This is only a preview of the May 2000 issue of Silicon Chip. You can view 32 of the 96 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Items relevant to "Building The Ultra-LD 100W Stereo Amplifier; Pt.2":
Items relevant to "Build A LED Dice":
Items relevant to "Low-Cost AT Keyboard Translator":
Purchase a printed copy of this issue for $10.00. |
50 Amp Mode
Controller – W
Are you into large or powerful radio-controlled electricpowered models? The ones where battery life is measured
in minutes, not hours? Here’s a controller which can handle
motor currents of up to 50A and is compatible with your
existing radio control equipment.
Article by ROSS TESTER – Design by BRANCO JUSTIC*
T
he obvious question is who
could possibly want a control
ler capable of such huge currents? After all, your typical radio
controlled model, say a race car or
buggy, has only a 7.2V battery and a
75W motor – ergo, 10A.
For this project, we’re not talking
your typical off-the-shelf radio controlled car or buggy. We’re talking
industrial-strength models powered
by, say, 12V motorcycle or even car
batteries. Large boats, electric-pow-
ered planes and big cars and trucks,
for example.
At the opposite end of the scale are
competition boats, cars and planes
which may not be very big but have
very powerful motors demanding a
lot of electrical power.
They might draw 10 or 20A or more
on load and therefore need significantly more in the controller department.
But 10 or 20A is a far cry short of 50A.
Why the brute strength? Couldn’t
we make it a bit simpler and save a
few bob? Yes . . . and no!
The problem lies not so much in the
typical load current of the motor, nor
even the start-up current (which can
be high). It lies in the stall current. A
motor loafing along at 10A might draw
ten times as much if locked up – for
example, when the car it is pushing
hits an obstacle and before the wheels
start slipping.
Another scenario is when a boat
runs through some underwater
greenery and gets its prop snagged.
Housed in a tiny plastic case the motor speed controller is small enough to fit into the vast majority of models. The 3-wire
rainbow cable on the left connects to the radio control receiver servo output while the wires on the right connect to a battery and the motor. You will probably need thicker cables. Note the six MOSFET tabs emerging through the case lid.
78 Silicon Chip
el Motor Speed
With Brake
Sometimes it will cut or power its
way through; other times it will be
locked up.
(We make no comment about what
happens when an electric plane’s prop
is locked up...)
That’s when you need a controller
capable of significantly larger peak
currents than you would otherwise
think were necessary. Sure, you could
take the risk and hope that you can
cut power before any damage occurs
– but that damage might occur in a
few milliseconds and for the sake of
a few extra MOSFETs valued at less
than $10.00 total, why would you?
Of course, if your application says
that the motor can never be locked up,
you can get away with fewer MOSFETs. Five MOSFETs handle 50A so
it follows we’re rating each at about
10A. But we’ll look at this in more
detail shortly.
This controller is significantly
cheaper than commercial units available and is also nice and small. Overall
the assembled PC board is only about
Three DC motors which
would be ideal for this
controller: the top one is
one we had in the “junk
box”and is rated at 12V
and draws about 5A off
load, rising quickly to 1520A under load. The two
smaller motors are both
from Oatley Electronics,
the middle one rated at
4-8V DC while the bottom one is 12V DC. These
motors sell for $8 each.
For more info visit www.
oatleyelectronics.com/
motors.html
25 x 35 x 60mm; even in its specified
case it’s only 80 x 40 x 27mm, including mounting feet. So it should fit in
the vast majority of models.
Radio control compatibility
This controller is compatible with
typical radio control equipment
which has servo outputs; ie, 99.99%
of commercial radio controls.
The radio control servo output (only
one output – there are three wires
but two of them are for power) gives
a pulse train between 1ms and 2ms
long, depending on the position of
the radio control “stick”, on a frame
Fig.1: the circuit uses a ZN409 servo driver IC not to drive a
servo but to drive MOSFETs which control the motor speed.
May 2000 79
Fig.2: the current
standard for radio controls uses
these wave-forms
to achieve forward, stop and
reverse in the
servos. The 20ms
repetition rate is
usually not at all
critical but the
pulse width is.
rate (or time between pulses) of about
20ms or so. At centre, or rest (in a
±stick), the pulses are 1.5ms long.
This pulse stream results (or should
result) in the servo adopting the centre
or zero position.
By the way, the frame rate isn’t
usually at all critical but the pulse
width is. We’ve seen frame rates of up
to 50ms and they work fine. However,
20ms is the current “standard” for
radio control systems so we’ll use that.
Push the stick in the positive direction and the pulses lengthen – up to
2ms at maximum travel which should
have the model’s servo in full forward
position.
Push the stick in the negative direction and the pulses shorten – 1ms
pulses at full travel will have the servo
in the full reverse position.
Fig.2 shows the waveforms of these
various pulse trains.
In practice, a small amount of
Fig.3: the component overlay, reproduced same size. Note the
lengths of tinned copper wire soldered to the tracks under the
MOSFETs to increase current capacity. Compare this diagram to the larger-thanlife photograph of the completed PC board below. Here you can also see the
external connection wires are soldered to the back of the board, not the front.
“trim” is usually required to achieve
the correct positions – the trim tabs
on the transmitter adjust the pulse
width slightly to make sure the servo
behaves as you intend (not as it sometimes wants to!).
Note that no provision for reverse
direction is made in this simple controller. It has only zero to maximum
(or 1.5ms to 2ms) capability. However,
moving the stick to the normal reverse
direction actuates the controller’s
braking circuit.
No radio control?
You’re one step ahead of us (or per80 Silicon Chip
haps we’re one step ahead of you!). Because the controller’s input demands
are relatively simple, a square-wave
oscillator capable of producing a pulse
between 1ms and 2ms every 20ms will
give full-range (zero to maximum)
control of the controller plus braking.
Such an oscillator is quite simple to
make with either discrete components
or, say, a couple of 555 timer ICs.
However, simple oscillators usually
drift a little with temperature so this
needs to be taken into account.
A suitable oscillator which simulates a radio control receiver output
is shown later in this article. This
oscillator can not only be used as a
“wired” controller but can also be
used to set your controller up.
The controller
We’ve already discussed the reasons for the number of MOSFETs
in the output but we haven’t yet explained what they do – or how they
get the information they need.
That information is all taken care
of by IC1, a ZN409 servo driver IC.
There are no servos in this circuit, of
course, but this IC is ideal because
it decodes the radio control receiver
“servo” pulse signal described above.
The ZN409 has its own reference
oscillator, producing 1.5ms pulses
every 20ms. The precise length of
these pulses can be varied slightly by
VR1. Incoming pulses from the receiver are fed to pin 14 and are compared
to this reference.
If the pulses are longer than the
reference the pin 9 output is taken
high and the pin 5 output is taken
low. Pulses shorter than the reference
have pin 9 low and pin 5 high. Pulses
equal to the reference have both pin
9 and pin 5 high. Remember that all
this is happening every 20ms or so.
Pin 5 is connected to three Schmitt
NAND gates wired as inverters in series, so a low on pin 5 will result in a
high on the gates of parallel connected
MOSFETs Q1-Q5 (and vice versa).
Pin 9 controls the “brake” MOSFET,
Q6, via another Schmitt NAND gate/
inverter and transistor Q7.
A low on pin 9 will result in Q7
being turned fully on, turning on Q6
which is wired directly across the motor. This effectively shorts the motor
terminals which in turn acts as a brake
on the motor armature.
If you don’t believe how effective
this is, try spinning the shaft of a
small, permanent-magnet DC motor
with your fingers, then short the terminals together and try spinning it
again. Notice the difference?
The length of time that pin 9 or 5
is held low is in direct proportion to
the difference between the incoming
(receiver) and reference pulses. A
pulse width equal to, or very close
to, the reference will result in an
extremely short “low” time on pin 5,
so the MOSFETs will effectively be
turned off.
Increasing this incoming pulse
width results in a longer and longer
“low” time until the point is reached
where at 2ms pulse width, pin 5 is
low for almost all of the 20ms cycle,
thus turning the MOSFETs fully on
for virtually all of the cycle.
Pin 9 operates in a similar manner
except that it controls the brake MOSFET. When the pulse length is between
1.0 and 1.5ms pin 9 goes low, and the
output of inverter IC2a (pin 3) goes
high. This turns on Q7 which connects
the Q6 gate to ground, turning it on.
As the pulse length approaches
1.5ms, Q6 on time becomes shorter
and shorter until at 1.5ms (centre
stick) the brake MOSFET is fully off.
MOSFET ratings
We’ve mentioned that the output
of the speed controller is handled by
five N-channel power MOSFETs, all
The completed project with the disassembled case in the background. The case “lid” is actually the larger piece – note the
cut-out in the case lid for the MOSFETs. If space is a real problem the PC board could be simply insulated in heatshrink
plastic and shoe-horned into a suitable area within the model.
May 2000 81
Parts List
1 PC board, 60 x 33mm, with
chamfered corners to fit case
Semiconductors
1 ZN409 servo driver (IC1)
1 4093 quad NAND gate (IC2)
5 IRFZ44 N-channel Power
MOSFETs (Q1-Q5)
1 MTP2955 P-channel Power
MOSFET (Q6)
1 C8050 NPN transistor (Q7)
Capacitors
2 10µF 25VW electrolytic
2 1µF 25VW electrolytic
4 0.1µF MKT polyester
1 .022µF MKT polyester
Resistors (0.25W, 1%)
1 68kΩ 1 47kΩ
1 33kΩ
1 4.7kΩ 3 1kΩ
Miscellaneous
Suitable case (if required)
Heavy duty hook-up wire (see
text)
Fuseholder and fuse to suit
Short lengths heavy tinned
copper wire
connected in parallel.
Like all semiconductors, MOSFETs
have a variety of ratings but there are
only a few which really concern us in
this application.
Of course, we must ensure that
the voltage rating is sufficient for not
only the battery voltage but also any
back-emf generated by the motor. And
this can be substantial. The IRFZ44
MOS-FETs specified have a VDS (ie,
drain-source voltage rating) of 55V.
Likewise, the current rating of the
MOSFET must be considered. In fact,
there are two ratings – a continuous
current rating (ID cont) which is 41A
and the pulsed current rating (IDM)
which is significantly higher (160A).
We are using the MOSFETS in a
pulsed mode but the limiting factor in
this speed control circuit is the heat
dissipation in the MOSFETs.
Most important of all, though, is
the MOSFET’s “on” resistance. When
turned on as hard as possible (ie, any
increase in drive to the gate results in
no further drain/source current) the
MOSFETs still offer some resistance
to current flow.
It is tiny – MOSFETs are significantly better than bipolar transistors
in this regard but even then, the
Speed controller rating. . . should it be 200A?
We have rated this speed controller at
50A and this is a continuous rating, to
suit the very high current motors used in
today’s electric flight models, as well as
those used in high performance model
cars and boats.
As noted in the text, we base this
rating on the drain-source resistance
of the specified IRFZ44 Mosfets. This
gives rise to two limitations in the speed
control circuit: voltage drop and power
dissipation.
For a 50A load, the circuit would
have a likely voltage drop of 240mV
and that means not much loss in speed
compared to running the motor directly
off the battery.
Secondly, the power dissipation for
a 50A load would be around 2.4W for
each Mosfet or a total of 12W. That is
quite a significant amount of power to be
dissipated in such a small package and
it is going need good ventilation which
is often difficult to provide inside the
fuselage or body of the model.
But if you purchased an equivalent
82 Silicon Chip
speed control from your local model
shop it would be rated at 200A or higher.
This is based on the peak current ratings
of the Mosfets.
Could a speed control such as this
withstand 200A?
The answer is yes but only for a second or two, as the likely total dissipation
of around 50W in such a small package
would not only blow the Mosfets but
would melt the solder off the back of
the PC board.
We should also note that some motors
that are likely to draw around 50A continuous could also draw as much as 200A
or more, at initial start and if the motor is
accidentally stalled. Under those conditions, a speed control like this one could
survive the very high current, provided
the overload condition did not last any
more than a second or two.
So when you see those 200A speed
controllers in model shops, remember
that, at best, it is only an instantaneous
rating. The continuous or “real” rating
is likely to be 50A or less.
small amount of resistance has to be
considered.
In fact, there are two important considerations: one is heat dissipation,
the other voltage loss.
The IRFZ44 has an on resistance
(RDS (on)) of just 0.024Ω. But as you
know, passing a current through any
resistor causes that resistor to heat
up. So it is with the “resistance” in
the MOSFET.
Our maximum current is about
10A per device, which equates to a
dissipation of some 2.4W. (P = I2 x R).
Even though well within the device
ratings that’s a significant amount of
heat for any component to get rid of
and we have five of these devices all
wired cheek-by-jowl.
The second problem any significant resistance causes is voltage loss.
Passing a current through a resistor
causes a voltage to develop across
that resistor – voltage which is then
not available to the load.
If for a moment we assumed a single MOSFET could handle the total
50A load, we would be losing almost
1.2V across it (E = I x R). That’s an
intolerable loss from a 12V supply and
will make the motor run significantly
slower.
But as you also know, when you
connect resistors in parallel the resistance drops. We’re connecting five of
these MOSFET “resistors” in parallel
so the equivalent resistance is just
.0048 ohms.
Using Ohm’s law again, 50A x .0048
is just 0.24V loss – a much better proposition. Remember that’s the worst
case; at say 20A the loss is only going
to be about 50mV.
The MTP3055 P-channel power
MOSFET used as the brake doesn’t
have to handle very high currents.
That’s fortunate, because P-channel
devices generally have a higher RDS
than N-channel devices (in this case
0.3Ω). Its 60V, 12A rating should be
more than adequate for this application.
Construction
All components are mounted on a
small PC board, nominally 60 x 33mm.
Before commencing construction,
make the usual checks for defects in
etching.
Also, if you are not building this
from the Oatley Electronics kit, you
will need to file the corners off the
board – to about 5mm in each direc-
tion – so that it will fit in the specified
case.
The Oatley kit, by the way, includes
the case, the wiring loom pictured
including fuseholder and fuse and, of
course, the PC board and components.
After checking that the board fits
the case, commence assembly with the
smallest components first. Note that
most of the resistors mount on end.
Our prototype used sockets for both
ICs but this is left up to you.
Use two of the resistor lead cut-offs
to form the two links required on the
board – both under where the MOSFETs mount.
The final components to be mounted should be the MOSFETs. Note particularly their orientation – all go the
same way but they must be the right
way around – and also the location of
Q6, the P-channel MOSFET. It mounts
closest to the BAT + and MOTOR terminals.
To keep the MOSFETs straight and
in position we lined them up with a
3.2mm drill bit through all their holes
and then soldered them in position.
Because of the significant current
drawn by the MOSFETs some short
lengths of heavy tinned copper wire
should be soldered along the appropriate PC board tracks (ie, under the
MOSFETs) to increase the current
carrying capability significantly. Just
remember that Q6 is not in parallel
with the rest of the MOSFETs!
Speaking of current capability,
the wiring used in the prototype for
battery connection was certainly not
rated at 50A! Our application called
for only a fraction of this capacity so
we used standard 10A hookup wire
and a 4A in-line fuse.
If you are powering anything larger,
not only will greater capacity cabling
be needed but you will also have to
think seriously about connections to
the PC board – soldered connections
may be inadequate. Some form of busbar may be required.
Some model shops sell silicone-coated hookup wire which is
specifically intended for high-current
applications such as this. It could be
worth a look.
Fitting to the case
The final step is to mount the complete assembly in its case. The case is
in two sections with the larger section
of the case actually the “lid”. The PC
board mounts upside-down in the
Where do you get it?
This project, including the circuit
and PC board pattern, is copyright
© 2000 to Oatley Electronics.
They can supply a complete kit
of parts, including the case, for
$35.00
They will also shortly have available a simulator (see next page)
suitable for use with this circuit.
Contact Oatley Electronics on
(02) 9584 3561, fax (02) 9584
3563, by email at sales<at>oatleyelectronics.com, via mail at PO
Box 89, Oatley NSW 2233, or via
their website www.oatleyelectronics.com
* Branco Justic is the Manager of
Oatley Electronics.
“lid” so that when you turn it over
it’s the right way up.
Double dutch? Not really, but the
photos might give a better idea of
what we’re saying. No screws are
necessary to hold the PC board in
place – it’s held captive by its leads
and the MOSFETs.
A hole needs to be cut through the
top of the lid for the MOSFETs – ours
was 27 x 10mm, centred 5mm from
one edge – and also a small hole
drilled to allow VR1 to be adjusted
from outside the case with a fine
screwdriver. A 3mm hole would be
about right, lined up with VR1 underneath.
With the external leads soldered
to the underside of the PC board
(ie, direct to the tracks) they emerge
from the assembled case through the
cable- ways provided. Significantly
larger leads will of course need larger
holes cut.
One feature of the specified case
worth noting is that no extra screws
are required to hold it together. The
two portions snap together and then
the same screws which mount the
case prevent it from coming apart.
The centres for the mounting screws
are 73mm apart.
Testing
You will need a radio control transmitter and a matching receiver with
servo outputs, a suitable DC motor
and a DC supply or battery equal to
the task.
If you don’t have a radio control you
may wish to build the radio control
servo pulse simulator described at the
end of this article.
We will assume you are using a radio control receiver but if not, simply
connect the wires to the simulator the
same way around.
Connect the three servo wires to the
radio control receiver output. The red
and black wires go to + and - on the
output while the brown wire goes to
the data output – usually the middle
pin and on “real” servos, usually
coloured yellow.
Set the radio control transmitter
stick to either minimum if it is a single
direction controller or to centre (off)
in a dual-direction controller, with
trimtabs set to the centre as well, and
turn both transmitter and receiver on.
Apply power to the controller.
You’ll almost certainly find the
motor starts to turn (be careful of the
starting kick on a large motor if it is
not secured in some way!) but when
you adjust VR1 you should be able to
stop the motor completely.
If so, move the stick on the radio
control transmitter and you should
find the motor turns with its speed
proportional to the stick position.
Full stick should give you full motor
speed, or very close to it.
What if it doesn’t work?
Obviously, there is an error somewhere. Perhaps as a starting point,
eliminate the radio control transmitter and receiver by connecting a real
servo to the receiver and make sure
it works properly. That ensures you
have the right sort of waveform coming from the receiver.
If it works, check your wiring and
component placement again – more
than 95% of faults in kits are due to
one or two wrongly placed or reversed
components or poor soldering.
Check that you have +5V coming
to the ZN409 supply rail from the
radio control unit. If you have an
oscilloscope, view the waveforms
at pins 14, 5 and 9 of IC1. If you get
what looks like a correct waveform,
look further along.
Otherwise the error is somewhere
around that IC. There should be a
positive-going waveform at approx.
50Hz from pin 11 of IC2, its width
varying with either the input signal
or the position of VR2.
Also check the inversion of signal
between pins 1/2 and 3, 5/6 and 4,
8/9 and 10 and finally 12/13 and 11.
May 2000 83
Manual motor control via a simulator
Earlier we referred to the waveform from a radio control receiver
– a square wave of 50Hz with a duty
cycle dependent on the setting of
the radio control transmitter stick.
At rest the pulse should be 1.5ms
wide and full forward it should be
2ms wide.
It follows then that if a waveform
of this type was fed into the input
the system would operate as if it was
attached to a radio control receiver.
All we need do is simulate that
waveform. Fortunately, that is quite
simple to do. Two suitable circuits
are shown below.
The first consists of two 555 timers (actually a 556 which is two 555s
in one package) – one connected
as an astable oscillator running at
50Hz (ie, producing continuous
20ms-wide pulses). This triggers the
second 555 wired as a monostable
which has its pulse width variable
from less than 1ms to more than
84 Silicon Chip
2ms by adjusting VR1. The output
from pin 3 of IC2 then is a series of
pulses, 20ms apart, which vary in
length from less than one to greater
than two milliseconds. Now where
have we heard that before?
A similar circuit was first described in SILICON CHIP in May
1994. It produced 30ms pulses –
which should work fine – but we’ve
adjusted the values to give approximately 20ms, just to be consistent.
The second circuit, from the same
issue, is even simpler and contains
just one 4001 quad NOR gate and a
few other components. Its drawback
was that due to its simplicity the
frame rate changed with the pulse
width but apparently that didn’t
cause any problems.
For a full description of these
circuits, refer to the May 1994 issue.
Copies of that issue are still available from SILICON CHIP Publications
for $7 each including postage &
packing ($7.70 after July 1).
We believe either could be used
but we must say that we haven’t
tried either with this circuit. PC
board patterns are shown for both
but as they are so simple these could
just as easily be built on a small
piece of Veroboard to save the cost
of a PC board.
You can use these simulators to
either set up your controller in the
absence of a radio control system or
you can use it to “hard wire” control
an electric motor (low voltage DC
only!). That’s up to you.
Note that you will have to arrange
a 5V supply for both the simulator
and the ZN409 circuitry in the speed
controller. This could most easily be
done with a 7805 regulator taking its
input from the 12V supply. (When
used with a radio control receiver
the speed controller takes its 5V
supply from the servo output of the
SC
receiver).
T
|