Electronic Building Blocks
By Julian Edgar
Quick and easy construction
Great results on a low budget
Introduction to linear actuators – Part 2
Controlling a linear actuator with positional accuracy using the programmable
Pololu Jrk 21v3 module.
I
n Part 1, we looked at how linear actuators can
be controlled by simple switches and relays. This approach is
fine if you want the actuator to be fully open or fully closed, or
you are controlling a progressive opening by physically watching
the movement of the actuator (eg, shutting a gate). However, if
you want more accurate control (for example, always opening the
actuator by exactly 100mm) then a controller that uses position
feedback is needed. So, this month, we turn to a fully programmable
control module that provides excellent positional accuracy, even
with varying voltage feeds and loads. In addition, the module
can also control maximum speed and acceleration – and more.
Note that to make use of the feedback function, you will need
to select a linear actuator that uses an internal potentiometer
to provide position information. These are less common
than actuators without feedback. That said, companies like
AliExpress do offer them from about £30.
Fig.1. The Pololu Jrk 21v3 is a tiny board – just 34mm square.
However, it can run brushed DC motors drawing up to a peak of
5A and can be controlled through multiple interfaces, including an
analogue input, serial, PC USB link or model RC pulse train. Once
programmed, it can be disconnected from the PC.
64
The controller
The controller we will use is the Pololu Jrk 21v3 USB Motor
Controller with Feedback. It is widely available for about £40.
Now, before you blanch at that price, let’s take a step back. Why,
you may be thinking, don’t you just use a cheap microcontroller
like an Uno with a H-bridge motor drive shield? After all, you’re
only controlling a DC motor! Well, the key difference is the
software. Unless you want to spend a lot of time writing code
– and I mean a lot – then the off-the-shelf software that is freely
available for the Jrk 21V3 controller provides a huge advantage
over a home-grown software solution.
But before we look at the software, what does the module
comprise? Measuring just 34mm square, the Jrk 21V3 is designed
specifically for controlling small, brushed DC motors. Part of
a family of motor controllers, the Jrk 21V3 has a maximum
supply voltage of 24V and a maximum continuous current of
2.4A, with a 5A peak rating. (Higher currents are available in
other controllers in the family.)
The Jrk 21V3 has four interfaces:
n USB for control via a PC
n TTL-level serial interface for microcontroller control
n Radio control (RC) PWM control for direct connection to an
RC receiver or servo controller
Fig.2. The connections for the Jrk 21v3 – a regulated 5V supply is
available from multiple pins to power control and feedback pots.
(Courtesy Pololu)
Practical Electronics | July | 2022
Fig.3. The Input tab. Note here that the Analog voltage input has been selected and that
the Target has been reduced from a full range of 0-4095 to 1400-2500, so limiting the
stroke of the actuator to suit the application. Visible near the top are the Target and Scaled
Feedback values – here they are identical, indicating that the actuator has stopped at
exactly the requested extension.
n 0-5V analogue control; eg, via a manual
potentiometer.
Here, we will be using the 0-5V analogue
control. The all-important feedback can
be via the following two means:
n A 0-5V analogue signal
n Tachometer (speed) up to 2MHz.
Alternatively, the system can be set to
have no feedback – that is, use openloop control. We will be using the 0-5V
feedback signal derived from the pot
inside the linear actuator.
Above, I mentioned software. Using
the USB connection to a PC (connection
is via a USB Mini B connection – no
cable provided), the following motor
parameters can be altered:
n PID tuning
n Max current
n Max duty cycle
n Max acceleration
n Error response
n Input learning for analogue and RC
control.
These may all mean little at this stage, so
let’s paint a word picture of what these
settings can achieve. Say you were pretty
excited by the picture in last month’s
issue, showing the TV that can rise out
of a cabinet at your command. Using
just a toggle switch to operate the linear
actuator and its internal limit switches
to turn it off at each end of its travel, you
will have the following outcome. When
the switch is operated, the TV will start
rising with a jerk as power is applied
to the motor. It will then rise at a speed
determined by the supply voltage and
actuator speed specification, before it
reaches full height and trips the internal
switch, stopping abruptly and probably
shaking the whole assembly. On the way
down it will probably travel a bit faster,
making an even bigger jerk as it stops at
the limit of its travel. OK, it does the job
but hardly an elegant solution.
Now let’s add the Jrk 21V3 controller.
This time, the actuator speed increases
p r o g r e s s i v e l y, s t a r t i n g a l m o s t
imperceptibly and then gradually rising
in speed. The maximum speed of the
actuator can be pre-set – for example,
to minimise noise, you can run it at a
maximum speed of only 30% full speed.
The point at which the actuator stops
rising can also be set (that is, the actuator
doesn’t need to hit the limit switch) and as
the actuator reaches that pre-set point, it
can be programmed to progressively slow
until it slides to a gentle halt. And that
stopping point can be set and achieved
with millimetre accuracy.
On the way back down, all these
parameters can again be set, but this time
taking into account the weight of the
TV that will be hastening the retraction
process. (In other words, you can have
different settings for ‘up’ and ‘down’.)
Now you can start to get a feel for
what this controller can do with a linear
actuator. And start to see the non-trivial
nature of the programming involved, but
which is built into the Pololu software
– in particular, building your own,
unconditionally stable PID controller is
not straight forward.
Connections
Fig.4. The Feedback tab – Analog voltage has been selected as the feedback type, and
because the feedback signal was working in the opposite direction to the Input direction,
Invert feedback direction has been ticked.
Practical Electronics | July | 2022
As bought, the board will probably come
with two loose 2-pin terminal blocks and
a male header strip. These can be easily
soldered to the board (this is not done at
the factory in case you want to fit angled
headers or different terminal blocks).
The terminal blocks are for power
supply and the two motor connections.
In addition, we are going to require
connections for the feedback signal from
the actuator, and the analogue control
signal to set the position we want the
linear actuator to achieve. (In our case,
we’ll use a simple pot to provide the
control signal.)
Therefore, we will need regulated 5V
outputs and grounds for the extremities
of the two pot connections – these are
available on the board and are well
65
Fig.5. The PID tab – the P, I and D parameters can all be altered. Note the unusual way in
which the data is input – either as a fraction with the denominator multiplied by the adjacent
numeral (unless you put in zero, whereupon it defaults to 1), or direct-entered as the coefficient.
Playing with these settings gives enormous control over the behaviour of the system.
labelled. (Incidentally, a very good 55page manual is available for the Jrk 21V3
– it’s worth looking at before buying, as
well as when building your board into
your application.) The feedback signal
from the actuator connects to the FB pin,
while – not so obviously – the control
input connects to the RX pin.
The board is protected against reverse
polarity, over-load and over-temperature
– so it should be difficult to kill it when
making these connections. However,
always turn off power first and check the
wiring before switching power back on.
Software
The software is available from Pololu
and is best accessed by clicking on the
link in the pdf manual. There is a large
section in the manual on installing the
software, but with Windows 10 Pro I just
followed the on-screen prompts and had
no difficulties.
Testing
Do your initial testing with the
actuator sitting on the bench, not
connected to the device that you wish
to mechanically move.
With the Jrk 21V3 plugged into the
PC, click on the Connected to dropdown box and you should be able to
select your device. Select the Input tab
and then choose Analog voltage as the
input mode. Ensure that as you turn the
input pot, the Target number alters. You
should also see a number associated with
Scaled Feedback – but at this stage this
won’t change since the actuator isn’t yet
moving. So, if everything is connected,
why isn’t the actuator moving when
the input pot is turned? It’s because the
PID settings are all still at zero. (PID –
proportional, integral, derivative – is
a control loop system that allows very
effective control.)
Move to the PID tab and start by setting
P to 20 – leave the other figures at zero at
this stage. (After any calibration change,
press Apply settings to this device to
upload to the module.) Now with P at 20,
when you turn the input pot, the actuator
should move, slowing abruptly when it
gets to the position you’ve selected. Now
Fig.6. The Motor tab – here maximum speed, maximum acceleration and brake durations
can all be set. These allow the system to be set up so that the actuator smoothly gathers
speed and operates at a maximum speed that suits the requirements of the mechanical
system. Current can also be limited – the system then specifies a motor duty cycle that
causes this current value to not be exceeded.
66
alter P to 5, and you should see that the
actuator more progressively comes to
a stop at the setpoint. To get the ‘last
little bit’ of movement accurately to the
setpoint, increase I (say to 0.125). You can
now check how accurately the actuator
is positioning itself at the setpoint –
compare the difference between the
Target and Scaled Feedback numbers.
It’s not hard to program the PID settings
to achieve identical numbers – that is,
the actuator has moved to precisely the
correct point.
If the system seems to work nothing like
the description in the above paragraph,
you may have the Input and Feedback
pots working in opposite directions. This
is easily fixed by ticking the Invert box in
either the Input or Feedback tabs (don’t
invert both or you’ll be back to where you
were before.)
If the actuator is being used across its
full stroke, the inbuilt limit switches will
come into play. That is, rather than the
actuator slowing progressively, it will hit
the limit switch and so stop abruptly. To
prevent this occurring, you can reduce
the target range of the Input pot. For
example, rather than the range being 0
to 4095, change the Target to a maximum
of 2500 and the minimum to 1400. This
will then cause the actuator to work over
a smaller stroke range.
So, now we have the actuator slowing
progressively as it reaches the setpoint,
but it still starts its movement with a jerk.
Go to the Motor tab and place a small
number (eg, 2) in the Max. acceleration
box. Now when you turn the knob,
the actuator will progressively gather
speed. In the same tab we can change
the maximum speed that the actuator
reaches – this is specified by Max. duty
cycle, where a count of 600 means 100%.
One thing that is not obvious in this tab
is the meaning of ‘Brake duration’. In this
application, it refers to the delay time
if you turn the pot from one direction
immediately to the other.
When setting up the PID parameters,
it is worth opening the Plots tab. This
window can then be dragged to one side
and enlarged. The Plots tab allows you to
graph a range of variables including error,
motor duty cycle, target and feedback.
The graph will quickly show if you have
control overshoots or undershoots. It’s
also fascinating to watch as the system
progressively reduces the error, the
motor duty cycle smoothly reducing as
the actuator draws closer to the correct
stopping point.
Once the actuator is working correctly
on the bench, install the actuator in the
mechanical system it is going to control.
Then, as the actuator is working under real
world loads, some settings will probably
need to be tweaked in the Jrk 21v3.
Practical Electronics | July | 2022
Multi-switch position input
Fig.7. The Error tab – a wide range of fault conditions can be monitored. They can each be
Enabled or Enabled and latched, the latter stops the motor until a new command is received.
Rather than using a pot, it’s easy to
organise a rotary switch to give a variety
of pre-set actuator positions. For example,
in a racing car where an aerodynamic rear
wing may need to be driver-adjusted in
angle to take into account different track
or fuel load conditions, a pot would be
too fiddly to use. Instead, a 4-position
rotary switch can be fitted – see Fig.10.
In this approach, each pole of the switch
is connected to the wiper of a 10kΩ, 10turn, trim-pot wired across regulated
5V and ground. The voltage available
on the wiper of each trim-pot can be set
by turning the trim-pot – and thus the
position of the actuator at each switch
position can be adjusted. With a linear
actuator controlling wing angle, four wing
angles can then easily be selected under
racing conditions. A similar ‘pre-sets’
approach could be taken if controlling
house curtain openings, how far down a
blind is to extend – and so on.
Conclusion
In the two parts in this series we’ve
covered operating a linear actuator with
the simplest and cheapest of approaches
– just a toggle switch – right through
to using a sophisticated programmable
controller. (How sophisticated? In fact,
in this article we’ve not covered every
available feature of the Jrk 21V3 – there’s
simply not space to do so.) With the
price of linear actuators having fallen
so dramatically over the last few years,
there’s no excuse not to automate a
whole variety of systems that require
movement. Have fun doing so – and
we’d love to see some of your projects
using these actuators.
Fig.8. The Plots tab – this tab graphs in real time the selected parameters. Here Target,
Scaled feedback and motor Duty cycle have been graphed. Note that what is shown
above is not a good representation of what you normally see – here the input pot has been
rapidly turned back and forth so that major changes could be seen over the maximum
five-second window. Normally a step change in the input is followed by the feedback and
target lines gradually coming together.
Fig.9. The Jrk 21v3
connected to a pot for
the input (foreground).
The green, yellow and
white connections are
to the feedback pot on
the actuator. The red
and black wires are
for power and ground
and the actuator
motor connections.
Here the Jrk 21v3 is
connected to a PC
via the USB cable for
programming. As can
be seen, to operate a
linear actuator requires
minimal connections.
Practical Electronics | July | 2022
JrK 21v3
input
10kΩ, 10-turn
potentiometers
4-way user
selection
switch
R egulated 5 V
0V (Gnd)
Fig.10. Here a multi-position switch has
been used with four trim-pots to give four
pre-set positions for the linear actuator. The
position of each pre-set can be adjusted by
rotating the relevant trim-pot.
67
