The other post. Both by Ken Biba.

I posted this in another discussion relating to competition measurement of altitude - particularly for TRA records that are now being set above 40k' MSL. It argues that hobby barometric altimeter systems currently available have inherent errors that are no less than 10% and likely approach 20% in aggregate - certainly in a flight regime above 35k'MSL. I will be posting a companion piece about GPS systems with an analysis illustrating a 95% vertical accuracy of less than .5% (30m or 100') above 35k' MSL.

There are a number of independent sources of hobby barometric altimeter error as used to provide geometric altitude. These need to be added to find the total error:

1. Inherent error in the sensor used. The older MPX4100A used in the AltAcc, ARTS, ARTS2, RDAS Classic/Compact, Missile Works, MiniAlt, GWIz MC, GWiz LC and others ... has a cliff below 20 kPa (which if you use the 1976 Standard Atmosphere Model) is the equivalent of 38.7k' MSL. With an error bound of roughly 10% on the voltage at the extreme range ... which translates to a 5.1% error bound on altitude using that same 1976 Standard Model.

The newer MPX2102 actually has no cliff in lower pressure, but the relative error at low pressure increases dramatically (per the data sheet) ... since a lot depends on how much additional noise is introduce by a needed preamp, I would guess that this device has at least the error bound of its older brethren and likely 2x. I would prudently guess 10%. This sensor is used in the new GWiz LCX and HCX and I suspect will commonly be used in new altimeter designs. It is a good sensor. Just unlikely to be precise, particularly at high altitude and low pressure.

The PerfectFlite HA45 uses an in between sensor that has a cliff at just short of 45k'MSL with a similar error characteristic of the MPX4100.

The Adept 60K is currently an unknown to me .. if a kind reader that has one will email the chip identity we can make some progress.

It MAY be POSSIBLE to individually characterize a particular example of a sensor .. but even if an individual sensor is calibrated - there are inherent errors in the process of converting pressure to geometric altitude that are just as large as we will see below.

2. Analog to digital conversion errors ... really depends on the altimeter. But the AltAcc uses an 8 bit ADC which will introduce perhaps an additional .5% error in the basic pressure measurement. Newer altimeters uses 10,12 or more bit ADC that further lower this source of error.

3. Software implementation errors. The de facto method of converting pressure to feet is a table lookup using a measured pressure difference between launch and apogee into the Standard Atmosphere Model to convert kPa to feet. Sadly there are multiple versions of the SAM that all disagree materially and we really have no idea which version an individual vendor is using. From personal experience, I personally suspect that early versions of the (otherwise outstanding!) GWiz LCX have suffered from this problem and have observed wide variances in reported altitude. I am assured that this has been fixed in v1.7 (the currently shipping version) and hope to personally validate this in a transparent manner soon. But despite using a better sensor (the MPX2102) ... software can still yield errors that make the reported geometric altitude unreliable. And in low volume usage, these errors can often take some time to find and fix.

4. Last and worst ... the inherent, fundamental flaws in the Standard Atmosphere Model. There are multiple disagreeing versions and even the current (1976) version is only a world wide average that does not recognize the fundamental local variances in the atmospheric pressure column that happen due to season, latitude, sunspots and local weather. And the variances are indeed HUGE in translating measured pressure difference to an asserted geometric altitude ... a range of 20% for a given altitude over the world and over the year.

Some of these variances are somewhat predictable ... season and latitude. But without local, timely radiosonde data I am skeptical that a conversion EVEN WITH THESE adjustments would be within 5%.

The accumulation of these independent sources of errors to me argue strongly against permitting baro altimeters for contest records at high altitudes. I can see no way that ANY of these products meet the 5% rule at high altitude.

And for altitude bragging rights ... baro altitude reporting does begin to resemble a fish story .. particularly at high altitude.

Someone may be able to construct a modern calibrated baro altimeter with a wide range ADC with validated software that uses current radiosonde data for conversion to geometric altitude and convince an independent committee convened on the Internet that it is accurate.

But I am skeptical of the physical and emotional cost/benefit of that approach.

And I think it is just easier, cheaper and more accurate to use a GPS. That has a well documented and independently verified vertical error of well under several hundred feet independent of altitude (well short of LEO anyway). A companion note will argue that the 95% error for vertical position in modern GPS is less than 30m - about .3% at 40k'MSL!

I guess I am just discriminatory of good equipment used for inappropriate applications.

I should add that I am a satisfied customer of AltAcc, RDAS, Missile Works, GWiz, PerfectFlite, Adept and others that I am now forgetting. For recovery deployment.

However, I rely for altitude on my BeelineGPS and my RDAS GPS. I have used GPS/Flight and I understand that ARTS2 also supports a GPS though I have not used it.

Maybe I need to get on the listserver, because the following is just pure B.S.:

The newer MPX2102 actually has no cliff in lower pressure, but the relative error at low pressure increases dramatically (per the data sheet) ... since a lot depends on how much additional noise is introduce by a needed preamp, I would guess that this device has at least the error bound of its older brethren and likely 2x. I would prudently guess 10%. This sensor is used in the new GWiz LCX and HCX and I suspect will commonly be used in new altimeter designs. It is a good sensor. Just unlikely to be precise, particularly at high altitude and low pressure.

Here is the datasheet for the MPX2102: http://www.freescale.com/files/sensors/doc/data_sheet/MPXM2102.pdf The only relevant spec in the datasheet that has any bearing on the accuracy of the installed sensor is the spec for 1% linearity at full scale. Whoever made that post either is just pulling numbers out of the air, or he is unclear on the concept of calibrating sensors after installation, and what that means for accuracy. With a properly calibrated altimeter, the error will go down with altitude, as the measured pressure approaches the calibration point.

As for the pre-amp noise, the Parrot's end-to-end measurement noise only 1-2 bits on its 13-bit measurement.

I don't argue that GPS units aren't accurate. But at the moment, they're large and expensive. I'm perfectly willing to talk about baro altimeter limitations, but it ticks me off when people just make stuff up. I stand by my statement that there is no reason for any baro altimeter to measure pressure with worse than 2% absolute accuracy.

The only point I would make is that as far as I know, no manufacturer does individual unit calibration on altimeters. I've spoken to one manufacturer who I know pretty well on a face to face basis and I know he doesn't individually calibrate altimeters to known, traceable standards. Perfectflight definitely doesn't either.

What I found interesting was the discussion on SAM's.

I think I'm going to start a discussion with the executive committee on altitude records though - above 25K in particular. I've already sent an email off to TRA.

Warren

Fascinating. I thought I read on some sites that they were. The Parrot altimeters are all definitely individually calibrated against a NIST-traceable standard. At two temperatures.

That's going to be one thing that sets your altimeters apart from the run of the mill Adrian. At this point I'm past wanting to fly your altimeter with another altimeter - I want to fly it with a GPS and see how it compares.

Warren

Thanks, Warren. I look forward to flying a Parrot together with a GPS unit to the highest possible altitude.

I've flown a few logging GPS altimeter shots, including one in my L3. All of them have significant spikes during shock events like lift off or deployment. The all re-aquired the signal relatively quickly. It's my opinion that any GPS altimeter used for a record attempt should be required to filter out abnormal shock event data points.

From my limited experience, the altitude data was not even close. Off by a well over a thousand feet - but this was based on one second NMEA strings. A one second lag followed by a shock event gives plenty of room for error. Once I cleaned up the data, it looked pretty cool displayed in Google Earth's 3D view.

While it's quoted that they can be pretty accurate, I sure as heck would not use them for deployment. Give me a tried and true altimeter any day.