Encoder-less Encoder!

Roman Black - April 2001

How to get speed feedback from a small DC motor.

This shows how to get speed feedback suitable for input
to a logic input (ie, PIC input pin) from a small DC motor
with no optical encoder needed, and no mods to the motor.
It can be used with no mechanical mods to the device.
It gives X pulses per motor revolution, X depends on
the motor type.


-------------
Requirements:
-------------

A small DC motor (brush/commutator type)
1x small NPN transistor
2x 22k ohm resistor
1x 5k ohm trimpot
some small ceramic caps
1x IN4148 small signal diode 
(see notes for parts details)
(multimeter and CRO handy for setup)


--------
Details:
--------

I have used this circuit to get speed information and
even count revolutions from a number of small DC motors.
It is not perfect, and won't give perfect position
feedback like a proper optical encoder disk, but it
is great for measuring motor speed with motors that
DON'T have an encoder disk. It's cheap, easy to do,
and works best on small cheap style DC motors like the
ones found in toys or small hand drills etc. These are
also the hardest motors to fit optical encoders to.


-------
Theory:
-------

The motor must be a brush/commutator type. Small motors
usually have two brushes and 3 or 5 commutator segments.
There are 6 or 10 places per rotation where a brush hits
the "join" between commutator segments and a small electrical
spike is generated.

This circuit uses a small capacitor to couple that spike
to a sense transistor and with every spike the transistor
turns on for a short time. The output from the transistor
can be coupled directly into a PIC input pin or other logic
type input (into any IC).

The spikes are short and fast, they contain very little
energy but they do couple well through a small capacitor,
even if the two grounds are not connected. The spikes are
NOT evenly spaced around the revolution of the motor, but
the number of spikes per revolution is constant and can be
used to give pretty good motor speed sensing if a little
software averaging is used.

This is a good way to speed-control a tiny DC motor,
like in a small robot or those little high speed hand-drills
(supertools?) that people use to drill PCBs. You don't need
to modify any of the mechanicals in the device, which
is great.


(Circuit diagram)



-------------
How it works!
-------------

As the motor turns the brushes pass over the joins between
the commutator segments. When a brush hits a join this is
like a little switch contact opening or closing, and as
the motor coils have inductance this switch effect makes
little back-emfs or "spikes". These spikes are short and
have very little energy. Generally a few microseconds.

Fast little spikes like this are about the only thing
that passes through a very small capacitor, like 330pF.
Slower waves and DC won't pass through the very small cap. The
spikes will even pass if the grounds of the two systems are
not connected. It works slightly better if they are connected.

The trimpot is for BIASING the transistor. It varies the
constant voltage at the base of the transistor between 0v
and about 1v. The transistor turns ON at about 0.7v.

The trimpot can be turned through 3 basic stages:
1. Transistor is always OFF.
2. Transistor is not-quite on, and sensitive to spikes.
2. Transistor is always ON.

Setup is quite simple, you turn the trimpot right down
(transistor OFF, output to PIC = +5v)
then run the motor and slowly turn it up until you get
good reliable triggering. With each spike, the transistor
will briefly turn on and there will be a short 0v pulse
to the PIC. Ten or twenty microseconds length is typical.


-------------
Important Notes!!
-------------

D1.
If the motor is small you don't need the diode. If the
motor is large, or a small high-current motor like
in dustbuster vacuum cleaner's etc, use the diode.
I would guess for under 2amp motor running you don't need
the diode.

R1.
The resistor R1 is not needed with most circuits.
Especially with DC motors in tools where the power leads
are long. If the 12v battery (etc), is close to the motor
you may need a small resistor here, try 0.5 ohm for
a small motor or 0.1 ohm for a larger motor. This makes
the spike more likely to go to the encoder circuit than
back to the battery.

C1.
The values of the two capacitors, C1 and C2 are fairly
critical. I can't give exact values as it all depends
on the type of motor you are using. A value of 100pF to
470pF is a good start for C1. If C1 is too small you will
miss some pulses, and they will be short and weak. If it
is too large you may get extra pulses.

C2.
You may not need C2 at all, but C2 will help reduce
the short fast spike to a "hump" that is lower and longer.
Hence it can increase the length of the pulse at the
output, which is a good thing. Try using a C2 that is
about the same value as C1, maybe going to twice as large
as C1. A CRO will help a lot when tuning these.

Motor.
This circuit works best where there is a bit of motor
load. This makes the motor draw more current and the
tiny spikes at the commutator are larger. You will find
that the circuit gives the most reliable output pulses
when the motor is under some load. Generally I have found
this circuit works well with almost every motor, even with
some unloaded motors like spinning a small mirror or paper
disk, etc. Do NOT put a capacitor directly between the
motor terminals. If RF filtering is needed use a capacitor
across the motor power, leaving R1 in series with the motor.

Adjustment.
The easiest way is to set the trimpot so the transistor
is fully off, (output=5v) and run the motor at it's
minimum load. Then turn the trimpot until the triggering
and output pulses become reliable. After that, when motor
load is increased the pulses will become slightly longer
and more reliable. When it is done you should check that
disconnecting the motor still gives 5v at the output,
so the PIC only gets 5v or 0v levels at it's input pin.

Filtering.
As the encoder senses fast spikes it may pick up extra
pulses from noise in the system. If the encoder is mounted
more than a few inches from the motor you should use coax
shielded cable for the main wire to C1. Only connect the
shield at the circuit end, not the motor end. If you
get the parts values tuned right for your motor this
circuit will perform quite well, and you will not get
many extra spikes.

Software.
A simple software averaging is enough, like measuring how
many pulses are made in 0.1 seconds, etc. With most motor
speeds this number will be in the hundreds and the rare
"suspect" extra pulse will not cause any great effect.
You should be able to control motor speed quite well and
even measure the number of revolutions over a given time
period.

:o)
-Roman

Questions: