Re: 1-wire to USB converter that can use 1820s directly

[Marc F Hult <MFHult@xxxxxxxxxxxxxxxxxxxxx>]

On Mon, 17 Apr 2006 15:31:02 +1000, "Rod Speed" <rod.speed.aaa@xxxxxxxxx>
wrote in message  <4agnheFspmn9U1@xxxxxxxxxxxxxx>:

{snip}

This thread is a good instantiation of the rise to hegemony in usenet of the
lower quartile  ...

To summarize:

1) The four-conductor RJ-11 plugs that the original poster suggests are
conventional and useful for 1-wire installation. However 8-conductor RJ-45
plugs and jacks are ubiquitous so they would seem preferable in many
situations especially considering that dependable whole-house 1-wire
distribution sh/would best consist in multiple channels (see below). A
4-conductor RJ-11 plug fits into a RJ-45 jack, so RJ-45 patch panels,
available for less than $1/jack on eBay, provide a flexible, dependable,
inexpensive solution that is adaptable wiring infrastructure needs other
than 1-wire (audio signals, audio control, RS-232, video, computer network,
and so on).

2) Twisted-pair wiring (eg CAT3, CAT5) is preferred to non-twisted-pair
especially if a whole house is to be wired as the OP proposes. The twisted
pair provides better immunity to noise and more constant impedance than the
("satin") telephone wire suggested by the OP. The use of non-twisted pair
wiring such as what is connected to the telephone cables that the OP
proposes seems to be asking for poor performance if used for 'long'
distances. Also note that some phone cords are stranded and may include
cloth-like fibers making them difficult to solder and durn near impossible
to punch down (and so are potentially unreliable for these reasons).

3) Best practices for Maxim/Dallas 1-wire does not include multiple channels
(= "strings" = "runs") on a single driver. However the OP perserverates on
this topology, perhaps because he under the mistaken impression that the
only issue is available power for the sensors. T'aint. Others has suggested
that he inform himself by reading the 1-wire documentation, but that doesn't
seem to be his MO.

4) Maxim has a DS2482 IC that is designed to drive multiple channels. Dr
Peter Anderson www.phanderson has developed and sells nifty, inexpensive
PIC-based systems that also drive multiple channels. OP has been repeated
referred to these resources by respondents, but doesn't seem to 'get it'.

5) Despite what OP writes, Dr Anderson's site does indeed include schematics
and explanations for the devices in question.

6) Despite the OP's imperviousness to the concept that USB, RS-232, 1-wire
and I2C all involve serial communication and that real devices (whether in
one package or not) necessarily involve translation betwixt them, a USB -->
RS-232 converter + PIC--> multi-channel 1-wire solution (such as Dr.
Anderson's) would meet his need. The multi-channel IC offering from
Maxim/Dallas has I2C input and so needs a USB front end to met the OP's
needs.

7) Despite what the OP now claims, reasonable people reading carefully would
conclude that he is looking for a pre-built device that includes software
drivers for about $30 or less. I don't know of any that are multi-channel
and am reluctant to recommend a single channel implementation of what he
proposes. It could be done, of course, by daisy chaining each of the strings
through a hub, ( IN1-->OUT2 -->IN2-->OUT3-->IN3 etc)  but it simply isn't
worth the aggravation of a flakey data acquisition system in my opinion. The
cost difference of $20 or so is small compared even to the price of the
sensors required by the system. When one also considers consequences of
errors and time expended, the cost difference is trivial. Among other
reasons, a single break or poor connection in a daisy chained system takes
down the entire system. Multiple drivers and channels ("strings") isolate
problems, are more robust than a single channel and are easier to
trouble-shoot and generally more reliable.

8) It is generally trivial to adapt an RS-232 device to USB using an
inexpensive adapter. This has been suggested by several respondents, but
apparently rejected by the OP for whatever reason.

9) In the experience and opinion of many (myself included) the most readily
accessible programming interface for 1-wire is an simple ASCII protocol that
the hardware device understands such as those used by Dr. Anderson. This
allows the hardware device to deal with 1-wire timing and electrical issues.
The OP wants a USB "driver" for VBA, Excel etc (whatever he means by that).
Three good, very different approaches/sources for communicating with
1-wire-aware devices are 1) Homeseer with plugs, 2) www.windmill.co.uk and
3) StampPlot www.selmaware.com (I've used them all and can recommend.)

HTH ... Marc
Marc_F_Hult
www.ECOntrol.org