Active Noise Reducer – Maybe the CFL Silver Bullet?

[Brian H]

Thank you for sharing the photo of the prototype.
Looks well designed.

[systemdm]

Thanks Jeff.   When do you think the kits will be available?

[William8]

Would this reduce noise only on the circuit it's plugged into, or the entire phase? I have one circuit that won't work when my Panasonic microwave is just plugged into the wall. That entire phase stops working when it's actually running. It's a new issue, just got the microwave a month or two ago.

I would be very interested in a simple plug in solution. I know I can get a 15amp plug in noise filter, but they are huge. The microwave is installed in a custom cabinet, and there isn't enough room for it behind the microwave.

[JeffVolp]

Thanks Jeff.   When do you think the kits will be available?

A few beta units should be available in a couple of weeks to those of you with serious noise problems.  Assuming they work as anticipated, the first batch will only be available as assembled units because these PCBs don't have silkscreen or solder mask.  Kits with silkscreened PCBs should be available in a couple of months.

Jeff

[JeffVolp]

Would this reduce noise only on the circuit it's plugged into, or the entire phase? I have one circuit that won't work when my Panasonic microwave is just plugged into the wall. That entire phase stops working when it's actually running.

It severely attenuates noise on the circuit it is plugged into.  When plugged into a receptacle near the distribution panel, it will attenuate noise on that entire phase.  But if the noise source is a long run from the panel, there may still be significant noise on that circuit near the source.

For a noise source like you describe, it should work effectively if plugged into the same receptacle as the Panasonic.

The Tuned Signal Sucker that I developed some time ago might also do the job.  It is just a passive device that severely attenuates out-of-band noise, and also attenuates in-band noise, but not as much.  Contact me privately if you would like to try a beta unit.

Jeff

[JeffVolp]

I have been testing the prototype, and words just don't do justice.  Here is a link to a video clip that shows its performance against a continuous in-band noise source being injected right onto the powerline.

Update - Changed to a low frame rate to get down to 9MB, but it still may take a while to download.

http://jvde.us/info/XTB-ANR_demo1c.avi

You can see on the XTBM that the XTB-ANR causes about a 20% reduction in the X10 signal voltage, but the noise level is reduced about 90%.  So there is a very significant improvement in the signal-to-noise ratio.

Update:  Measuring signal loss from my XTBR repeater on the other phase showed the signal attenuation was much less than 20%.  Without the XTB-ANR, the voltage read 2.11Vpp on the XTBM, and 1.95Vpp after the XTB-ANR was plugged in.  That is only an 8% reduction in the signal voltage, with a 90% reduction in the noise voltage.

Jeff

[Brian H]

Very impressive.
I hope the beta tests go well and the production models will be available shortly after that.

[JeffVolp]

I hope the beta tests go well and the production models will be available shortly after that.

I'm looking for a few beta testers who are plagued with serious noise problems.  If interested, contact me privately.

Jeff

[Oldtimer]

Jeff: 

Very nice piece of work!

[JeffVolp]

My Insteon controller arrived today, and I tested how well the XTB-ANR handles those signals.  It is working exactly as it should, and turns off the attenuation for the duration of any Insteon signal bursts.  So it should have very little effect on Insteon commands in a combined X10/Insteon system.

I did find one unexpected result though.  If I inject noise near 131KHz directly onto the powerline, the XTB-ANR can make that look like a series of Insteon commands even though they don't contain any actual data.  As a result, the XTBM displays that noise as potential Insteon commands. 

Since the XTBM just looks at the signal envelope, I don't think it will be possible to discern the difference between a real Insteon command, and noise that has the same envelope due to actions of the XTB-ANR.

Jeff

[dhouston]

Jeff,

If you can sense a phase shift, you can probably tell noise from signal but maybe not in time to do much about it.

I published these scope screen shots shortly after Insteon was introduced...

• http://davehouston.org/insteon.htm

Hmmm, maybe you can handle it - it's been too long for my ancient brain to recall detail but I think all commands start with a 1 followed by 0 to establish phase references.

Also, you might be able to take advantage of the fact that the powerline is clear for 1/2 cycle after a code in order to relay it via RF to the other phase. That, however, may have changed as they've gone through 3-4 design changes since it was introduced.

[JeffVolp]

Thanks for the suggestion Dave, but unfortunately that is not possible in the XTBM.  There is a precision rectifier that tracks the envelope of the signal burst.  The PIC A/D is running flat out over the burst making multiple readings, which it converts to an average.  Independently, the T1 timer counts the number of cycles in the measurement window to calculate the frequency.

The XTBM identifies an Insteon signal by it being at the right frequency, and the burst beginning almost a millisecond before the zero crosssing.  The XTB-ANR switches off the attenuator at that point so not to corrupt an Insteon signal.  As a result, noise sitting near 131KHz looks just like an Insteon signal.

Since the XTBM is essentially a X10 measurement tool, we will have to accept the fact that after the XTB-ANR is installed, the Insteon indication could mean there is noise near the Insteon frequency.

You are correct in that I could use the 5 zero-crossing bursts, followed by a gap to qualify for the Insteon indication.  But extended Insteon commands take 11 zero crossings before the gap.  So 5 or 11 could be potential Insteon commands, and anything else would be noise.  I’ll think about including that in the Pro version.

Jeff

[Brian H]

Insteon White paper says 800 us before zero crossing.

[JeffVolp]

Insteon White paper says 800 us before zero crossing.

Correct.  To allow for possible differences between the zero crossing detectors, I back away from that 800uS transition to make sure there are clean samples before and after the Insteon signal switches on.  That works perfectly because it accurately recognizes the Insteon command and leaves the attenuator off for the entire Insteon burst.  It must switch the attenuator back on before the zero crossing if there is no Insteon signal to guarantee a good X10 zero if there is no X10 burst.

Jeff

[JeffVolp]

I have been doing a lot of testing with the XTB-ANR.  It works well for most ordinary noise sources.  However, since noise as low as 30mVpp centered right at 120KHz can block a X10 module, it may still be necessary to isolate a powerful noise source like that with an in-line filter.  The Lumoform 4W LED light I reported on in the past is one such noise source.  It is perhaps the worst one that I have run across.

The Lumoform pegs the XTBM (over .99Vpp) when plugged into the same outlet, and its frequency ranges from 121KHz to 123KHz in the X10 transmit window.  My digital scope shows the frequency does vary throughout the AC half cycle, but only what happens inside the zero-crossing X10 transmit window is important for X10 communication.  Plugging the XTB-ANR into the same outlet as the Lumoform pulls the noise level down to .20 to .30Vpp at that outlet, but that was not low enough for an adjacent X10 appliance module to respond.

Moving the Lumoform to another outlet on the same circuit across the room reduced the XTBM reading to .56Vpp.  Then plugging in the XTB-ANR reduced the XTBM noise reading to .05Vpp (50mVpp), which was then low enough for the adjacent X10 appliance module to respond.

In summary, the XTB-ANR can help even with a powerful in-band noise source, but only if there is some separation between the noise source and the X10 devices.  A powerful in-band noise source may still need to be isolated if adjacent to X10 devices.

The beta XTB-ANRs are out there now, and I am waiting to hear back results.

Jeff