Gas/Water/Electricity/Solar System/Temperature/Humidity sensor reading & logging

Jee Labs Talk » Let's hook it up

Gas/Water/Electricity/Solar System/Temperature/Humidity sensor reading & logging

(47 posts) (7 voices)

Tags:

No tags yet.

mars
I'm new to JeeStuff, have done a lot of daily blog reading and think JeeStuff can be very usefull for me.
In short: I want to monitor gas/water/electricity consumption, solar system production and possibly temperature/humidy sensors.
I have seen some examples of gas/water/electricity monitoring based on a CNY70 which should work in my case also, except for the monitoring of electricity: due to the solar system, the rotating disk of the electricity meter can rotate both ways. It even wobbles when the sunlight changes in strength and/or electrical appliances are switched on/off!
So counting the rotations of this meter is a bit more complicated, but should be doable with 2, nicely aligned CNY70's and building a quadrature decoder:
```
    ---------------------------------> disk rotation
    00000000011111000000 -> Left sensor value
    00000000001111100000 -> Right sensor value
```
```
    disk rotation <---------------------------------
    00000000001111000000 -> Left sensor value
    00000000011110000000 -> Right sensor value
```
The combined two-bit value can be 'decoded' to obtain:
- the direction (consumption/production of electricity) of the counter. - enabling/disabling of the counter .
A single JeeNode USB happens to have 4 ports, so this should be enough to hook-up 4 CNY70 sensors.
So far so good, but now comes the part I don't fully understand: The rotating disk might spin very fast, so I want to use the interrupt functionality for all sensors. Of what I have read, the 4 ports use the same interrupt. This means that I can't use the interrupt value (up/down) as input for the quadrature decoder, but that I have to tie the data pin of the CNY70 to the digital input pin instead, which leaves me with the question how to generate the interrupt for all 4 sensors if the data value changes??? In the ISR, I would then read all digital input pins, store those values, and have the mainloop watch for changes in the values, decode, count, and finally send the counter values in intervals to the system that logs these values. Is that the right assumption?
Furthermore, for logging wireless temperature/humidity/electricity sensors on the 868Mhz & 434Mhz bands, I need some receivers and connect them to a JeeNode. From the blogs in April, I understand that this is possible using the ELV RX868-3V, 868 Mhz receiver: (€13.55), ELV Shop A 434 Mhz reciever/transmitter seems to be unavaliable at the moment at both Conrad and ELV stores for some reason. Can I use any other receiver/transmitter? May it require 5V or should it be a 3V model?
If I would need both receivers, can they be attached to a single JeeNode, or need I two of them, one for each receiver??
And last but no least: logging & viewing. I don't want a full-blown computer running 24/7 to log all the values, therefore the small SheevaPlug thingies seem very nice for this purpose. The problem is that these run some Linux distrubution, and I don't understand Linux very well (understatement), but would it be very difficult to get JeeMon running and adapt it to the sensors I would like to use?
The alternative would be a JeeLink I guess that stores all the messages, attached to my Windows Home Server (which is only awake on demand!!) which collects the data from the JeeLink during the times it is awake. This solution would not be the preferred way, since the data/graphs are not really easily available and displayed on a remote display or anything else.
An alternative for JeeMon, and still using something like a SheevaPlug would be (as mentioned by somebody else) translating the messages from the JeeStuff to xPL formatted messages and using the standard available xPL utilities / databases / web-frontends to store & display the logged values.
Suggestions are welcome!
Posted 05/05/2010 12:36:22
jeroen

I tried the cny70 for my gasmeter, but I had lots of trouble (the reading distance is max 5 mm), now I am using the QRB1134 (reading distance 5-15mm), works fine. I have a old laptop running 24/7, for logging data, webserver, databaseserver, nas, printserver creating graphps, etc. Laptop uses 13 watts (avarage), and the OS is OpenSolaris. (great for data snapshots). I dont use the jeemon sw for this (most is in php and some C/perl/scripting),For me it is easier and more fun to write the code myself instead of using a toolkit with very limitid documentation. Logging data is done with rrdtool, which give me very nice graphs.
regards Jeroen

Posted 05/05/2010 13:52:46
jcw

Yep - quadrature encoding with two light sensors would be really neat. I'm going to have to look into this as well, since I'm planning to add solar here as well, and then I have the same issue.
I think you basically have the right approach w.r.t. interrupts, but don't underestimate the speed of an ATmega doing nothing :) - what I mean is, that the ATmega will poll at well over 1000x / sec, I think. So you've already got millisecond resolution. Which is probably enough for any amount of electricty normally going around in homes.
SheevaPlug + JeeMon is trivial (working right now).
One use of JeeMon I haven't gone into yet, is to have it push data to anyhing you like, including another xPL server (I didn't know much about xPL - just started reading...).
W.r.t. OOK receivers - I am looking into making some 433 and 868 units available in the shop (probably Conrad units). You can use other transmitters as long as they are compatible with 3.3V JeeNodes. You can power them from 5V if that's what's on the PWR pin - just make sure the output signal is 3.3V compatible (or add a 1 k? resistor in series to make it compatible).
As to how many units - it's a matter of counting I/O pins. Right now, I'm planning to use one JN in a box for 4 pulse counters / light sensors, i.e. electricity up/down, gas, and water - and a second box with some OOK receivers and maybe also transmitters. In my case, it makes sense to split them in two, because the meter is on the ground floor, but I want the OOK signals to be picked up roughly halfway up at Jee Labs, so it also catches the weather sensors on the roof.
@jeroen - thx for sharing. I've ordered a separate electricity meter to test with. Need to figure out a way to sense rotations and counter changes. As for "a toolkit with very limited documentation" - yep, unfortunately, that's currently the case :( - I'm glad you found a way around it.

Posted 05/05/2010 14:23:55
mars

@jeroen - thanx for that hint about the QRB1134 / CNY70!
@jc Very nice to read that you'll be trying to do the same as I would want to do. Your set-up for gas/electricity/water and a second with some receivers half-way the house, combined with weather sensors matches my wishes completely ;-)
Any idea when you will start that endeavor?
Furthermore I still would like to use interrrupts and have made some calculations about the pulse speed the quadrature decoder will face:
- The disk of the electricity meter is about 10cm in diameter, ie 31,4cm outer side - It takes 187,5 r / kwh, so 1kWh will spin the disk each 3600 / 187,5 = 19,2sec - If I take a rather high consumption of 6kW, the disk spins around every 3,2sec - 31,4cm every 3,2sec = a speed of 10cm / sec - the black part on the disk is about 1cm, ie is passed in 0,1sec or 100msec - if the two sensors are focussed 3mm apart, this means pulses are about 30msec apart
If I calculated the above right, it should be feaseble to use scanning, but I'm not quite confident on finding spurious misses if they occur.
Since I still don't understand how to use the same interrupt on all 4 ports, I wonder if a timer interrupt could do wonders; if the timer is set for say 5msec, and scans the input ports in every ISR, checks for changes, and saves the changes (possibly in a simple buffer) for the main program to process this, I guess it should work, yes?
W.r.t. xPL: well go for it, get a vendor id, write basic sensor messages (or more vendor (== you) specific records) holding the data, and the xPL Hub will do the rest for you (at least: it should...). I have seen versions working under Windows, so can't say anything about the xPL-Perl, xPL-Java and xPL-Linux versions nor anything about how simple/standard it is to store entries in a database and view them using a web interface.
Maybe iRanger would be a nice place of inspiration...
It would be so nice if one could use an Android tablet, controlling music and these kind of stuff. Some time ago there where ideas using Android smartphones that implemented xPL to control everything in/around the house.
Extending you shop with OOK receivers would be very nice to be able to offer a complete package to ppl that want to build their own FS20/KaKu/Oregon/etc receivers, but once again: I can't find these receivers in the Conrad shop (NL), but that's not the first time I can't find it, altough they do sell it...

Posted 05/05/2010 17:48:27
jcw
Starting is not the issue... following through and finishing is!
Thx for the xPL / iRanger pointers. Lots of new stuff to look into.
The interrupt setup is meant to be used with peripheral chips which you can scan. So if each port has say some I2C chips pulling the interrupt line low, then the code would scan all the devices to find out which one caused the interrupt. I'm not sure how far it scales, but for a few chips it should work. Haven't used it in this way yet, though.
The timer interrupt you propose ought to work IMO, but it's not very different from writing:
```
    MilliTimer myTimer;
```
```
    void loop () {
        ...
        if (myTimer.poll(5)) {
            ...
        }
        ...
    }
```
You won't get more performance from running interrupts. Nor even more resolution, if the code inside loop() takes little time.
Posted 05/05/2010 20:28:35
mars

I'm pretty familiar with that: starting, but never ending...
Nice to read that the timer should work, but might be not necessary. The fact is that I have no idea how fast the JeeNode is, and how much time the mainloop would take (interpreting data, make packet, send packet) en therefore could miss a sensor change for the quadrature decoder.
As for the 433 & 868 receivers/transmitters: I found them again in the Conrad.nl shop. According to your blogs about the OOK receivers, the Conrad versions seem to give better results than the ELV receivers, altough the real reason is unknown as you stated (sensitivity of the unit is less, or external antenna is responsible for this result, or ...), but the units do indeed work, and that's my main concern for now ;-)
BTW, how do I communicate with the JeeLink/JeeNode in Windows? Must the USB connection be 'configured' as a serial device for the communication to work?

Posted 06/05/2010 08:31:25
jcw

Windows - you need to install the FTDI drivers (which come with the Arduino IDE for example), IIRC. Then it shows up as COM port.

Posted 06/05/2010 09:34:51
mars

I've placed an order in your shop, so I've started this journey not knowing where it will end, but hopefully with a nice working setup somewhere this year ;-)
BTW, you did test the Conrad Cost Control a while ago before you had the OOKscope. Is there a chance that with the OOKscope you might discover the protocol they're using?

Posted 07/05/2010 18:42:53
mars

Oops, I meant the Conrad Energy Count 3000 not the Cost Control...

Posted 07/05/2010 20:30:06
jcw

I don't remember - did I try the Conrad Energy Count 3000? Do you mean the EM10? That works.

Posted 08/05/2010 11:06:00
mars

I meant this one, the ELV Cost Control, but Conrad has a different name for it.
And I did order 2 EM-1000-FM in Germany, but they cost almost tripple the price of a single Cost Control unit, so the ability to handle the Cost Control protocol would be a very nice addition to the JeeNode functionality!

Posted 08/05/2010 13:00:26
jcw

Alas, haven't been able to identify those packets yet. It's either using a different modulation mode, or its using much faster pulses. It would indeed be a great one to decode. High resolution, frequent data transmissions, and very low cost.

Posted 10/05/2010 10:07:48
sysrun

Need this?
http://fhz4linux.info/tiki-index.php?page=EM+Protocol

Posted 10/05/2010 10:34:26
jcw

Thx, but no - those are the FS20-like EM sensors. I've got those covered (thanks to that page, btw).

Posted 10/05/2010 11:05:45
mars

Pitty! Assuming the units use the 868Mhz band, and have the choice between ASK/OOK and FSK, we should be able to decode this nice little and cheap unit. If the OOKscope can't give any insight information, would it be a possibility to use the RF12 to make an FSKscope?
Ah well, first things first, otherwise I will start things again without ending...
BTW: What 1-wire temp sensors are supported? Can I use the DS82B20, or is the S version the only supported one?

Posted 10/05/2010 13:03:53
jcw

DS18B20 should work.

Posted 10/05/2010 13:17:11
mars

Great! Will order them from Ebay. Conrad asks €5 or more for the "S" version already. Shouldn't be more than €1,50 orso...

Posted 10/05/2010 13:31:52
fireson

Hi Mars,
Basically I want to achive the same as you read the gas, water and electricity meters. I was wondering if you could help me with a list of used components and some tips and tricks. I haven't touched a soldering iron for 30 years and not very familiar with electronic components. So any help would be greatly appriciated!
Regards,
Jan

Posted 18/05/2010 21:29:16
mars

I'm trying to 'run' 3 different projects:
1. Gas/Water/Electricity monitoring 2. Solarsystem monitoring 3. Central heating system monitoring / optimization .
All the information should come together in a several JeeNodes connected wireless to a JeeLink/PC combo. Since some things are new to me, combined with rusty soldering skills, and the fact that some parts are still on their way from Hongkong/Thailand/Germany, the projects won't be finished tomorrow I'm afraid...
If I focus on point 1:
- The watermeter is of Type 1, which causes a problem for the read-out: I haven't figured out yet how to monitor this one; - The gasmeter contains a nice silver rectangle on the '0' position, so a nice object for a photo reflective sensor; - The electricity meter is a ferarris type meter: very nice for my solar system, since it turns both forward and backwards. On the other hand this imposes a small problem because I can't just count the number of rotations, but I also need the direction. Using two photo reflective sensors and a quadrature decoder should solve this problem.
Hardware:
- 1 JeeNode; - 1 Case (Either the one from Dick Best as JC is using, or the probably similar one from Conrad); - 3 QRB1134 with leads - 3 stereo jack plugs (Conrad 746717 - 89 or 304314 - 89) & case connectors (Conrad 734136 - 89) for the sensors - 1 220V to 5V USB unit for powering the JeeNode.
I haven't figured out yet a very easy way to position the sensors before the meters...
Software libraries:
- RFM12b library to send data wireless to my JeeLink connected to a PC;
Custom software:
- Quadrature decoder for the ferraris meter: counting consumption and production - Simple counter for the gasmeter
As for progress:
Hardware: - JeeNode is in; - QRB1134 are in; - Case, jack plugs and power supply not ordered yet;
Software libraries: - RFM12 library available from this site
Custom software: - Quadrature decoder for electricity seems to work in a simulated environment; Not yet tested with the actual sensors...; - Gascounter same thing; - RFM12 library not yet used. Is next on my list.
I hope this helps ;-)

Posted 19/05/2010 11:41:25
tor

I've got a slight variation on the Gas monitoring meter reading - I could not find a way of reading the meter itself, but as most of our gas usage is from our 2 boilers I use a LDR in front of the pilot light window to sense when the boiler turns on (and off). When we had the boiler serviced the engineer gave us a printout of the flow of the gas through the boiler - so now I can very accurately measure the gas usage.
The advantage is of course that I do not need to work outside the house in the meter box.
Easily done with 2 Jeenodes and some Ruby software on the server side.

Posted 19/05/2010 19:35:19
fireson

@Mars, thanks for the info! I will try to set up mine too as soon as possible. Keep us posted on your results!
Jan

Posted 19/05/2010 20:19:00
mars

After 3 weeks the RF power monitors for my solar system finally came in:
Now finding time to solder the Conrad 868Mhz receiver, getting the OOK receiver code working, send it to the JeeLink, and reading it into the PC...
I guess that will take more than 1 hour of my spare time ;-)

Posted 25/05/2010 13:09:59
mars

Ok. Small problem...
I got the Conrad 868Mhz receiver hooked up (Analog Input, Port 2), and loaded the ookDecoder sketch, but I don't receive anything...
Looking at the specs of the receiver, it says LOW 0v, and HIGH +0.8V. The sketch mentions " // use analog comparator to switch at 1.1V bandgap transition".
Might it be that the +0.8V is not recognized as a logic "1", and hence I don't receive a thing as the JeeNode always detects a logic "0" ?
It is weird though, as the 433Mhz receiver from Conrad has the same specs, but that's the one jc used in his daily weblog (April 2010).

Posted 25/05/2010 19:45:53
jcw

Did you enable the pull-up resistor? Might just pull it over the edge...
One way to check what's happening is to turn a LED on and off to match the input signal. If it doesn't blink, you've really got no usable signal. A transistor or op-amp may be needed to resolve this.
I don't recall running into this sort of problem. Strange.

Posted 25/05/2010 20:24:01
mars
Yup:
```
    pinMode(13 + PORT, INPUT);  // use the AIO pin
    digitalWrite(13 + PORT, 1); // enable pull-up
```
...is what the Setup in the sketch does (I have nothing changed in the sketch from svn).
Posted 25/05/2010 20:35:34
joop

I am not sure what kind of receiver you use , do you use the RX868-3V 0552670c

Posted 26/05/2010 06:32:14
mars

Nope, not that one. JC uses that one, but I'm using Conrad #190939:
The data output circuit of the receiver looks like this: .
Might be the diode that's responsible for the 0.8V HIGH level??
And BTW: I have no idea what for transistor circuit could solve the level problem (if that's the problem, and not some soldering mistake of mine)...

Posted 26/05/2010 11:51:40
jcw

Another thing to try, if you haven't already: power the receiver from PWR i.s.o. +3V.

Posted 26/05/2010 12:06:46
mars

Ok. I will try that.
However, the problem seems to be very old as you can read in this dutch posting from 2004 about the Conrad receivers
A simple single transistor level conversion will also flip levels, so that's also something I don't want actually...

Posted 26/05/2010 12:26:06
joop

Mars, here is some more info from elektor I think it's almost the same circuit,but maybe you found this already.
The logic high level is at about 2/3 of the supply voltage, i.e., between 3 V and 4.5 V. this is different from your conrad device.

Posted 27/05/2010 05:35:18
mars

I tested some more with the 868Mhz receiver.
I soldered (with my bad eyes a real conquest) a simple schmitt trigger (two BC550's and 3 resistors) between the DATA OUT of the receiver and the AIO.
Then I disabled the pull-up resistor, and just put a resistor between the +3V and AIO: this does trigger the analog interrupt.
Then I tested the schmitt trigger by holding one of the leads of my multimeter against the input: and yes I got interrupts again.
Last but not least: connect the receiver again: NOTHING!
Grrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrrr..
So either I have a faulty receiver, the antenna is wrong, or I made some other mistake, but I don't have any equipment to check more than I already have I guess or does anyone have suggestions?!?!?!

Posted 27/05/2010 21:08:07
mars

@joop: The Elektor diagram looks very much the same as the one from Conrad.
Who knows: maybe the HIGH=0.8V should have been Vss - 0.8V which would be exactly in line with the Elektor documenation...
But still mine doesn't work. Maybe I'll try the 433Mhz receiver too to see if that one is working (as it is exactly the same, just other frequency) and I just have a faulty 868Mhz receiver...

Posted 28/05/2010 08:05:55
jcw

Maybe both - schmitt trigger and pull-up on its input. Clutching at straws here...

Posted 28/05/2010 11:42:14
mars

@JC: About the Conrad 433Mhz receiver: you didn't do something special I guess while connecting it to the JeeNode? Just connect the GND to GND, the + to the +3.3v, and the DATA out to the AIO?
Just to make sure I'll end up with exactly the same approach as you did and showed that it works using the ookDecoder.pde sketch!
BTW: while using the schmitt trigger, I disabled the pull-up in the sketch, but that has no result at all: I receive nothing, not from the EMx transmitters, nor from the JeeLink manually sending FSx commands on 868Mhz...

Posted 28/05/2010 11:53:50
jcw

I've added some extra pictures to the OOK plugs weblog post.

Posted 28/05/2010 12:05:21
mars

Great!
What I see is that you connected the ELV units to the +3.3V and directly to the AIO, where the Conrad unit is wired to the PWR, and you've used a resistor (1k?????) between the DATA output and the AIO input.
I didn't add the resistor in my first testing: might that have blown the OPAMP on the 868Mhz receiver resulting in the problem I have now?

Posted 28/05/2010 13:45:49
fireson

Hi Mars,
My equipment has arrived! But now I am wondering how to connect the QRB1134 to my JeeNode. Since I do not have a electrical background a probably need a simple 'how to' :) Do you have any tips and tricks? Maybe pictures or schemes...
Thanks in advance!

Posted 28/05/2010 22:24:17
jcw

Google is your friend - look for "qrb1134 datasheet" and "qrb1134 circuit". It has an LED and an opto-transistor.
I used a qrd1114 in my setup, which was built in 2008. See http://news.jeelabs.org/2008/12/08/ir-sensors/ and http://news.jeelabs.org/2008/12/12/better-mousetrap/
Basically it has a LED which needs to be powered through a 470 ? or 1 k? current-limiting resistor, and a photo-transistor, which is connected to an I/O pin with it's internal pull-up enabled.

Posted 29/05/2010 09:45:37
fireson

Hi Jean-Claude,
Would this schema be correct?

Posted 29/05/2010 21:06:38
jcw

Yes, that looks good. Make sure the pinout corresponds with your specific sensor. This will also work with 3.3V, I expect.

Posted 29/05/2010 22:47:41
jeroen

I have the same sensor. Using a 1k resistor for the diode, 10k resistor for the transistor. colorcoding = Anode=orange, Cathode=green, Collector=white, Emittor=blue
regards Jeroen

Posted 30/05/2010 07:52:41
mars

Note that the resistor value for the diode depends on the distance between the diode and the reflecting surface. The gasmeter normally is a short distance sensor, the ferraris meter on the other hand might need more power from the diode, hence a smaller resistor.
The datasheet should give the max/min values and afaik also a diagram distance versus mA for the diode.
PS. @JC: Could you give me the resistor value used for the 433Mhz Conrad receiver?

Posted 30/05/2010 10:20:21
jcw

I used 470 ?, but 1 k? would also be fine. It's not a pull-up but in series between data out and the ATmega pin. Reason is that this received is powered from PWR, so I added this to reduce the current in case the output signals swings above 3.3V.

Posted 30/05/2010 10:26:28
mars

Ok, thx. Going to try if this receiver does work.

Posted 30/05/2010 12:48:36
mars
Mars is happy!
The 433Mhz receiver seems to work. Used the images from JC, and added a 1k2 resistor between the data out of the receiver and the AIO pin of the JeeNode on Port 2.
If I set the JeeLink to 433Mhz, and send "88,4,0 k" for a KaKu command, I receive:
```
    HEZ C4DC455DC410
    HEZ C4DC455DC410
    HEZ C4DC455DC410
    HEZ C4DC455DC410
```
A mix-up with Home Eazy, but I think JC mentioned that already.
I assume I receive the right data now, but haven't figured out yet that this is indeed the same as the KaKu command I'm sending...
Gonna resolder the 868Mhz unit, who knows if it starts working after that ;-)
Posted 30/05/2010 14:45:57
mars
Attempt #3 seems to pay off: the 868Mhz receiver does receive the EMx protocol.
However, I stumble on 2 problems:
- Sensitivity / range is VERY BAD: 3-5M is the max. range. I have changed antenna length / antenna direction etc. but no luck
- The bits or nibbles of the EMx protocol in the ookDecoder seem to be reversed:
```
    EMX 20B0E0408001020408009000
is unittype 2 (shows '20') at address 5 (shows 'B0'). Address 6 reads as 'D0', although I'm not sure which unit is #5 and which is #6, so I might be wrong on that point.
```
I can't test the range for the 433Mhz receiver, since I don't have equipment yet that sends data on that band.
JC already mentioned the fact that either the Conrad receiver is better in surpressing noise or is just a lot less sensitive than the ELV receivers.
I now got the choice between the ELV RX868-3V (€13,55) and the superhet RX868SH-3V (€39,95). Some report that the RX868 is better, others that the RX868SH is better. I guess even the cheapest of the two is already much better than the Conrad version...
Are there ppl using the superhet, or is everybody happy with the normal one? And can anybody tell anything about range of these two?
Posted 30/05/2010 18:21:02
jcw

The decoders in the ookDecode.pde sketch all use a new common DecodeOOK class, which stores all results in the same way. My plan was to rearrange the bits and decode the different fields in either JeeMon or some additional ATmega code (either on the same unit, or on a receiving unit if this thing is used as a relay into a RF12 packet network). I didn't get to that yet - I admit that at the time, my only concern was proof-of-concept for a range of protocols. So its all still early days...
Can't help with the choice of receiver - I don't know much about the different ones.

Posted 30/05/2010 19:12:19