This project has been on the back burner so it's time for an update :-)
In fact after my experiments with the digital hydrometer I very nearly abandoned this project.
The problem is
creep, i.e. the tendency for load cell measurements to drift gradually when subjected to a continuous load. This makes it very hard to monitor accurately the weight of a load that is left in place on the scales for long periods of time.
In the hydrometer I fixed this problem by using a servo to lift off the weight between readings. But with a 25kg Corny that would be decidedly inconvenient.
Fortunately I think I may see a simpler solution. My thoughts go something like this:
- The effect of creep should look pretty different to that of drawing off a pint.
- Creep probably looks like a very slow weight increase - maybe of the order of 1 gram per minute.
- In contrast, pouring some beer should look like a much more rapid weight decrease of 50-100 grams per sec.
- So by disregarding any slow weight changes I should be able to track accurately the amount of beer drawn off each time and the running total.
- This leaves the question of the initial volume. However in this case creep is not an issue because the load cells are initially unloaded.
Next step is to do a little testing to establish whether these assumptions are true.