Archived Forum

I believe I've seen something for this - basically it's a high pass filter that gives you a short between phases at the carrier frequency. But have you tried it without? There might be enough capacitance in any 3-phase cable to transfer the signal anyway. TP-link seems to think it will work: http://www.tp-link.com/en/article/?id=406

[Edit: I knew it! Use your Google-fu on "powerline phase coupler"]

Robert,

thanks. I've tried it as it stands, and did actually get a connection, but it died and haven't been able to reestablish it. The fact that it worked ( even for a little while ) is some good news 

I've also seen those couplers and the faq, and agree with you that you'd expect them to be coupled anyway , capacitively speaking... The client is 80-100km away from me but just found out that one of their clients is using a Solwise powerline adapter so I suspect I need to give my adapter a unique 'name' - I'm checking with tp-link to see how to configure them to avoid interference.   Thanks a lot for your input - much appreciated.

would a powerline adapter on each phase near the fuse box work if you use a xover patch cable between then?

just a thought

Yeah; fair point you need to be able to get cat5 cable to each adapter and then back to the router; not possible for me.

no no

if 2 of the phases are close just use a crossover cable between them... no need for a router or switch

it should work as a link

Surely the dedicated adapter will be better/cheaper even if only two phases are needed, and definitely cheaper than two pairs of adapters for 3 phases?

if it was me...

and lets say my fuse box is at one end of building and my internet is the other end i would take 2 of these adapters somewhere near the fusebox and place them at each phase and just just an ethernet crossover between them...

if 3 phases i would just get the cheapest 4port or less switch and connect each of the 3 adapters to it, at internet end another adapter and you should have internet no matter what outlet you pick... 

@ Robert: by 'dedicated adapter' did you mean the phase coupler? 

@ Bo : The electrical switch room doesn't have any distribution smaller than 240mm2 Aluminium 4core swa for 2 of the 3 phases so it's not possible to 'link' them. I've worked on that size cable in the past but in this case it's simply terminated off a busbar and going straight out of the room :-)

thanks,

Eamonn

Eamonn:

"@ Robert: by 'dedicated adapter' did you mean the phase coupler?" Sorry, yes. I've got a head full with cold at the moment, my brain is on a go-slow.

Is the rule still in place that allows you 2 m of unprotected (unfused) cable inside switchgear / a cabinet? If so, could the coupler be installed in the busbar chamber? Failing that, (1) a set of busbar fuses, or (2) is there anywhere else where the phases come within reach of one another. There's nothing to stop you running the coupler off 1.5 mm² or less provided you have the appropriate protection in place.

Robert,

 hope you get well soon.....

You are correct, and I must take another look at the busbar; small outage might well be worth it. 

thanks for the ideas.

Eamonn

I was at the IBC broadcast technology exhibition in Amsterdam yesterday and Devolo, the HomePlug power line networking company, were there. With this thread in mind I asked if their products would work in a three phase environment. Without being able to provide particular detail the salesman said they would, due to induction between the phases.

When I got home I found these links that might be of interest:

http://www.devolo.com/uk/dLAN-Professional/dLAN-Powerline/dLAN-200-AVpro...

http://www.devolo.com/products/Professional-dLAN-Powerline/dLAN-200-AVpr...

Regards,

Steve

Steve,

 I really appreciate that. Thanks a lot for asking that question of them. I'd been thinking about the Devolo products,having heard they're performance is a lot better. I'll certainly test them out and let ye know how it goes.

Thanks again,

Eamonn

I used the Devolo dlan 550+ but it didn't work. I was able to occasionally get internet access in the electrical switch room, but could not get data to emoncms.

One thing I am going to verify (hopefully tomorrow) is that the root cause is lack of ethernet connectivity rather than some error between the emontx and emonbase......that would be embarrassing !

Do you guys know if I can use the oemgateway image to log data to SD card, so that I log locally while also attempting to send the data online?

The bufferedwrite (https://github.com/emoncms/emoncms/blob/bufferedwrite/readme.md ) solution is based on emonhub and that would require me to update the emontxV2 sketch and change the payload. Need to verify if I can change the continuous watthours sketch to add current, VA for each phase and voltage rather than just power.

Eamonn.

Hi Eamonn

How is your payload structured ? emonHub is intended to be a direct replacement for OEMGateway.

There maybe a distinct advantage in using emonHub if you suspect possible Ethernet/network issues, oemgateway's single thread can get bogged down trying to deliver to an unavailable target and miss incoming data, where as emonHub runs an additional thread for each target so if it is waiting for a http request timeout the main thread continues to log data.

Also emonHub's logging is far more verbose and may help identify any issue with connectivity.

Paul

Thanks Paul, makes sense to move forward(emonhub) rather than hang on to the past(oemgateway)  :-) 

Payload is as follows ; (from emontx CT123 Voltage sketch)

"typedef struct { int power1, apparent1, irms1, PF1, power2,  apparent2, irms2, PF2, power3,  apparent3, irms3, PF3, Vrms; } Payload1; "        

It's a long payload, and the addition of PF is overkill (given I've real/apparent power anyway).

I thought I'd need to change the payload as Trystan had done here to include msgNumber etc in the correct format, though he's motivation was prob to use the continuous sampling sketch ... http://openenergymonitor.blogspot.ie/2014/08/monitoring-solarpv-heatpump-and-house.html

The best approach for me may be to use the continuous sampling sketch, with emonhub, and shorten the payload to get the real power demand working. I can then update the sketch for current and apparent power later. 

Eamonn

If it wasn't for the fact emonHub is so heavily based on OEMGateway and my involvement with emonHub development I would be hesitant to change too, I was a big fan of OEMG.

emonHub will process a payload of 13 ints no problem using the default datacode settings and only if you choose to add any other datatypes would you then need to declare the nodes payload structure just as Trystan has in the example. 

Paul

Paul,

 thanks again for your input ( terrible unintentional pun....)

Could I just check 1 thing..... 

(i) in the emonhub.conf : to set the frequency, one should add '433' for 433Mhz as per https://github.com/emoncms/emoncms/blob/bufferedwrite/docs/setup.md and not just the shorthand '4' as it was with the oemgateway instructions at http://emoncms.org/site/docs/raspberrypigateway

Thanks,

Eamonn

Yes, the full 433 rather than just 4 is used in emonHub, the lack of documentation is on the radar but not yet tackled I'm afraid,

Paul 

great... just wanted to check.

I have deep (painful) memories of keeping system documentation up to date. It's practically impossible to innovate as fast as this and keep everything in sync......

thanks,

Eamonn

I'm trying to get logging working here locally (completely offline sports clubhouse( patchy gsm coverage ))

The date was way off on the pi ( 1970) and I needed to manually bring it closer using <date -s "3 Oct 2014 13:18:00">

Apparently if the date is too far off, raspi-config or dpkg configuretzdata won't fix it and this was what I was seeing. 

I'm now seeing an error "PHPFina:post timestamp out of range" in the emoncms.log but not sure how to correct it. 

also the feeds are updating but I can't visually see the data.

ok; she's logging now. This is really an amazing setup(emontx/emonhub/emoncms).

Think my error in the log file above was because I tried to export data and it didn't like the date timestamps....

In any case I'll give it some time and return next week to check accuracy with the Elster A1120

Also seeing negative kWh and kWhd feeds, though the power inputs are positive. Just logging the Wattage and then using the Power to kWh/d process to generate the feed. Anyone seen this before?

Eamonn

Hi Eamonn,

The time and date thing is a right royal PITA as both emoncms and emonhub rely on a time signal,the Pi's timekeeping (or lack of in some instances) is quite complex and is an issue for the read-only SD card image development.

From what i can make out the problems are predominantly apparent where there is no working internet connection at boot up, but cannot be absolutely sure as the exact symptoms or conditions are not always clear.

Jörg (JBecker) and I had a long PM conversation about this sometime ago and in his case the Pi was not reading the symlinked copy of /etc/fake-hwclock.data on the data partition at boot up and the Pi would default to unix time 0 (1970) until it had established a working internet connection, at which point the current time would be corrected and all would be good. but if a data frame was received by the rfm2pi before that time the incorrect timestamp is already logged by emonhub and passed to emoncms, which could cause some disruption to feeds. We got around that by deleting the symlink and recreating the original file, which I assume is still working for him but makes his install totally dependent on a internet connection as the fake-hwclock file is located in the RO file system so cannot be automatically updated,

I'm not 100% sure of the exact problem so I don't have a definite fix but the core of the issue is the read-only file system and lack of internet connectivity.

In your situation it may not be feasible to do the same fix as the internet connection maybe slower to achieve and therefore may benefit more from being able to update the fake-hwclock file in it's own time rather than being able to find the file at boot time, but I'm unsure if that's feasible or not.

I'm guessing the fact you can see the feeds update but not see the data maybe because emoncms is accepting the data and the feed is seen to update but then the PHPFina module is preventing the data being written because it has a acceptable window of -5yrs to +24hrs (from memory) so it's discarded.

Could the negative values also be a result of erratic timestamps?

If the sketchy internet connection is going to be for a prolonged period it may be worth considering a RTC or in the short term there are a couple of things you could try, like reverting the fake-hwclock to it's original location and add some form of method to update the time, adding the "rpi-rw" & "rpi-ro" commands to the cron entry for updating the fake-hwclock (relies on a good internet conn at the time the cron runs) or the shutdown procedure (no use if Pi's power is interupted) or a custom script to run a timer since last time update and then keep trying to update until successful (extra work) none of these are perfect (or even tested) but may help.

Paul

Paul,

 thanks that's really interesting. I have 2 scenarios:

(i) My 'powerline adapter install' is completely offline(posting to localhost - 8Gig SD card). I was able to use by iphone hotspot for a while( GSM coverage is patchy) to post above on this thread and try to debug the date issue. I think I could use that to make sure the inital setup is correct. 

(ii) Online (cat5 connection to router, posting to online server): With this pi, I changed the fake-hwclock.data file earlier, but when I rebooted the pi it reverted - can't 'make it stick' so your ideas above are worth trying. I also saw negative kWh feeds when all inputs were positive, but after I did an apt-get update & upgrade on my ubuntu 12.04 emoncms server earlier today it cleared up, though the date/time was always perfect on the server. So, not 100% sure if it was the root cause but clearly I had let it go 'stale'...need to manage the server better.

Is there a reason why I can't expand the filesystem with emonhub like I was able to with oem? I would like to do likewise(update&upgrade) with my offline installation in (i) above.

Also, feedwriter.log needs to be started manually - is that ok or could it point to a problem?

I'll do some testing and see if I can better document my exact use cases & workarounds .

thanks,

Eamonn

Is there a reason why I can't expand the filesystem with emonhub like I was able to with oem? I would like to do likewise(update&upgrade) with my offline installation in (i) above.

It's not emonHub but the the way that the SD card is set up for emonCMS that causes an issue, raspi-config doesn't know how to handle the 3rd partition and there is no room to expand the file system unless you move the 3rd partition to make a hole to expand into see RaspB+ and SD card Extend to manually expand the file system.

Also, feedwriter.log needs to be started manually - is that ok or could it point to a problem?

I think there might be an omission in the ready-made image, Jörg also had to carry-out these steps from this guide

sudo cp /var/www/emoncms/run/feedwriter /etc/init.d/
sudo chmod 755 /etc/init.d/feedwriter
sudo update-rc.d feedwriter defaults

The fake-hwclock seems to have a will of it's own most of the time, I've done extensive reading on npt and fake-hwclock, and it still baffles me :-)

Paul

thanks a lot for that. 

Eamonn

Hi Eamonn,

I did some research on the time issues see Read-only image time issues for some more info.

Paul

Paul,

 great; I'll validate this manually over the next few days since I've a recent use case.

Eamonn