DS1820, DS18B20 Support

 I have two DS18B20 probes that I connect to my 1wire bus. I have noticed they only display 0.5C resolution in my WC registers. As I understand it both models only have 0.5c accuracy but the DS18B20 has selectable 9-12 bit resolution.
 
If I connect a DS18B20 part and only get 0.5c resolution,
- does this mean only the DS1820 part is supported?
- is support for the "B" part coming in the near future?
- is there a way to change the resolution of the communication?
- was the 0.5c resolution kept intentionally?
- is support for the DS1820 being or has been discontinued?
- am I missing something on this unit's usage?
 
Used for outside temperatures 0.5c resolution is just fine IMHO but for inside temperature measurements 0.5c is too coarse. O.1c is needed for many applications like, duct temperature measurements, room temperatures or room temperature proportional balancing.
 
I understand this device only offers 0.5c accuracy but the last few 0.1c is relative to comfort or comparison anyway and can be human compensated for.

12bit temp sensors are 0.5C accuracy, according to Maxim datasheet. That is in DS18B20 datasheet.
If you consider the temp range of -55C to +125C, that is the max resolution they provided to us.  We can not do more than what they provided.
To have higher resolution, DS2438 has 13 bit resolution, that allows 0.03C increment, in the display, we show 0.1C resolution in WebControl gui.

However some confusion still exists. The min/max scale is not the resolution transmitted or portrayed at the WC end.
12 bit resolution provides 0.0439c resolution minus any unusable overscale.
 
      (125c - -55c) / 2^12 = 0.0439c
 
Surely they don't scale the raw output for +/-1,024b.  (0.5c x +/-2^11) and only use +125c/-55c ???
 
This is still unanswered why more resolution is not displayed in the WC.
 
Thanks!

If you look DS18B20 datasheet, its first page stated the sensor accuracy is +/- 0.5C
In our code, we read all bits back, then based on the bits value convert them to the 1/10th C resolution and store them for display.
However, in reality, most DS18B20 sensors only show 0.5 temperature jumps. 
 
Same logic being used for DS18B20 also used for DS2438, you can see that change is in 0.1C increment.

Thank you for that info.
 
I wonder why they would offer a new chip with 9, 10, 11 and 12 bits if they only use 9 bits??
        +/-128c / 0.5c resolution = 512 = 2^9
 
Maybe a compatibility between parts on the bus  issue?
 
I don't know the 1wire protocol. Is the Tx bit length being set by WC drivers?

DS18B20 has been allowing 9,10,11, and 12 bit from its start.  By default, DS18B20 is working at 12bit resolution.
Supposedly 12bit would get 0.0625C, or 0.1C, but in reality, that sensor only reply back to 0.5C change, matching its max resolution description.
I think that sensor internally might do some average to reduce noise level.  Its real useful resolution is probably 9bit.
 
DS2438 does work in 0.1C/F resolution reliably from our testing. For higher resolution, that might be worth to consider.

The few DS18B20 probes I have will be good for basic room (freeze detection and interest) and weather but not a good choice for HVAC usage.
 
Thanks for the heads up for all to see, here.

Larry, Good discovery for the temp sensors in bits! For anyone wanting to get 0.1C control, your discovery will be helpful.

Accuracy != Resolution
 
The controller logic must write a configuration register to the DS18B20 to make it change from its default resolution of 9 bits to full 12 bit resolution.  The conversion time will be greater with 12 bits so the update time will take longer.  More information here http://www.maximintegrated.com/en/app-notes/index.mvp/id/4377

http://datasheets.maximintegrated.com/en/ds/DS18B20.pdf

Automate,
 
From DS18B20 datasheet page 8 first paragraph, it stated:
The power-up default of these bits is R0=1 and R1=1(12 bit resolution).
 
So we did not write anything to change that.  Unless Maxim datasheet is wrong, that is what we are based on.

What kind of update rate do you get?  Their table shows 3/4 of a second for 12 bit.

I don't honestly understand what you guys are claiming.
Firstly, as has been pointed out,   "accuracy" != "precision".
Secondly, the DS18B20 sensors I've been using for many years, paired with the WC8 boards, are perfectly capable of (and always have) 0.1 deg C resolution.
 
12-bit resolution is 1/16 degrees C, or 0.0625C, the WC software chose to round that to 0.1 degrees which is still adequate for the majority of purposes.
I've dataloggers and temperature monitoring and control stuff all over the world using these sensors and the WC8 boards and they're ALL operating perfectly happily with 0.1C resolution/precision.

Automate,
 
We do not pull the result at fixed time.  We sent conversion command to the sensor and wait till it is ready, read time slot reply 1, then we read its value.
I don't know if reading before the conversion finished would get less bits, before the conversion time finished, it probably won't generate correct CRC.  We do CRC checking on each temperature read from the sensor.

Firstly, as has been pointed out, "accuracy" != "precision".

Secondly, the DS18B20 sensors I've been using for many years, paired with the WC8 boards, are perfectly capable of (and always have) 0.1 deg C resolution.


Semantics aside, my two DS18B20 sensors with WC v3.02.18a and v3.02.18b only report in 0.5c resolve accuracy on my WC status page.  My DS2438Z reports 0.1c increments.  Is it possible that if I was sold DS1820 instead of DS18B20 parts, could they even work in WC? Is it possible?
 
The specs sheets seem to avoid clarification on this. I see accuracy stated as 0.5c and 12 bits of Tx, implying better resolution, but no actual statement giving resolution accuracy of sensing.
 
Maybe we have stumbled on to something here. What WC version are you running that reports better than 0.5c increments from the DS18B20 sensors?

LarrylLix said:

Semantics aside

Click to expand...

 
Sorry, you cannot just sweep aside critical differences as "mere semantics".
Accuracy has nothing to do with Precision.
 
If I wave my hand around to feel the air temperature and pronounce it is "36.2266451 degrees", that is PRECISE but it is NOT ACCURATE.
If I said the air temperature when I went outside to start work was "2 degrees", it is NOT PRECISE but it was ACCURATE.
 
 

LarrylLix said:

my two DS18B20 sensors with WC v3.02.18a and v3.02.18b only report in 0.5c resolve accuracy on my WC status page.

Click to expand...

 
Here's a couple of different versions:
Version: v03.02.07

Temperature Sensors

48.1 C


42.5 C


39.6 C


40.4 C


41.1 C


35.6 C


46.6 C


57.1 C

 
 
Version: v03.01.02

Temperature Sensors

61.6 C


56.3 C


52.1 C


72.0 C


48.8 C


50.5 C


49.5 C


24.6

 
Version: v03.02.18a3

Temperature Sensors

7.1 C


7.5 C


unbound


unbound


unbound


unbound


unbound


unbound

 
 
 
 

LarrylLix said:

Is it possible that if I was sold DS1820 instead of DS18B20 parts, could they even work in WC? Is it possible?

Click to expand...

 
I *THINK* the two parts are interoperable, but 1820 I *THINK* only does 9 bits, the 18B20 does up to 12 bits (selectable)
 

LarrylLix said:

I see accuracy stated as 0.5c and 12 bits of Tx, implying better resolution, but no actual statement giving resolution accuracy of sensing.

Click to expand...

 
"resolution accuracy" is a nonsense term. It doesn't exist. Please refrain from using it as it merely confuses everything.
Metrology is a science. Precision and Accuracy are different things and cannot, nay MUST not be used interchangably.