No announcement yet.

xAPmcs1wire Temperature resolution?

  • Filter
  • Time
  • Show
Clear All
new posts

  • xAPmcs1wire Temperature resolution?

    I've suceeded in getting xAPmcs1wire version 1.7.0 installed and it seems to work well and reliably. However, I notice that the step increment of the temperature is only approximately 1 degree F, using DS 18B20 sensors. In other words, if the displayed temperature is 73.4 degrees, the next step increment upward would be 74.3 degrees. Unfortunately, I could really use more resolution.

    Do you take advantage of the 12 bit resolution mode in the 18B20, or do you have it set to use only the 9 bit resolution mode? If 9 bit, do you have any plans to change it to 12 bit?

    I noticed a setup parameter of "significant digets", and tried changing it. The only change I noticed was if I set it to "0", the displayed temperature has "0" decimal places, but if it's set to anything over 0, the displayed value has one decimal place. Is this working as expected?


    Gary Sanders

  • #2
    I did not implement lower resolution selection for conversion. It should default to max resolultion supported by the device. The database retains two decimal digits. The display is controlled by the setting that you identified.

    Take a look at the IO window to look at the raw data received. I will typically see numbers such as 5.125 for a B20 temp sensor where the value shows in degrees C.

    I can implement resolution selection, but it just add user setup complexity with the only real value being a reduction in the conversion time at the device.

    Take a look at the IO window data and see if a value with 3 significant decimal digits appears. This will confirm resolution is at max. The next step would be to look at any rounding that may be applied subsequently. You should be able to correlate the raw data with the delivered data messages to see if there is a problem in this area. Some of my sensors do look a little choppy, but most are smooth so I did not investigate a potential sw bug.


    • #3
      While following up on your suggestions, I looked a little closer at the sensors I'm using, and I want to make sure I didn't mislead you in my first message.

      One sensor is indeed a DS18B20. However, 2 of the sensors are DS1920's. The significance didn't hit me until I noticed in the IO window what ALL of the sensors are identified as DS1920 family, including the DS18B20. I did a quick Google search, and see that there are apparently significant differences in the way higher resolution solutions are aquired in the two sensor types.

      If temperature sensors are mixed in a single, parisitic powered network, will xAPmcs1wire identify the correct type, or do I need to stay with one type?

      I did the following:
      With 1 - DS18B20 and 2 - DS1920's connected, I looked at the IO window and noticed that all readings were either an integer, or ended in .5. In other words, readings at approximately room temperature were 21, 21.5, 22, 24, 24.5, etc. If I recall correctly, this would indicate low resolution mode for a DS1920. The Browser window, when set for centigrade, tracked the IO window exactly (same value, with one significant decimal digit) after each refresh. Changing the "significant digits" field to 1, 2 or 3 digits made no difference, unless set to "0".

      I then disconnected the DS1920's, leaving only the DS18B20 connected. The results stayed the same.

      Thinking that perhaps the earlier connected DS1920's created an incorrect configuration, I erased all ini files in the config directory and restarted xAPmcs1wire. Results stayed the same as the first test.

      I then used regedit to see if there was anything in the registry for xAPmcs1wire that would indicate a misconfiguration. I found nothing.

      I am concerned that I may have inadvertently misled you initially, but am still confused at the above results when I disconnected the DS1920's.

      Gary Sanders


      • #4
        It turns out that the resolution on the family 10 (DS18S20/DS1920) defaults to 0.5 C. Conversion time is about 200 milliseconds in this case. The resolution on the family 28 (DS18B20) can be programmed to use up to 12 bits resolution with conversion times going up to around 700 milliseconds for max resolution.

        I thought the default was max resolution for the DS18S20. I updated the software to explicitly request the highest resolution.

        Take a look at the 16 character serial numbers. The last two characters define the family. I'm guessing that all your devices are family 10. I have a mixture of both and run parasitic power.


        • #5
          Yes, all three of the sensors that I'm testing with are of Family 10. I just dug out the lighted magnifier and verified that the sensor I assumed was a DS18B20 (because all my spares and permenenlly mounted sensors are) is actually an apparantly older DS1820 that I got from Dallas Semi as a sample a LONG time ago. It was a challenge to read, because it's crimped into a male RJ45 connector. Sorry (again!) for the mis-identification of the part.

          I installed your revised code, and there is a drastic difference in the data output now. The IO window and xAP data now have 4 decimal digits of resolution, and the step interval is correspondingly much smaller.

          One last question - The "Significant Digits" field in Setup - Does it affect just the formatting of the displayed data in the browser window? I noticed that it has no effect on the transmitted xAP data.

          Thank you for providing the update to xapmcs1wire in such a timely manner. It's greatly appreciated.

          Gary Sanders


          • #6
            Yes the setup page resolution is only for the formatted text. It should be on the status browser page and in the DisplayText key of the xap message. The Text or Level key of the message will have the raw unformatted data delivered.

            There are some who are more concerned about iteration rate than resolution so I may be hearing from them now that the sample time just increased. My intention was to make it max resolution so if I need to deal with max sample rates it will be a new discussion.


            • #7
              so are you saying that if I load this newest code I cant have the resolution at .5 c anymore? I had some issues with james XAP plugin so I went back to MCStemp for a short time but that program drastically slows down my whole homeseer system so Ive loaded MCSXAP which works well and was about to update to the lastest mcs1wiretemp code till i saw this thread... the speed for me is VERY important seeing as how im monitoring supply air directly from a furnace and that changes very quickly... or is there an INI file on this such as in MCStemp where i can program my resolution? I really DONT want 3 decimals of resolution..
              PerfecTemp - the Most advanced HVAC system I've ever Built - and its in my House


              • #8
                I did not remember if you were using this one since there was not resolution settings such as I gave you for mcsTemperature. I'll add an ini setting to toggle between max and min resolutions.


                • #9
                  yes what i had done was to start up MCStemp long enough for the sensor to request a conversion at the low resolution once as the resolution setting is held in the sensor so that if it is once programmed at a particular resolution it doesnt need the resolution request each time a conversion is requested.. so for the xap1wire if i had one set at a hig hres it stayed there.. if i had sone set at a low res it stayed there regardless... I have 2 separate 1 wire networks... 1 with the furnace supply sensor by itself and another with all my other 1 wire stuff. im not sure what had happened with james plugin but for some reason all my XAP stuff was locking up... id see it on the hub fine but never in HS.. if I reset it would work for a while.. all seems fine with mcsXAP now.. will the revision get posted on this thread or another?
                  PerfecTemp - the Most advanced HVAC system I've ever Built - and its in my House


                  • #10
                    It is posted at the in the xapmcs1Wire subforum.