Archived Forum

Wow, that's a lot of spikes! I think 85 dec C is an error value. Check the connections between the sensor and the emonTx. Have you got another sensor you can try?  

 Interesting ; I'm seeing spikes also though I hope I'm not diverting from your problem which seems different.... For my application, I'm measuring 2 temps and every 10 hours, the reading spikes....see attached. Have you verified your connections - did you get this from the shop here or wire it up yourself? I would strip it down and check it ; as well as swapping to another emontx if you have one spare. Do you have a long cable run?

Eamonn

Actually, my spikes were at 10 hours intervals initially, but  to me, (for my issue) it looks like interference from the heat exchanger relay or something else that's 'kicking' in, in my installation. I bought the 2 sensors from the shop here but have extended the cable using standard 3 core non shielded cable. I think I need shielded cable....

Eamonn

Hi Eamonn

Sorry been away for the weekend, yes I bought the sensor in the shop. I've looked at the soldering connections and all look OK. Will be trying the emonTX I have in another location to see if it makes any difference. I'm wondering if it's the jack sockets on the board which are a little fiddly?

Alastair

Hi,

I've had the same issue in the past.  I think you'll find that it registers 85c when there is no connection.  As in, it thinks all the bits are low.  If you are only using 2 wires I suggest moving to 3.  Also I found having a larger gap between the readings can help.

If you have a look on the arduino play ground under the one wire sensors they have some info about these kinds of issues there.

The easiest option is to just discard anything > 80.

I'm in the process of moving my 1-wire network over to an Emontx (once I can get it working - see other thread). What you are seeing is not uncommon on a 1-wire network, rather than necessarily Emontx related. I currently run a 7 sensor daisy chain network over telephone wire into an eepc running ubuntu - and I was getting similar issues. Some people have reduced the problem by powering the 1-wire network separately from the data line (i.e move to 3 wires). I decided to just ignore >50C values for the house and outside sensors. For the solar HW and cylinder sensors (which may well legitimately see >80C values) I compare with the previous reading and throw away if there's a >10C difference. I read the network every 10 minutes.

OK, just looked at your PDF - that is a lot of spikes. I was only getting them 4-5 times in 24 hrs. What cable are your using,and what is the arrangement (daisy chain etc), are you operating with 2 wires ('parasitic') or 3 wires (powered)?

Dont know about the network topology (lenght of wiring, star or line) and does that meet 1-wire standards to comment directly on this problem, but as overall comment having built a 1-wire network of around 15meters in lenght with ~10 sensors for monitoring heatpumps and house heating system.

If parasitic design (two wires) there according to my memory are severe limitations and design recommendations to make it working ( http://www.1wire.org/Files/Articles/1-Wire-Design%20Guide%20v1.0.pdf ). Therefore I would only run the networks as powered (three wires) and even still  ensure there is adequate voltage in the network. 

Interested in to follow up this topic as starting to move the network gradually to emonTX and definately will only continue to use the powered (3 wire) design.