X10: On again, Off again

• Measure X10 signals
• Measure power-line noise
• Help find noise sources
• Help find "Signal Suckers"

• Allow oscilloscope to view and measure X10 signals
  
• View and measure power line noise
• Safely isolate the power line voltage
• High-pass filter to reject 60 Hz and harmonics
  
• Bandwidth about 10KHz to a couple of MHz
• Quick, cheap, and easy to build.
  
• I want it NOW
  

I threw this circuit together with parts I had on hand using the "Ready. Fire. Aim." approach to circuit design. I guessed that 10 turns per winding was about right for a 120KHz transformer. A toroid would work, but I didn't have any small toroid cores. So I used an old 1970's pot-core I had: Ferroxcube 2213,  3B7 ferrite, no gap. Any modern ferrite with Al of 4000-6000nH would work. I had a 3 winding bobbin which allowed the 2 windings to be separated, but a 2 winding bobbin would also work fine. I used 30AWG wire-wrap wire for the windings, figuring 2 layers of wire-wrap wire insulation would easily withstand 120VAC plus power line surges. I taped the core half's and bobbin together. The transformer inductance is 460uH, which should be fine. I tested the transformer frequency response using a function generator 1V p-p sine wave, and a scope. I had some film caps, 0.022uF, 400VDC. The circuit did an excellent job of isolating and reducing the AC 120VAC, 60 Hz. I started with a load resistor R1 of 1K. 499/510  ohms is better. 

The response at 100KHz is very good. I measured the 20KHz response and it was about 1/20 the 100KHz, so about -26db. There is a bit of resonance, so it's not really a high-pass filter. So it doesn't measure noise well, but does a good job at X10's 100KHz.

To scope the raw power line voltage, I just use a second scope probe on the raw AC Line (in x10 mode. Confusing, huh). Not the safest, but if I build this into a proper box, I'll add a voltage divider and just BNC outputs so I won't have to mess with probes. This is not a super-portable solution, but since modern scopes are small, it's not so terrible to lug one around the house to make some remote X10 measurements.



Here is a scope shot of X10 activity. This thing works great! This test is in my lab, where I have 4 power strips (signal-suckers) on 2 outlets. X10 works within this room, but the signals are not strong enough to pass to devices on the other phase. Note that at this scale, the X10 bursts cannot be seen at all on the 120VAC line. They are there, I promise. The test circuit is doing a fine job of filteing out the 60Hz. Also note the spike and other low level power line noise.



I was surprised to see three 1mS bursts per half-cycle, separated by 1.7mS. All of my X10 sources do this. After a bit of head scratching, I realized that this is to accommodate 3-phase power. These 3 pulses line up with the zero-crossings of 3 phase 60Hz power.  I have searched and searched for an X10 specification, and found none. Where is a bloody X10 electrical spec?

The signal above is about 500mV p-p. If you measure closer to the X10 source, the signal is about 2V p-p.

A lot has changed since the '70s

When X10 was designed in the late 70's, a home probably had no electrical loads that used switching power. Nowadays every wall wart, LED light bulb,TV, PC, charger, and dozens of other devices contain switching power supplies that operate between 40 and 200Khz. If they have a (valid) CE or FCC sticker, they should be OK. It's a miracle that X10 works at all. We do have CE testing, so thanks for that. Unfortunately, for nearly every switcher, there is one or more X safety rated capacitor across the line to filter the switching noise. These act as "Signal Suckers". Their low impedance at 120 KHz reduces the X10 signal. 

BTW I recently read the excellent The Art of Electronics X Chapters which has a section on p477 on TVS use in power strips, showing several power strip implementations.

I found that the main cause of my X10 loss is my numerous power strips. I use several large 12 outlet strips, and each one has a surge protector and filter. Here is a simplified schematic of such a power strip. The basic, non-surge, 6 outlet power strips generally have no filter or surge components.



X10 signals are carried on the line and neutral as "normal-mode" signals. The common-mode choke is there to reduce common-mode noise, and does not significantly reduce normal-mode (differential) signal of X10. The X safety rated, across the line, capacitor CX is the only component in the circuit that causes a significant reduction in X10 signal. Here is a table of capacitor impedance vs. capacitor values at 120 KHz, the X10 frequency, calculated using Xc = -j  / (2 * pi * 120KHz * C).

• Belkin, 12 outlet, surge, 315nF
• Belkin, 12 outlet, surge, 450 nF
• APC, 7 outlet, surge, 460 nF
• Universal, 6 outlet, surge, Model 71645, 24nF
• Amazon Basics, 12 outlets, surge, Model S9NC01RB00, 932 nF
• Four different 6-outlet power strips, all < 1nF

• Brother color laser printer, 553 nF
• Canon flatbed scanner: 210 nF

I measured about 10 different switching power supplies from CUI, TDK, Meanwell, and Condor, plus a few PC power supplies. They all measured between 500 and 1200nF. Any of these is a significant signal-sucker.

I measured a handful of small phone and USB chargers and they all measure about 100pF. Apparently these do not contain large X capacitors.

I removed the 470nF X cap from a Belkin power strip that powers several X10 modules. X10 works better, but not on both sides of the power line though. This was not a simple operation. Many power strips are held together with one-way screws.  You need to push hard to get them out. I try to replace these with regular phillips screws. Here is the Belkin strip I modified. 



Here is the surge board with TVS covers removed. The 470 nF X-cap is the large grey rectangle. There are 6 orange TVS and 3 blue ones. Not sure how exactly they are wired.



Ready to re-install with the X-cap removed and the funky TVC covers back in place.

FilterLinc

I added a FilterLinc X10 Filter, model 1626-10 to my Lab / Office Computer circuit. This circuit is one of my worst Signal-Suckers with 2 power strips, a PC, Laser printer, Scanner, Monitor, and a host of other stuff. The results were excellent! X10 now works much better in the office and and even couples to the other phase throughout the house reasonably well. The X10 signals in his room and on this circuit went from marginally visible to quite strong.

I opened the FilterLink and it has high quality components, a proper fuse, and good construction. The design is a T- filter with 2 17 uH inductors and a considerable capacitor: 2.2uF. There are 2 capacitors bridging the inductors.

Conclusions

Building a DIY X10 receiver??

• Measure and quantify line noise and X10 signal strength
• Decode all X10 commands
  
• Open hardware and software using Arduino CPU
• Reasonably low cost
• Do both isolated and non-isolated versions
• Be the front-end for various controls: Relays, AC Light dimmers, DC light controls
  

• Basic diode
• Precision rectifier
• Integrated AM demod (AD8361, 8361, etc.)
• Mixer / demod  (NE602 / SA602A)
  
• Synchronous demod
• Diode bridge
• CMOS Switch (Tayloe demodulator)
• Analog Multiplier

• Uses a low-cost CMOS 4:1 analog mux such as 74hc4053
• I and Q outputs filtered by simple capacitor or RC
  
• 120KHz I and Q clocks can be 480KHz oscillator plus dual D flip-flop
  
• I and Q outputs then filtered and digitized by slow (1mS) 10 bit ADC on Arduino
• Amplitude detect uses simple Euclidean detection (square root of sum of the squares of I and Q) in firmware
• X10 signal threshold is set using baseline noise level
• Decoder uses time filter by knowing when in the power line cycle the X10 signals should appear
• This approach also directly measures signal and noise levels
• Provisions for serial port monitoring, and small (OLED) status display
  

An isolating transformer can be fairly simple, consisting of a toroid core with a reasonable number of turns of insulating wire. This can be used for both receiving and transmitting x10.

Using an Arduino allows fancier and better dimming control such as LED trailing-edge dimmers. These use FETs instead of Triacs and allow better control of LEDs. Here's a video on these types of circuits. And another Youtube. The circuit is more complex than a simple Triac dimmer, but is more flexible. This could make possible X10 color lights, 12V lighting...

References