It also turns out that I have the relay layout mirror imaged, so that is a good find.
I've fixed the parts list and the spreadsheet.
The off board connections will use screw terminals. There is a power jack for an unregulated supply and an RJ-45 socket for a connection to the 1-wire network.
The board will be double sided with a VCC plane on one side (actually I'm now not fond of this idea as it will make the soldering rather more difficult) and the Gnd plane on the other (with a nice gap to make cutting the board in half easy). There is no solder mask or silk screen. Crucial items will labelled in the metal layer, but assembly is encouraged with a printout of what the silk screen would say. Pin 1 (or the marked pin) is sometimes marked with a different shaped pad. Otherwise you have to know that all the chips point the same way! Also all the resistors and capacitors are 1206 form factor. While this is more expensive that 0803, they are easier to work with -- and the few cent savings will not even buy a cup of coffee!
There are many options -- bits of the circuit do not have to be built.
I'm using Eagle PCB to design and layout the board. This is not ideal, but it is free for this size of project.
The current version of the schematic (200k PDF file) includes the functionality below. The corresponding top and bottom layers for the PCB are also available (actually I lie, these are from the version 1 layout). These are double scale. They are each around 100k bytes.
I have also made up a parts list that corresponds to the this version of the board.
There is now an Excel spreadsheet that calculates which components you need for which options. I don't guarantee that it is right, but it looks as though it works. It also does approximate pricing on most of the parts -- except the MPX4115A and the parts from Dallas.
I have now ordered my parts for four boards -- in various combinations. You can see what I am doing in the Excel spreadsheet above.
There are also four holes for Vcc, VccUn, DQ, Gnd. These can be used for attaching wires directly or using screw terminals. The Vcc, DQ and Gnd are also available on another three holes for daisy chaining.
There is fairly comprehensive transient protection. This consists of a zener on the data, and a zener on the unregulated power. Note that not all these need to be populated.
There is a undershoot protection diode and a parasite power takeoff diode. There is a jumper that chooses whether power comes from the RJ-45/power jack or whether parasite power is to be used.
There is a capacitor on the power supply.
There is a regulator that can be used to provide 5Volts from an unregulated supply.
The adjustable voltage source can either be a 10-turn pot, or a 1-wire digital pot (DS2890). Given that Dallas will provide two samples of anything, I'd go with the digital pot. This approach of using the current sense improves the resolution at the expense of range -- but the range still exceeds any atmospheric pressure change (at a specific altitude).
In the circuit below, either the pot (10-turn) or the DS2890 should be populated -- but not both.
The frequency is measured by a DS2423 (counter). Actually, the pulses are counted and the host determines the frequency.
The capacitor chain is connected from two holes in the board. A CMOS 555 is used so that the circuit can run on parasite power. This uses channel A of the counter.
A jumper is provided that selects either the unregulated or regulated power (5V) input. A position for a pullup resistor is provided if one is required. A jumper block is provided so that the generic counter can be attached to either channel of the counter.
An alternative is provided in the form of an opamp that will act as a constant current source. The DS2438 can be used to measure both the current and the voltage.
I also updated the spreadsheet with some more useful items -- like a power supply transformer.