|
What you need is the National Semiconductor servo chip,
NE544, or something like that. I'll look it up, and post it here if you
refresh this topic on Monday to remind me.
The circuit you want, which is simply an on/off, with feedback being, i
presume, some sort of limit switch detector, can be readily accomplished
with a few components, and probably not more than $10.00. If I have an
application sheet for the chip in question, I'll make a copy and see if
I can get it too you.
It's been done before.
Randy
|
| Just got this from the UUCP net.
Bye --+--
Kay R. Fisher |
---------------O---------------
################################################################################
From: [email protected]
Subject: Servo Tutorial (Long--Good)
Date: 20 Jan 90 01:48:54 GMT
Tutorial on servos:
Once, a year ago or so, I posted an explanation on the innards of
an R/C servo. It would seem that it might time to post it again. I
did not save a copy of the last posting, so, will do so from the
beginning again. Some history is included.
I give first an operational discription, followed by a crude logic
drawing/schematic.
The stages within the servo are:
1--A pulse generator.
2--A pulse comparator.
3--A pulse stretching circuit.
4--Output drivers to the motor, in both minus and plus polarity.
5--A feedback or "look ahead" circuit.
6--Motor/geartrain/pot, most servos contain four gears, not counting
the pinion on the motor. The pot is connected to the servo output
shaft. If the output turns 90 degrees, then the pot wiper moves
90 degrees.
In operation, the incoming pulse from the receiver or pulse source
of your choice triggers the pulse generator internal to the servo.
The time constant of the internal pulse is controlled by the pot
in the servo. In the past some manufacturers used negative pulse
servos, but at this time, I believe most all use positive pulse
types.
In the beginning, servos were built up with discrete parts, until
Phil Kraft paid Texas Instruments something in the order of $250,000
to lay up the first IC servo chip. After that, many others followed.
The first servos were called 4 wire servos, improved design brought
about the bridge configuration in the output stages and they became
3 wire servos. Loss of a single cell in the battery pack did not cause
slow run in one direction and normal speed the other direction.
Also, in the early days, Heathkit marketed a servo with a variable
capacitor as the time adjustment in the servo. This removed the dirty
pot blues, but limited the possible reduction in physical size that
could be achieved. It fell by the wayside as a result. I have always
felt that a magnet and magnetic field sensing logic might be the way
to go. Hall effect transistors might be employed.
The incoming pulse is a positive pulse, the one generated within
the servo is a negative going pulse. The two pulses are fed to
a summing circuit (comparator). If no pulse exists and voltage is
applied to the servo, the summing output is 1/2 of applied voltage.
Assuming applied supply to be 5 volts, the output at the summing
circuit is 2.5 volts.
When a pulse arrives at the input of the servo, which is positive,
it triggers the negative pulse in the servo. With both pulses
present at the summing circuit, the output is still at 2.5 volts.
When the shorter of the two pulses goes away, the output of the
summing circuit changes to the polarity of the remaining pulse.
If the input pulse were shorter, the output of the summing circuit
would be 0 volts as the input pulse went away, until the negative
pulse timed out, whereby it would return to 2.5 volts.
If the input pulse were longer, the output of the summing circuit
would be 5 volts after the internal negative pulse timed out, until
the input pulse went away.
The output stages are a totem pole arrangement, where input to them
at the 1/2 voltage level turns nothing on and the motor does not
run. The summing circuit drives the output stages, so when it is
at the 1/2 voltage, there is no drive. When either at the 0 or 5 volt
level, the summing circuit will activate a pair of drivers. There
are four drivers, arranged in pairs, such that either lead of the
motor can be pulled to plus or minus. The arrangement is such that
one of each gets turned on. If one motor lead is pulled to 0, the
opposite will be at 5 volts.
The servo actually runs on an error in timing between the two pulse
lengths, when they equal, no activity takes place. Modern servos can
detect an error as small as 4 or 5 micro seconds, which is the dead
band time of the servo.
The pulse stretching is usually a capacitor connected to the summing
curcuit output, such that it will remain at the other than 2.5 volt
level for a little time, after both pulses are gone (provides drive
to the motor between pulses). The drive voltage from the summing/stretch
circuit rises quickly as the error occurs, but decays at an exponential
rate. If the servo is updated at a low rate, the torque at the servo
output is less than if it is updated more often. This is because the
drive voltage does not get the chance to decay to a low point on the
exponential curve before it gets updated again.
The feedback or look ahead circuit is a lead tied to one of the
two motor terminals, which has a resistor going to the summing circuit
output. This is usually done on the PC board, so you will not see an
additional third motor lead. The purpose of the feedback is to shift
the output of the summing circuit by a few millivolts, in one direction
or the other, depending on which way the motor is running. This causes
the motor to shut down slightly before the two pulses are actually
equal, and allows for the coasting of the motor as it comes to a stop.
If the motor shuts off at the correct time, it will not coast PAST the
desired point, generating another error in the wrong direction causing
the servo to then back up. In practice, the servo would "dither", or
chatter.
As a note of knowledge, the "Autopilot" developed by Mr. Hill amplified
the difference in electrostatic voltages, wingtip to wingtip, and the
output of the amp was fed to the summing circuit, such that it could
cause the servo to move based on wing tip elevation. This fact indicates
that if one is clever, the servo CAN be moved via a DC voltage input at
that point. For those interested in this, but not fimiliar, the AMA
magazine, plus others carried articles on this some years back, and
were entitled "An Electrostatic Autopilot", by Mr. Meynard Hill.
Logic/schematic:
---
| |
Input pulse _____| |______
>-------------------------------------
| \
| Summing resistors------> /
| \____________
| / | |
| ------ \ | |
| |Pulse | / | ___
-------->| |---------- | --- <--Pulse stretch
+5V | Gen | ---- ---- | |
| ------ | | | |
/ ^ | | | Gnd
\ | --- |
/<------ |
Position pot->\ ^ |
tied to the / | |
gear train. \ -Feedback goes |
| here in some |
Gnd designs. |
|
------------------------------------------------
| | |
| \ |
| 5V /<--Feedback 5V |
| | \ | |
| | / | |
| ------- | ------- |
| | | | | | |
| | Pos | | | Pos | |
| ->| | | | |<- |
| | | Amp | | | Amp | | |
| | | | | | | | |
| | ------- | ------- ------- | |
| | | | | | | | |
->| |-------->| Motor |<---------| |<-
| | | | ^ | |
| ------- ------- | ------- |
| | | In early servos- | | |
| | Neg | this motor pin | Neg | |
|->| | went to battery | |<-|
| Amp | center tap and | Amp |
| | the totem pole | |
------- to the right did -------
| not exist. |
| |
| |
Gnd Gnd
In the above amplifiers, they are multiple stage. Some designs
have all transistors contained within the servo IC, others have
the preamp in the IC and the drivers are external to the IC.
The outputs going to the motor above COULD in fact drive external
darlingtons to run a really big motor/gear train and the position
pot could be external to the above logic, gear reduced such that
several turns of the output shaft would be possible. A tiller
servo on a 12' R/C sail boat might be an example.
Just dream on...........
Al Irwin
Univ of Illinois
Dept of Comp Sci
[email protected]
|