forked from richardghirst/PiBits
-
Notifications
You must be signed in to change notification settings - Fork 5
Richard G. Hirst's 'servoblaster' userspace PVM Servo controller for Raspberry Pi
License
srcshelton/servoblaster
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
ServoBlaster This is software for the RaspberryPi, which provides an interface to drive multiple servos via the GPIO pins. You control the servo positions by sending commands to the driver saying what pulse width a particular servo output should use. The driver maintains that pulse width until you send a new command requesting some other width. By default is it configured to drive 8 servos, although you can configure it to drive up to 21. Servos typically need an active high pulse of somewhere between 0.5ms and 2.5ms, where the pulse width controls the position of the servo. The pulse should be repeated approximately every 20ms, although pulse frequency is not critical. The pulse width is critical, as that translates directly to the servo position. In addition to driving servos, ServoBlaster can be configured to generate pulse widths between 0 and 100% of the cycle time, making it suitable for controlling the brightness of up to 21 LEDs, for example. The driver creates a device file, /dev/servoblaster, in to which you can send commands. The command format is either <servo-number>=<servo-position> or P<header>-<pin>=<servo-position> For the first format <servo-number> is the sevo number, which by default is a number between 0 and 7, inclusive. For the second format <header> is either '1' or '5', depending on which header your servo is connected to, and <pin> is the pin number on that header you have connected it to. By default <servo-position> is the pulse width you want in units of 10us, although that can be changed via command line arguments, and can also be specified in units of microseconds, or as a percentage of the maximum allowed pulse width. So, if you want to set servo 3 to a pulse width of 1.2ms you could do this at the shell prompt: echo 3=120 > /dev/servoblaster 120 is in units of 10us by default, so that is 1200us, or 1.2ms. Alternatively, using the second command format, if you had a servo connected to P1 pin 12 you could set that to a width of 1.2ms as follows: echo P1-12=120 > /dev/servoblaster As an alternative to setting absolute servo positions, you can specify adjustments relative to the current position via a '+' or '-' prefix. For example, the following would increase the servo pulse width by 10 step units: echo P1-12=+10 > /dev/servoblaster Beware that within the driver the servo position and any width adjustments are rounded down to the nearest step-size, which is 10us by default. So, if you continually issued "0=+1us" commands, the servo would never move, and "0=-19us" is treated the same as "0=-10us". In addition, relative adjustments are silently truncated to keep the servos within the allowed min/max range. If you set a servo width to 0 it turns off the servo output, without changing the current servo position. The code defaults to driving 8 servos, the control signals of which should be connected to P1 header pins as follows: Servo number GPIO number Pin in P1 header 0 4 P1-7 1 17 P1-11 2 18 P1-12 3 21/27 P1-13 4 22 P1-15 5 23 P1-16 6 24 P1-18 7 25 P1-22 P1-13 is connected to either GPIO-21 or GPIO-27, depending on board revision. Command line options described later in this document allow you to configure which header pins are used for servo control, and what servo numbers they map to. Note that servoblaster must know at startup which header pins it is control of. If you want to use a header pin that is not in the default list above, you must use the appropriate command line option to inform servoblaster. The driver also creates a /dev/servoblaster-cfg file, which describes which pins servoblaster is currently configured to use. When the driver is first loaded the GPIO pins are configure to be outputs, and their pulse widths are set to 0. This is so that servos don't jump to some arbitrary position when you load the driver. Once you know where you want your servos positioned, write a value to /dev/servoblaster to enable the respective output. When the driver is unloaded it attempts to shut down the outputs cleanly and revert the GPIO pins to their original configuration, rather than cutting some pulse short and causing a servo position to jump. The driver takes note of how many servos you have configured and distributes the start time for the servo pulses evenly across the cycle time. This way the driver aims to ensure that only one servo pulse will be active at a time, which should help minimise total drive current needed. In the following description it refers to using the PWM peripheral. For the user space implementation it can instead use the PCM peripheral, see below for details. Using PCM is typically a better option, as the 3.5mm jack also uses the PWM peripheral, so ServoBlaster can interfere with sound output. The driver works by setting up a linked list of DMA control blocks with the last one linked back to the first, so once initialised the DMA controller cycles round continuously and the driver does not need to get involved except when a pulse width needs to be changed. For a given period there are two DMA control blocks; the first transfers a single word to the GPIO 'clear output' register, while the second transfers some number of words to the PWM FIFO to generate the required pulse width time. In addition, interspersed with these control blocks is one for each configured servo which is used to set an output. While the driver does use the PWM peripheral, it only uses it to pace the DMA transfers, so as to generate accurate delays. I used Panalyzer running on one Pi to monitor the servo outputs from a second Pi. The pulse widths and frequencies seem very stable, even under heavy SD card use. This is expected, because the pulse generation is effectively handled in hardware and not influenced by interrupt latency or scheduling effects. The DMA channel used depends on the SOC type. If the Pi has a BCM2711 (Pi4B only, at present) then it uses channel 7; otherwise it uses channel 14. It uses PWM channel 1. It makes no attempt to protect against other code using those peripherals. It sets the relevant GPIO pins to be outputs when the driver is loaded, so please ensure that you are not driving those pins externally. I would of course recommend some buffering between the GPIO outputs and the servo controls, to protect the Pi. That said, I'm living dangerously and doing without :-) If you just want to experiment with a small servo you can probably take the 5 volts for it from the header pins on the Pi, but I find that doing anything non-trivial with four servos connected pulls the 5 volts down far enough to crash the Pi. The user space daemon --------------------- To use this daemon grab the servod.c source and Makefile and: $ make servod $ ./servod --help Usage: ./servod <options> Options: --pcm tells servod to use PCM rather than PWM hardware to implement delays --idle-timeout=Nms tells servod to stop sending servo pulses for a given output N milliseconds after the last update --cycle-time=Nus Control pulse cycle time in microseconds, default 20000us --step-size=Nus Pulse width increment step size in microseconds, default 10us --min={N|Nus|N%} specifies the minimum allowed pulse width, default 50 steps or 500us --max={N|Nus|N%} specifies the maximum allowed pulse width, default 250 steps or 2500us --invert Inverts outputs --dma-chan=N tells servod which dma channel to use, default 14 --p1pins=<list> tells servod which pins on the P1 header to use --p5pins=<list> tells servod which pins on the P5 header to use where <list> defaults to "7,11,12,13,15,16,18,22" for p1pins and "" for p5pins. p5pins is only valid on rev 2 boards. min and max values can be specified in units of steps, in microseconds, or as a percentage of the cycle time. So, for example, if cycle time is 20000us and step size is 10us then the following are equivalent: --min=50 --min=500us --min=2.5% For the default configuration, example commands to set the first servo to the mid position would be any of: echo 0=150 > /dev/servoblaster # Specify as a number of steps echo 0=50% > /dev/servoblaster # Specify as a percentage echo 0=1500us > /dev/servoblaster # Specify as microseconds echo P1-7=150 > /dev/servoblaster # Using P1 pin number rather echo P1-7=50% > /dev/servoblaster # ... than servo number echo P1-7=1500us > /dev/servoblaster Servo adjustments may also be specified relative to the current position by adding a '+' or '-' prefix to the width as follows: echo 0=+10 > /dev/servoblaster echo 0=-20 > /dev/servoblaster $ sudo ./servod Board revision: 1 Using hardware: PWM Using DMA channel: 14 Idle timeout: Disabled Number of servos: 8 Servo cycle time: 20000us Pulse increment step size: 10us Minimum width value: 50 (500us) Maximum width value: 250 (2500us) Output levels: Normal Using P1 pins: 7,11,12,13,15,16,18,22 Servo mapping: 0 on P1-7 GPIO-4 1 on P1-11 GPIO-17 2 on P1-12 GPIO-18 3 on P1-13 GPIO-21 4 on P1-15 GPIO-22 5 on P1-16 GPIO-23 6 on P1-18 GPIO-24 7 on P1-22 GPIO-25 $ The prompt will return immediately, and servod is left running in the background. You can check it is running via the "ps ax" command. If you want to stop servod, the easiest way is to run: $ sudo killall servod Note that use of PWM will interfere with 3.5mm jack audio output. Instead of using the PWM hardware, you can use the PCM hardware, which is less likely to cause a conflict. Please be aware that the PCM mode is very lightly tested at present. Some people have requested that a servo output turns off automatically if no new pulse width has been requested recently, and I've had two reports of servos overheating when driven for long periods of time. To support this request, servod implements an idle timeout which can be specified at module load time. The value is specified in milliseconds. Typical small servos take a few 100 milliseconds to rotate from one extreme to the other, so for small values of idle-timeout you might find the control pulse is turned off before your servo has reached the required position. idle-timeout defaults to 0, which disables the feature. By default ServoBlaster attempts to protect your servos by setting minimum and maximum values on the pulse width of 50 and 250 (500us and 2.5ms). If you want to generate pulse widths up to 100% for some other purpose, you need to specify the minimum and maximum values you want (probably as 0 and 2000, for 0ms and 20ms). If you are connecting some external drive circuitry you may want active low rather than active high outputs. In that case you can specify an option to invert the outputs. If you want finer control over your servos, you can change the step increment size from 10us to some value as low as 2us, via --step-size. Note that if you do change the step size then any min, max, or servo pulse widths you specified in terms of step size units will also have to change. You'll probably find it less confusing to switch to using microsecond values everywhere. If you are driving LEDs, for example, you may want a shorter cycle time, as flickering may be visible with a 20ms (50Hz) cycle. You can change the cycle time via the --cycle-time option. The final options relate to which header pins you want to use to drive your servos. On a Rev 1 board you can use up to 17 pins on the P1 header, and on a Rev 2 board there are an additional 4 pins available on the P5 header. The default option is the equivalent of specifying --p1pins=7,11,12,13,15,16,18,22 As another example, if for some reason you want only two servos but you want them to be referenced as servos 4 and 5 (perhaps you have existing software that uses those servo numbers), you can use '0' as a placeholder for unused servo IDs, as follows: $ sudo ./servod --p1pins=0,0,0,0,15,16 ... Using P1 pins: 0,0,0,0,15,16 Servo mapping: 4 on P1-15 GPIO-22 5 on P1-16 GPIO-23 If you wanted to refer to the servos by their P1 header pins you could do something like this, which gives you servos 7, 11, 12 and 15 on P1 header pins 7, 11, 12 and 15: --p1pins=0,0,0,0,0,0,0,7,0,0,0,11,12,0,0,15 If you specify both --p1pins and --p5pins, then the order in which you specify them is relevant because servo numbers are allocated in the order the parameters are specified on the command line. For the full set of P1 and P5 header pins you can use, please refer to the GPIO section of the following web page: http://elinux.org/Rpi_Low-level_peripherals It should also be clear from the servod.c source which header pins ServoBlaster will allow you to select. Clearly if you tell ServoBlaster to use pins that are normally used for some other purpose, then that other functionality will not be available while servod is running. If you want servod to start automatically when the system boots, then you can install it along with a startup script as follows: $ sudo make install You may wish to edit /etc/init.d/servoblaster to change the parameters that are specified in that script (e.g. the idle-timeout, which is set to 2 seconds in the shipped version of that script). Related projects: Ville has written a simple Qt wrapper for servoblaster, which you can find here: https://github.com/vranki/kittinger/blob/master/servocontrol.cpp & .h Todd wrote a nice script to provide a simple user interface to control your servos, see his sbcontrol.sh script here: http://www.raspberrypi.org/phpBB3/viewtopic.php?f=37&t=15011&start=25#p187675 Richard Hirst <[email protected]> December 2013
About
Richard G. Hirst's 'servoblaster' userspace PVM Servo controller for Raspberry Pi
Topics
Resources
License
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published
Languages
- C 89.4%
- Shell 9.1%
- Makefile 1.5%