The FAA says that a DME ground station sends a response exactly 50 microseconds after having received an interrogation from an aircraft. For what reason is there a 50 microsecond delay?
-
$\begingroup$ An important question is: does it say (and mean) typically, or is that a specified delay? $\endgroup$– FattieMar 14, 2015 at 5:24
-
1$\begingroup$ if milliseconds is a typo, why doesn't someone edit it? $\endgroup$– FattieMar 14, 2015 at 16:47
-
$\begingroup$ It would appear that the "typically" here is totally incorrect. (See Henning below.) Someone should edit it. $\endgroup$– FattieMar 14, 2015 at 17:00
-
3$\begingroup$ @JoeBlow: At Math.SE, where I spend most of my time, we're reluctant to edit questions to remove misconceptions by the OP. Clearing up the misconceptions should rather happen in answers, because getting that cleared up could be the entire point for the asker (here the OP might be asking because he thinks 50 ms is a strangely long delay to be using -- who knows?), and having it pointed out in the form of an edit is much less visible than an actual answer. I don't know if the standards are differently on aviation.SE, but this is why I didn't edit. $\endgroup$– hmakholm left over MonicaMar 14, 2015 at 17:12
-
1$\begingroup$ I'm afraid I don't agree Reirab, you'd never use "typically" to describe an exactly specified part of a system (when it's deliberately varied shorter you're still exactly "simulating" that precision spec-specified time). $\endgroup$– FattieMar 16, 2015 at 17:04
2 Answers
The delay is not 50 milliseconds, but 50 microseconds.
It takes some time for the ground equipment to decode the incoming signal, decide it needs to reply, and fire up its transmitter. In order for the distance measurement to be reliable, this inevitable delay needs to be precisely defined and the same for all DME stations. It also gives the airborne interrogation some minimal time to shut down its transmitter and start listening even when it's right on top of the ground station.
50 µs must have felt like a suitable interval to standardize on, long enough for it to be reasonably possible to engineer the ground stations to meet it (given the technology of the time the design was made), yet short enough that one DME station can service an acceptable number of aircraft without needing to handle overlapping delays between different interrogations.
The standard delay also means that a DME station used for ILS can be located midway between the runway's thresholds and adjusted such that aircraft see the distance to the threshold rather than the distance to the physical antenna, by reducing the delay appropriately. (Such a DME does not give useful readings for aircraft not aligned with the runway).
Note, by the way, that the 50 µs is the time between receiving the second pulse of an interrogation pulse-pair and transmitting the second pulse of the resply. So the ground station actually has to start transmitting the reply less than 50 µs after it knows it has received a complete interrogation. For Y channels where there are 30 µs between the two reply pulses, the ground station actually has only 20 µs to react.
-
$\begingroup$ A small point: the overlapping issue would be irrelevant, I think. If the delay was (let's say) 20 seconds, it would be utterly trivial to still send each one out exactly 20 seconds after the relevant arrival, even if as it happened many were arriving at the same time. In other words regarding this issue: "handle overlapping delays between different interrogations", it's quite true the soft3are engineers would have to "handle" that but (A) it makes utterly no difference what the specified delay is, and (B) as it happens, it is trivial anyway. $\endgroup$– FattieMar 14, 2015 at 5:28
-
$\begingroup$ {Any obscure racetrack bugs which would appear - would appear anyway regardless of the set delay.} $\endgroup$– FattieMar 14, 2015 at 5:28
-
$\begingroup$ @JoeBlow: It is indeed possible to handle overlapping delay period if you have a computer to keep track of them -- not utterly trivial, because the exact timings of replies have to be executed with something approaching nanosecond precision, but certainly doable. However, remember that the system dates back to the 1950s. At that time you wouldn't have a computer around to control the ground station; the delay would be implemented with analog circuits. In that setting, keeping several different delays active at the same time would be an entirely different problem. $\endgroup$ Mar 14, 2015 at 10:01
-
2$\begingroup$ @JoeBlow: It is precisely 50 µs (still not 50 ms, which is one thousand times longer than 50 µs). The principle of DME is that the aircraft measures how long it takes for the reply to arrive, and multiplies that with the speed of light. If the ground station replies with less than the specified 50 µs of delay, the aircraft will think it is closer to the DME that it actually is. For example, responding 12 µs too early will lead to a reading one nautical mile too short. $\endgroup$ Mar 14, 2015 at 16:48
-
2$\begingroup$ @JoeBlow: As for "70 years ago": The existing protocol works, and changing it would require all aircraft to carry receivers for two different systems in a transition period -- it could be years or decades until the ground stations that implement the old protocols had all been replaced. The cost of this would be phenomenal, so a change won't happen unless it provides concrete benefits. Not just because "we might have chosen a different number if we started over from scratch today". $\endgroup$ Mar 14, 2015 at 16:53
the purpose of the 50 microsecond delay is to eliminate the possibility of uncoordinated operation when the aircraft is very close to the ground station
From AVweb - DME Basics
Addendum
I have not found a clear explanation for what is meant by "uncoordinated operation" in the context of DME. What follows is speculation:
If you look at patents related to DME, it seems this predictable delay ensured that the initial transmission from the aircraft would have been completely received before the reply was transmitted by the base-station and received at the aircraft.
I believe a DME response has to match the pattern of the DME interrogation so that the receiver in the aircraft can identify which response to apply timing to if multiple DME-equipped aircraft are in range.
The pulse spacing is 12 µs1 and this is the same order of magnitude as the round-trip transmission time at 1 mile. A single DME interrogation contains several pulses. It may be difficult to design ground station equipment that can start transmitting a matching response if it has not yet received the pattern of data it has to reproduce. The alternative would be to start transmitting the response as soon as possible after detecting the end of receipt of a valid DME interrogation.
There are delays in processing within the ground station equipment which may vary (at least from station to station depending on equipment) and which therefore need to be made constant for accurate distances to be calculated by the airborne receiver. Any variations would be more significant at closer distances.
It might also be easier to design airborne equipment if it does not have to start receiving the response before it has finished transmitting the interrogation.
References
I believe DME was invented around 1944-45 but cannot find any patents that early. There are quite a few patents on this subject.
-
1$\begingroup$ Although it does not provide the detail, this should be the accepted answer. Specifically, the "uncoordinated operation" would occur if the transmitter was still sending the second interrogation pulse when the first reply was received from an aircraft at close range. 50 µs corresponds to approx 4 NM within which, without the delay, aircraft would show unreliable readings. $\endgroup$– SimonMar 14, 2015 at 11:50
-
$\begingroup$ @Simon: It is not the correct answer. The DME interrogator on board determines the slant range by measuring the time for the signal to reach the ground station (transponder), plus the time for the ground station to send the response (call this one T), plus the time for the response to reach the on board interrogator. Then the interrogator which is interested only in the time for the signal to travel back and forth, needs to remove T. If T is variable (dependent of the station design) then it is not possible. Hence T must be standardized. $\endgroup$– minsMar 14, 2015 at 12:01
-
$\begingroup$ Actually, I've just realised this "reference" is just from a pretty crappy, 20 year old general-interest magazine article! Sorry I at first commented that it seemed to be factual. $\endgroup$– FattieMar 14, 2015 at 16:51
-
$\begingroup$ Hi mins. Do you have an actual reference that states the 50ms is standardised. What's the allowed error for equipment makers? (if it's an ISO standard or something, it's inconceivable these days it wouldn't be given with an error range.) $\endgroup$– FattieMar 14, 2015 at 16:53
-
1$\begingroup$ @JoeBlow: The 50 µs (not ms) is standardized in ICAO Annex 10, volume I, section 3.5.4.4. (This doesn't seem to be publicly available, but there are pirated copies floating around if you google hard enough). The specified tolerance is ±1 µs (section 3.5.4.5), which corresponds to an error of 500 feet in the result. Most installed DMEs do better than that. $\endgroup$ Mar 14, 2015 at 17:03