More interesting Results

[Tony]

Rick, this should be quite practical to do.

I don't know how significant it will be, but in theory, there may be the possibility of some stiction and/or hysteresis effects in the transducer when measuring an absolutely constant fixed non varying static pressure.
Software averaging of many readings will not work quite so well either if there are only 1024 noise free ten bit steps to resolve.

It should not be quite that bad, but a bit of deliberately introduced noise or vibration into the transducer may help things along quite a bit.
Something like a bit of constant low level vibration from an electric motor fed into the mounting of the transducer should tickle it enough to introduce enough random scatter into the readings to give the software averaging something to work with.

This may all be totally unnecessary, but I just thought I would suggest it if you are having trouble with repeatability or end up recording some lumpy curves.

These transducers are not recommended to be used directly with water (or corrosive or electrically conductive fluids) but that recommendation is only for very long term permanent installation.

The pressure port is protected with some kind of inert waterproof jelly filler to keep the exposed internal electronics dry.
There should not be any problem using water with the sort of short term testing this experiment will involve.

[larrycavan]

Tony,

Do sensors have a sweet spot?...

Would be interesting to see what happens when an analog manometer is hooked up in parallel with the digital unit on a 1020....

I really don't know the answer to that Larry, you guys are really pushing the limits....

But I will theorise that the biggest sources of initial (straight out of the box) error will be the zero pressure point, and full scale calibration point. The sensor is internally corrected for temperature, and we ourselves can ensure the dc supply voltage to the sensor remains absolutely constant.

After zero and full scale pressures have been tweaked in software, what is left will be some non linearity and hysteresis due to the mechanics of the strain gauge part of the sensor.
We can ignore hysteresis, because the random pressure fluctuations occur evenly in both directions, and after some really brutal averaging, what we are left with will be a simple very slight curvature away from true linearity.

This will be made worse because we are exploiting all of the full maximum possible pressure swing to minimise the effects of other possible errors.

Not certain about the cause of this non linearity, but my guess would be that a cross section of the measurement diaphragm acts something like a beam in bending mode, and the strain gauge part measures the tension/compression changes along top and bottom of the flexing diaphragm.

It is all geometry related, and providing the actual mechanical flexing is kept very small, the linearity should be quite reasonable.
But it would be easy to imagine how, if something like this were made of rubber, how blowing it up into a massive dome shape could introduce some dramatic non linearity into how far the rubber stretched versus the applied pressure.

If you can detect and quantify this non linearity, correcting for it should be possible, and I would expect the correction to remain fairly valid from sensor to sensor, or at least it should get you very close.

This is all armchair speculation on my part, but it is what I would be looking for.

Fluid manometers will be very difficult to read to the required degree of accuracy because of the meniscus effect, surface tension, and how the fluid clings to the wall of the tube.

It may be easier to generate an accurate static pressure rather than trying to read or compare existing pressures by different means of measurement, each of which also have their own set of problems.

How about a header tank full of water and a hose leading down into the pressure transducer.
The header tank could be raised on a lead screw to increase the static pressure by known fixed amounts without any significant fluid movement if the surface area of the reservoir were kept relatively large. It may with care and repetition, be possible to plot this non linearity and then do something about it in software.

Again all just crazy wild thinking on my part.

Thanks for that input Tony.

Not that it matters one bit...but I'd curious to see if the 1020 benches vary as much as the 600's seem to.

This flow bench comparison is akin to running numbers through various software when customers send Dyno numbers to me from different Dynos. All the measuring tools in the world only amount to a specific set of values that you can ponder and hope that your hunch's are correct when you suggest a change to some component.

For all that we know / think we know, a large part of the performance game is still educated guessing a lot of the time.

[Brucepts]

These transducers are not recommended to be used directly with water (or corrosive or electrically conductive fluids) but that recommendation is only for very long term permanent installation.

The pressure port is protected with some kind of inert waterproof jelly filler to keep the exposed internal electronics dry.
There should not be any problem using water with the sort of short term testing this experiment will involve.

A couple of sensors can be sacrificed in the name of R&D

[Tony]

A couple of sensors can be sacrificed in the name of R&D

No problem Bruce.
Just pointing out that in the application notes they say not suitable for water.
They will definitely work, and probably for quite a long time.