I have looked into the available sensors a little further with a view to testing this. For the record, for anyone looking at this avenue in future, here is the search process I've gone through and my conclusion (in a lot of depth, so I recommend skipping unless you're actually looking to solve the same issue). Forgive me if I've misunderstood anything; I'm at a very basic beginner level in electronics.
Updating the pressure range and sensitivity requirements first. Let's say the minimum depth of wort anyone might want to measure is 30cm (my application is more like 60cm) and the maximum in a homebrewing application is 135cm (e.g. Ss brewtech 1bbl unitank). The highest level of sensitivity required would be 0.0004psi (per gravity point at min sensor depth) and the maximum pressure it would ever need to read would be 2.2psi (1.120 gravity wort at 135cm depth).
I first looked at analogue gauges, as this wouldn't interface with a computer but would save having to take gravity samples. They do actually sell gauges within the right range (0-3psi), but I couldn't find any that were food safe and didn't enjoy the idea of sanitising them.
The Honeywell sensors (
https://sensing.honeywell.com/honey...essure-mpr-series-datasheet-32332628-d-en.pdf) have a load of versions, but of course finding the right one in stock somewhere that can ship low volumes to you is tricky. They are a good place to start because they are waterproof (the sensor bit), temperature compensated, cheap, have integrated analogue/digital conversion and there are pre-made breakout boards that will connect them up to an arduino uno. I've looked through loads of other options and they seem to form three groups:
1. Industrial application sensors, which are great but very expensive. There's a homebrew company that sells one of these for $400, applying it to measuring liquid depths.
2. Small plastic - sometimes barbed - sensors intended for PCB mounting. These come in a variety of forms but tend to be for dry gas pressure applications and not going to be easy to clean if you get gunk in them, if they even survive.
3. Barometer sensor boards. Inappropriate for this application as the sensors are the wrong shape and spec.
Looking further into the Honeywell sensors: a 'gage' (gauge or differential, as opposed to absolute) version is necessary because I don't want to have to measure and compensate for atmospheric pressure. I want one with the closest possible range to 2.2psi (152mbar/15.17kPa/114mmHg) because the accuracy will be similar over the whole range between sensors so a larger range means more precision. This means one of the following:
0160MG or 0016KG - perfect (160mbar/16kPa max measurement)
0250MG or 0025KG - still fine (less than 2x the range we need)
0400MG, 0040KG, 0005PG, 0300YG - not great (all are a bit under 3x the range we need)
I then want to specify the version with L (long sensor, to put a tube over) and S (silicone coating for the end of the sensor; it seems this is the food grade version, though not 100% clear) in the part code.
Regarding precision of measurement: this has been tricky to work out. The datasheets quote something called 'transfer function', which I believe translates to the steps at which it measures. As noted above, it needs to measure in at least 0.0004psi (or 4 x 10^-4) steps to be sufficiently accurate. The transfer function tells you the number of measurement steps it can distinguish between (I think) and where the lowest measurement starts and the highest measurement stops. For the sensors above, I have calculated what I think the smallest measurement steps are:
0160MG/0016KG - 1.7 x 10^-7psi
0250MG/0025KG - 2.7 x 10^-7psi
0400MG/0040KG - 1.7 x 10^-6psi
0005PG - 1.5 x 10^-6psi
0300YG - 1.7 x 10^-6psi
So all are 100x to 1000x more accurate than we need.
On the other hand, the datasheet also tells you the minimum (worst case, I assume from the context) 'resolution' in bits for the various transfer functions for each sensor. Using these numbers (assuming x bits gives you 2^x possible values - I think that's right), the bottom few sensors are all just about scraping the level of accuracy needed to discern single gravity point changes in 30cm depth of wort. Less good, but still fine.
There are also error figures (plus or minus 2.5%) in the datasheet, but I'm not going to be able to interpret what these mean as they are clearly more complex than just 'your reading may be 2.5% under or over the actual value'. My hunch is they describe the full error range across radically different measurement conditions and so are not relevant here.
Conclusion: all these sensors seem OK. Possibly. What about sourcing?
None of the above are available in the UK as separate sensors individually online (YMMV in the US). Conveniently, however, the 0300YG sensor is sold pre-mounted on a PCB here:
https://www.digikey.co.uk/product-d...2rbboFwo2PuPo4c-JquL5tblMfn2aY4YaAsVeEALw_wcB
Honeywell also sell testing boards that can be used with a wide range of their sensors in case you can find an individual 0160MG/0016KG sensor and want an pre-made way to mount it:
https://www.digikey.co.uk/product-d...uctivity-solutions/SEK002/480-7110-ND/8024261
Looks like I'm going to get a Honeywell 0300YG sensor on a breakout board, connect it up to an Arduino Uno and do some fiddling. I'll need tubing and a through-fermenter-wall seal that will connect to the sensor (which is 2.5mm in diameter and only 3.7mm long). These bits are going to be a pain to sort out. If I can get tight enough silicone tubing and arrange it so that it won't slip off the sensor (maybe glue the tube to a rigid case enclosing the Arduino and breakout board), I can remove and clean the tubing when wort gets in.