power supplies that were used in car radios prior to the transistor. They used a vibrating reed (designed to vibrate at 60Hz) to switch the 6V (prior to ~1955 most cars used 6V) DC voltage on/off, generating a square wave that was fed to a step-up transformer to create 120V/60Hz. A standard power supply circuit then converted this into the high DC voltages needed by the vacuum tubes used in the radio. (Many baby boomers exist because of vibrators and back seats.)

Switching power supplies have replaced the vibrator with electronic switching operating at much higher frequencies (20kHz-1MHz, no standardization). The higher frequencies mean any transformer and other components can be much smaller. They can be highly efficient (but many are not) in terms of energy use and they can use a wide range of input voltages so one PS can be used worldwide. Linear power supplies use heavy transformers and are inefficient energy users.

Switching power supplies can cause problems for X-10 in two ways.

1. They can output noise in the 120kHz neigborhood (You've seen such noise from a CFL.) which an X-10 receiver sees as continuous logical 1s at each half cycle. This tends to jam all X-10 signals. (NOTE: An X-10 receiver counts transitions in a 650µS window starting at ZC+250µS. 48 or more transitions is a logical 1. See
   formatting link
   - first paragraph.)
2. They may filter their output to prevent noise from reaching the mains. The filters may also filter 120kHz thus draining the X-10 signals.

There's no easy way to tell whether a switching PS is X-10 friendly other than trial and error. They seldom provide much information beyond input/output voltage/current.

The noise you saw from a CFL using the ESM1 is probably ~120kHz from its switching power supply. Since it's a square wave, it contains a lot of harmonics so it may be a harmonic of the fundamental switching frequency.

AFAIK, all of the diagnostic tools available have similar frequency responses. If Dan Lanciani is right, its possible to have a noise source that will not show up on them that will still jam X-10. He said he had a ~200kHz noise source that interfered with X-10 but did not register on the ESM1. I'm not sure whether he was using the ESM1 that I loaned him (fullscale=10Vpp) or an ELK model (fullscale=5Vpp). It might be that it would register on the latter but not the former. It's hard to say because he really didn't provide much detail about the actual frequency and amplitude.

John Galvin indicated that the Monterey responded to out-of-band noise that did not affect the ESM1 (again, my ESM1 on loan) but he didn't provide any further details on frequency/amplitude.

X-10 specs are for 120kHz ±2kHz so most test equipment is designed around that. With the relatively recent proliferation of switching power supplies, currently available test gear may prove incapable of detecting all the frequencies that will jam X-10.

For further reading:

|8882122804884371552/169937909/6/7005/7005/7002/7002/7005/-1|5298542958250215617/169937910/6/7005/7005/7002/7002/7005/-1&FOLDER%3C%3Efolder_id=538847&ASSORTMENT%3C%3East_id=333133&bmUID=1120478128741 NOTE: If you follow the link in the first URL to the ATX power supply schematic, the 220nF capacitor used across the mains presents 6 ohms impedance to 120kHz.

If a PS causes problems only when it is on, it's most likely a noise problem. If it still causes problems when off, it's most likely a capacitor across the mains. However, there may be cases where the capacitor is behind a switch so that complicates the analysis.

A bandstop filter (e.g. all of the X-10 filters) deals with both noise and the capacitor. It blocks noise from reaching the mains and blocks X-10 signal from reaching the capacitor.

If it's a noise problem, the noise will be constant. It will not come and go in sync with the ZC in patterns that repeatedly turn on selective lights in the middle of the night. The X-10 noise fairy does that.

From memory, I think the ESM1 was designed for ~175kHz.

Anything greater than ~74kHz should meet the 48 cycles within the 650µS window.

Obviously, the designers of the various test gear didn't look into this. I cannot find any bandwidth specs for the Monterey but the manual says noise greater than 110kHz might register as a false "1".

I haven't done anything with it yet. It's still on my list but I can't predict when or if I'll get to it. I have new health issues that are taking priority.

No doubt! But thanks for taking the time to offer guidance anyway.

That was back before alcohol was known to have serious fetal effects. I am sure combining all three elements led to the baby boom! :=)

Am I right in assuming if it runs warm to the touch, it's probably a linear power supply and not a switching type?

I understand this part pretty well. The "shouting in windy tunnel" effect. If there's a constant random signal present, it will render a structured signal of near equal strength unintelligible. One way to overcome this may be with a repeater, which would in a sense act as a megaphone in the windy tunnel.

I'm still pretty fuzzy on exactly how an X-10 filter can block both unwanted transmissions of noise and isolate signal sucking components as well. Those seem like disparate functions. From what I've read these X-10 filters perform the equivalent of unsoldering a capacitor from across the power leads. I think I'll have to read up on the design of X-10 filters before I venture deeper into unknown waters and drown.

Would it be fair to say an X-10 filter plugs a hole in the X-10 network and that hole, before plugging, would let the X-10 signal "leak" out of the net and noise from something like a bad power supply leak into the net. The tradeoff is that no good X-10 signals can reach beyond that point.

Opps. Gavin's the actor from Pyscho who became Amb. to Mexico, Galvin's the X-10 man. Erica Gavin was the actress with who starred in Russ Meyer's classic "Vixen" and had a 38" DD bustline in an age long before silicone when 38" MEANT something!

That's not heart-warming news. I assume that an oscilloscope could detect that noise even though commercial X-10 testers won't.

|8882122804884371552/169937909/6/7005/7005/7002/7002/7005/-1|5298542958250215617/169937910/6/7005/7005/7002/7002/7005/-1&FOLDER%3C%3Efolder_id=538847&ASSORTMENT%3C%3East_id=333133&bmUID=1120478128741 OK, professor. I promise to do my homework. I'll admit I got sidetracked by that copper.org detective story. It's nice to know that even professional sparkies often take several tries to get to the root of the problem. It was scary enough to make me wonder whether the person that said he saw better work from a DIY than from a licensed electrician was telling more truth than hyperbole. I'll try to stay on point. I may take a screwdriver to my ACT and X-10 filters to see what's inside. I confess to having to actually tinker with something for it to make sense to me.

What sort of impedance levels should one be seeing around the house? Is this a number that's going to vary from outlet to outlet? What little I've read is that impedance varies depending on the signal frequency. (I was just about to read some more when I realized this is exactly where I left the rails and got lost in tracing circuits in Columbus Ohio buildings . . .)

Must of today's PC's are never really "OFF" anymore unless you yank their cables.

I assume that simply removing such capacitors is not a good idea since that would likely change the power supply from a signal sucker to a noise generator. I just wish there was was a better solution to the ever-widening number of devices that impinge on X-10. If UPB can use an increased voltage, could an X-10 transmitter similarly put an increased voltage on the line. I know it violates the EU Cenelec rules, but is it technically possible? Would it interfere with other devices if the power was boosted beyond the 10V P to P of the current generation of devices?

I'll have to Google on filters to understand more. It seems the X-10 filters are doing the electronic equivalent of removing the soldered in PS capacitor from the circuit. I've got a pretty fair understanding of how SPS's create noise now.

I can only offer what I see on my ESM-1 when I leave it plugged it. From time to time, for no apparent reason that I can discern, the first bar of the ESM-1 lights up steadily with no corresponding X-10 good LED. It happened in the kitchen twice and the living room once. I turned off every light in the house but the bar persisted. It even came back after unplugging it but eventually it left as mysteriously as it came. I am hoping the Monterey will be able to confirm that there's real noise on the line and not just a metering anomaly.

The one thing the Lynx did show was how RF collisions (activating two different TM751's on different housecodes on different circuit branches via two different keychain remotes) created what they called "code fragments." Using B, F and D housecodes I was able to reliably generate code fragments that were for housecode M. From what I can deduce, Marrick avoids using "Bad Start Code" by considering all start codes as good and then labelling what follows a good command or a code fragment based on whether it's complete. If it's OK with you, I'll email a copy of the screenprint.

-- Bobby G.

Probably. But linear supplies may be cool depending on the design. Weight is probably a more reliable indicator. A 60Hz power transformer is heavy (relatively). A switching supply may have no transformer or one that is for a much higher frequency which is usually quite light. Compare the size/weight of the little transformers (silver rectangular cans) inside an X-10 module with the heavy transformer inside a linear power supply.

A repeater doesn't help if it also repeats the noise. It needs to selectively repeat only the signal. Making every module an intelligent repeater (i.e. Insteon) makes some sense as it fills the network with a

3.16V signal but it remains to be seen whether their implementation works as advertised. They also have Insteon filters and repeaters so...

They block the passage of higher frequencies while allowing lower frequencies through. There's no perfect analogy but you might view them like a furnace filter which blocks large particles but passes small particles. But, electronic filters don't get clogged.

Yes.

You would need a real analog oscilloscope. A DSO (Digital Sampling Oscilloscope) usually has low limits on voltage input levels. Its sampling rate may be too low to accurately reproduce higher frequencies. It's difficult to see a high frequency signal riding on the 60Hz as you cannot lower the baseline and the 60Hz goes offscale before the gain is high enough to see the higher frequency signal. You need to filter out the 60Hz and then the characteristics of the filter have to be considered.

I'm not sure what ACT uses in their Scope-Test2. A schematic on their web site shows a DIY version that merely uses a capacitor as a low-pass filter to block 60Hz while passing higher frequencies. That should pass the frequencies of interest but, if they use a bandpass filter, its bandwidth is the crucial parameter.

I believe you can see what's inside some of them on Ido Bar-Tana's site.

There's no simple way to measure. That's why I bemoan the loss of the web site that had all the white papers on powerline communications. There are network analyzers designed for testing 50/75 ohm coaxial networks. You can buy one used starting at about 10K.

Removing the capacitor _may_ turn it into a noise source. It depends on the type of equipment. You may find capacitors on AV gear where the intent is to block external noise from entering.

The effects of higher 120kHz voltages on other devices is unknowable. I think it's a bad idea. I also think UPB's 40V noise pulses are a _very_ bad idea that might cause problems for other gear.

No, the filter merely isolates (selectively, depending on frequency) the part of the circuit that contains the capacitor from the part of the circuit on the other side of the filter.

For a few weeks, I kept my ESM1 plugged into an outlet where it was visible out of the corner of my eye whenever I was at my desk. I saw something similar periodically for short periods and only at certain times of the day. I did not investigate further. It may be a neighbor's device or maybe the electric utility is experimenting.

The ESM1 does more or less the same. If any 1110 is followed by a sequence that appears to be valid X-10 (without any complex analysis), it lights the "X-10 Good" LED. That's also the way X-10 modules work.

I'll try to look at the screenprint if you send it.

[...]

It's been more than 40 years since I learned electronics and I never really worked in the field after leaving the Air Force in 1962. I've forgotten most of what I think (or at least claim) I once knew. I also need to "pull the wings off", etc. to understand how things work.

There are some really good web sites that help. Don't let the title of this one scare you away. It is very well organized with short segments and good illustrations.

Chapter 1 covers everything you'll ever need to know (plus a lot you'll never need to know) on filters.

I'm gonna latch onto this topic since I think my problem is also line noise or a signal sucker....

I've got a couple x10 wall switches which do not work when the PC in my living room is turned on. Note that if the powersupply switch is on, but the PC itself is not booted, things work fine. I don't know if that matters.

Anyway, I purchased the smarthome 10amp filter and have been able to verify that the filter works by plugging modules into it and verifying that i couldn't control them, but when I use this filter on the PC, there seems to be no effect at all...the x10 switches still only respond when the PC is off (but the powersupply can be on or off).

I'm assuming this is just a really noisy/signal sucking PC or something, but has anyone experienced anything like this where even a filter isn't sufficient? I've tried plugging the PC directly into the filter and then into the wall (no surge protector) and that didn't matter either.

A filter designed for X-10's 120kHz should "fix" a signal sucker so I think you probably have a noise source. The fact that you only have the problem when the PC is on also points to it being a noise source.

It may be that the frequency of the noise is beyond the bandwidth of the filter but not beyond the bandwidth of the affected switches. In another thread, Dan Lanciani has indicated X-10 modules and switches will respond to and can be "jammed" by frequencies above 200kHz. The filters may not be designed to block such high frequencies.

Many Digital Volt Meters >I'm gonna latch onto this topic since I think my problem is also line