PF MAWD Data Analysis

[bandman444]

So I learned a lot. thanks for posting!

I found out that for some reason it doesn't like files with a space in them (rocket on I.txt) so just changed to (rocketonI.txt) and I got it to work.

cool stuff

 

[rocketboy16]

Alright,

Here is the new program. This should fix the errors that were encountered, and find the real apogee.It will take me some time to get the next V2.0 of this program out. It will hopefully fully eliminate the irregularities of brute force integration on the position time graph for velocity, and possibly graph. If graphing is not easily feasible certainty exportation of velocity/acceleration data for graphing elsewhere, excel.

Peter

View attachment FAV1.1.zip

 

[Robaust]

G'Day Peter,

Great file, i've used both versions.

I am a big MaWD fan and believe it is a great device. I developed and evolved a MS Excel program to extract and graph similar data that your program offers. However, my MS Excel version takes manual manipulation to generate the same numbers.

I find the descent rate very handy tool to understand if I have the chutes correctly sized for the rocket. Your tool and my Excel table came out with the same data.

I had been struggling with determing max speed and youv'e done that - many thanks.

Keep up the development - Great stuff.

Regards,

Rob Winchcomb

 

[JDcluster]

The new program is still giving junk data for decent rate.
You need to incorporate places that might be below sealevel. It said launch site altitude was -116ft.


JD

Alright,

Here is the new program. This should fix the errors that were encountered, and find the real apogee.It will take me some time to get the next V2.0 of this program out. It will hopefully fully eliminate the irregularities of brute force integration on the position time graph for velocity, and possibly graph. If graphing is not easily feasible certainty exportation of velocity/acceleration data for graphing elsewhere, excel.

Peter

 

[rocketboy16]

JD,

Could you possibly upload that file so that I may work on getting data like it working? It would be alot of help!

Thanks
Peter

 

[JDcluster]

Ok, here it is:



JD

View attachment Enforcerspark.txt

 

[Zeus-cat]

You should include the readme file with each version. I only downloaded the 1.1 version (with no readme) and I couldn't get it to analyze any data. I finally decide to strip out the header info and save the file as a txt file and it worked.

You may also want to advise people to clean up their data by eliminating spikes. I used data from a very low flight (under 500 feet) which must have a spike when the rocket landed. Your program reported the following:

There were 584 data points

The apogee was 440ft, and it took 6.4(sec) to reach apogee

The maximum velocity was 3035 ft/s

The altitude at which the maximum velocity was reached at was 0ft,
This was at 29.2 (sec) after launch.

The descent rate under main was 25.53 ft/s

The descent rate under douge was -1.#J ft/s

I'm pretty sure it didn't reach 3035ft. per second after it landed. :bangpan: I'm not sure what one cause the drouge error. I'll clean up my data some time and try it again. Thanks for posting the program.

 

[rocketboy16]

Sorry, I have gotten side tracked with a lot of things. I am finishing up my senior year, and tying up college things. I have been working on some algorithms to smooth barometric altimeter data. I am not sure when i will have the completely new V2.0 with smoothed data, possibly to the point of fitting an equation to the data for more accurate derivation and double derivation for acceleration data.


Peter

 

[sandmantoy]

I am going to wait to try this but it sounds so good. I have a few MAWDS and to get more out of them would be awesome Once you get the bugs out and make it more user friendly. I can not wait to try it.

 

[JDcluster]

I'm not sure if anyone knows this or not but; If the unit loses power it will beep out 20,000ft with 4,600 data points. I guess that's the max it can do.
On my flight last weekend it did just that. You can still get the real data from it but it might mess up this little program?


JD

 

[Rocketjunkie]

I'm not sure if anyone knows this or not but; If the unit loses power it will beep out 20,000ft with 4,600 data points. I guess that's the max it can do.
On my flight last weekend it did just that. You can still get the real data from it but it might mess up this little program? JD

Also, after the point at which the problem occurred, the data from the previous flight is still there.

 

[Zeus-cat]

I am not sure when i will have the completely new V2.0 with smoothed data, possibly to the point of fitting an equation to the data for more accurate derivation and double derivation for acceleration data.

I would be interested in the techniques you would be using to do this. A few years ago I tried this myself using data from an Alt15K. I was not impressed with the results I achieved and did not pursue it beyond a few samples.

Anyway, if I remember correctly, for the data smoothing I tried using simple three point averaging, five point averaging, weighted three point averaging and weighted five point averaging. The data smoothing worked OK, but the derivative data not so good and the second derivative considerably worse. If I can find the results I will let you know which one worked best.

You might even want to consider something like seven or even nine point weighted averaging to really smooth out the data. The altimeter data that I have is rather jumpy. Here is an example for a low altitude flight (apogee was 482 feet). You could not attempt to derive velocity or acceleration data from this flight without some serious data smoothing.

Time Altitude
0.40 58
0.45 68
0.50 77
0.55 77
0.60 87
0.65 96
0.70 96
0.75 115
0.80 115
0.85 115
0.90 134
0.95 134
1.00 154
1.05 154
1.10 154
1.15 173
1.20 173
1.25 183
1.30 192

And just for the record I am an electrical engineer with a masters degree in Operational Research (in laymen's terms that would be applied statistics).

 

[rocketboy16]

That is why it is taking me so longer, i am trying to figure out if it is even a possibility.Along with the functionality of the the program, i was struggling to create a more user friendly interface. Remember i am a high school student with only C/C++ classes that have only covered to about 3d arrays ,file accessing , bubble sorting.Nothing we have learned has been oriented towards making an actual "program", I considered learning Java as i was told it was more oriented towards my needs.So i asked myself why learn java to write hundred of lines of code when there is Excel? Excel has curve fitting, which will give me a polynomial to the X degree.Then it is possible to take a mathematical derivative. Also excel is more commonly used and a program in excel would more likely suit the average rocketeer over one in C.


Peter

 

[Zeus-cat]

I did find the spreadsheet I used for data smoothing. I agree rocketboy16 that doing it in Excel is a lot easier than writing a program. If anyone is interested I can post the smoothing formulas otr even the whole spreadsheet.

 

[brianc]

If anyone is interested I can post the smoothing formulas or even the whole spreadsheet.

Please post both. Do you know if your spreadsheet works with
Open Office?

 

[Robaust]

Hi JD,

I have encountered the "20,000ft with 4,600 data points" issue before. This was raised in a previous thread. In one of the previous software upgrades (1.20) Perfectflite released a firmware upgrade to signify a power loss;

https://www.rocketryforum.com/showthread.php?t=16044

https://www.rocketryforum.com/attachment.php?attachmentid=32860&d=1288179401

Hi Zues,

I'm an MS Excel fan, love to see your smoothing formula's ? I have the Al15K and MaWD so often clean the data in MS Excel, a smoothing formulae would be just the trick.

Regards,

RobW

 

[Zeus-cat]

The file contains two smoothing formulas, columns C and D. It has been a long time since I played with it, so I don't remember a lot of details.

The flight data is from a Perfectflite Alt15k. The rocket is a BT-60 design, just over 3 feet tall launched on this flight with a cluster of two C6-5 motors.

View attachment Tonka Flight 3.xls

 

[Larry Curcio]

I would be interested in the techniques you would be using to do this. A few years ago I tried this myself using data from an Alt15K. I was not impressed with the results I achieved and did not pursue it beyond a few samples.

Anyway, if I remember correctly, for the data smoothing I tried using simple three point averaging, five point averaging, weighted three point averaging and weighted five point averaging. The data smoothing worked OK, but the derivative data not so good and the second derivative considerably worse. If I can find the results I will let you know which one worked best.

You might even want to consider something like seven or even nine point weighted averaging to really smooth out the data. The altimeter data that I have is rather jumpy. Here is an example for a low altitude flight (apogee was 482 feet). You could not attempt to derive velocity or acceleration data from this flight without some serious data smoothing.

Time Altitude
0.40 58
0.45 68
0.50 77
0.55 77
0.60 87
0.65 96
0.70 96
0.75 115
0.80 115
0.85 115
0.90 134
0.95 134
1.00 154
1.05 154
1.10 154
1.15 173
1.20 173
1.25 183
1.30 192

And just for the record I am an electrical engineer with a masters degree in Operational Research (in laymen's terms that would be applied statistics).

ftp://ftp.rocketryplanet.com/Numerical_Methods.pdf

 

[Zeus-cat]

Thanks for the link to Numerical Methods Larry. I downloaded it and will give it a look.

I actually was thinking about something like this last night after posting the Excel file. The launch to apogee phase could be simulated by a polynomial determined using regression analysis.

 

[Zeus-cat]

I read through the Numerical Methods. Yikes; some pretty hard core stuff in there. Teh paper essentially came to the same conclusion that I did; accurate velocity and acceleration curves are very difficult to obtain from barometric altimeter data.

The data analysis I did was with data from my Perfectflite Alt15K. That is a better choice than using data from the Perfectflite MAWD (I also have one of those).

Key points:
1) The sampling rate of the Alt15K is half that of the MAWD (10 samples per second versus 20) so noise is less of an issue for Alt15K data than it is for the MAWD data. You can read the article for a detailed explanation, but the point is that the random error is not as significant for slower sampling rates when you use adjacent data points to estimate velocity and acceleration. Thus, data from the Alt15K produces more accurate velocity and acceleration data than does the MAWD.

2) The Alt15K (16 bits) is calibrated up to 15,000 feet, while the MAWD (? bits) is calibrated to 25,000 feet. If they use the same hardware then the resolution for the Alt15K is nearly twice that of the MAWD. Based on the data I have collected I suspect the MAWD uses less than 16 bits per sample.

A number of other factors that can cause inaccuracies for barometric altimeters were discussed in the paper, sampling port sizes and placement, ambient temperature versus calibration temperatures, av bay temperatures versus calibration temperatures, and others.

It is probably not worth the effort to try to estimate velocity and acceleration curves from the altitude data collected by these altimeters. Most of these altimeters cost $100 to several hundred dollars. They were designed to use barometric data to determine when to fire ejection charges. They perform that function very, very well. They are not ultra-sensitive scientific measurement devices, which is what you need to determine accurate velocity and acceleration data.

 

[BEC]

Please post both. Do you know if your spreadsheet works with
Open Office?

Just tried it in Open Office Calc 3.2.1 (Windows) and it seems to work just fine. I was able to grab the data from one of my Alt15k/WD flights from yesterday and basically paste it in to the spreadsheet and got very plausible results indeed.

I haven't tried the OP's program yet.... maybe that's next.