Water heater eating X-10 signal

Hello,

I've had a modest X-10 system running in our house for a few years. It took me a while to get it working reliably enough to get the WAF to a level where my wife wasn't cursing "HAL" every other day.

Part of that effort involved installing an active phase coupler in our mains panel. Everything was working pretty well, until we recently added an electric water heater to the panel.

The water heater is wired to a double pole 30 amp breaker in the panel via about 60' of 10/2 w/ground (no neutral connection on the heater, just the two hots). After I installed it I noticed that our X-10 controlled outside lights weren't coming on in the evenings. After a little troubleshooting I discovered that if I switched off the breaker for the heater, the X-10 system went back to normal. Switching the breaker on makes the problem reappear.

I'm guessing that the connection to heater and/or the heater itself is "sinking" the X-10 signal.

Is there a wired, in-line filter available that I can fit in after the breaker to block the X-10 signal from the heater and its wiring? Is that the correct approach in this case?

Thanks.

Reply to
graftonfot
Loading thread data ...

I don't believe there is any commercially available filter that can handle

30 amps. The largest is the big XPF, which is rated for 20 amps. It would take 4 of them to filter both hot leads running to the water heater.

A heavy electrical load across the two phases normally helps X10 signal distribution, so there must be something more here than just the heater element. There may be some sort of surge protection shunting the X10 signal to ground. Perhaps your signals were marginal before adding the heater.

It might be a good idea to invest in a X10 signal level meter, such as the ESM1, to find out what is really going on. Since filtering a high current load is not very practical, the only suggestion I have if it is the water heater is to increase your signal strength beyond what your active phase coupler can do.

Jeff

Reply to
Jeff Volp

What happens if you disable your active coupler?

formatting link
snipped-for-privacy@yahoogroups.com

Reply to
Dave Houston

| A heavy electrical load across the two phases normally helps X10 signal | distribution, so there must be something more here than just the heater | element.

He mentioned that he has an "active phase coupler." If by that he means what we normally call a repeater and if the repeater is of a design that drives the two legs out of phase then a heavy leg-to-leg load will diminish the apparent signal level as the carriers interfere destructively.

Dan Lanciani ddl@danlan.*com

Reply to
Dan Lanciani

Hi Dan,

I had thought of that, but I didn't think any repeater would drive the two legs out of phase due to that exact issue. Any 240V resistive load, such as a dryer or stove, would squash the signal. But it is possible...

Jeff

Reply to
Jeff Volp

| > | A heavy electrical load across the two phases normally helps X10 signal | > | distribution, so there must be something more here than just the heater | > | element. | >

| > He mentioned that he has an "active phase coupler." If by that he | > means what we normally call a repeater and if the repeater is of a | > design that drives the two legs out of phase then a heavy leg-to-leg | > load will diminish the apparent signal level as the carriers interfere | > destructively. | | Hi Dan, | | I had thought of that, but I didn't think any repeater would drive the two | legs out of phase due to that exact issue. Any 240V resistive load, such as | a dryer or stove, would squash the signal. But it is possible...

The only repeater whose circuit I've traced (ACT CR230) drives them out of phase. My understanding has always been that this is the standard practice in order to accommodate 240V modules which would otherwise have a difficult time seeing the signal. I assume it is also the reason for the standard warning about using a repeater in conjunction with the blocker/coupler. Have you ever encountered a repeater that drives the legs in phase (ignoring your own designs, of course :)? I have the original Leviton unit (i.e., the one before the one before the current one) but I'm not sure it's worth the bother to take it apart and trace the circuit.

Dan Lanciani ddl@danlan.*com

Reply to
Dan Lanciani

That's an interesting thought regarding 240V modules. Of course, the percentage of those is miniscule compared with 120V modules used on split-phase systems.

Since most X10 modules work fine down to 100 mV, or even lower, it would take almost perfectly balanced attenuation in both legs for in-phase signals to null enough at the 240V device to cause a problem. With the small percentage of 240V devices in service, it would seem to make more sense to use in-phase drive so 240V resistive loads would not attenuate the signal. You are correct in your assumption that the XTB-II drives 120KHz in-phase to both legs. When I tested the XTB-II here, I measured 30Vpp on one leg, and

25Vpp on the other leg due to an imbalance in loading. So, a 240V circuit would start off with a 5Vpp differential signal at the panel.

The commonly used .1uF passive coupler will drive both legs in-phase, but the second leg will always be lower in amplitude. That again would provide sufficient signal to a 240V module bridged across both legs.

I have a couple of the old Leviton 6201s kicking around. If I get a chance, I'll scope it to see what it does.

Jeff

Reply to
Jeff Volp

| That's an interesting thought regarding 240V modules.

Incidentally, it isn't my thought; I'm sure I read it somewhere though I can't remember where. So I wasn't surprised when I saw how the CR230 was built.

| Of course, the | percentage of those is miniscule compared with 120V modules used on | split-phase systems.

The percentage of 240V modules in my home is exactly 0; hence my interest in a repeater that drives the legs in phase for less contention with the blocker/coupler.

| Since most X10 modules work fine down to 100 mV, or even lower, it would | take almost perfectly balanced attenuation in both legs for in-phase signals | to null enough at the 240V device to cause a problem.

I'm not sure it's that simple. The imbalance required depends on the strength of the signal at the receiver. If it is already near the sensitivity threshold then a 50% imbalance could still be insufficient to avoid trouble. This is basically the generic carrier interference argument where if the signals are strong enough to begin with then even their difference is likely sufficient to operate a module. It's why you typically don't see a carrier interference problem with multiple synchronized transmitters collocated at a receiver. For a simple protocol, X10 can lead to some pretty complicated analysis...

| With the small | percentage of 240V devices in service, it would seem to make more sense to | use in-phase drive so 240V resistive loads would not attenuate the signal.

I certainly can't see any downside, but then I don't have any 240V receivers. Possibly the repeater designers know something we don't. I suppose if I were building a repeater product I'd be tempted to make it a switchable option. There's something unaesthetic about driving what amounts to a differential line pair in phase and hoping for random impairments to make it work. :)

| You are correct in your assumption that the XTB-II drives 120KHz in-phase to | both legs.

Not an assumption; I asked you before. :)

| The commonly used .1uF passive coupler will drive both legs in-phase, but | the second leg will always be lower in amplitude. That again would provide | sufficient signal to a 240V module bridged across both legs.

It depends. What if the sensitivity of the 240V module is 100mV, it is currently seeing 100mV on one leg and 0V on the other, and the passive coupler causes 50mV to appear on the previously-0V leg?

I keep meaning to check whether the official passive coupler flips the phase. It uses transformers on both sides so it certainly could. My brief experience with it was quite negative with previously working transmitter/receiver pairs failing once the coupler was switched in. Something funny was going on and since there was already significant passive coupling I wonder if it was somehow producing cancellation. Of course, this was at a time when I was lucky to have 50mV levels at some locations.

I'm pretty sure that the blocker/coupler cannot flip the phase because of the way it blocks, and it was with that device that they started warning about combination with a repeater.

| I have a couple of the old Leviton 6201s kicking around. If I get a chance, | I'll scope it to see what it does.

I have the original (two black wires and one white; does not support extended codes). There was a second rev that supported extended codes and could drive a third phase but had an absurd power-on sequence requirement. They were both the same part number, right?

Dan Lanciani ddl@danlan.*com

Reply to
Dan Lanciani

I am having a hard time with the phrase "in phase" here. Why the worry over "the phase of the signal" between different legs of the circuit? I get the feeling you both think the reason that neutral is at 0 volts potential is that, the voltage waveforms on each leg destructively interfere with each other due to phase cancellation and not simply as a consequence of vector addition. I mean that's fine but its hardly the standard model for analysis of a split-phase power circuit. I suppose you could look at that way but this has nothing to do with the X10 signal.

As to 240V modules, I'd be surprized if they just didn't listen on one leg only, since there is no return path for the signal, if one listened only between legs as that is isolated from neutral--the designated return path of the source which IS on one leg.

Slammer

Reply to
Slammer

| I am having a hard time with the phrase "in phase" here. Why the worry over | "the phase of the signal" between different legs of the circuit?

It's germane to the question that started this thread, i.e., why would adding a leg-to-leg load to an otherwise working system that uses a repeater induce failure?

| I get the | feeling you both think the reason that neutral is at 0 volts potential is | that, the voltage waveforms on each leg destructively interfere with each | other due to phase cancellation and not simply as a consequence of vector | addition.

The neutral is often defined as being at 0V potential by convention (and, pragmatically, because it is usually bonded to a grounding system). It really has nothing to do with interference or the vector addition that can represent same. In reality, only potential _differences_ are meaningful and you could just as well choose something other than the neutral as your

0V reference if it makes the analysis easier. Many X10 modules use a hot leg (rather than neutral) as their logic 0V reference. It is convenient to adopt that convention when working on such circuits since otherwise you have to deal with Vcc being a 5V DC signal added to the 120V line supply with respect to neutral.

| As to 240V modules, I'd be surprized if they just didn't listen on one leg | only,

How exactly would you propose that a 240V module listen on one leg only? Remember, you can't measure potentials; you can measure only potential _differences_.

Dan Lanciani ddl@danlan.*com

Reply to
Dan Lanciani

Phil Kingery mentions phase-to-phase cancellation and 240V modules in at least one of his coupler/repeater articles.

formatting link
formatting link
snipped-for-privacy@yahoogroups.com

Reply to
Dave Houston

I thought a bit more about this since my earlier post. A 240V load should be on its own circuit. There shouldn't be any "signal suckers" on that circuit other than potentially the 240V load itself. So the starting signal will be the differential level at the panel. In an installation with balanced loads, a repeater would produce essentially the same signal level on both legs. So, that could present a problem for 240V loads if both phases were driven in phase.

The differential line pair only refers to driving a 240V load, which you agree is rare. I think the better approach is to provide best performance for normal 120V loads, which are probably over 99% of X10 devices. I don't like the possiblity of a 240V resistive load possbily causing havoc with signal levels. If the installaton does include a 240V load that has a problem with signal levels, it could be addressed by placing a .1uF capacitor to neutral on one of the hot leads to imbalance the signal levels.

Yes, it does depend on the loads on the other phase. However, with a reactive impedance of 13 ohms, there should be significant signal drop from one leg to the other in a normal installation. With no other "signal suckers" in the path to the 240V load, the only thing to contend with is the signal loss in the cable run itself.

That was an easy test to do - just confirmed an old Leviton 6299 does flip the phases. Interesting is that its peak coupling is at 134KHz.

My two have two blacks, a white and a red. So that must be the newer 6201. I'll test it after I get a couple more XTB-IIs assembled.

Jeff

Reply to
Jeff Volp

| I thought a bit more about this since my earlier post. A 240V load should | be on its own circuit. There shouldn't be any "signal suckers" on that | circuit other than potentially the 240V load itself. So the starting signal | will be the differential level at the panel. In an installation with | balanced loads, a repeater would produce essentially the same signal level | on both legs. So, that could present a problem for 240V loads if both | phases were driven in phase.

Yes, I suspect this is why they decided to give the advantage to 240V modules. The "difference of large signals" argument applies just as well from the point of view of 120V modules with the legs driven out of phase, but it is probably even more unlikely that the drop on one leg will be exactly matched by the net coupling from the other leg through the various 240V loads. As I've often said, the analysis here is more complicated than it might at first appear. :)

| The differential line pair only refers to driving a 240V load, which you | agree is rare.

I agree that it's rare in my house. If I had a few 240V modules that weren't behaving I might sing a different tune. Having a switch at least gives you a choice.

| I think the better approach is to provide best performance | for normal 120V loads, which are probably over 99% of X10 devices. I don't | like the possiblity of a 240V resistive load possbily causing havoc with | signal levels.

It is not at all clear that this is common occurrence, though. If (as I suspect) it turns out that all commonly available repeaters drive the legs out of phase we would probably have to do a lot more investigation to determine why it (may have) happened in this case. Also, keep in mind that driving the legs out of phase does not make 240V loads a special problem wrt reducing the signal; they will have approximately the same effect as two 120V loads of similar power. If a 240V load causes a problem it may simply mean that the signal levels were already on the edge. Driving the legs in phase mostly takes the 240V loads out of the picture, but not necessarily without a cost:

Consider that driving the legs in phase means that every 120V load will be coupling the carrier in-phase to the neutral, reducing the apparent leg-to-neutral signal available to all 120V modules. In a distribution system where the 120V loads are reasonably well distributed between the legs the loss of the ability to cancel the carrier current in the neutral may dominate any advantage gained by taking 240V loads out of the picture. It would take a lot more analysis to convince me that driving the legs in-phase is in fact (at least on average) the way to provide the best performance for 120V loads.

| If the installaton does include a 240V load that has a | problem with signal levels, it could be addressed by placing a .1uF | capacitor to neutral on one of the hot leads to imbalance the signal levels.

You'd probably have to put the capacitor at the panel since pure 240V loads don't normally have a neutral available. In addition to the code issues you wouldn't have the advantage of the series impedance of the line feeding the device to help drop the level, so you would be trying to unbalance the whole system.

| That was an easy test to do - just confirmed an old Leviton 6299 does flip | the phases.

That certainly confirms the behavior I saw.

| Interesting is that its peak coupling is at 134KHz.

Maybe just poor QC?

Dan Lanciani ddl@danlan.*com

Reply to
Dan Lanciani

Many houses have 240V high-current resistive heating loads (ranges and dryers). The inductance of those circuits would be relatively low, so the effect may be more significant than a variety of 120V loads that sum up to the same current.

You are correct that 120KHz distribution over household electrical wiring is a complex problem. When I was working on the XTB-II output stage I built a rudimentary simulation using 8 circuits with line inductance, a variety of loads, and a few signal suckers. It mirrors the decay of signal ampltude that is normally seen as the distance from the panel increases. Running a frequency sweep shows peaks and nulls all over the place, along swith some significant phase shift on some of the circuits.

At first I thought you had a point here, but remember that each neutral goes back to the distribution panel common bus. It doesn't make any difference on each of the 120V circuits what the phase is because there is no summing together of the return signals on a common netural. Of course, that assumes the transmitter is relatively close to the distrbution panel so the summed return signals in its own neutral can be ignored.

That is true if there is no netural at the 240V X10 load. Receptacles for all our 240V appliances ARE wired with both neutral and ground. I had to buy a new dryer cord for our old Maytag dryer. I was told that is a NEC requirement now.

Jeff

Reply to
Jeff Volp

I have not tried removing the coupler, I'll give that a shot and post my results. By the way, it's a Leviton coupler (sorry, don't have the exact model # at hand, but I do remember it has a black face.)

I do happen to have an ESM-1 - I'll try to use that to supply more info.

Thanks very much for the help - and for the interesting technical discussion (although I must admit I was left in the dust about halfway through.)

Reply to
graftonfot

Simply put, it's possible the hot water heater, when on, is acting like a passive coupler and cancelling your active coupler on one of the phases. It's an unusual problem.

formatting link
snipped-for-privacy@yahoogroups.com

Reply to
Dave Houston

| Many houses have 240V high-current resistive heating loads (ranges and | dryers). The inductance of those circuits would be relatively low, so the | effect may be more significant than a variety of 120V loads that sum up to | the same current.

Just to clarify something: when you talk about the effect of 240V loads is the concern that they lower the impedance of the whole network as seen at the repeater significantly compared to the impedance of the repeater's drivers such that you can observe a lower carrier level at the repeater itself?

| At first I thought you had a point here, but remember that each neutral goes | back to the distribution panel common bus. It doesn't make any difference | on each of the 120V circuits what the phase is because there is no summing | together of the return signals on a common netural.

You are ignoring multi-wire branch circuits and sub-panels (themselves essentially big multi-wire branch circuits). I don't know whether this is a reasonable simplification to make in general, but it would almost certainly not be appropriate for my house which has more than its share of both.

| Of course, that assumes | the transmitter is relatively close to the distrbution panel so the summed | return signals in its own neutral can be ignored.

As long as we are assuming that, it would probably also be reasonable to assume that we can ignore any loss in the repeater's hot leg connections along with any losses in the single panel's bus bars, breakers, etc. With this model, it appears that the only way a load on one circuit can affect the signal level on another circuit is by driving the level down at the repeater itself by presenting a relatively low impedance compared to that of the repeater's drivers. This conclusion seems to conflict with my observations where the level at a repeater can remain constant (high) while switching a given circuit in and out changes the level seen on another circuit. This makes me worry that the model is now _too_ simple. :(

N.B. I'm still very interested in the possibility of a repeater that drives the legs in phase (especially for use where a blocker/coupler is required) as I mentioned the last time this topic came up over a year ago:

formatting link
But you are going to have to work a little harder to convince me that ACT/X10/etc. got it wrong for the general case, even if you found their choice surprising.... :)

| That is true if there is no netural at the 240V X10 load. Receptacles for | all our 240V appliances ARE wired with both neutral and ground. I had to | buy a new dryer cord for our old Maytag dryer. I was told that is a NEC | requirement now.

Electric dryers and most electric stoves are not pure 240V loads; they are 240V/120V loads and they have always had a neutral. For many years the NEC allowed dryers and stoves to ground their chassis to the neutral in order to save copper (think wartime). This is why their receptacles and cords could in many cases be 3-wire. The exception did not apply if the appliance was fed from a sub-panel and it never applied in mobile homes, so happily all existing appliances had to be made to adapt to either 3- or

4-wire hookups. Recently the NEC removed the exemption for dryer and stove circuits (in new construction) so now all 240V/120V circuits must be installed with separate neutral and ground. You may continue to use existing 3-wire circuits even for new dryers and stoves.

None of this has anything to do with pure 240V circuits of the type for which X10 makes modules. Such circuits never had a neutral and never will. See for example:

formatting link
Dan Lanciani ddl@danlan.*com

Reply to
Dan Lanciani

I'm not sure I understand your question. I think a low-impedance 240V resistive load with relatively low inductance can act as a "super signal sucker" if the two legs are driven out of phase at 120KHz. I would expect some decrease in signal level at the repeater when confronted with such a load, especially if that repeater has a transformerless power supply.

By multi-wire branch circuits, you are referring to 120V circuits fed off both legs with a common neutral. Yes, the X10 signals would sum on that common. I suppose it would be possible that each leg might have several X10 loads, but I think that is reaching pretty hard to find something that might have a problem.

I did not consider multiple sub-panels. It would be intersting to find out how much signal is lost in the heavy cable run between sub-panels.

Actually, here I saw the combined loads have a significant effect on the XTB-II output. The XTB-II will output over 40Vpp with no load. I measured one leg at 25Vpp, and the other leg at 30Vpp when it was connected to the panel. With only a couple of feet of wire between the XTB-II and the panel, the levels were the same at the XTB-II and inside the panel itself. While I don't have measurements to back it up, I would expect that switching off all breakers on either leg would increase the level on that bus back to 40Vpp.

Now that I've finally finished with my tax return, I've started back on the enhanced repeater version of the XTB-II. That will drive both legs in phase.

BTW, thanks for the info on the stove/dryer receptacles. I had thought the

3rd prong was ground.

Jeff

Reply to
Jeff Volp

| I'm not sure I understand your question. I think a low-impedance 240V | resistive load with relatively low inductance can act as a "super signal | sucker" if the two legs are driven out of phase at 120KHz.

Often when referring to "signal suckers" people are talking about local loads that effectively form a voltage divider in combination with the line back to the transmitter (repeater, whatever). Usually the transmitter itself is not noticeably loaded. Under the simplified model we discussed a

240V load on a dedicated circuit would pretty much have to impair the signal at the transmitter to cause a problem.

| I would expect | some decrease in signal level at the repeater when confronted with such a | load, especially if that repeater has a transformerless power supply.

So you are thinking that even though the repeater's driver might have an impedance in the fractional ohms there isn't enough power available from the supply to maintain voltage? That might be reasonably easy to instrument directly with a meter on the supply rail of my spare/repaired CR230 (which uses the typical reactive power supply).

| By multi-wire branch circuits, you are referring to 120V circuits fed off | both legs with a common neutral.

Yes, also knows as Edison circuits.

| Yes, the X10 signals would sum on that | common. I suppose it would be possible that each leg might have several X10 | loads,

It's not just X10 loads; it's any loads. And it doesn't take more than one per leg, though obviously for this to be meaningful in the case of one load per leg at least one load has to be interested in hearing X10.

| but I think that is reaching pretty hard to find something that might | have a problem.

I'm not looking for a problem; I'm looking for pros of each approach. I probably should not have phrased it in terms of giving something up. So far, out-of-phase drive appears to offer 240V module compatibility and the ability to take advantage of shared neutral configurations in much the same way the power distribution system itself does. In-phase drive allows you to use a smaller power supply and/or higher impedance drivers in the face of 240V loads and may offer better compatibility with in-phase couplers.

These are all pretty minor features, and I'm not convinced that the problem you fear from 240V loads is any more an issue in real life than the ability to take advantage of multiwire branch circuits. I brought it up only because the original poster's fact pattern was unusual. At least to me the optimal choice is far from obvious and I would still be inclined to make it an option if I were building a repeater product.

| Actually, here I saw the combined loads have a significant effect on the | XTB-II output. The XTB-II will output over 40Vpp with no load. I measured | one leg at 25Vpp, and the other leg at 30Vpp when it was connected to the | panel.

Was this due to the output impedance of your drivers or a sagging power supply?

Dan Lanciani ddl@danlan.*com

Reply to
Dan Lanciani

That is really the only way signal suckers can effect other circuits on that phase.

Since the transformerless power supply cannot provide that much energy, I don't think the output impedance is in the fractional ohm region.

I thought we were discussing summing of in-phase X10 signals on the neutral reducing the end-point signal strength. That would only be true if there were multiple X10 loads (or signal suckers) on both legs of that circuit.

The XTB and XTB-II do not use a smaller supply or higher impedance drivers. They both use a 6-watt transformer supply. The supply is unregulated, and drops about 25% during transmission. The TW523 spec says it will deliver

5Vpp across a 5 ohm load. On the bench I measured the XTB-II delivering 30Vpp across a 5 ohm resistive load. That's actually about 35 times the power of a normal X10 transmitter. The output impedance looks like about .5 ohm.

Since a common power stage drives both output coupling networks, it is due to the fractional ohm impedance across the coupling networks. Note that the higher level was on our "X10 phase" and the lower level was on the phase used to power "unfriendly" loads.

The voltages above were measured at the panel. Signal levels decrease as they propogate away from the panel due to the combined effect of line inductance and attenuation from distributed loads.

Jeff

Reply to
Jeff Volp

Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.