I build my first emonTx over the weekend and running it with a Rasberry basestation. Firstly - what a nice system you guys have here, keep it up!
Now for the questions:
1) What is the lowest voltage level the emonTx would function at? I am currently powering the emonTx from two rechargable AA batteries, and after about 16 hours it stops functioning. After swapping batteries and fiddling with the (very low quality) battery compartment it starts up and keeps working again for about 16 hours. I've narrowed the cause to either the low quality battery container, or voltage level from the batteries dropping too low (2.7v based on the output stream before emonTx stops functioning.)
2) One of my CT channels seems to have a shifted bias - when not connected, or connected to a CT on a 0 current line it keeps reporting values between 36 and 40 W. Is this acceptable - or where/how can I check if I made some soldering errors? My other two channels do not seem to have this problem (one channel, measuring power usage to my Water Heater reports between 1-10W when no current is flowing).
Re: First build questions - operating voltage and 'floating' CT measurement
I cannot comment on (1) - but I think you'll find quite a bit of information on battery operation if you do a search (Google: "[search term] site:openenergymonitor.org")
As for (2), your problem is quite likely to be noise picked up on that input. It may be mains pickup from the cable itself (can you put the c.t. on the neutral rather than the line?) or from adjacent cables, or digital noise from inside the processor. If you download the appropriate tool from Robin's Mk2 Code variants and associated tools, that will show whether the input is sitting in the middle of the range and what the input actually is.
40 W is a bit high if you're not measuring voltage, quite a bit higher than we normally expect if you are. There's a lot of reading you can do about this:
http://openenergymonitor.org/emon/buildingblocks/measurement-implication...
http://openenergymonitor.org/emon/emontx/accuracy
http://openenergymonitor.org/emon/node/1385
http://openenergymonitor.org/emon/node/2311
Trystan did some wonderful animated pages demonstrating how the errors vary here: http://vis.openenergymonitor.org/dev/adc/noiseless, here: http://vis.openenergymonitor.org/dev/adc/noisy and here: http://vis.openenergymonitor.org/dev/adc/noise.mp4.
Re: First build questions - operating voltage and 'floating' CT measurement
Awesome, thanks for the detailed reply, will surely investigate. I'm not measuring voltage at the moment, and the offset seems to be coming from on the board itself (said input shows the same offset regardless which CT is plugged in, or even when no CT is plugged in.
Now if I can just organise some spare time to properly debug :)
Re: First build questions - operating voltage and 'floating' CT measurement
On question (1), you can try with a step-up booster. It's only a suggestion, because I don't have any experience with them.
On question (2), I have the same problem, and, at the moment, I didn't find a solution.
After reading the posts that pointed Robert, I have ran Robin's RawSamplesTool sketch with the following results
average of voltage samples 508.43
CT1 average current samples 510.85
CT2 average current samples 512
CT3 average current samples 514
I figure that differences are due to the two biasing resistors are not exactly equal, and that ideally the average reading should be 511 (512) on all channels.
Using the only_current sketch, with nothing plugged as load, I measure CT1=0 W, CT2= 15 W, CT3= 43 W. The readings seems consistent to me.
But running the CT123_Voltage sketch, with no load, the readings are CT1 around 30 W and often changing (+/- 4 W), CT2 and CT3 exactly zero (and not varying at all).
My questions: are the bias resistors the problem? Would be a good practice to select pairs of resistors? Why not use a common 2.5 V DC line for all analog inputs?
If the problem is noise, how determine the source (wires, pcb, atmega)? And how solve it (if it's possible)?
Thanks in advance
Re: First build questions - operating voltage and 'floating' CT measurement
If you read the first of the block of links, you'll see that if the bias sits in the middle of an ADC step, it needs a larger input (noise?) to switch levels and so create a change in the digital output. If it sits right on the edge of a step, it requires only a tiny change to switch. It's worse if you're not using the voltage input because the squaring process in rms turns all the changes into positive power. Nobody has found a certain solution yet. I think part of the problem might be that the digital and analogue supplies to the processor are common. Additional decoupling on the 3.3 V supply might help.
So, no, the bias resistors aren't the problem, but they can assist it!