Am 29.04.2015 um 18:35 schrieb Paul W Panish:
>
> Thanks for that, I'll implement the verification on write, I didn't
> realize that I could read from the output register without affecting
> state. This may be enough to get the levels of reliability I need.
>
Actually, the DS2408 PIO write command has the verification step already
built-in, see the datasheet, pp. 32-33. OWFS checks the 0xAA byte the
chip sends. You get an EINVAL if the verification fails.
But that isn't enough as the chip may have been power-cycled at any moment.
So verifing the chip's state any few seconds or so is required. In my
host programs, I have a (minimal) state representation of each chip and
there's a function which checks the correct state of the real device
against this model from time to time and reapply it if something failed.
For some chips as the DS2408 it's easy as the have a POR flag, for some
chips (like the DS182x) you have to simulate a POR flag by setting a
custom mode which is reset to default on power-cycle.
You know you want that, right? Because it gets you rid of the irritating
85°C reading - that's the power-cycle default of the temperature
register on the DS182x.
I cite myself again:
--------------------------------------------------------------------------
For other applications where 85°C may be a valid readout, I once posted
a way to tell the valid value away from the POR case:
* don't do 12-Bit conversions. This is the default in EEPROM, so it acts
as our POR flag.
a) do a 9, 10 or 11 bit conversion.
b) check the value returned
c) !=85 -> valid reading, return temperature
otherwise check further:
d) read scratchpad
c) byte 4 != 0x7f -> no POR -> valid reading, return temperature
otherwise
e) Power-On-Condition, need to repeat conversion.
If you really *need* to do 12-Bit conversions, you'll have to change the
default configuration register value in the EEPROM. There is currently
no way to do this with owfs, bus we could change that.
--------------------------------------------------------------------------
> Note that the only element in the system that gets hot, or even warm for
> that matter, is the DS18B20. Heating of connectors isn't a problem. When
> I first discovered this issue I replaced all the wiring, connectors, and
> devices to make sure everything was well connected and working properly.
> The results were identical.
>
Hmm, that's where I'd switch the chip and see if the problem magically
disappears.
> The HobbyBoards I/O board is where I'm inserting power for the bus. It
> also grounds pins 1, 2, 5, and 8 on the RJ45. I could ground the rest,
> but it doesn't seem likely that this would cause a temperature
> sensitivity such as what I'm seeing even if it introduces some
> additional noise through capacitive coupling.
>
The problem when powering the components through the bus is ground lift,
that's worse than the additional noise. But if your only loads powered
through the bus are a couple of DS182x, it shouldn't be the problem.
>
> I'm not doing anything programmatically with owcapi that would scan the
> bus other than at initialization when I read the directory. However, I
> don't know what owfs is doing in the background.
>
Nothing, really. As far as you don't use the kernel's w1 interface. That
one system may scan automatically, depending on its settings.
AS A SIDE NOTE, you have DISABLED the kernel's w1 drivers if you don't
use them? If not, you'll get a conflict between the kernel driver and
OWFS accessing the host adaptor at the same time, leading to exactly the
scenario you've described (magically disappearing chips).
> The BAE0911 sounds interesting, but seems to be severe overkill for the
> purpose, especially if I want to commercialize any of this.
>
Well, I use the BAE0911 commercially. Pascal has moderate prices, ask
him - he needs these commercial projects. And you don't have to use all
the functions of the BAE, it "just works".
Would be a pity if we had such a good onewire homebrew slave chip and
everybody is still playing around the limitations of the crappy DS2408
et al.
> It also
> doesn't seem to be available from distributors as a chip or board in the
> US, and I can't find anything on the RailBox Board availability
> anywhere. The only reference is the Brain4home.eu site which appears to
> be down.
>
Pascal is reading on this list. Or just ask him directly:
pascal.baer...@gmail.com
Kind regards
Jan
------------------------------------------------------------------------------
One dashboard for servers and applications across Physical-Virtual-Cloud
Widest out-of-the-box monitoring support with 50+ applications
Performance metrics, stats and reports that give you Actionable Insights
Deep dive visibility with transaction tracing using APM Insight.
http://ad.doubleclick.net/ddm/clk/290420510;117567292;y
_______________________________________________
Owfs-developers mailing list
Owfs-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/owfs-developers
