Steve_B
70cc twin V2
There has been some heated discussion on the forum about what radio gear is the best and one thing that comes up a lot is 'latency'.
So what is this latency thing?
Technically in RC terms latency is the time taken between moving the stick on the Tx and the servo achieving the commanded position. The higher the latency the longer the time taken therefore we want our latency to be as low as possible to give us faster responding controls.
For RC model purposes latency is usually measured in milliseconds (ms). One ms is one thousandth of a second (0.001s). To put that in context: 'the blink of an eye' is about 300ms long, so one ms isn’t very long at all!
So what causes latency?
Well lots of things add up to cause latency. The time taken for the servo to move to position after it receives the signal is probably the biggest one so to reduce latency we should use the fastest servos possible, no big shock there.
If for the sake of this discussion on radio system latency we exclude the servo and look only at the part of latency that’s due to the radio system itself. That is the delay between the stick moving and the servo receiving a command to move.
Some might reasonably assume that the delay is mainly due to the time it takes for the radio waves to travel between the Tx and the Rx. Surprisingly this is not the case. The time taken for the signal to travel is virtually zero. The radio signal travels at the speed of light, which is incredibly fast. Looking at the numbers
Speed of light = 300,000,000 metres per second (rounded up)
So if our 3D model is 300 metres away time taken for the signal to reach the model is:
300/300,000,000 = 0.000001 seconds, or 0.001ms one thousandth of a millisecond. Cutting to the chase, to all intents and purposes the signal gets from the Tx to the Rx instantly, so we can forget it.
If the signal is instant what else causes the delay? Again there are several things:
• Modern radios have computers in them. The computer receives a signal from the stick that is in the form of a variable voltage. The electrical signal travels through wire almost at the speed of light so is virtually instant. When it gets to into the computer chips things slow down. The computer has to take the voltage signal from the stick and do lots of complex math, factoring is stuff like end point adjustments, rates, exponential setting, mixes etc. etc. so it can figure out what the required signal that needs to be sent to the plane. For instance you might move the stick 50% travel but by the time rates, exponential and end points and mixes are all factored in the signal that goes to the plane could be maybe 33% travel. The computer has to crunch the numbers and this takes time. The computers in transmitters are not all equal. Some do this number crunching faster than others.
• From the computer the data is sent to the part of the radio that actually sends the signal. The data has to be converted into radio signal and transmitted, this takes some time. Some radios have to convert data to old fashioned PPM ‘frames’ that adds additional delay (the radios with add on 2.4GHz modules all suffer from this but some others do too). Lower frame rates improve this and some radios have ditched PPM altogether which is better still.
• When the signal arrives at the Rx it has to be decoded and a signal sent to the servos. This decoding takes time.
• Most radios can’t send signals to all servos at the same time so the servo has to wait its turn for data to come to it. This again takes time depending on where the particular servo is in the queue when the signal arrives at the Rx.
Only now after all this data conversion and computerised number crunching does the signal arrive at the servo. It’s actually amazing it all happens in a tiny fraction of a blink of an eye.
What’s the latency for my radio?
This thread gives some very interesting data on typical latency figures for current radios: http://rc.runryder.com/helicopter/t172571p1/
Here is the latency table reproduced from that thread:
There are some pretty interesting and quite surprising results in that table.
But what does latency really mean, to me?
Ok.. so now we know what latency is, what causes it and what radios have high and what have low latency. Trouble is ‘milliseconds’ are fairly abstract things to get your head around. What difference does it make if my radio has 10ms latency as opposed to say 55ms latency? Will I even notice a 45ms difference if the blink of an eye is 300ms?
How far the plane would fly between the stick moving and the servo receiving a signal maybe helps put some reality into pretty much meaningless talk of a few ms here and there. Of course it totally depends on how fast the plane is flying! 3D model aren't really very fast so what about if we take 30mph as an average speed half way between walking speed harrier and WOT high speed manoeuvres?
30mph is 44ft per second (5280/60^2)
How far the plane travels before the servo receives a signal to move of some typical systems (latency data from previous table):
• Airtronics SD-10G (10.3ms average latency) - plane travels 0.44ft (5.3")
• Futaba 8FG (14.1ms average latency) - plane travels 0.62ft (7.4")
• Spektrum DX8 (19.2ms average latency) - plane travels 0.84ft (10.1")
• Futaba 14MZ (47ms average latency) - plane travels 2.1ft (25.2")
• Hitec Aurora 9 (53.5ms average latency) - plane travels 2.35ft (28.2â€)
So the difference between the fastest and the slowest radios at 30mph flying speed, everything else being exactly equal, to be near enough 2ft (24"). That could very easily be enough to make the difference between a crash and a near miss especially when you might only be flying 6ft above the ground to start with. There doesn't seem to be much correlation between radio price and latency. In fact due to the complexity of the computation being carried out in the higher end systems it’s possible that latency of that stage is increased. Some systems may achieve lower latency when operated in alternative modes (the 14MZ for example).
Of course radio latency is only part of the latency story. As noted earlier servos are really important too but the slowest part of the overall control system is the guy holding the sticks. Human reaction speed is 200ms and more even when you are fully alert!. So as usual we find that the human is the weakest link in the chain.
So what is this latency thing?
Technically in RC terms latency is the time taken between moving the stick on the Tx and the servo achieving the commanded position. The higher the latency the longer the time taken therefore we want our latency to be as low as possible to give us faster responding controls.
For RC model purposes latency is usually measured in milliseconds (ms). One ms is one thousandth of a second (0.001s). To put that in context: 'the blink of an eye' is about 300ms long, so one ms isn’t very long at all!
So what causes latency?
Well lots of things add up to cause latency. The time taken for the servo to move to position after it receives the signal is probably the biggest one so to reduce latency we should use the fastest servos possible, no big shock there.
If for the sake of this discussion on radio system latency we exclude the servo and look only at the part of latency that’s due to the radio system itself. That is the delay between the stick moving and the servo receiving a command to move.
Some might reasonably assume that the delay is mainly due to the time it takes for the radio waves to travel between the Tx and the Rx. Surprisingly this is not the case. The time taken for the signal to travel is virtually zero. The radio signal travels at the speed of light, which is incredibly fast. Looking at the numbers
Speed of light = 300,000,000 metres per second (rounded up)
So if our 3D model is 300 metres away time taken for the signal to reach the model is:
300/300,000,000 = 0.000001 seconds, or 0.001ms one thousandth of a millisecond. Cutting to the chase, to all intents and purposes the signal gets from the Tx to the Rx instantly, so we can forget it.
If the signal is instant what else causes the delay? Again there are several things:
• Modern radios have computers in them. The computer receives a signal from the stick that is in the form of a variable voltage. The electrical signal travels through wire almost at the speed of light so is virtually instant. When it gets to into the computer chips things slow down. The computer has to take the voltage signal from the stick and do lots of complex math, factoring is stuff like end point adjustments, rates, exponential setting, mixes etc. etc. so it can figure out what the required signal that needs to be sent to the plane. For instance you might move the stick 50% travel but by the time rates, exponential and end points and mixes are all factored in the signal that goes to the plane could be maybe 33% travel. The computer has to crunch the numbers and this takes time. The computers in transmitters are not all equal. Some do this number crunching faster than others.
• From the computer the data is sent to the part of the radio that actually sends the signal. The data has to be converted into radio signal and transmitted, this takes some time. Some radios have to convert data to old fashioned PPM ‘frames’ that adds additional delay (the radios with add on 2.4GHz modules all suffer from this but some others do too). Lower frame rates improve this and some radios have ditched PPM altogether which is better still.
• When the signal arrives at the Rx it has to be decoded and a signal sent to the servos. This decoding takes time.
• Most radios can’t send signals to all servos at the same time so the servo has to wait its turn for data to come to it. This again takes time depending on where the particular servo is in the queue when the signal arrives at the Rx.
Only now after all this data conversion and computerised number crunching does the signal arrive at the servo. It’s actually amazing it all happens in a tiny fraction of a blink of an eye.
What’s the latency for my radio?
This thread gives some very interesting data on typical latency figures for current radios: http://rc.runryder.com/helicopter/t172571p1/
Here is the latency table reproduced from that thread:
There are some pretty interesting and quite surprising results in that table.
But what does latency really mean, to me?
Ok.. so now we know what latency is, what causes it and what radios have high and what have low latency. Trouble is ‘milliseconds’ are fairly abstract things to get your head around. What difference does it make if my radio has 10ms latency as opposed to say 55ms latency? Will I even notice a 45ms difference if the blink of an eye is 300ms?
How far the plane would fly between the stick moving and the servo receiving a signal maybe helps put some reality into pretty much meaningless talk of a few ms here and there. Of course it totally depends on how fast the plane is flying! 3D model aren't really very fast so what about if we take 30mph as an average speed half way between walking speed harrier and WOT high speed manoeuvres?
30mph is 44ft per second (5280/60^2)
How far the plane travels before the servo receives a signal to move of some typical systems (latency data from previous table):
• Airtronics SD-10G (10.3ms average latency) - plane travels 0.44ft (5.3")
• Futaba 8FG (14.1ms average latency) - plane travels 0.62ft (7.4")
• Spektrum DX8 (19.2ms average latency) - plane travels 0.84ft (10.1")
• Futaba 14MZ (47ms average latency) - plane travels 2.1ft (25.2")
• Hitec Aurora 9 (53.5ms average latency) - plane travels 2.35ft (28.2â€)
So the difference between the fastest and the slowest radios at 30mph flying speed, everything else being exactly equal, to be near enough 2ft (24"). That could very easily be enough to make the difference between a crash and a near miss especially when you might only be flying 6ft above the ground to start with. There doesn't seem to be much correlation between radio price and latency. In fact due to the complexity of the computation being carried out in the higher end systems it’s possible that latency of that stage is increased. Some systems may achieve lower latency when operated in alternative modes (the 14MZ for example).
Of course radio latency is only part of the latency story. As noted earlier servos are really important too but the slowest part of the overall control system is the guy holding the sticks. Human reaction speed is 200ms and more even when you are fully alert!. So as usual we find that the human is the weakest link in the chain.
