X10 Interference from the Cellet Cellphone Charger

The X10 Good LED only checks for 9 ones and 9 zeros so something like

1110111111111000000000 or 1110000000000111111111 might cause it to light.

First Draft:

Comments and questions appreciated.

Jeff

It looked like it was pounding out a pretty good series of solid "1"s.

Jeff

Thinking more about the ESM1 rather than the XTB-II, I believe that uses envelope detection. I guess it depends on where it samples the envelope. If the ESM1 ZCD is slightly skewed to one polarity, it might catch alternating 1's and 0's. That doesn't explain why that green LED stayed on most of the time, since the start pattern needs three 1's in a row.

Jeff

If you look at the screenshots at

'll see the ZC signal is skewed. It just uses the 12VAC signal on a PIC pin (through a 33K current limiting resistor), relying on the PIC clamping diodes to limit the voltage. It's really not terribly sophisticated.

Looks like an INSTEON killer too (INSTEON self-repetition/signal boosting notwithstanding).

And suggests another product that would provide a public service and have some commercial potential.

Up to now most folks plug new AC devices into the powerline 'communication medium' in their house and then retroactively try to determine what the problem is with X-10 and(or) INSTEON as noise and signal attenuation accumulate, in varying degrees, with each device.

A testing tool could provide a more proactive approach by quantitatively assessing potential detrimental effects _before_ plugging a prospective new electrical device into the 'system'.

It c/would incorporate an isolation filter, noise measurement, signal attenuation and visual indicator in one unit. Presumably the 'signal attenuation' portion would consist in a transmitter and a receiver that would register 100% in the absence of a signal sucker.

For this purpose, the isolation and power to the DUT and the tester itself could be provided by a pair of low-voltage transformers with their secondaries connected together (with or w/o a capacitor). I've played with this arrangement with INSTEON and it works well for isolation (and FWIW, for transmission of INSTEON/X-10 signals over a low-voltage AC line when the transformers are bypassed with capacitors and the secondaries connected with up to 200 feet of Cat5).

The test function might be incorporated into a future version of the XTB-xx transmitter with the front AC outlet that is usually just a pass-through being the outlet into which one plugs in the Device Under Test. Indicators could be as simple as one row of LEDs for attenuation and one row for noise. Or a small LCD or via RS-232.

...Marc

Marc_F_Hult

It may be time to migrate to a physically larger form-factor than the CM11a/ACT-TI103/XTB wall-wart model.

Using an enclosure based on an AC outlet power strip would provide needed volume for filter/isolator and circuitry and allow the outlets to be used for different purposes including device testing as discussed above, X-10 isolation ( aka "filtering") and even ON-OFF switching. The latter is a common need for controlling (eg) routers and cable/dsl 'modems'. What better place to add an X-10 or INSTEON appliance module for that purpose than immediately adjacent the transmitter on the protected ('firewalled') side of an isolation device (ala Lightolier Compose).

The volume made available by using a power strip might allow for incorporation of existing XTB-xx PCBs with little modification -- add-on PCBs providing additional functionality. The extra volume would also allow for use of a conventional L-C-L-C-L filter ala X-10/Leviton and ACT that could be available through one or more of the AC receptacles on the strip.

... Marc Marc_F_Hult

For the record, the Cellet charger that I sent to Jeff never created any problems for my Insteon devices. It wasn't till it was plugged in at our other home equipped with 10 year old WS12A switches that the problem arose and even then it was localized to a couple of switches just a few feet away from the charger and presumably on the same circuit.

Marc_F_Hult wrote: snip

snip

While Insteon claims 10mV sensitivity, which would tend to make it more sensitive to noise, it also has AGC which, in the presence of a stronger signal, raises the threshold and suppresses noise. The repeater feature of each Insteon module means there will always be a relatively strong (local) signal (the more Insteon devices, the more ubiquitous the signal) when compared to X-10. And Insteon uses Phase-Locked-Loop (PLL) which can recover weak signals in a noisy environment. PLL also means Insteon has a narrower bandwidth in comparison to X-10 devices (some of which can be blocked by frequencies as high as 200kHz).

The bottom line is still what I said in my initial review of Insteon - while Insteon is not immune to noise, noise will be less of a problem than it is with X-10. OTOH, signal sinks will still be a (largely localized) problem for both Insteon and X-10.

Note also that Jeff has not documented a valid spontaneously generated X-10 signal from the charger. It's still more likely that the noise somehow operates directly on the microcontroller in the wall switches.

Jeff has implied (on the JDS forum) that the "X-10 Good" LED is evidence of "spontaneous generation" but I think it's stronger evidence that the ESM1 does not check for manchester validity. Of course, I'm further inclined that way since Paul Beam, who designed it, told me that it doesn't check manchester validity. It might even be that the same phenomena that cause the wall switches to turn on also cause the "X-10 GOOD" LED to turn on.

I'd like to see the relationship between the 60Hz signal and the charger noise. A lightly loaded SMPS tends to draw current near the peaks of the

60Hz, midway between ZC po>For the record, the Cellet charger that I sent to Jeff never created any

Yesterday I added a couple more photos at different loads to the report, but I did not include one totally unloaded. In that case, there were narrow bursts spread randomly throughout the half cycle, not just at the peaks where you might expect.

I also didn't take a photo with a 10 ohm load, but that spread the 125KHz throughout. Amplitude was max just after zero crossing, falling off to about half that throughout the remainder of the half cycle.

Jeff

Some interesting suggestions Marc.

I don't know what is down the road. I'm getting pretty burnt out right now. My ToDo list includes porting some of the "R" features back to the 8-pin PIC in the XTB-II. And there are a couple of features I had wanted to include in the "R" firmware that were put on hold to just get the unit done. One is a possible way to repeat extended commands while avoiding the overlap issue.

The RS232 link is something I had thought about. There is so much info available in the XTB-II relating to line noise, traffic, etc. It would be nice to be able to access that information. However, since the unit is referenced to the powerline, it would take a fair amount of additional circuitry for an isolated RS232 port.

Having looked at what is in the ESM1, and the fact that it costs as much as it does, I had thought about making an alternative X10 test unit. If I ever do that, features such as you describe could be included.

FYI, the receptacle on the XTB & XTB-II is powered through a L-C-L filter to totally isolate it from the powerline. That is essential to prevent a feedback loop when the XTB or XTB-II is transmitting. Since it was only intended to provide power to a X10 transmitter, the inductors are sized for that current.

Jeff

(some snipped)

the overlap issue.

One approach that might at first appear to be a kludge, but I think has merit, is to use coupling through a custom LED-phototransistor interface. One could use a standard MF DB-9 connector shell pair with an LED and phototransistor on each side for TX and RX respectively. Capabilities built in to the XTB-xx could be made accessible if the owner also sprung for the custom/oddball computer-side adapter. The computer-side adapter could steal power for its LED from the hardware handshake lines and be available as USB or RS-232 (Easy for me to say because I'm not gonna do it ;-)

Seems to me that a HOW-TO on a qualitative setup to test device attenuation and interference on X-10/INSTEON signals would be useful. I may cobble together one with a pair of transformers connected by their secondaries for isolation, a CM11a for signal generator and ESM1 and ACT scope for receivers. I already have a test-stand of sorts at the entrance panel where the X-10/INSTEON couplers , PLC modem/transmitters and ACT Scope-Test2 reside. One of the receptacles could be dedicated to test mode.

...Marc Marc_F_Hult

I recently added what I thought were going to be "X-10 filters" to the relay controlled air-gap disconnect and filter unit that I use to protect and isolate (i.e., "filter") X-10 and INSTEON ~120khz signals from the AC circuit used for audio and instrumentation.

These are X-10 PR brand , model " XPF 20-amp Wired in-Filter " with a documentation sheet dated 10/98.

To my surprise, they are band-stop, rather than low-pass, filters.

The low-pass part of the design is L-C-L-C-L with 7.7 uH ferrite stick inductors (measured at 1 kHz) in series and 1.0uF capacitors to neutral. But the three inductors are each bypassed with 0.22uF capacitors. My need and intention is to filter high frequencies from the power line so I removed them.

Interestingly, ACT's popular AF120 filter uses an L-C-L configuration with

8.0 ufd toroidal inductors, 2.2uF cap that presumably has a lower cutoff frequency and lower slope. It also has 0.22uf bypass caps on each inductor. There is a 1-amp fast-blow AGC fuse in series with the 2.2uF capacitor's signal path to neutral. The AF120 is marked as "No User Serviceable Parts Inside Tampering Voids Warranty" on the sticker/label that serves to seal the case. So if the internal fuse were to blow, filtering performance would suffer but the user would have no indication of decreased attenuation. (The AF120 also has a user-accessible 15amp fuse in series with the load but that's not what I am referring to.)

I haven't done a SPICE model or other calculation of these devices/ schematics in part because I haven't located the reference that I had on measured/estimated typical impedances of household AC networks/circuits.

My question is why are these filters bandstop and not lowpass? Is this for circuit protection as implied by AF120's internal fuse. Or to purposely transmit X-10 across the filters? Or ?

... Marc Marc_F_Hult

various snips

('firewalled')

Several X10 filters are bandstop, including the little Leviton 6287. You can make a pretty efficient 120KHz trap with a small inductor and capacitor. However, a low-pass needs much larger components to have the same attenuation at 120KHz. The Leviton 6288 and XPPF are low-pass with an initial break at 13Khz. As you know, they are rated 5A, but run warm (and stink) when pushed that hard. FYI, the AF120 schematic I have shows the inductors are 7.2uH, and all caps are 0.22uF. The fuse may be needed for UL approval if the center cap is not rated for across-line applications. Interestingly, the center cap on the little 6287 is listed as 1.0uF.

Since these filters were designed as companions for X10 modules, they probably take advantage of the bandpass characteristics of the modules themselves. The module's input network will attenuate out-of-band signals, so the in-line filters only have to block in-band noise.

Jeff

Thanks Jeff. It makes sense that folks interested in a clean X-10/INSTEON signal would optimize the depth of the notch at 120/121khz.

The 20-amp version of the CORCOM R Series filters that I use

\CORCOM R Series schematic.htm

in my air-gap switch and power conditioning unit

\power_conditioning.htm

is down 44db at 500khz but only -4 db at 125khz. This should clean up any X-10/INSTEON 2nd harmonic, but doesn't do much for the fundamental and lower. I would be better off leaving the 0.22uf caps in place in the

8-component X-10 filters I am adding if my only concern were to attenuate the X-10/INSTEON fundamental but there's also sub-120khz garbage from triac dimmers and various switchers so I think I'll leave them in.

Folks looking for a less expensive way to filter small, noisy power supplies like the Cellet Cellphone charger than the standard X-10/Leviton/ACT options might consider the smaller (eg) CORCOM filters or equivalents. The 1 and 3amp versions of the filter series I cited are spec'd as 25-30 db down at 125khz and so should do a good job cleaning up in-band X-10/INSTEON noise. They are typically available as New Old Stock for a couple of bucks on eBay. I have a big box full I should put up for sale in my porch sale.

However, the capacitor input (C-L-etc) filter versions such as the CORCOM's I cite and use may attenuate significantly ("signal suck") compared to the inductor-input (L-C-etc ) of the commercial X-10 filters unless they are isolated by an inductor in a cascade of filters as in my case.

... Marc Marc_F_Hult

I used those Corcom filters in most of the industrial automation equipment I built over the years. As you say, some of the capacitive input filters are not at all friendly to X10. The low-pass XPPF 5A plug-in X10 filter can be found on eBay for about 10 bucks including shipping, and even less if you buy several.

That air-gap is an Interesting idea on isolating your system. We have a Leviton surge protector at our main panel, but I haven't found any "earth" ground for the house. Our water main is plastic due to the alkalinity of the soil. All utilities come in underground, and the ground reference must be out at the underground utility bunker, if there is one at all.

Jeff

For a residential ground, steel rebars encased in concrete would serve well in alkaline soils that would otherwise corrode scratched, copper-plated steel ground rods placed directly in the soil. Solid copper rods would work too but are too soft to install in hard or rocky soils. Google: Ufer ground The air-gap isolation is used in combination with other isolation techniques. My HA PC and Web Server, for example, are in a surplus equipment rack that is on rubber wheels. Ethernet to the outside world from the rack is over fiber. Serial RS-232 and RS-485 comms are via ethernet-->

RS-xxx hubs on the other (grounded) side so they too are isolated from the rack. And audio to the AV rack is also fiber. Most of this is done with high-quality surplus gear that is a generation or two out of favor. EBay reigns in equipment recycling.

I found a 24vdc-input PC ATX power supply for the HA PC. (It isn't big enough for the latest power-hog CPUs and so it has placed a welcome, arbitrary cap on power consumption. Sometimes restraint is imposed on us by externalities and we are wise to acquiesce ;-) It is powered by a pair of transformer-based 12vdc chargers that are also connected to a pair automotive-sized, deep-cycle batteries. So the power supply is galvanically isolated even if the air gap relay and SSR aren't opened.

What's still missing is a lightning detector to isolate the system automatically -- hopefully even intelligently and not doing the right thing at the wrong time ;-)

... Marc Marc_F_Hult