Alright, against my better judgement, I'm gonna lay out the reasoning of PfT and their cohorts, and why they are wrong, and why what they believe cannot be true.
Here is the "equation" for the time-slip error for the last measurement X of some event until impact:
T0: The time at which an event actually occurs (plane is X feet above ground)
T1: The time it takes from T0 for an instrument to generate a measurement
T1.5: The time it takes for that measurement to be digitally buffered at the recorder (this is effectively 0, so it can be ignored)
T2: The time it takes from T1 for the measurement to be included in the bit-stream. This is the time that the measurement remains in the buffer waiting for his turn to enter the bitstream.
T3: The time it takes from T2 for the measurement to be written to the media. This is the time it takes to encode the bitstream, compress it, and store it
T4: The time from T3 for when that sub-frame is "completed" on the media, ie, you have a "complete" sub-frame
T5: The time from T4 when data was lost for whatever reason (any and all missing sub-frames, including the impact one)
T1 you could get from the RADALT's datasheet. T2 can be estimated in the worst case easily. T3 comes from the spec of the recorder. T4 would be known via the data frame layout. T5 is the most difficult to estimate.
So, the equation that tells us how long a measurement was "accurate" until the impact is T1 + T2 + T3 + T4 + T5.
Turbofan has argued that T1 + T2 + T3 + T4 + T5 < 500ms because of regulation and an L3 salesperson telling him so:
What would be a typical time lag between the sensor signal being
generated (for example aileron angle) and the data being logged to the
protected memory of the recorder?
L-3 Response: Per ED55, it shall not exceed 0.5 seconds,
The question, as posed, essentially asks what is the answer of T2 + T3.
That means Turbofan is already wrong as this 500ms does not represent T1 + T2 + T3 + T4 + T5. At best, it represents only T2 + T3. How long from when it is generated, sits in the buffer, and makes it onto the media. And the answer that comes back is 0.5 seconds.
I do not have a copy of ED-55 but I will bet money that ED-55 does not say what this alledged L-3 person has claimed, namely, that T2 + T3 < 500ms
I am willing to do this because I can demonstrate it trivially by proving that T2, alone, can be longer than 500ms. T2 being the maximum time a measurement might wait in the digital buffer. This can be proven as follows...
NTSB said:
The output is a continuous sequence of four-second data frames. Each frame consists of four sub-frames..."
What that means is that a parameter sampled once per frame (once every 4 seconds) can sit in the digital buffer up to T2 = 4 seconds. 4 seconds > 500ms. Now, RADALT I believe was sampled every second, so for RADALT this number (T2) is 1 second.
Please note that 1 second > 500 milliseconds. So unless one of the steps requires negative amounts of time, T2 + T3 cannot be less than 500 ms.
If you were to look up the actual regulation in ED-55 my suspicion is that it would say that ONLY T3 alone < 500ms. That means from the moment the sample leaves the buffer and enters the bitstream, it is recorded within 500ms. THIS PARAGRAPH IS JUST PURE SPECULATION. But I have absolutely no doubt that what the L-3 person has allegedly said, in response to the precise question, is not true.
If anyone has a copy of ED-55 and wants to look it up, let me know what you find.
ALL THAT BEING SAID:
Turbofan and others still conviently IGNORE T1, T4, and T5 as if they do not exist. As far as they are concerned, the RADALT generates measurements instantaneously (it might) and that _zero_ frames were lost at the end of the recording due to all the things that happen between dumping the data and storing it into computer files. Like, for example, loss of frames because of synch word corruption, and removal of partial frames, and whatever else might have happened.
I can't tell you how long T5 is but I can tell you that PfT and Turbofan are assuming it is 0 and have provided no justification for that belief, at all.
They keep quoting a regulation that pertains to, at BEST, T2 and T3, and pretending it applies to T5. It doesn't.
