On 4 September 2016 at 19:59, Jan Kandziora <j...@gmx.de> wrote: > Am 04.09.2016 um 19:08 schrieb Colin Law: >> >> I have it to 3.3V ok. So 4.7k is ok for 3 wire mode but for parasitic >> it should be 1.5k >> > Yes. > >> >> To summarize then, for parasitic mode I should not need an external >> strong pullup and should be able to use just a 1.5k pullup and to >> specify >> dtoverlay=w1-gpio,gpiopin=4 >> and >> sudo modprobe w1-gpio pullup=1 >> >> With in addition the 3.3 to 5V level shifter as in your first post if >> any devices on the bus need 5V >> > Yes.
Excellent, thanks. I will write this all up somewhere ones I have got it all fully working. > > >>> There are two drawbacks when you use such a strong "weak" pullup. First, >>> the power consumption. I think that's negligible. Second, heating up the >>> sensor from inside by its bus output transistor. That's negligible as >>> long as you don't measure low temperatures. >> >> My rudimentary knowledge of thermodynamics tells me that if one >> provides a certain amount of power from inside as you suggest and that >> heats the device by (for example) 0.1 degrees when the device is in >> ambient 0C, then if the device were in ambient 50C then the heating >> would still be the same (0.1 degrees). Is that wrong? >> > It's a constant heat amount. Not a constant temperature amount. > Temperature inside the sensor is a derivative value. > > That said, when someone mentions the absolute temperature of something, > you can immediately say it's about radiation. Heat radiation works much > better at higher temperature. It goes with T^4. > > So, at lower temperature, changes of the sensor temperature in reaction > to ambient temperature will be much slower, and internally generated > heat can make a difference when sampling the temperatures often. > > For your application, you always have to check if radiation is dominant. > If your answer is yes, be aware absolute temperature matters. It is purely academic in my case as I will only be reading once a minute or so, and what you say is correct if radiation is dominant. Whether radiation is, in fact, dominant in the typical case I don't know. If conduction, either through the case or along the leads were dominant then I think my original suggestion holds and the temperature rise will be largely independent of absolute temperature. When estimating device temperature for devices mounted on a PCB the convention is to use a linear model (see [1] for example). This applies to devices such as transistors mounted above the PCB on leads as well as to surface mount devices. Quite how one would estimate the conduction/radiation relationship for a sensor on the end of a wire I have no idea. [1] http://www.rohm.com/web/eu/tr_what7 Cheers Colin ------------------------------------------------------------------------------ _______________________________________________ Owfs-developers mailing list Owfs-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/owfs-developers
