Not the Mercdog again

[larrycavan]

Not everyone has a need for 600cfm@28"

That's right - even 600 is way overkill for nearly everybody -, but think about a flowing a 300cfm head. You can now do it at 38"

A 600cfm plate at 16" is 3.326" diameter, a 600cfm plate at 2" is 5.593"

A 300cfm plate at 16" is 2.35" diameter, a 300cfm plate at 2" is 4.0" - well within everyone capacity -.

Using 38" I'm allowed 7.5" on the orifice with my 46.6 limit, and that equates to 2.85", which gives me some range to overlap orifices.

8.5" was less stable and creates a large orifice hole at 600cfm, 40" was to much depression we felt, 16" seemed to work out fine with good hole sizes in the 6" square aluminum plate. We did not do any scientific studies I built a DM using each sensor, tested them out and we discussed the findings and settled on the 16" sensor. A 16" sensor allows you to use the whole sensor range on a given plate, if you use a 40" sensor on a 16" range or 2" as you suggest you are only using a portion of the sensors range.

Just looking at this sheet on page 3 the graph shows that the sensor is more accurate at 1" than 16", so I don't see any need to change sensor, but I do think more work is needed in getting the orifice more stable sooner.

http://www.datasheets360.com/pdf/6060859169405517462

There appears to be more to this where sensors are concerned....

Through research and discussions regarding pressure sensors certain things that I had previously taken as gospel / accepted on faith from lack of personal desire to dig deeper have come to my attention. What works has variables attached to the approach and one of those variables is cost, another is sensor accuracy. The sensors themselves play a huge role in the approach in terms of cost, hardware requirements and the coding to pull it all together for end users...

It's interesting to note that the Performance Trends Box uses a 1 PSI/28"W/C sensor and a 5PSI/140"W/C sensor. The 3rd sensor being optional. Accuracy and stability of those units I have no personal experience with. Still, their reputation precedes itself in the industry..

There are always different ways to a solution & costs vary with the approach. However, sometimes the initial costs buy you other benefits.

[larrycavan]

On second thought the hole size might not be that big of a problem. If one hole gets a pulse of high pressure another hole will bleed off that pressure before the sensor sees it, and when you drill soft plastic the hole probably closes up a touch when the drill is removed.

The averaging tube is probably a better idea because it works first time, where as a single hole tube will have to be moved around and tested numerous times - I spent one day doing it -.

Persistence is not something you're short of!

[1960FL]

Grey,

First thank you for taking the time to 1, put this together and perform the test but 2, sharing it with us all.

I have attached a spreadsheet with some assumptions on what you were doing I/E were you said at 28" it was a mathematical CFM from the tested 40" flow, but that really does not affect my point. The sheet just takes the data as output and averages it plain and simple. I have also included in the sheet two screen shots of one of my tools that shows the calculated CFM for the 300 PAP at .62CD.

My point is that though your dampening has not really effected the output that you visually see the fact is that the internal programming of the unit and the NOISE it receives from the sensor create spikes in the data the better averaging the the smoother the numbers. The sensor is nothing more than a glorified microphone and it gets input (sound Pressure waves) from all over the bench not just actual pressure differentials. Data cleansing, averaging and smoothing Make the customer interface!

Think of it like this, if you took an analog Volt meter to your Mains it might read 230VAC +/- YOU interpret this form the meter scale, Now you hook up a Fluke digital volt meter and it reads 231.45VAC Ahhh higher resolution more accuracy, then you hook up a digital oscilloscope to the mains fine tune the axis adjustments and WOW you see voltages all over the place some as high as thousands of volts! Noise, leakage, EMF etc. The tools we use may often confuse us and cause focus in an area that may not need attention, maybe just a little more Bass and a little less treble

I think your solution is to "POST Process" the data from the DM you are using, Feed the output to an array; Put it in a FIFO array, clean erroneous data before putting in the array and I think you will have what you are looking for.

Rick

[Old Grey]

There appears to be more to this where sensors are concerned....

Through research and discussions regarding pressure sensors certain things that I had previously taken as gospel / accepted on faith from lack of personal desire to dig deeper have come to my attention. What works has variables attached to the approach and one of those variables is cost, another is sensor accuracy. The sensors themselves play a huge role in the approach in terms of cost, hardware requirements and the coding to pull it all together for end users...

It's interesting to note that the Performance Trends Box uses a 1 PSI/28"W/C sensor and a 5PSI/140"W/C sensor. The 3rd sensor being optional. Accuracy and stability of those units I have no personal experience with. Still, their reputation precedes itself in the industry..

There are always different ways to a solution & costs vary with the approach. However, sometimes the initial costs buy you other benefits.

There seams to be a lot of talk of the benefits of different sensors ideas, but now I'm kind-of leaning the other way toward averaging as being more important.

I can see how a 16" sensor over 5 volts is technically a better idea than the 40" over 5 volts in my 100" FP1, but really how much is it better. I reduced the error by 50% just by subtracting the highest and lowest figures from the data - it wasn't even full averaging -, and I think that will be way more than any sensor difference.

0.040" restrictor on end of tube encased in a housing at 28" max to min = 8.5cfm
Averaging tube at 28" max to min = 11.2cfm
Each set of readings had erroneous blips that that accentuated the difference between max and min, so if the highest and lowest was removed it would be 5.9cfm and 5.5cfm.

I reckon if you put a 40" sensor in the PTS DM it's robust averaging will make the difference nearly neglectable, or at least very small.

[Old Grey]

Grey,

First thank you for taking the time to 1, put this together and perform the test but 2, sharing it with us all.

I have attached a spreadsheet with some assumptions on what you were doing I/E were you said at 28" it was a mathematical CFM from the tested 40" flow, but that really does not affect my point. The sheet just takes the data as output and averages it plain and simple. I have also included in the sheet two screen shots of one of my tools that shows the calculated CFM for the 300 PAP at .62CD.

My point is that though your dampening has not really effected the output that you visually see the fact is that the internal programming of the unit and the NOISE it receives from the sensor create spikes in the data the better averaging the the smoother the numbers. The sensor is nothing more than a glorified microphone and it gets input (sound Pressure waves) from all over the bench not just actual pressure differentials. Data cleansing, averaging and smoothing Make the customer interface!

Think of it like this, if you took an analog Volt meter to your Mains it might read 230VAC +/- YOU interpret this form the meter scale, Now you hook up a Fluke digital volt meter and it reads 231.45VAC Ahhh higher resolution more accuracy, then you hook up a digital oscilloscope to the mains fine tune the axis adjustments and WOW you see voltages all over the place some as high as thousands of volts! Noise, leakage, EMF etc. The tools we use may often confuse us and cause focus in an area that may not need attention, maybe just a little more Bass and a little less treble

I think your solution is to "POST Process" the data from the DM you are using, Feed the output to an array; Put it in a FIFO array, clean erroneous data before putting in the array and I think you will have what you are looking for.

Rick

Thanks for that, it clears up the mud.

I was looking at your sheet and seeing that the averaging tube reads a bit higher at 28" - can't compare the actual cfm because the deps are different -, and I was thinking the AT must be out. But then I remembered that I calibrated the figures to my single hole tubes, making that fact arbitrary. Now I'm thinking the AT is better for accuracy than my SHT because there is less chance that I have found the exact best spot with only one hole in a tube.

What do you think?

Going by the POS bench video it looks like the air bounces around the walls of the chamber until it slows down enough to fall into the centre orifice, so I was thinking of placing the AT in a circle half way between the walls and the orifice because that area would be calmer.

I think you're right about the fluctuations being just noise, and it kind-of backs up that the averaging is more important because you can't eliminate all of it.

[1960FL]

I reckon if you put a 40" sensor in the PTS DM it's robust averaging will make the difference nearly neglectable, or at least very small.

It was, the first test version of the board had 3 40" sensors and the only reason we went to 16 was that it showed better linearity across a range on an orifice as we has no intention of a 40" Delta P.

Rick

[Old Grey]

I reckon if you put a 40" sensor in the PTS DM it's robust averaging will make the difference nearly neglectable, or at least very small.

It was, the first test version of the board had 3 40" sensors and the only reason we went to 16 was that it showed better linearity across a range on an orifice as we has no intention of a 40" Delta P.

Rick

I've been thinking about the non linearity of orifices over large ranges.

In this table the high flows are spot on, but the low flows are off, and if you think about it the high flow should be spot on, because it was probably calibrated at high flows.

Now this begs the question whether the DM is reading off because it just can't do it, or that the mechanical aspect of flow through an orifice isn't that liner through all velocities - it needs a certain velocity to form the correct cone that makes it linear -. I'm banking on the latter, and have ideas to make it more liner over a larger range, but it's probably not possible because someone would have worked it out by now. I'm now thinking of having individual cd's for each flow. Every time you flow the PAP plates you have to fudge a single cd so the it falls between all of them. Why not get the 3 PAP flow points and work out an equation that you add to the spreadsheet so that every lift gets the prefect cd, making every point as best as it can be.

[Tony]

There are basically two very different problems when trying to accurately read very low flow with respect to a given orifice size.

The first is that as you reduce pressure drop across the orifice, it becomes much more sensitive to unstable flow due to up stream turbulence.
Bench design is paramount, especially the size and detailed design of the settling chamber.

The second problem is that because of the inverse square orifice pressure law, you are trying to measure a very small pressure differential.
Any sensor errors will become a progressively larger problem when what you are trying to measure becomes vanishingly small.

Software averaging is an extremely powerful tool, but it can only take you so far.
Once you drop below some threshold capability of the bench, and the pressure transducer, the software averaging has nothing tangible left to work with.

[Old Grey]

The second problem is that because of the inverse square orifice pressure law, you are trying to measure a very small pressure differential.
Any sensor errors will become a progressively larger problem when what you are trying to measure becomes vanishingly small.

Is there a way to prove it not the electronics?

The table shows that a 100cfm test orifice on the 3.490" internal orifice reads 107cfm.
What if it always repeats 107 cfm every time, would that prove it's not the electronics.

[Tony]

Is there a way to prove it not the electronics?

You could hook up a fluid manometer and measure the orifice pressure differential that way.

A 500CFM orifice measuring 100CFM would be 1/5 the flow and 1/25 the pressure.
If its a 16 inch calibration pressure orifice at 500 CFM, expect to see 0.64 inches rise at 100 CFM.

How about fitting a manometer tube beside a transparent plastic school ruler into the focal plane an old 35mm slide projector laying on its side, and project the image of both side by side onto a wall.
It will look absolutely huge next to the ruler scale, which might have sixteenths of an inch or millimeter graduations.
When in sharp focus, its amazing what you can see if the image is made big enough.