RRC3 velocity log is off

I just got back from XPRS and downloaded my RRC3 Flight data. Everything looks good to me, except velocity. It says it went 2450 MPH for a brief time. I don't think this is possible because my rocket was a 3in diameter flying on an L1000 that was simed to mach 1.2. Yes, I made sure I was in the correct units!

I used 4 vent holes (3/32 in size) as recommended by: https://www.vernk.com/AltimeterPortSizing.htm and a few other sources. Would this kind of data indicate too much/little pressure in the altimeter bay? I attached all the files from the RRC3. If anyone could spare the time to look at my data, that would be great!

View attachment XPRS_Scratch_2015.rff

Your simulation is likely more correct. Getting velocity from your barometric altitude measurement is inherently inaccurate and needs manipulation to be useful. Is the RRC3 data raw or filtered in some way?

What is the rff file? I can't open it. A simple text file that can be parsed in Excel would be easier for people to open and inspect.

Velocity from barometric info is inaccurate. At best, with manipulation, it is very rough.

You want velocity and acceleration get a unit with an accelerometer.

The rff file is the raw data file from the RRC3 , which opens in it's own software. Here is the PDF version:View attachment XPRS 2015 Flight Data XPRS 2015 Flight Data.pdf


I always thought Baro was inaccurate, but to be 3x inaccurate seems a bit much. ~800mph to 2450mph seems like a huge gap to me. With that said, what altimeter would you recommend that has an accelerometer?

The pdf doesn't help much. Doesn't the RRC3 program export txt or csv?

Anyway, I did manage to copy and paste the data around the peak velocity into Excel and plotted it. If this is raw data, then there is no way that velocity peak comes from that altitude curve. Something else (averaging, filtering) must be going on.

Killachrome said:

The rff file is the raw data file from the RRC3 , which opens in it's own software. Here is the PDF version:View attachment 272517


I always thought Baro was inaccurate, but to be 3x inaccurate seems a bit much. ~800mph to 2450mph seems like a huge gap to me. With that said, what altimeter would you recommend that has an accelerometer?

Click to expand...

There are a number of them. The Raven, the Marsa, certain Altus Metrum products.

Your altimeter pressure data is extremely noisy, and varies by 1800' altitude before lift-off! Was the wind gusty? What was the diameter and length of the altimeter compartment? This noisiness appears to be in the first second or so of the flight so you velocity is going to be way of just because of this, and that doesn't even account for transonic pressure gradients....

Bob

bobkrech said:

Your altimeter pressure data is extremely noisy, and varies by 1800' altitude before lift-off! Was the wind gusty? What was the diameter and length of the altimeter compartment? This noisiness appears to be in the first second or so of the flight so you velocity is going to be way of just because of this, and that doesn't even account for transonic pressure gradients....

Bob

Click to expand...

There was maybe 5 mph of wind, so not much. The compartment was 9in long x 2.59 in diameter. This is the 5th time I have flown this unit and all the other times were fine. Should I add the extra 1800' to the flight as well? I will have to look into using a new altimeter setup and perhaps use this as a back-up unit.

Killachrome said:

There was maybe 5 mph of wind, so not much. The compartment was 9in long x 2.59 in diameter. This is the 5th time I have flown this unit and all the other times were fine. Should I add the extra 1800' to the flight as well? I will have to look into using a new altimeter setup and perhaps use this as a back-up unit.

Click to expand...

I have a terrible confession to make. I seriously have no idea how to tell, from the data (not from the prescribed formulas), if the static ports are placed, sized, and trimmed correctly. Having no idea of that, I also have no idea how the formulas were derived.
Why do I have these problems? Because I have NEVER (not once) seen barometric data that make total sense at high speeds. And, by high speeds, I don't mean supersonic speeds or trans-sonic speeds. Just fast sub-sonic speeds will do. (And yeah, I've heard of Bernoulli, Poiseuille and even enthalpy.)

Some such data make sense in the altitude graph, but if you tease them out a bit more, you discover weird stuff going on around maximum speed. Not just noise, here, but also distortion. For example, I have ways of analyzing barometric data along with inertial data to refine speed and to derive sines of angles WRT the horizontal. Those sine graphs all have funny stuff going on near maximum velocity. Some are worse than others, but the effect is always present.

I also don't know how a very fast rocket can have an instrument compartment that is in pressure equilibrium with the outside at all times. Pressure changes take time and some rockets go very far up in very little time. You want to resize those holes so it don' take time? Explain this thing to me. (Really. Please!)

Altimeter data are also typically smoothed with recursive filters, which work through weighted averages with previous readings. Trouble is, during ascent, all of the readings with the POSSIBLE exception of the current reading are from lower than the current altitude, so the filtered readings will be biased low, and that bias will be worse at high speeds.

As for noise, numerical differentiation makes noise explode. The higher the sampling frequency, the more the noise explodes.

Anyway, the result is that barometric velocity data tend to get squirrelly near maximum velocity - which is exactly where they are most interesting.

Now. You got a problem there too? Well I'm here ta tell ya, things are tough all over!

OK. OK. Sorry. Couldn't resist. The spike can probably be removed, but the filters that remove such spikes (e.g. Savitzky-Golay) have personalities of their own. They will want to round that point, and the amount of rounding will have as much to do with the filter as with the data. It'll look prettier. Be happy to do it if you post the data.

Luck, Regards, and... Relax
-LarryC

Larry Curcio said:

OK. OK. Sorry. Couldn't resist. The spike can probably be removed, but the filters that remove such spikes (e.g. Savitzky-Golay) have personalities of their own. They will want to round that point, and the amount of rounding will have as much to do with the filter as with the data. It'll look prettier. Be happy to do it if you post the data.

Luck, Regards, and... Relax
-LarryC

Click to expand...

Here is the txt and xls file (In a zip file because of upload size) that I think you would need:

TXT: View attachment XPRS 2015 Flight Data.txt

XLS in a ZIP: View attachment XPRS 2015 Flight Data.xls.zip

This is the plot of the first 5 seconds of your flight data.



1. Note the negative altitudes (higher barometric pressures) for the first 0.6 second. Highly unusual.

2. Note the Mach transitions in the altimeter trace starting at 1 and 3 seconds. Possibly indicates velocity breaking and dropping thru Mach respectively.

3. Note the anomalously high velocity appears above Mach.

4. There is definitely an erroneously high calculated velocity derived from the pressure altitude.

Bob

Killachrome said:

Here is the txt and xls file (In a zip file because of upload size) that I think you would need:

TXT: View attachment 272590

XLS in a ZIP: View attachment 272591

Click to expand...

OK. Yeah. You have noise and some supersonic stuff going on, as Bob K. points out. I'm wondering if the initial filtered value in the data isn't the first value identified in the launch detect. Then the noise might be the result of the filter stabilizing. (Oddly, I have seen supersonic flights where there is relatively little effect at maximum velocity, and these are extremely puzzling to me. This one is not one of the exceptions.)

I used a Savtizky Golay filter with a window size of 61 and 101 - that's way over-filtered in an attempt to shave that peak. To get around the initial point problem, I reflected the data set around its first point. It's a standard rough-and-ready technique for smoothing and differentiating when the derivative at the first point (velocity, here) is expected to be zero. Not a whole lot of effect on maximum velocity. The smaller window actually had a higher max velocity. (FWIW, the order of the polynomial was 4 - that gives enough information to repeat the procedure.)

The shape of the velocity curve is different, mostly because the underlying altitude curve is smoothed. (Smoothed altitude is labeled SALTITUDE)

SF works by fitting a polynomial to the point in a moving window. Fitting is by ordinary least squares. The polynomial is differentiated at the middle point in the window, and that is the numerical derivative at that point. The window is shifted over by a point and the process is repeated. Sometimes this works well, but barometric data... GIGO, as we used to say.

A picture is easier to explain Larry's reply. This is the plot of the initial 5 seconds after filtering. No amount of filtering is going to work...as bad data gives bad answers...

Bob