Instructions on how to update firmware can be found in the manual.
The new firmware can be downloaded here.
Source is here.
This (upper cabinet of refrigerator)
and this (lower cabinet of refrigerator)
First of all, this is quite a complex way to do this, it could have been solved in an easier way. But I wanted to test the modules Grodans Paradis AB made and put them to work.
The Kelvin NTC10K (on nickname = 5) have one thermistor connected to input one which is located in the upper cabinet of the refrigerator and the one thermistor connected to input two which is connected to a sensor that is located in the lower cabinet. Both has been set to send out temperature every ten seconds
As noted also the on-board sensor send out temperature data, It sends out the data with a longer interval (0x3C = 60) which is once a minute. This value can be used to sense the room temperature but is always a bit high as the sensor is placed on the pcb. One can adjust this in the sensor calibration registers
Most thermistor needs a calibration and here it can be set with a two complement number that can calibrate a temperature with an accuracy of two decimals.
If you want temperatures sent to some sort of central server this is actually all you need to configure. The values will be sent with the set interval. The default unit is degrees Celsius but this can be changed. In our case we also need some other features enabled. This configuration we do in the control registers for each sensor
Temperature Sensor 0 and temperature sensor 2 just send out the temperature with the set interval. They are set to 1 which means they send the temperature in degrees Celsius.
For temperature sensor 1 more features has been enabled. It is configured with 0xB9 which is 10111001 in binary.
As Bit 0/Bit 1 forms 1 the temperature will be presented in degrees Celsius as for the other sensors.
Bit 3 is set so low alarm is enables.
Bit 4 is set so high alarm is enabled.
Bit 5 is set which means that CLASS1.CONTROL, TurnOff events and CLASS1.CONTROL, TurnOn events will be sent instead off CLASS1.ALARM, Alarm occurred. So an OFF event will be sent when the temperature goes below the low alarm level and an ON event will be sent then the temperature goes above the high alarm level.
Bit 7 is set which means alarm events will be sent continuously, not just when a limit is passed.
For this to work we need to set the low alarm level
which is set to four degrees Celsius. And also the high alarm value
which is set to six degrees.
The hysteresis value is set to two as default and we leave it at that.
So what will happen here?
If the temperature is below four degrees Celsius CLASS1.CONTROL, TurnOff events will be sent out every second. This will continue until the temperature is above the low alarm temperature + the hysteresis, that means at 6 degrees. When the temperature go over 6 degrees Celsius CLASS1.CONTROL, TurnOn events will be sent out every second until the temperature goes below the set high alarm point minus hysteresis, that means 4 degrees Celsius.
In addition to the TurnOn/TurnOff events the temperatures will also be sent out for us to diagram or collect in a database or do other things with.
Before we setup the Beijing module to handle the TurnOn/TurnOff events we need to mark out TurnOn/TurnOff events so they can be identified to the refrigerator. We can do this with the nickname id (node id) which in this case is set to 5. But this value could be changed if a module is replaced but it can also be good not to use this id for other purposes as well. So instead we use the zone/subzone. For the Kelvin NTC10K we can set a zone/subzone for each sensor. And we set
the zone to hex 23 (35 decimal) and the subzone for each temperature sensor to it’s index. So temperature sensor 0 get subzone= 0, sensor 1 get subzone=1 and so on.
There is no magic here. One can set up any number schema one likes.
This is how a temperature event looks like for sensor 1
And this is how a CLASS1.CONTROL, TurnOn event looks like
Beijing I/O module
The Beijing module can do many things. But as we have a relay that control the compressor of the refrigerator connected to I/O channel we are only interested in configuration of this channel.
The I/O channels of the Beijing module is set as outputs by defaults
We have set bit 1 as an input here for a task we want to go through in this write-up (refrigerator door sensor).
Note also that the nickname (node id) is set to 4.
We also need to match the zone with the one we set for the Kelvin NTC10K (0x23) and also the subzone (0x01) for I/O channel 0 .
That is all we need to configure.
To complete the task we also need to program the decision matrix of the Beijing module. In this matrix we program what the Beijing should do when certain events is received by the module.
In our case there is the two events from the Kelvin NTC 10K module we are interested in.
If CLASS1.CONTROL, TurnOn is received we want I/O to be turned on or stay on, that is the compressor of the refrigerator should be turned on.
If CLASS1.CONTROL, TurnOff is received we want I/O zero to be turned off or stay off, that is the compressor of the refrigerator should be turned off.
The decision matrix have actions it can perform when an event matches a selection criteria. For us the SET and CLR actions is what we need.
The rest of the actions are described here.
We program the decision matrix like this. One row for each action we want to be carried out
In the first row, which looks like this
we enable the row, don’t check the originating address as we discussed above, and the originating address is not hardcoded, the zone of the received event should match the zone of the our module (0x23) and the subzone should match that for the channel set in the action parameter (I/O 0 = 0) or the module subzone depending on chosen action.
the class mask tells which bits of the class filter that is of interest. Here all is of interest (0x1FF) and the class filter is set to 0x1E which is 30 and is the code for CLASS1.CONTROL
The type mask tells which bits of the type filter is of interest. Here all is of interest (0xFF) and the type is set to 0x05 which is the code for CLASS1.CONTROL, TurnOn
So if a CLASS1.CONTROL, TurnOn event is received with zone=0x23 and subzone=0x01 the Beijing module will carry out the chosen action. We want I/O 0 to be turned on so the compressor is turned on and the action set output of I/O channel high will do that for us. This action expects the I/O channel as action parameter so we set it to x00.
The DM row for the CLASS1.CONTROL, TurnOff is then setup in the same way.
Thats it. We have now made a self contained unit consisting of two modules that can work autonomously.
The fridge project explains how to share the generated data to the cloud. I will do another write-up about this later. But it’s not more to it then to program the decision matrix of the VSCP daemon to send data to the cloud. You have live data for this setup here as a sample. This data is on a CAN4VSCP bus in this house and is collected by a Raspberry Pi and is sent up to ThingsSpeak.
VSCP have other means of displaying data and more will come in the next version.
The refrigerator again. It takes more time then expected of course. It always do. This time because I have a lot of other things to attend to at the moment. I understand if you ask some questions about this project. So I answer them here (before you actually ask them)
Why this project?
Our refrigerator broke down two weeks ago. And we need one also in my family. Of course we do, open source programmers may not afford much buying stuff but we also want our milk to be cold. Luckily it is winter here now and we have some other storage that is at about four/five degrees so it is no immediate panic, but as weather is getting better (the spring is coming) this is not a good long-term solution.
When our fridge broke down some years ago I decided to try to rescue it and if it was possible I would take standard VSCP modules to do so. The compressor was OK and if it is this is a simple project mostly. This project has worked perfectly well since then and became a nice VSCP demo project.
Yes, Yes, YES it’s an overkill using VSCP standard modules for this. An Arduino or some other simple board would have done the job. But I wanted to use the modules to see if they did their job also over time. The best way to actually see if things are working the way you thought is to put them in some critical system in your house (the heating in our house is already controlled this way). If there is some bugs or design problems things will be solved by necessity. To have them in a demo system on a workbench is just not the same thing.
So for the refrigerator the first question was to check if the compressor still worked. If it was dead it is just a pieces of junk and not a refrigerator. But just as with the fridge it was the control logic that had broken also this time. Make me wonder how many perfectly fine refrigerators/fridges are thrown away out there that could have new control added for a few bucks and do the job they were designed for for many years more.
As with the fridge I decided to go for standard VSCP modules for the refrigerator project to make it a demo project. It is easier this time as I have the CAN4VSCP bus connected to the fridge that stands next to the refrigerator. I just need to connect the two together.
So I put in a Kelvin NTC10K module and a Bejing I/O module in the refrigerator and hook it up to the CAN4VSCP bus so I am able to monitor temperatures and change temperature settings and get possible alarms from the unit. As with all VSCP modules they form a self-contained unit. There is actually no need for a server after they have been configured. The second video here show the configuration process and the first show how the modules are connected together.
The Kelvin NTC 10K module is a module that one can connect a couple of temperature sensors to. It can be programmed to alarm at low and high temperatures and as in this case turn on and off things when certain temperature thresholds are reached.
In the fridge project a Paris module is used together with the Kelvin NTC module. The Paris module is constructed to control relays and have all the electronics on board to do that. It also include protection timers and a lot more.
In the refrigerator project I decided to use the Beijing module instead of the Paris module. Mostly because it did not matter to have standard I/O channels as I use a solid state relay to switch on and off the compressor, but my main reason was that I got a chance to use it in a real life demo system.
So the Kelvin NTC10K module sense the temperature in the refrigerator and the Beijing module is used to switch the compressor on or off and to sense if the refrigerator door is open or not and if it is light the lamp inside the box.
It is possible to sound s siren if the door to the refrigerator is opened between midnight and six in the morning. But some things just hit you right in the face so I will not implement that functionality. At least not now.
Connect cables and putting it in the refrigerator where the old control logic was located. No programming is needed just configuration. The second video above show a bit of this configuration but I will go through it in a more precise way in a future post. Hopefully things will work as thy should tomorrow. The VCSP modules are very flexible and can be used in most control situations to form self-contained systems that can be connected together and be connected to the world in a safe and secure way.
Testing Kelvin 1-wire
In the lab, brewing, the Kelvin 1-wire module. Four channels times two sensors plus one on board sensor. That is 8 + 1 temperature channels with all the features on the Kelvin NTC10K. Hopefully, if all goes well (yes it works already but manuals, mdf’s, testing is needed), it will be available in the FrogShop from next week.
I ordered the PCB’s for this module series for almost 1 1/2 years ago. Thinking to much of my ability to finish stuff. One learn. Hardware and firmware and software takes time when you are a small company. The Frankfurt BLE (bluetooth) and the San Francisco (barometer) module is still left to do in that batch of PCB’s. And many other such as Frankfurt Eth is in the queue.
Conclusion, life is to short…
2016-03-30 AKHE – Made code work for PIC18F26K80
2016-03-29 AKHE – Raw A/D register values was displayed in wrong order. Fixed.
In the manual the sensor calibration part has been corrected and extended. It previously described the indexed calibration value version which was present in the first version of the module.
- The Paris module module firmware has been updated to version 1.1.5.
- The Kelvin NTC10K temperature module firmware has been updated to version 1.2.5.
- The Beijing I/O module firmware has been updated to version 1.1.5.
- The Accra counter module firmware has been updated to version 1.0.2.
- The Odessa expansion module firmware has been updated to version 1.0.2
- The Vilnius A/D module firmware has been updated to version 1.0.3.
Please update your module.
Instructions on how to update the firmware is available in the modules manual.