James,

>
> Interested in hearing from people who know the DS1820 temperature sensor
> quite well.
>
> I've stuck two next to each other and read them using my PIC code.  I
> get different results.  How different should these results be?

  I've got a handful of these that I tested at both 0.0 & 25 degrees C.
  I found some major differences on the order of 3.6 C. I checked with
  my good old Mercury thermometers and they were all within 0.1 C.
  Now I'm not the greatest at soldering, so it could be argued that I caused
  some damage to them. I found once I applied a "correction" factor to
  the reported temperature they appear to work well. I'm surrounding
  them with a marine epoxy and using as submersible thermometers in an
  aquaculture project.

  I should note that the water baths used for the calibration were well
  mixed and I haphazardly chose the order that the ds1820's reported
  the temperature. So now I have an array of corrections factors that
  match to a particular romid.

>
> They say they are factory calibrated.  They claim a few things in the
> data sheets that I don't fully understand.  I calculate from observation
> that the relative measurement error is about one seventy fifth of a
> degree at about 25 degrees centigrade.  I know they make a claim about
> the absolute accuracy; and I've seen the pretty graph which shows the
> manufactured sensor population culling they might be doing.

Good Luck & Happy New Year,

James Widman
Research Fishery Biologist
National Marine Fisheries Service
jwidman@mi.nmfs.gov