Servo and pulse widths

[dam74]

Hi, I'm just having a bit of a browse into PWM control of servos and but I think a lot of the information I'm reading are refering to industrial servos as they talk about servos operating at 50Hz with a pulse width of 20ms. From the Align Trex450 manual I see that the DS420 has a pulse width of 1520us which by my calculations would mean it would have a frequency of around 657.89Hz (rounded down). Does this sound about right or am I calculating it wrong?

Are there any good docos out there explaining the frequency, pulse widths and control pulses. From what I've read a 50Hz (20ms pulse width) Servo has 1.0ms pulse for full left, 1.5ms pulse for middle and 2.0ms pulse for full right.

[serpent887]

There is endless stuff out there but here is a basic doc for starters. http://www.rcheliwiki.com/Servo_protocol

Remember some Futaba servos have a frame rate of 760uS.

[GravityKills]

there's the distance between pulses and the pulse width.
For example, analog servos have 1/50 = 20 ms pulse distance and about 1..2 ms pulse width.

A picture tells more than a 1000 words (source)
It shows
- analog pulse rate (radio)
- a common digital pulse rate (swash servos)
- a higher pulse rate for the tail servo

[dam74]

Cheers Serpent, yeah in the Align manual it mentions for the GP750 setup about 1520/760us pulse widths. I'll take a look at that doco.

[dam74]

there's the distance between pulses and the pulse width.
For example, analog servos have 1/50 = 20 ms pulse distance and about 1..2 ms pulse width.

A picture tells more than a 1000 words (source)
It shows
- analog pulse rate (radio)
- a common digital pulse rate (swash servos)
- a higher pulse rate for the tail servo

Cheers, I expected the tail servo to commonly have a higher pulse. So a 1520us pulse width is the actual pulse signal length not the distance betweeen pulses?

[Dedcalm]

the centre position of a standard servo (most) is 1520uS. Most servo's work in the range of 1000uS to 2000uS or 1ms to 2ms if you change the decimal place.

Again most servos work on a ~20ms refresh rate, i.e. 50Hz. Although if you scope out different receivers you get different rates. i.e. 16Hz for example.

Digital servos can cope with high refresh rates, 166Hz or 250Hz or even 333Hz. Also some of the more specialised (tail) servos such as the 9256 use a 720uS centre. I have not played with any of these though (as I don't have any).

Let me know if you need more help/info on this.

Rob

[GravityKills]

Cheers, I expected the tail servo to commonly have a higher pulse. So a 1520us pulse width is the actual pulse signal length not the distance betweeen pulses?

yes, that is correct: At 50 Hz, the distance is 20 ms = 20000 us.

Historically, the reason for the narrow pulses is that a number of servo channels (8?) could be multiplexed onto a single radio channel in stone-age PPM radios.

[dam74]

K, so the pulse width specified is the center signal of the servo.

I haven't got an oscilliscope yet was looking at some of the ones at farnell and maplins but haven't got the money to fork out on one. It's something I'd like to add to my electronic equipment that I have. I've looked at some of the PC based ones but they're not really a cheap alternative either.

[serpent887]

K, so the pulse width specified is the center signal of the servo.

Correct

[Dedcalm]

You don't necessarily need a scope to investigate servo signals - for example I have a PIC with a 4 x 20 LCD screen on a bread board and that measures the signals and can then tell me the frequency in Hz as well as the pulse length and display them on the LCD. Its actually far easier to use than the scope for measuring servo signals Also the code was simple to write and I can modify it to do what I want/need as I like.

[dam74]

I'm planning on doing something similar with my AVR and Batron (think 40x4 backlit lcd). Just digging up info on them, still haven't got around to writing USB comms to the AVR which is the next thing on my list to do. I might tap you up for the code for the PIC as I have one of the very first ones floating about

When I pulled apart the DX6i I was pleasantly surprised to see an AVR chip powering it You see a lot of them about in industry as well such as VSDs think either ABB or Danfoss use them.

[Dedcalm]

Think my unit was a 16F684 running 20MHz with a 44780 based LCD.

[pchristy]

Beware of confusing pulse width with frame rate. When Doug Spreng designed the original pulse-width servo, he used a variable pulse width of 1-2mS to determine the servo position. This pulse is repeated at a rate of about 50-60 Hz (20-16 mS).

In a PPM system - all that there was 45 years ago, if you exclude analogue systems - then a 20 mS frame rate gives you room for up to 8 pulses at their maximum length (8x2=16) PLUS a 4mS "sync" period. Since the sync pulse was twice as long as the maximum "signal" pulse, it was easy to separate out using a simple integrator. This was used to reset the decoder ready for the next frame of information.

Its a tribute to Doug Spreng's genius that we are still using essentially the same system nearly 50 years after he designed it!

PCM systems tend to have a slower frame rate than PPM. An 8-bit system has to transmit 8 pulses per channel (putting it very crudely) instead of 1. Bandwidth restrictions on 35 MHz limit how fast you can transmit information - typically you are looking at fall/rise times of around 100uS for each pulse. PCM gains a bit back as it doesn't have to pause for a sync period, but it is still slower than PPM, and always will be. The situation gets worse the more bits you transmit - a 10 bit ystem will be slower than an 8-bit! However, what you lose on the swings you gain on the roundabouts in terms of added security - a PPM system has no error checking whatsoever!

When we talk about high frame rate servos, bear in mind that the transmitter CANNOT transmit any faster than it already does - due to the aforementioned bandwidth restrictions! However, the GYRO - which is hard wired to the servo - is under no such constraints and can send information as fast as the servo can accept it.

Of course, 2.4 GHz is under no such bandwidth constraints, and can transmit far faster than is possible on 35 MHz - which is why users report the system feeling more responsive. However, to maintain backward compatibility with existing servos (which may not be high rate ones) the frame rate is still restricted to around 16-20 mS.

There's a long out-of-print book by Paul Newell called "Model Radio Control". Its an excellent read for anyone interested, and you should be able to get your local library to get a copy.

--
Pete

[Angelos]

There is some info here regarding frame rate and pulse width Spartan RC

PCM systems tend to have a slower frame rate than PPM. An 8-bit system has to transmit 8 pulses per channel (putting it very crudely) instead of 1. Bandwidth restrictions on 35 MHz limit how fast you can transmit information - typically you are looking at fall/rise times of around 100uS for each pulse. PCM gains a bit back as it doesn't have to pause for a sync period, but it is still slower than PPM, and always will be. The situation gets worse the more bits you transmit - a 10 bit ystem will be slower than an 8-bit! However, what you lose on the swings you gain on the roundabouts in terms of added security - a PPM system has no error checking whatsoever!

Whilst the background theory is correct there are cleaver methods that can give better performance. PCM1024 for example has a frame rate of 71Hz which is faster than PPM transmission. They achieve that by:

1. Re-mapping the bit pattern in sequences that single bit 0s and 1s don't exist. Obviously these sequences are a little larger than the original message but because no single bit pulses exist the bandwidth requirement is halved. The required bandwidth is defined by the highest freuqency of the data stream = the shortest pulse.

2. Sending absolute position data in one frame, and a relative correction on the next. For example odd (radio) frames have channels 1,3,5,7 absolute and 2,4,6,8 relative. Even frames have 2,4,6,8 absolute and 1,3,5,7 relative. Channels 9 and 10 are single bit = two fixed positions.

In PCM1024 the shortest pulse is 150uS.

-Angelos

[Edders]

I bought this

TextStar 'Stick-On' Serial LCD Display with 'GraphSmart' | Cat's Whisker Technologies


Edders