Archived Forum

For (3), you were looking in the wrong place. The jump leads have to be downstream of your c.t! ! ! Has the cat put the fire on in your shed when he got cold overnight?

My money would be on 2 or 4 (4 being the emoncms database has got corrupted). Are we talking about emoncms.org or a local copy running on your RPi? If it's local, what is the database on (physically)? Have you expanded the graph and checked that it really is continuous high power on every reading?

Is it significant that it's Sunday - Monday - Tuesday both times? I'm not sure, it didn't happen at the beginning of this week.

I have read somewhere (possibly Martin Harizanov or JeeLabs) that the RFM12B can be overloaded if the transmitter and receiver are too close. I can't remember where I saw it, nor any numbers, but it might be worth trying to find a reference. However, I'd expect that to be a permanent problem, not a few days at the start of a week. If it is that, the transmit power can be turned down in software.

Other than that, as you say the chief suspect would be the mobile phone repeater. If you have a longer Ethernet cable, I'd move that further away as a start.

Robert,

Thanks so much for the reply, so for not coming back to you- been a bit distracted by other projects.

The data I showed you is from the emoncms.org feeds, not the raspberry pi. I was hoping to look at the pi and see what data was there but I'm having other problems now! (A new forum post may follow!).

I have now got a longer ethernet cable so hopefully this will eliminate the potential issue with the mobile signal booster.

I'll let you know if I get the problem again.

Chris

All,

I have recently got my OEM system up and running again after resolving issues such as SD card failure and assembling various bits need for real power and temperature monitoring.

Sadly the problem I raised at the beginning of this thread has reappeared. After running for 14days, I now have a power reading permanently above 2.5kW. I know that this isn't right, because I have a cheap energy monitor which says I'm using significantly less than this.

Some updates;

• Now using RPi Gateway method, just using emoncms as the server
• Mobile phone booster is a couple of metres away, but it is now off, and the problem is still there.
• I try resetting the RPi and the emonTX (including reuploading the firmware) to no avail.

Some more info;

RPi is right next to the wireless router- is this an issue?
EmonTX is 1.5m away, sitting on top of distribution board

I have tried reducing the transmitter power on the emonTX as per here;
http://harizanov.com/2013/07/reducing-rfm12b-transmission-power/

Can anyone suggest what could be the problem?

•  Problem on the server?
• Problem with the RMF12B? (on the emonTX or the RPi?)
• Wireless router too close?

Chris

You've got to do some systematic, and quite likely meticulous, checking. I'd suggest you carefully and coldly check absolutely everything, including all the things that could not possibly go wrong. These are the sorts of things I would look at:

Start with your emonTx. Are you certain that it is sending the correct values? Have you plugged a laptop in and checked what it is sending? You said you only changed temperature sensor addresses and node numbers. Are the node numbers inside the legal range of 1 - 30? Is the group correct? Is the frequency correct?  Can you reprogram it to send test data?

If that's OK, Is the RPi receiving data in line with the emonTx transmissions? and (bearing in mind I don't know the ins and outs of a RPi) is the RPi receiving the correct data? Is it receiving only from your emonTx? Are your Node ID's correct? Does everything go quiet if you switch the emonTx off?  

Can you send test data by hand from a browser, and is that received correctly by emonCMS? Can you send test data by hand from the RPi, and is that received correctly by emonCMS?

I've had a NanodeRF a foot or so away from a wireless router and not seen a problem, however a NanodeRF isn't a RPi. It just means I didn't have a problem and the radio part is probably OK.

Hi Robert,

Your message made me realise something very obvious - the temperature readings from the emonTX are fine. So it can't be an interference issue, no a problem with the RPi, as this would affect the temp as well. I didn't use to have temp sensing when the problem first started, so hadn't occurred to me to check this.

That also shows that there is nothing wrong with the network group settings on either device.

Now since the system has been working for the last 14 days, that kind of rules out a problem with the code in the emonTX, which wouldn't suddenly stop working.

So that points in the direction of something wrong with the CT clamp. I just pulled on the cable (which goes into my distribution board), and pushed on the jack into the emonTX. The power reading dropped to 250W, which is what I expect it to be!

I'll do some more investigation but it seems like there is an issue with either;

• dodgy connection into the emonTX
• dodgy CT clamp
• CT cable picking up inference from other cables in the distribution board? Induction? 

This would explain why I have struggled to repeat or be able to predict when the problem would start again. Sometimes uploading fresh code to the emonTX seemed to fix it (sometimes not)- in reality it seems I was probably just moving the cable around when I repositioned the emonTX such that it stopped the problem.

Thank you!

Chris

Chris,

Your mention of pushing on the CT cable plug reminded me of an instance where I had the same problem. (abnormally high readings) it turned out the plug was not fully seated in the jack. At first glance, it looked OK, but closer examination revealed it was not fully inserted into the jack. After I seated it fully in the jack, the reading returned to normal.

It was only after I looked at the connection closer, and saw part of the plug sleeve did I realize it wasn't fully seated.

Regards,

Bill

Why the second thread http://openenergymonitor.org/emon/node/3381 ? You're confusing me.

I think the favourite is dodgy soldering. Which c.t. is it?

Unclip the c.t. from its mains cable, unplug from the emonTx and dissect.  The cable screen on the plug end might/might not/might occasionally touch the sleeve connection. I think it's best if it is properly connected there. The two inner cores should connect to tip and sleeve. There should be no connection to the ring.

If it's a YHDC, at the c.t. end, open the c.t. wide, and you'll see two plastic clips hooking over the bobbin cheek. Using three hands, hold the c.t. in one, push up on the cable with the other and with the third ease the clips away and the bottom half-core and bobbin should slide out without damage. Inspect the joints and resolder if necessary. The screen should not connect to anything.

Piccy: http://openenergymonitor.org/emon/sites/default/files/yhdcreport/yhdc06.jpg of the 100 A version. If it's one of the others with an internal burden, then you should have a resistor instead of the two zener diodes. If it's that which is dodgy, there's your answer.

Bill's point is a good one too. If you can see any sleeve, the plug is not fully home. New sockets can be very stiff.

Edit:

This plug is not fully home. In the bottom picture, I've unscrewed the cover to make it clearer. If you look closely, the contact springs for the sleeve and ring are resting on the insulation between the contact surfaces.

To make it easier for someone with the same issue to find it (rather than it be buried at the bottom of a longer thread). Maybe it's not- my bad.

Thanks Robert and Bill. I will have a look at this. Seems feasible.

Why would this lead to high readings (approx factor of 10 out)? What's going on here?

Chris

I think one side (sleeve) of the c.t. is floating, and the tip has a connection to the ADC input.

Which c.t are you using? If you have a c.t with an internal burden, it's simple: without a low impedance or indeed any connection to ground to you're squirting pickup straight in to the high impedance ADC input, and you probably get a huge input. If you have the 100 A c.t and the burden on the pcb, then you're squirting the pickup voltage into (effectively) a voltage divider formed by the stray capacitance to wherever and the burden resistance (assuming the "earthy" end of the burden is a reasonably low impedance to ground, which it is). It would be hard to know what the exact effect would be.  I certainly wouldn't dismiss it.

3.5 kW  ~= 15% full scale ~= 150 mV at the ADC input.