Hi Paul,
 
> The DS28EA00 is supported in OWFS for temperature and PIO.

Yes, I noted that and it kinda encouraged me down this path as I knew that
having some extra I/O available at the sensing point would be a way to
achieve the control I expect to need if I don't go down the custom 1-wire
slave path initially. 
 
> Chain mode hasn't been implemented, since there was no demand 
> and since it fits a little awkwardly in our filesystem metaphor.

Yeah, I had stared at this long and hard and while it looks to be useful, I
figured that I'd just add a SMT LED to the board that I can control remotely
would be another way to identify sensors a bit like id done with racks of
servers and disks etc that can enable a special flashing LED etc so you can
find things when there are a lot of them that all look the same.

> There is a DS1821 thermostat chip. It's only sort of a 1-wire 
> device, with no ID and needing external circuitry to return 
> to 1-wire control.
> You would be better off using a microprocessor 1-wire slave.

Yeah, It's not really clear in my mind yet how best to do the actual control
aspect of my project. 

I have a number of sensors in the same areas as I have 19 under-floor
heating zones in total, that are fed from 3 manifold cabinets (4, 6, 9 zones
respectively). 

I've got at least 1 sensor in the concrete for each zone and several on the
walls in those same areas. I thought I'd also try and measure the return
temperatures for each zone and average that with the measured floor
temperature to get an idea of what that zone is doing in terms of providing
a heat source. Then consider the ambient room temperature before deciding to
what should happen next. So it's not just a simple on/off thermostat 1
sensor 1 control output problem.

Also as the system is now there is really only 1 heat on/off control per
manifold and you just tune the flows through each zone to hopefully balance
all the zones relative to each other so that overall the whole area is at a
comfortable heat. I have about 5 zones that I have to pay particular
attention to as the flooring material (laminated wood) must not change
temperature too fast or invalidate my warranty. So this was my main
motivation to consider all this. 

> Very nice. Even if your wire runs didn't return to the same 
> location, you could have joined the 1-wire networks with 
> owserver over TCP/IP.

Yes, but as everything cabling wise runs back to this central location and
as I found the LinkUSB adaptors I figured it would solve the potential star
network problem and also terminate each LAN into the one device.

> The NSLU2 certainly works. Something more recent like the 
> SheevaPlug is the same price and power consumption and has 
> 512MB ram vs 32MB.

Yes, although living in NZ means international shipping which can vary in
cost and the NSLU2 is close at hand. However, I have my eye on the GuruPlug
+ JTAG variant, but it's not out yet so I figured the NSLU2 should be
adequate in the meantime and may be good for the long term. I need this type
of control to be solid and reliable and if the NSLU2 can just sit there
doing the sensing and control part of the job reliably, that would be ideal.
I've got several other PC's and a NAS that can do the logging over time and
web serving etc.

> I'd use owserver as the interface to the LINK adapters, and 
> then communicate with owserver using the various OWFS 
> programs. This allows ad-hoc queries and debugging, while 
> also running a data collection and control process separately.

Ok

> Linking the two adapters is effortless, just list them both 
> on the command line.
> /opt/owfs/bin/owserver --LINK /dev/ttyS0 --LINK /dev/ttyS1 
> (Here I used the LINKs in LINK mode rather than DS9097U 
> emulation mode, and used the default port of 4304) Depending 
> on serial port permissions, this may have to be run as root.

Excellent! 

I was thinking I might have to run two owservers and then somehow merge the
readings in the owfs somehow but having owserver handle multiple LinkUSB's
makes it all simple again.

> To see your network:
> /opt/owfs/bin/owdir
> 
> To mount it as a filesystem on /mnt/1wire (with proper 
> permission) /opt/owfs/bin/owfs -s 4304 -m /mnt/1wire Then get 
> all your temperatures with grep -r '.' /mnt/1wire/*/temperature
> 
> You should also look at the alias file entry to give each of 
> the sensor ID a human readable name (like the location of the sensor).

Again excellent. This seems like it might be easier than I had thought -
however past experience tells me that there is _always_ some issue that will
crop-up along the way. I just don't know what it is yet.

> As for making web interface, and storing in a database, I'll 
> let other people chime in.

Cool

Alex


------------------------------------------------------------------------------
Download Intel® Parallel Studio Eval
Try the new software tools for yourself. Speed compiling, find bugs
proactively, and fine-tune applications for parallel performance.
See why Intel Parallel Studio got high marks during beta.
http://p.sf.net/sfu/intel-sw-dev
_______________________________________________
Owfs-developers mailing list
Owfs-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/owfs-developers

Reply via email to