Re: Lynx X-10 Analyzer (WAS Re: Decora HCPRF)

[nobody@xxxxxxxxxxxx (Dave Houston)]

"Robert Green" <ROBERT_GREEN1963@xxxxxxxxx> wrote:

>I was afraid that might be the case.  A more primitive way to get "signal
>health" feedback might be to design a "sensor" using a combination of  one
>of the filter designs floating around the net and an appliance module.  If I
>can find a filter design that offers variable attenuation, then I can
>balance the signal to an appliance module (for convenience, I'll call it an
>ASTM for Attentuated Signal Test Module) so that it just managed to fire.
>These modules could serve as Go/No Go signal indicators.

There would be no need for a filter. The PLC_IN circuitry in the appliance
module already has a tuned filter. It has the advantage of being X-10's own
design so it should pass the range of frequencies that all X-10 made devices
are affected by.

The disadvantage is that it is not isolated from the mains.

>Any disturbance to the powerline from a new noisy appliance or a signal
>sucker would likely make at least one of the ASTMs deaf.  All I need to do
>is run a signal wire back from the ASTM to a central panel.  If the house
>wiring's OK, all ASTM's will respond to their appropriate signals.  If
>there's something amiss, one of the modules will fail to operate, no closure
>will occur and I can be alerted.  HomeSeer could run the diagnostic a few
>times a day.
>
>I'd like to implement something like this because on occasion something will
>get plugged in that will have only marginal effect on the main house wiring.
>It's the second signal sucker that causes real trouble and makes it hard to
>trace the problem.  I've had a number of "sucker hunts" and they are
>miserable, time-wasting processes.  I'm willing to spend some money to lock
>them down early.

Why not just put a module that will respond to STATUS REQUEST on each
circuit and poll them periodically? That requires a signal to traverse the
route in both directions.

If I ever finish some other projects, I've thought about creating some
modules that would report temperature and ambient light levels using
extended codes that could be triggered by a STATUS REQUEST.

>Truth be told, I'd rather have modified ESM1 sensors, though, because I'd
>get actually voltage levels and not just a "set point" gauge.  It may turn
>out that the ASTMs would be a better, more automatic way of checking signal
>health on a daily basis.

The best thing to do would be to replace the PIC16C711 in the ESM1 with a
newer pin-compatible PIC (e.g. PIC16F88) and adding a serial output on one
of the unused pins. The code would be fairly simple. But, if all ESM1s are
like mine, desoldering the PIC is very difficult. I gave up after struggling
with it for about an hour without freeing a single pin. I think it requires
more sophisticated (and more costly) tools than are in my kit. I've never
had to give up on one before this.

>>
>>      http://www.mbx-usa.com/xtc798.txt
>>
>> The first paragraph explains why the cycle count is important.
>
>I'm a little confused.  Are they saying that not only the peak voltage is
>important, but the length of time the signal is at or near that peak?

No. They're saying what that paragraph says. If the signal is weak, you may
miss some 120kHz cycles.

>For instance, when looking at:
>
>http://www.mbx-usa.com/x10-sig.htm
>
>and the image dated 8/20/01 2:37PM would it be reasonable to say that each
>of the separate vertical magenta trace lines represents a "count?"

No. Each of the magenta bursts represents ~120 cycles or counts. The
horizontal resolution in that picture cannot show the detail. I have other
screenshots that show ~1ms but I'd have to hunt them down.

X-10 says transmitters should send continuous cycles of 120kHz for 1ms,
starting as soon as possible after ZC. 1ms of 120kHz gives 120 cycles (or
counts). The receiver window is 650µs wide starting 250µs after ZC. Only 78
cycles can fit in the 650µs window. 48 cycles represents 61.5% of the
expected 78. If the amplitude is marginal, some cycles will be missed but
the accuracy will improve as the signal strength increases.

The TW523 does it differently, presenting a demodulated data envelope. Most
applications that interface with it check for the presence/absence of the
data envelope mid-way through the acceptance window.

>I can imagine a need for a 220V capable analyzer in dealing with phase
>problems in a house with a number of 220V appliances.  Not a big need, but
>it could be useful.  It's nice to know I could eaily modify the ESM1 to read
>signal strength on a 220V line

Maybe there's something that I don't understand but I don't see how that
tells you anything of value.