gregorypye / trimix-analyser Goto Github PK

Need to better handle multiple screens, as they currently link directly to each other, so are hard to maintain. This could include refresh cycles that allow screens to update themselves (via knowing whether they are refreshed or newly created, though that refresh load avoidance may not be worth doing)

Tidies up code

There are three basic options for a waveguide in my mind at the moment, assuming that for accuracy reasons we need some large fraction of a metre of length, and for practicality reasons it needs to be some convolution rather than just a straight length.

TL;DR on planning assumption - option 3, with most of the length being straight tubes, and 3D printed reflecting end caps

Option 1, likely the cheapest, is to use loops of relatively rigid wall tubing with an adaptor at each end that connects to an ultrasonic transducer, and also has a gas port for the sample gas. The total cost is likely to be a few pounds for the adaptors (less if the printer cost is ignored - the plastic cost will be a pound or less each) and 1-2 pounds for the tubing. The potential challenge is that the length down the inside of the circle will be shorter than that down the outside of the circle, so the pressure pulse from the transducer will get extended which is likely to reduce accuracy. For a radius of (say) 6mm, the extra distance would be 2 x Pi x R ~= 36mm difference per loop. It might be that we can detect the front of the pulse enough that the stretching is manageable, but it doesn't feel ideal.

The second is to use straight wave guides with exact 45 degree angles at the end . There is bound to be some scattering on reflection, but at least the straight line path length is the same right across the area. Option 2 is to 3D print the entire thing. Or, possibly, via milling from a block, though it might be hard to get clean inside angles. This option would keep the part count down, which might also help robustness. And, the cost wouldn't be a disaster, though on a semi-commercial basis where print time is costed I'd guestimate the cost as £10 - 30

Option 3 is an intermediate model which uses lengths of straight rigid wall tubing, along with a printed set of end caps that have exact 45 degree chamfers, along with adaptors for the ultrasonic transducers. This gets the benefits of commodity tubing costs, with precision reflector angles. It also makes it fairly easy to test option 1 as the adaptors could be the same.

Direct use of the o2 sensor show too much Jitter,11.8 mV through to 12.6 mV (7%)

Fairly major change as the format has moved