We did some hacking on the source code.  With a very basic hack to
change the delay on a simultanous temperature command, alarming on
bus-powered 18b20's now work.  I think this hack is not
production-ready, but it does appear to be a fairly easy fix.

--Jim

On Tue, Jan 4, 2011 at 3:34 PM, Jim Kusznir <[email protected]> wrote:
> Hi all:
>
> I've been reporting for a while that alarming was not working with
> DS18B20 temperature chips.  Today we have finally gotten the traces
> and compared to the Dallas data sheets to figure out exactly why
> they're not working.
>
> Thanks to the relatively new bus traffic debug feature, we were able
> to get this.
>
> It appears that all the alarming setup stuff works.  We did find a bug
> in the reading of the alarm values: owfs only reads from the
> scratchpad; it does NOT first copy from the alarm EEPROM to the
> scratchpad.  Therefore, unless you just set the alarm value, the value
> OWFS reads out is actually not correct.
>
> After setting Thigh and Tlow, we entered into a loop of echo "1" >
> simultaneous/temperature followed by a ls of the alarm directory.  In
> our case, the temperature chip always shows up in the alarm directory,
> even when it should not.  However, if we cat fasttemp for a chip, it
> will disappear from the alarm directory.
>
> Today's research revealed that the issue is that OWFS is NOT handling
> the bus powered situation correctly.  It looks like the alarming
> function would work perfectly if all of our DS18B20s were externally
> powered.  However, we run all of ours parasitic.  I suspect that some
> logic needs to be added to OWFS to "know" if it has at least one
> bus-powered chip (there's a simple OW command that can be run to do
> this, IIRC), and then it would need to execute the strong pullup after
> a convert command.
>
> Thanks!
>
> (BTW: our test version of OWFS 2.8p3 with owtraffic, used with the
> dallas USB bus master blows up randomly with cannot reconnect errors;
> 2.8p2 worked stably.  We did not check 2.8p4.)
>
> --Jim
> CASAS Lab Manager @ Washington State University
>

------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl
_______________________________________________
Owfs-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/owfs-developers

Reply via email to