Author Topic: Hard wired communication link?  (Read 19253 times)

GAR

  • Full Member
  • ***
  • Posts: 131
  • Karma: +0/-0
Re: Hard wired communication link?
« Reply #15 on: August 13, 2011, 10:09:29 PM »
110813-13421 EDT

tazer:

With the extension cord method I am using and no in line filter I do not have a major problem. There are just random times and for moderately short periods there is loss of data. Some of the loss may actually be the MTU failing to send data. Had one time when the MTU quit for hours. I think I switched MTUs, but I might have just cut power to reboot the MTU.


jrwalte:

There is nothing that would prevent a fiber cable from going inside any cabinet. I will admit that ease of installation is probably the driving force behind the use of the power line. But then one has to go thru all these gyrations to solve miscellaneous problems. Most houses do not have many circuits that serve only one or just a few receptacles, and those circuits with only one outlet are usually dedicated to high power appliances. My extension cord across the basement floor is a simple present solution. The circuit to the bench where the RDU is located has several computers and other equipment running on the circuit. A separate circuit sometime is a possibility, but it is not a high priority. 

One distributor on the Internet said that they quit selling TED systems because customers were encountering too many problems. Thus, is ease of installation more important than a high reliability system and customer satisfaction?

.

SherlockOhms

  • Full Member
  • ***
  • Posts: 168
  • Karma: +0/-0
  • Sherlock
Re: Hard wired communication link?
« Reply #16 on: August 16, 2011, 06:26:17 PM »
@Gar Everyone is pretty much correct in their reasoning.  The reason for PLC has to do with electrical code. Remember the system is used in a lot of different configurations. As some one said by code you can't mix low voltage and hi-voltage wiring. Specifically "all conductors must be rated for highest possible voltage used in the panel" so for US it would mean 300v typically. While it may seem so simple and make sense... its not that simple. As some one else mentioned wireless will have hard time going through metal box. Filtering a circuit and isolating communication link is really the best approach.

GAR

  • Full Member
  • ***
  • Posts: 131
  • Karma: +0/-0
Re: Hard wired communication link?
« Reply #17 on: August 16, 2011, 10:53:07 PM »
110816-1424 EDT

SherlockOhms:

I do not follow your logic.

Nothing prevents you from using 600 V wire.

Further in the Installation Guide Example 2 shows the MTU on the outside of the panel. Now you presumably already have 600 V wire to the inside, and outside 600 V is not required.

There are the other possibilities I mentioned previously, and others to get data from inside to outside.
1. Build the MTU so that a communication connector is on a hub that is put thru a knockout hole.
2. Use a fiber optic cable inside to outside.
3. Use ultrasonic or magnetic coupling from inside to out.
4. And I could think up other ways.

Power line communication may in reality be satisfactory, but its problems and limitations should be known by the customer beforehand.

.   

SherlockOhms

  • Full Member
  • ***
  • Posts: 168
  • Karma: +0/-0
  • Sherlock
Re: Hard wired communication link?
« Reply #18 on: August 17, 2011, 01:07:05 AM »
Let me know where we can purchase 600v Ethernet rated cable.

Your also assuming a single MTU installation.  In that case yes your ideas are fine and would work.

Now let's throw a curve ball.... how will you handle 4 MTU's across entire house?  Your willing to invest into fiber-optic line ran through the house? That may not always be an option.  Again it may be simple in your case personally... the goal however is to have a system that is easily installed in as many situations as possible.  Not many people are skilled enough to do the required fiber-optic wiring. At same time your making the system even more complicated and more expensive. 

Don't give up on PowerLine communication yet :) 

GAR

  • Full Member
  • ***
  • Posts: 131
  • Karma: +0/-0
Re: Hard wired communication link?
« Reply #19 on: August 17, 2011, 03:04:09 AM »
110816-1845 EDT

SherlockOhms:

If the MTU is outside the panel, then why do you need 600 V Ethernet cable?

If the connector is on the outside of the panel, but the MTU is on the inside, then again why do you need 600 V wire for the Ethernet cable? All you need is the surface of the panel as an isolation between the inside and outside wiring. Obviously I am assuming galvanic isolation between inside and outside in case of an internal short.

Further you could leave the MTU inside, use 600 or higher voltage wire to come outside the panel to the Ethernet isolation transformer, then regular CAT-5 cable.

Also I am not assuming an Ethernet source in the MTU, but that might be OK.

Basically in the 5000 system you go to Ethernet cable following the Gateway anyway.

.

PV-Skip

  • Jr. Member
  • **
  • Posts: 67
  • Karma: +0/-0
Re: Hard wired communication link?
« Reply #20 on: August 20, 2011, 07:59:33 PM »
Read through all of this I am still a little bit confused.
My idea was also to run a separate, filter isolated power cable through my attic to my office and provide the gate way with power and data from the MTUs.


Now here you guys mention a direct connection form the MTU to the Gateway.
Installing the MTU outside of the main panel should not provide a big problem. But how can I run a ethernet cable from the MTU to the gateway. There is no connector on the MTU for that, right ??? ???

GAR

  • Full Member
  • ***
  • Posts: 131
  • Karma: +0/-0
Re: Hard wired communication link?
« Reply #21 on: August 20, 2011, 10:42:35 PM »
110820-1313 EDT

PV-Skip:

When I started this thread I was asking why there were not hard wired versions of the TED system.

There are not any MTU to RDU or Gateway hard wired solutions except to use the power line.

The TED designers chose to to use power line communication. They have given some reasons for this. One reason not given is cost and the belief that power line communication provides for flexible and simple installation.

Simple is probably true if filters and experimentation to solve problems did not become a factor in installation or questions on why something was not working.

Cost is also probably true if unintended consequences are ignored.

Your only solution, without modification of the original components, is to install an in-line filter at the MTU location. On the output side of the filter connect the 120 V input to the MTU, and a new separate circuit to the location of the RDU or Gateway. This new circuit should have nothing on it except the RDU or Gateway. You have to operate in the 120 V mode because the in-line filter is only for one phase. But I would operate in the 120 mode anyway.

With the Ethernet connection from the Gateway to whatever (computer or network) you have electrical isolation, usually thru a coupling transformer, between the Gateway and the computer. Thus, the AC power to the computer can be and probably should be from some other circuit. This will reduce the noise from the computer power supply that gets into the power line communication cable.

The 1000 series RDU is a different problem. In this case, if the computer power cord has three prongs, then it should be powered from the power line communication cable so the computer EGC (equipment grounding conductor --- ground pin in the plug) is tied to the EGC of the the power line communication cable. This may require a plugin filter for the computer to minimize noise from the computer getting to the communication path. This ac supply source for the computer is to minimize ground path noise that might interfere with the USB communication path.

I have just received one each of these filters from Energy, Inc. These are in fact X!0 filters and probably can be obtained anywhere X10 is sold. The X10 part numbers are XPPF for the plug-in, and XPF for the in-line.

If you are using a single gateway, then all MTUs have to be on the load side of that single in-line filter.

.
« Last Edit: August 20, 2011, 10:47:27 PM by GAR »

PV-Skip

  • Jr. Member
  • **
  • Posts: 67
  • Karma: +0/-0
Re: Hard wired communication link?
« Reply #22 on: August 22, 2011, 06:50:24 AM »
Thanks Gar,
that was a excellent summary.
After digesting all the stuff yesterday I came to the same conclusion and ordered two in-line filters. I found a cheaper source though. . .
I just came from the attic installing a separate power-line for the gateway, running from the panel to my office and are waiting now  for the filter(s) before the final connection.

Two questions remains. Does it matter, if I connect both my MTUs and the Gateway on that filter, or should I have a filter for every MTU?
Secondly, I also have a Enphase microinverter system and an Envoy gateway for that, which sends its data over the house power lines.
So far they don't seem to have problems together.
My intention is, to have this Envoy also running on this special, filtered power line.
Do you think, that would create a problem?

Thanks for your help!
PV-Skip

GAR

  • Full Member
  • ***
  • Posts: 131
  • Karma: +0/-0
Re: Hard wired communication link?
« Reply #23 on: August 22, 2011, 05:34:50 PM »
110822-0814 EDT

PV-Skip:

Last week I ordered 1 in-line, and 1 plug-in filter from Energy, Inc. These arrived Saturday. These are actually X10 filters in X10 boxes with X10 instructions. In a separate thread I commented on my analysis and experiments to date.

Changing subject for a moment. The data sent from the 1000 MTU is in bursts of 100 milliseconds (0.1 second) every 1 second. The data burst is around 200 milliseconds for the 5000 and it appeared to be a lower signal level. I had sync problems getting the leading edge without other noise triggering. Also note that the 5000 sends data only once every 2 seconds. That is why the exported second data from the 5000 appears to provide duplicate data in 2 second blocks.

There is an intercept probability of two or more MTUs on the same circuit overlapping their transmission time. If this occurs there is loss of data or errors. The time clocks in the MTUs are relatively accurate, therefore the drift rate of one clock to another is rather slow. Suppose two MTUs on the same circuit start out with one transmitting just 0.1 second before the other. As the slow drift occurs let the first drift into the second and there will be errors for a long time depending upon the drift rate.

All the TED systems use 132 kHz for their carrier frequency. X10 apparently uses 120 kHz. Others may use slightly different frequencies.

I have no idea what two way communication TED uses, nor whether any collision detection and retransmission occurs on errors.

Back to your question. I would start with one in-line filter to a single dedicated circuit. On this circuit would be 1 MTU and 1 RDU or Gateway. Get several days of 1 second data. About 3.8 megabytes of exported data in a .CSV file per day. In the exported data there are 42 bytes plus line feed per second for the 1000. Each of these records has about 15 unnecessary bytes, and if the year had been excluded a few more.

Analyze this data for lost seconds.

Next add a second MTU and receiver to the circuit. Put these current sensors on the same wires that the first set of sensors are on. Collect data from both receivers and compare the lost seconds. This may provide an indication of the likelihood of intercept problems. However, it is all chance on what the relative drift rates are between the MTU clocks. The MTU in the 1000 series does not contain a real time clock. That is in the RDU. The timing of the 1 second rate of transmission will come from the MTU's microprocessor clock (its basic time source - probably a crystal oscillator). TED could add some randomized jitter to the 1 second to minimize collisions when two MTUs were on the same circuit. But all this is conjecture.

I do not run multiple MTUs, but running just one on a fairly clean circuit I get loss of data at times and for many seconds at times.

.   

PV-Skip

  • Jr. Member
  • **
  • Posts: 67
  • Karma: +0/-0
Re: Hard wired communication link?
« Reply #24 on: August 23, 2011, 05:44:12 AM »
Thanks again Gar,
got it so far. (Hopefully) ;D

I still don't know, how I can look for lost data or seconds.

PV-Skip

GAR

  • Full Member
  • ***
  • Posts: 131
  • Karma: +0/-0
Re: Hard wired communication link?
« Reply #25 on: August 23, 2011, 06:37:33 AM »
110822-2150 EDT

PV-Skip:

One second export from a 1000 system looks like the following:

"10/2/2010 11:11:59 PM","1.67","0","125.2"
"10/2/2010 11:12:01 PM","1.68","0","125.2"
"10/2/2010 11:16:29 PM","1.61","0","125.3"
"10/2/2010 11:16:30 PM","1.61","0","125.3"
"10/2/2010 11:16:31 PM","1.61","0","125.3"
"10/2/2010 11:16:32 PM","1.61","0","125.3"
"10/2/2010 11:16:33 PM","1.61","0","125.3"
"10/2/2010 11:16:34 PM","1.62","0","125.3"
"10/2/2010 11:16:35 PM","1.61","0","125.4"
"10/2/2010 11:16:36 PM","1.61","0","125.4"

After I remove some of the unnecessary stuff I get a result similar to this

23:16:32, 1.61, 0, 125.3

In the 5000 system they got rid of all those unnecessary quote marks. I think they still use AM and PM which makes no sense for data processing, or even thinking, and midnight should be 0.

You simply write a basic program to find a discontinuity in the sequencing of the 1 second increment data.

When a discontinuity in the seconds occurs, then output the full record before and after he occurrence. You don't need to remove the junk for this purpose. I do it to reduce the file size and make manual reading easier. What I also do for plotting is convert to decimal hours. So 23:16:23 becomes 23.8000 . Converting to decimal hours would not be useful for your purpose, but you will need to work in a mod 60 number range. Thus, 0 = 60 when comparing with 59. I think sufficient useful information would obtained from comparing seconds. Otherwise convert minutes to seconds and add to the seconds. Now you would work modulo 3600 .

.

 

GAR

  • Full Member
  • ***
  • Posts: 131
  • Karma: +0/-0
Re: Hard wired communication link?
« Reply #26 on: August 23, 2011, 04:52:14 PM »
110823-0821 EDT

PV-Skip:

Following is a sample of my data where there is lost data:

"7/22/2010 8:52:07 PM","3.51","0","123.7"
"7/22/2010 8:52:08 PM","3.51","0","123.7"
"7/22/2010 8:52:09 PM","3.51","0","123.7"
"7/22/2010 8:52:10 PM","3.51","0","123.7" ....... start of loss
"7/22/2010 8:52:53 PM","4.01","0","123.5" ....... end of loss
"7/22/2010 8:52:54 PM","3.78","0","123.7"
"7/22/2010 8:52:55 PM","3.5","0","123.6"

Notice the lost time between 52:10 and 52:53 . This is what you would be looking for in the data. Only one MTU is running.

This problem could be Microsoft (XP-Pro), MTU, RDU, or external noise interference.

One time I had the 1000 MTU quit sending data completely, the MTU runs without a command for data, and therefore every second should send data. I replaced that MTU with another one, but later, on the bench, the non-functioning MTU began working again. Note: just power on reset did not correct its problem. So I have no idea about the cause of this failure.

.


GAR

  • Full Member
  • ***
  • Posts: 131
  • Karma: +0/-0
Re: Hard wired communication link?
« Reply #27 on: August 23, 2011, 06:04:45 PM »
110823-0935 EDT

PV-Skip:

Another consideration on loss of data. What is the TED algorithm for calculating energy? If this is simply summing the 1 second power values divided by 3600 over some time period, then loss of data within that time period reduces the accumulated energy for that period.

If it is assumed that a good estimate of the power during the loss time is the average of the starting and ending powers of that period, then that average power value divided by 3600 and the result multiplied by the decimal time in hours of that loss period could be added to the other accumulated energy.

These problems would not need to be considered if the communication path was rock solid.

.

jrwalte

  • Jr. Member
  • **
  • Posts: 60
  • Karma: +0/-0
Re: Hard wired communication link?
« Reply #28 on: August 24, 2011, 12:55:44 AM »
Two questions remains. Does it matter, if I connect both my MTUs and the Gateway on that filter, or should I have a filter for every MTU?
Secondly, I also have a Enphase microinverter system and an Envoy gateway for that, which sends its data over the house power lines.
So far they don't seem to have problems together.
My intention is, to have this Envoy also running on this special, filtered power line.
Do you think, that would create a problem?

Both MTUs, if communicating to 1 gateway, MUST be on the same side of the filter as the gateway. The purpose of the filter is to separate the electrical noise on either side of the filter and prevent pass-through to/from the breaker. If a MTU isn't on the same side of the filter as the gateway, it would get filtered out and wouldn't be able to communicate with the gateway.

A second benefit to me having to install an in-line filter where my home entertainment connects, it is stopped my LED lights from occasionally flickering. Apparently my electronics were causing some noise the LEDs were reacting to.
« Last Edit: August 24, 2011, 01:01:29 AM by jrwalte »

PV-Skip

  • Jr. Member
  • **
  • Posts: 67
  • Karma: +0/-0
Re: Hard wired communication link?
« Reply #29 on: August 24, 2011, 06:36:10 PM »
Thanks again Gar,
I exported the second data from the TED 5000.
had to format the cells  to se the seconds and it looks very good.
I did not go through all of them (would need an hour for that) but I compared the cell numbers with the second numbers in order to see, if there are seconds missing.
So far, so good.
But there could be some noise at times I did not look at. . .
And all that on the normal house network.
Since I have installed the dedicated power line over the weekend and are just waiting for the filters, I'll install everything on that line anyway.

For installing the filter(s) and going back just to the black wire on the MTU, I suppose both my MTUs have to be on the same side of the phase. A or B, right?