PTS Flowbench Forum

I think a lot of what determines best velocity is determined on two factors. How much cam you put to it both lift and duration wise. How many RPM you spin it. To run big lift, long duration cams in 4 cylinder motorcycle engines things need to be matched very well. I've seen times when backing down on cam gave better results at the drag strip. Engines setup to make a dyno happy don't always work best in the real world. Good example was a Busa that a guy I know was racing. Too much cam. It would go from bog to blow the tire away in about 400 RPM range. It never did work right at the strip and he wouldn't listen to anyone about trying a cam with less duration....I'm certain it would have improved his ET....

Flowsheet fail.

2 days of learning enough VBA to add averaging, and excel doesn't live update when a macro runs.

https://www.youtube.com/watch?v=yrfjVnf ... e=youtu.be

Grey,

You are starting to see the complexities of serial data computation, first your array work should be in a module and you would then think like this. your FP1 will stream the data at full absolute fastest possible to three cells in excel DEP, CFM, Air Speed these are used as some type of place holder variable just to collect data. Your excel VBA becomes the communicator or intermediary, in serial programming we use a event called OnComm but you do not have it so you need to build one. it is in this case a simple timer that runs in a loop say ever 500 milliseconds (Sample Rate) it will do X

what is X it is your code to get new data from cells A, B, C and send to your averaging function which will process the data and output it to the screen/ output cells. You will need to understand "Variable Scope" within Excel as you will need your array counter variable to be Public or global so they do not reset every time the function is called. remember this time is not a TIME timer just a delay or counter mechanism in this application we do not care if it is 100 milliseconds or 125 it is just what responds well to our data averaging and output smoothness.

The link below is just one concept on a google search for "calling a VBA function on a timer"

http://stackoverflow.com/questions/2319 ... -120-secon

I know this may sound deep but you have gotten this far why quit.

Rick

The problem is that I don't even know the basics of VBA - it took me a while just to work out how to paste with a range -, and that is annoying.

I'm basically just butcher coding from samples that I can roughly work out - more than 5 lines and I'm lost -.

I couldn't really work out the array thing so I just copied the 3 cells at 500ms intervals to a table concurrently, and just averaged the 10 samples.

If you think I can get Excel to do updating in my ½ sec time delays, I will have another go when I get my mojo back - I'm having motor problems currently, so I'm walking away for the time been -.

I guess I don't understand what you're trying to accomplish.

Are you bypassing FPcom.exe altogether in your workbook? Unless you're doing that I don't understand or see any need to average numbers that were already [without need of tweaking] as accurate as any other device of that type available. In fact if you ARE using fpcom.exe as the data broker to excel, then any averaging routine you introduce would be "after the fact". There would be no value to averaging numbers that have already been displayed...

Conversely, averaging processed numbers pre-display could have merit. But in order to do that you have to access or bypass the programming functions of the FP1...aka...bypass fpcom.exe altogether for display purposes. I am not certain but in conversations with the previous owner of the business I do recall mention of the software having been developed in some flavor of C language...

Other guys had those boxes hooked up in line with the manometers on their bench, some were SF600 benches and everything corresponded correctly using FPexcel and FPcom as the broker. Additional averaging was not required.

As far as coding goes....yes, it's a learning curve. Some days it all flows well and sometimes you get stuck on something for a day or two. Programming rates right up there with working on cars for a living in my book.. not my first choice....but sometimes you have to bite the bullet and do it.

Regardless of anything else...you may find this helpful in some way as you tweak your worksheets in the work book.

http://excelexperts.com/copy-values-vba

larrycavan wrote:Are you bypassing FPcom.exe altogether in your workbook?

I'm not that good . You can see FPcom.exe on the left behind the excel sheet.

https://www.youtube.com/watch?v=yrfjVnf ... e=youtu.be

I only know SF benches, that's all I've used for 20 years, and this bench, with a single orifice and the FP1, isn't as good. It's adequate for what the owner wants, but I wouldn't be happy unless he gets a SF equivalent or better.

Most of the head test leakage that I have seen is between 0 and 1.25 cfm - I would always make sure it's under 1.5 cfm -, this FP1 single orifice bench only reads 2.5 cfm minimum. I would love to see the PTS DM reading that low mainly to see if it can see 0.5 cfm.

The bench doesn't repeat well enough.
3 tests using the FP1 set to sample #2(FP1 goes from 0-20) - which provides good control of the bench because the FP1 responds quickly -, there was a difference of 1.6cfm at 0.700" lift.
3 tests using the FP1 set to sample #5 - which provides bad control of the bench because the FP1 responds so slowly that it's hard to set 28" at low flows -, there was a difference of 1.3cfm at 0.700" lift.
By testing it says that the higher the sample rate, the more repeatable, but the problem is that it makes the response so slow, you can not control the bench graphically.

Averaging is my solution to making the 'fast response low sample rate setting' more accurate in final tabulation.

I could probably get the repeatability spot on with setting #20, but the hysteresis in finding 28" would be so bad it might take 15 minutes to turn the knob, wait for the number to change, turn the knob, wait for the number to change, etc


Here are the figures I got, the first 3 are at sample #2, the next 3 sample #5
1----2 2----2 6----2 3----5 4----5 5----5
0.000 0.000 0.000 0.000 0.000 0.000
32.60 34.70 34.40 35.10 34.80 33.70
72.10 72.50 72.30 73.00 72.90 71.60
133.9 134.9 134.4 134.1 133.9 134.6
183.5 185.4 184.7 184.8 183.7 183.5
223.7 224.9 222.7 223.9 224.5 223.2
240.8 240.7 241.1 240.4 241.3 240.6
240.8 243.3 243.0 242.7 242.9 241.6
246.0 244.6 245.0 246.4 246.0 247.3

This video compares the FP1 sample rate, to controllability.
The left is set on #1 and right is set on #10, and you can see that it takes half the time to hit 28"

http://youtu.be/C2pCtoaxB4E

After testing I can see why Tony said "Best smoothing with the fastest response, will be with very fast broad band data sampling, and averaging the numbers in software."

Here is head leak testing from 0.030" down. I set sample rate to #1 and now I see 1.3cfm@0.022", but it's still not as nice and steady as a SF manometer

http://youtu.be/gWNgFhEeMJA

Old Grey wrote:
Here is head leak testing from 0.030" down. I set sample rate to #1 and now I see 1.3cfm@0.022", but it's still not as nice and steady as a SF manometer

Water gauges dampen the readings, they can not react as fast as the digital sensor. So they will always be smoother, digital is able to react fast to the changes so you see the bouncing readings. You can smooth out the numbers using averaging in digital. It's all going to depend on how you are able to manipulate the samples via the software.

The PTS software allows, sample rate, baud rate, and averaging changes. So you can tune the display to the end-user liking. The speed of the computer plays into this also. I have found it's better to slow everything down and run a slower baud and sample rate on the PTS DM depending on your computer speed. I personally run an old laptop and have found my DM likes a speed 0f 1200 with a low average and sample rate. The new Win 8 machines like a 57600 baud with a 240 sample rate and higher averaging.

It's been a long time since I have used an FP1 and I've pretty much not given it any thought as to how they achieve the end numbers via the software. As Larry said it's hard coded in C on the chip, the PTS DM is all software driven so updates to the DM do not require sending back to the factory for changes.

Old Grey wrote:

larrycavan wrote:Are you bypassing FPcom.exe altogether in your workbook?

I'm not that good . You can see FPcom.exe on the left behind the excel sheet.

https://www.youtube.com/watch?v=yrfjVnf ... e=youtu.be

I only know SF benches, that's all I've used for 20 years, and this bench, with a single orifice and the FP1, isn't as good. It's adequate for what the owner wants, but I wouldn't be happy unless he gets a SF equivalent or better.

Most of the head test leakage that I have seen is between 0 and 1.25 cfm - I would always make sure it's under 1.5 cfm -, this FP1 single orifice bench only reads 2.5 cfm minimum. I would love to see the PTS DM reading that low mainly to see if it can see 0.5 cfm.

The bench doesn't repeat well enough.
3 tests using the FP1 set to sample #2(FP1 goes from 0-20) - which provides good control of the bench because the FP1 responds quickly -, there was a difference of 1.6cfm at 0.700" lift.
3 tests using the FP1 set to sample #5 - which provides bad control of the bench because the FP1 responds so slowly that it's hard to set 28" at low flows -, there was a difference of 1.3cfm at 0.700" lift.
By testing it says that the higher the sample rate, the more repeatable, but the problem is that it makes the response so slow, you can not control the bench graphically.

Averaging is my solution to making the 'fast response low sample rate setting' more accurate in final tabulation.

I could probably get the repeatability spot on with setting #20, but the hysteresis in finding 28" would be so bad it might take 15 minutes to turn the knob, wait for the number to change, turn the knob, wait for the number to change, etc


Here are the figures I got, the first 3 are at sample #2, the next 3 sample #5
1----2 2----2 6----2 3----5 4----5 5----5
0.000 0.000 0.000 0.000 0.000 0.000
32.60 34.70 34.40 35.10 34.80 33.70
72.10 72.50 72.30 73.00 72.90 71.60
133.9 134.9 134.4 134.1 133.9 134.6
183.5 185.4 184.7 184.8 183.7 183.5
223.7 224.9 222.7 223.9 224.5 223.2
240.8 240.7 241.1 240.4 241.3 240.6
240.8 243.3 243.0 242.7 242.9 241.6
246.0 244.6 245.0 246.4 246.0 247.3

Looks to me like you have turbulence issues.

Do you get similar results when testing with an orifice mounted?

larrycavan wrote:

Old Grey wrote:

larrycavan wrote:Are you bypassing FPcom.exe altogether in your workbook?

I'm not that good . You can see FPcom.exe on the left behind the excel sheet.

https://www.youtube.com/watch?v=yrfjVnf ... e=youtu.be

I only know SF benches, that's all I've used for 20 years, and this bench, with a single orifice and the FP1, isn't as good. It's adequate for what the owner wants, but I wouldn't be happy unless he gets a SF equivalent or better.

Most of the head test leakage that I have seen is between 0 and 1.25 cfm - I would always make sure it's under 1.5 cfm -, this FP1 single orifice bench only reads 2.5 cfm minimum. I would love to see the PTS DM reading that low mainly to see if it can see 0.5 cfm.

The bench doesn't repeat well enough.
3 tests using the FP1 set to sample #2(FP1 goes from 0-20) - which provides good control of the bench because the FP1 responds quickly -, there was a difference of 1.6cfm at 0.700" lift.
3 tests using the FP1 set to sample #5 - which provides bad control of the bench because the FP1 responds so slowly that it's hard to set 28" at low flows -, there was a difference of 1.3cfm at 0.700" lift.
By testing it says that the higher the sample rate, the more repeatable, but the problem is that it makes the response so slow, you can not control the bench graphically.

Averaging is my solution to making the 'fast response low sample rate setting' more accurate in final tabulation.

I could probably get the repeatability spot on with setting #20, but the hysteresis in finding 28" would be so bad it might take 15 minutes to turn the knob, wait for the number to change, turn the knob, wait for the number to change, etc


Here are the figures I got, the first 3 are at sample #2, the next 3 sample #5
1----2 2----2 6----2 3----5 4----5 5----5
0.000 0.000 0.000 0.000 0.000 0.000
32.60 34.70 34.40 35.10 34.80 33.70
72.10 72.50 72.30 73.00 72.90 71.60
133.9 134.9 134.4 134.1 133.9 134.6
183.5 185.4 184.7 184.8 183.7 183.5
223.7 224.9 222.7 223.9 224.5 223.2
240.8 240.7 241.1 240.4 241.3 240.6
240.8 243.3 243.0 242.7 242.9 241.6
246.0 244.6 245.0 246.4 246.0 247.3

Looks to me like you have turbulence issues.

Do you get similar results when testing with an orifice mounted?

Get back to basics.......The sensor probes live in an environment that is chaotic. The box sensors are very sensitive. Both situations present problems. THINK...."which can I actually do something about"?

It appears the attempt to "after the fact average" needs some tweaking to work properly. I have some ideas on that but they require answers to specific questions first.

The chaotic environment solution may be the better way to go here. Installing the baffle in the top chamber and adding the top section improved my bench significantly. With water gauges, every time the inclined would approach 90% the gauge bounced dreadfully.

Regarding your averaging attempts. Two things come to mind immediately. The actual code used to perform that function and how you are using FPCONFIG.TXT to distribute brokered data in Excel. It gets a bit involved and there are ways around this but a simple statement in your code to turn screen updating off eliminates screen flicker and the momentary freezing while processing is taking place. Study the link I posted for you yesterday as there is some information about that in there as well..... KEEP PLUGGING AWAY...YOU'LL GET THERE