Pressure (air) sensor best practices

Washing machine/diswasher water-level switches?

No, I'm looking at pressures in air handling systems.

In systems that I've designed previously, someone else selected (very expensive!) transducers and I just had to interface with them.

But, those were industrial systems with dedicated staff to keep them running, calibrated, etc. You cared about the actual "numbers" from the sensors because you were controlling "regulated" processes and needed to document compliance/exceptions.

I'm trying to leverage some of those algorithms but in a cheaper consumer market where stuff "just has to work". I don't care about numbers as much as *trends*, using adaptive algorithms to sort out what's "normal" from "indicating".

They operate on air pressure. The rising water level closes off the bottom end of a chamber; thereafter any further increase in water level causes an increase in air pressure within the chamber. A very flexible rubber diaphragm presses aginst a spring and eventually operates a microswitch.

By substituting a movement sensor for the microswitch (and with the use of a suitable spring) you could make a proportional air pressure sensor of good sensitivity which held its calibration to a reasonable degree of accuracy.

I have also used a differential air pressure sensor that used a thin silvered tensioned plastic diaphragm as a concave/convex mirror with a light emitter and sensor on each side, connected to a balancing circuit. It was exceptionally sensitive but not particularly robust or repeatable.

I am using an ebay BMP085 module for air pressure, works OK:

Slightly stretched beryllium copper membranes are pretty robust. They have to be stretched enough (well within their elastic limits) to keep them flat, and clamped between suitably dished mounting plates to prevent them being stretched beyond their elastic limits by large over-pressures.

Capacitative sensors can be used to track the movement of the membrane accurately and with good stability. They've been commercially available since the 1950's, but they weren't cheap.

Fast modern integrated circuits will have made the electronics cheaper.

For small pressure differences you can use charged plates on either side of the grounded diaphragm to compensate for the pressure difference across it. A big enough voltage difference to compensate for atmospheric pressure or anything close to it would break down the gaps.

Yes, I know that from tinkering with old washing machines. But, the mechanism was large (the size of a lemon).

But, there are semiconductor devices already that do this for a buck or two (in tiny quantities). They would be considerably easier to deploy -- in large numbers. E.g., one typically instruments an AHU in many places as the air is "processed" within -- each observation point lets you close a loop or infer something about that process.

The same problem (as above). You can use a chilled mirror to detect dewpoint -- or, by changes in capacitance of polymers. Big difference in size, cost, precision, durability, etc.

When you move away from thinking of sensors as needing "calibration", you can approach problems with more flexibility.

I've had some fun with teeny tiny BME280 sensors. They have a sub-1% absolute accuracy and enough resolution to detect the pressure change between your head and your feet. They measure temperature and humidity too. The interface is I2C.

Jeroen Belleman

Of course they might be detecting the difference in the mean molecular weight of the local atmosphere. ;)

Cheers

Phil Hobbs

Even the smelliest feet don't make a significant difference to the mean molecular weight of the local atmosphere around your feet. The extra CO2 in the exhaled air around your head will raise the mean molecular weight more, but still not enough for a pressure gauge to register.

If by "no tubing" you mean drill a hole in the duct fit a rubber grommet and then plug a barbed sensor into the grommet. that could work, putting the whole sensor inside the duct seems fraught.

At 20 feet I2C would be tricky, but Analog sensors are available.

Puting a sock over the connector will keep most junk out. maybe somethign like the outer covering of parachute cord. I'm not sure where you buy that. There's probably somethig better available.

Installing the sensor with the aperture facing downwards will help too.

Grmpf ;-)

While I was graphing the outputs, I was surprised by the efficiency of humidity readings to detect human presence. The curve was quite smooth with no one in the room, but got quite noisy with someone present.

jeroen Belleman

We’ve used the very nice Sensirion SHTC3 (“schtick #3?”) T/H sensor to calculate the dew point to monitor desiccant performance and prevent condensation. Super good medicine.

Cheers

Phil Hobbs “Three French hens….”