WebControl by CAI Networks is a $35 TCP/IP-based (web) device that I've used for about a year and can recommend... [snip]
.. I also tested various 1-wire temperature devices for them and based on that they added the more accurate DS18B20 1-wire device than the "econo" DS1822 device which was the only device they previously supported.
The DS18B20 looks to be sensitive to the input voltage when used in the simple interface mode that CAI is using. This means that one needs to be concerned about the voltage drop over the length of wire between the 5V source and the sensor. I have four different production lots of DS18B20's that I have tried. The most recent one I obtained on a recent ebay purchase does not work at all and I suspectg it is actually a -PAR device. I have one lot that tracks very nicely with other sensors that use a DS9097U interface in parasitic mode. I have two lots that produce unreliable results. Typically too high of a reading unless locally injected and then sometimes locks up when injected. I have tried both mods to increase the DQ midpoint and the Vdd and these did not make as much difference as changing the sensor. I suspect that a redesign to accomodate the parasitic power mode is not being planned by CAI so one's success will depend upon the DS18B20 they happen to have.
The ones I found that work with the Web Control are
http://cgi.ebay.com/Digital-Thermal-probe-...=item35a7edfb6e
Power can be be from the WebControl directly for short distances. For long distances where the voltage drop is excessive then a stable source of a litle under 5VDC can be used.
The DS18B20 interface on the WebControl is its weak point. I would recommend the WebConrol for just about any local control application, but not a great unit just to interface temperature sensors.
Several observations and some planned actions:
1) As I understood it, the initial problem with migrating from the ' "econo" DS1822 device which was the only device they previously supported' to the more accurate DS18B20 was a timing issue. (More bits need to be received from the DS18B20 than frm the "econo" version). For this and other reasons I am sceptical of the problem being fundamentally a power-supply issue.
2) I initially tested the Webcontrol with a full array of DS18xxx devices obtained directly from www.Maxim-IC.com. Two individual DS18B20's worked with the new fimware. But (as several others have also) noted, it doesn't with the pile of DS18B20's I purchased at low prices from various eBay distributors.
Without claiming to have reached any conclusions, I would suggest folks read Maxim's discussion about the rampant problem of substandard, off-spec and counterfeit ICs here:
www.maxim-ic.com/sales/counterfeit_parts.cfm
I note that the only ebay-purchased DS18B20 device that Michael and others have found works (Is this assertion correct? Any one else have other observations/data?) is the "Digital-thermal probe" referenced above and that it is the only one on eBay that explicitly describes the provenance of the IC (" Maxim-made"). Make of that what you will ...
3) It seem likely (to me at least)that the sensitivity that others have found with cable length is not (barring outright component failure) primarily power-supply related, but timing related owing in part to the increase in capacitance with increasing cable length. This has long been a bug-a-boo with 1-wire devices, in part because timing and signal waveform strongly depend on transmitter design as well as load capacitance. Maxim has a discussion and a variety of fixes here and elsewhere.
http://www.maxim-ic.com/app-notes/index.mvp/id/4104
http://www.maxim-ic.com/app-notes/index.mvp/id/148
Quote: A proper match among network components (i.e., master, network cabling, and 1-Wire slave devices, "slaves") is the precondition for reliable 1-Wire operation. A 1-Wire network is a complex arrangement of 1-Wire devices, communication lines, and connections. Every 1-Wire network is different, often both in topology (layout) and hardware.
4) I've ordered a fresh batch of known-authentic DS18B20's and will scope the output pf these and known problematic eBay ICs to assess whether there is an obvious timing/rise-time signal error. I may also see whether an output buffer and/or RC filter helps as described in the second Maxim Ap Note cited above help.
( Why I continue spend time on fixing/dealing with 1-wire performance issues after a decade+ of grief is something that maybe only a shrink can explain ....)
HTH ... Marc