X10 Community Forum

General Home Automation => Other Automation Systems => Powerline Control Reliability => Topic started by: JeffVolp on December 03, 2011, 10:31:02 AM

Title: Possible New Filter Design
Post by: JeffVolp on December 03, 2011, 10:31:02 AM

Most of you know I have been working the X10 signal integrity issue for several years.  The signal level problem has been pretty much solved with the XTBR and XTB-IIR.  I’m now trying to address the noise problem that is affecting more and more of our installations as things like CFLs, LED lights, and noisy “wall-wart” power supplies proliferate.  And now we have a new attack on our system reliability from the smart meters being installed by utility companies.

Pretty much everyone is familiar with the inexpensive X10 XPPF filter.  While the going price now is around $12, it had been available through last summer for under $5.  (That is about the price of just the plastic case here in the U.S.)

The XPPF is a low-pass filter, and works very well blocking noise and isolating signal suckers.  However, even though rated 5 amps, people have found it will overheat when pushed that hard.  So it is a good idea to stay under 4 amps.  Fortunately, that covers most of the “problem” devices in our homes today.

The higher current filters (XPF from X10, and various units available from SmartHome and ACT) are all notch filters.  While they can be effective blocking noise right at the X10 bandpass, they will not sufficiently attenuate very strong noise that is just outside the X10 passband.  As a number of people have reported, that can still interfere with X10 communication because X10 modules are not very frequency selective.

I am thinking about offering a 10 amp filter that combines a notch to block high-current signal suckers, and a low-pass to block broadband radiated noise.  The problem is that components alone are much more expensive than the XPPF.  A kit using a bare-bones PCB (no solder mask or silkscreen) would cost about $30, and an assembled version about $10 more.

Before I head down this path, I thought I should get some feedback as to whether there would be enough interest to make it economically viable.

Jeff
Title: Re: Possible New Filter Design
Post by: Knightrider on December 03, 2011, 02:47:12 PM
Jeff,

I'm having a hard time wanting to improve my x10 system, if the module supply runs out.  If I can't expand my setup, I'll have to look into other technologies.  Not sure I would build new filters for a legacy protocol.

It could be, in the coming days, that the X10 pipeline will flow again, but I'm not placing any money on it.  I could also be that Smarthome saves us all with their X10 only devices, but I fail to see any advantage to that.  Is A10 even around anymore?
Title: Re: Possible New Filter Design
Post by: JeffVolp on December 03, 2011, 03:50:48 PM
While testing the XTB-232, I bought a 2-way SmartHome Insteon/X10 module.  While expensive, I was impressed by its capabilities.  The automation software (HCA in this case) knew whenever it was turned on locally.  And it responded to query checks, indicating the actual dim setting (in the preset dim format ignored by ActiveHome).

Since SmartHome is apparently offering X10-only devices again, we should continue to have a source for X10 modules.  So even if the company X10 disappears as a source for X10 equipment, we should be able to maintain (and improve the performace of) our systems for years to come.

But the underwhelming response to my filter query leads me to believe this is not a path to pursue.

Jeff
Title: Re: Possible New Filter Design
Post by: Brian H on December 03, 2011, 04:04:31 PM
Would the filter be so sharply tuned. The Insteon 131.65KHz. power line signals would be effected?

Title: Re: Possible New Filter Design
Post by: JeffVolp on December 03, 2011, 06:21:50 PM
Since X10 is very sensitive to in-band noise, maximum rejection and minimum signal loading would be right at the 120KHz bandpass.  But since it would be a low-pass filter, it would block noise over a wide range of frequencies.  For example, only 1% of noise at 80KHz would leak through, and the rejection improves as the frequency increases.

Tuning will be an issue.  Large low value inductors have relatively poor tolerance (typically +/- 20%).  So there will have to be a selection process during kitting to match the correct tuning capacitors with each inductor for maximum rejection centered at the X10 bandpass.

Jeff

Title: Re: Possible New Filter Design
Post by: William8 on December 03, 2011, 10:29:40 PM
I'd be interested, but it would have to handle 15 amps, for my troublesome microwave.
Title: Re: Possible New Filter Design
Post by: Myronbg on January 19, 2012, 03:06:17 PM
Where is the filter placed? Would a licensed electrician be necessary?
Title: Re: Possible New Filter Design
Post by: JeffVolp on January 19, 2012, 07:48:42 PM
Where is the filter placed? Would a licensed electrician be necessary?

It would be a plug-in module like the XPPF, but bigger.

Jeff
Title: Re: Possible New Filter Design
Post by: Myronbg on January 20, 2012, 12:47:22 PM
thanks, but "plugin" where? I "assume" not a plugin to an electrical wall plug. Somewhere between the supply and the power panel?
Title: Re: Possible New Filter Design
Post by: JeffVolp on January 20, 2012, 01:56:52 PM
thanks, but "plugin" where? I "assume" not a plugin to an electrical wall plug. Somewhere between the supply and the power panel?

Yes, it would plug into an electrical outlet like the XPPF.  But because there has been very little interest in this, it is on the back burner for now.

Jeff
Title: Re: Possible New Filter Design
Post by: JeffVolp on April 22, 2012, 01:24:29 PM
Most of us use a number of XPPF filters, and we know to keep well under its 5A rating.  The XPPF is a very good low-pass filter.  The other higher current filters that are available are notch filters that block signals right at the X10 passband, but let lower frequency noise pass through.  Simulations show the ACT AF120 has a big pass-through spike around 86KHz, which is where the Echelon smart meters are transmitting.

I'm working on a higher current low-pass filter.  It would be rated for at least 10A.  The problem is that the components will cost about $25.  So a kit would probably be $29, and an assembled unit $39.  I wonder if there would be enough demand for such a unit to make it worth ordering PCBs.

Jeff
Title: Re: Possible New Filter Design
Post by: Brian H on April 22, 2012, 01:32:08 PM
Well the Smarthome 10 amp FilterLinc is $29.99. So you are in the same price range as the FilterLinc and I believe it is a notch style.
Title: Re: Possible New Filter Design
Post by: JeffVolp on May 02, 2012, 03:07:40 PM
In case anyone is interested, here is a frequency sweep showing the response of the XTB-F10 and the X10 XPF.  The sweep is linear from 20KHz to 220KHz, with 120KHz (the X10 frequency) at the center.

(http://jvde.us/info/F10_XPF_sweep.jpg)

The excitation scale factor was 2V per cm.  The filter outputs were 200mV per cm.  A resistor was inserted in series with each filter to prevent loading down my signal generator and causing interaction between the two filters.  (Without the resistor, the peaks on the XPF are much larger as that series-resonant circuit sucks energy from the signal generator.)

Jeff
Title: Re: Possible New Filter Design
Post by: Brian H on May 03, 2012, 06:35:44 AM
Looks very interesting.
Thanks for your continued work on improving X10 reliability.
Title: Re: Possible New Filter Design
Post by: Noam on May 03, 2012, 08:24:31 PM
Looks very interesting.
Thanks for your continued work on improving X10 reliability.


Not being an electronics wiz (although I have been dabbling in the Arduino for the past few months), I didn't understand anything in Jeff's post past the "In case anyone is interested" part. But the picture is pretty!  ;)
Title: Re: Possible New Filter Design
Post by: LouisJr on May 06, 2012, 03:08:08 PM
I use quite a few XPPF filters and several AF-120 filters.  This filter looks like it would cover more problems than the AF-120 at enough current for some of the bigger loads like UPS units, home theater power buss, etc. up to maybe 1,000 watts?  I have smoked a few XPPF filters at what I thought were acceptable current levels.  Jeff, I guess you will post if and when you decide?
Title: Re: Possible New Filter Design
Post by: JeffVolp on May 06, 2012, 03:24:25 PM
Jeff, I guess you will post if and when you decide?

I expect to receive the first batch of PCBs in another week, and the filter should be added to my product line around May 20.  It will be fused at 10A to handle most applications.

Jeff
Title: Re: Possible New Filter Design
Post by: JeffVolp on May 10, 2012, 06:08:13 PM
Here is a photo of the prototype:

(http://jvde.us/info/XTB-F10_prototype.jpg)

Jeff
Title: Re: Possible New Filter Design
Post by: Brian H on May 10, 2012, 06:32:24 PM
Jeff; Thank you for the update on the filter project.
Looks good and I am sure it will be a great help to X10 users with noise problems.
Title: Re: Possible New Filter Design
Post by: JeffVolp on May 23, 2012, 10:58:26 AM
Update:  These are now listed on the XTB Ordering page both in kit and assembled form.

I am also considering a higher current version with two receptacles on the cover if there is enough interest.

Jeff
Title: Re: Possible New Filter Design
Post by: JeffVolp on November 09, 2012, 12:53:17 AM

There have been several requests for a higher current version of the XTB-F10, and I will be offering a 15A version.  The in-band filtering of both filters is similar, but the out-of-band performance of the F10 is slightly better because of its lower Q.  That is a function of the values available for the higher current inductors.

The higher current version does not contain an internal fuse.  Actually, the inductors are rated in excess of 15A, and the limit is due to the AC receptacle itself.  (There is not sufficient clearance for two receptacles on the cover.)

Looking at the fuse curves, higher current fuses don't really offer much protection.  It takes a lot of heat to melt the heavy conductor in the high-current fuses, and they dissipate about 4W at max current.  Both 15A slow-blow and fast-blow fuses will typically pass 30A for 1000 seconds before they pop.  Since a 15A fuse doesn't offer as much protection as a typical circuit breaker feeding the circuit, I chose to eliminate the fuse and save the dissipation.

Jeff
Title: Re: Possible New Filter Design
Post by: pomonabill221 on November 11, 2012, 04:07:09 PM
Your new filters look great!
One question... are the copper traces heavy enough to carry the 10A?
Title: Re: Possible New Filter Design
Post by: JeffVolp on November 11, 2012, 06:14:12 PM
One question... are the copper traces heavy enough to carry the 10A?

There are duplicate traces on both sides for the high current paths.  The first run of PCBs has 1 ounce copper, which is rated for 10C rise at 14A at that trace width.  That is less than the temperature rise in the inductors at max current.  The 15A version has 4 ounce copper, and the temperature rise at max current will be insignificant.

Jeff
Title: Re: Possible New Filter Design
Post by: pomonabill221 on November 11, 2012, 07:35:03 PM
Great!  Thanks for the info.
As always, Jeff designs VERY good tools and products!  Very professional and well built!