I'm currently using a couple of old H&M controllers with an even older
Triang one on "standby duty" for a yard area.
I would like to replace them with a modern PWM controller - with (as I have
a few tight bends) feedback. I am considering building my own - I do have
prior experiance of electronics and have some good curcuit diagrams but
having PCBs made up is making me think twice ...
I've been looking at the various adverts for Guagemaster, KPC etc and they
are all quite loud on the subject of whether they provide feedback, inertia
and all that but none of the adverts state if the controllers are PWM or
Any guidence folks?
It's not strictly the width of the pulse but the time interval between
full power spikes.
The idea is that the rotor's coils get a kick as they go past the pole
pieces so the motor speed synchronises with the pulse interval.
It's not used very often these days because it can cause modern motors
to overheat - in traditional motors the rotor has an iron core that
acts as a heat sink.
So (silly question time) what makes modern controllers so good compared to
the older ones, IOW how do they achieve their performance? I was under the
impression that PWM was the dogs 'do-dahs' - subject of course to the
problems with coreles motors.
Voltage regulator, which delivers max current even at the lowest
voltage - with an old fashioned rheostat, the resistor draws the
Feedback, basically the same but it also turns the voltage off
momentarily, and the motor becomes a generator for a tiny instant. It
monitors this back EMF and adjusts things accordingly to maintain the
same speed. Useful where there are gradients. Early ones weren't
suitable for modern coreless motors but most are these days.
Some of the better older controllers were actually pretty good.
Variable Transformers could deliver current at all voltages (H&M
Safety Minor and the later generation of PowerMaster). Transistor
controllers (H&M Electran?) were also good.
Somebody (Buehler?) still makes variable transformer controllers.
At my club (in Kingston, New York - I'm an ex-pat O-gauger from
Manchester) we still use a very old setup with variable transformers.
Feedback controllers would be nice because our long, heavy trains need
real driving on the gradients. You can't leave them running on their
own while you're distracted. They also pull a huge current - fellow
members wouldn't believe how little current my 517 and 6-wheel
On Tue, 14 Mar 2006 14:30:05 -0500, Christopher A. Lee
look Chris I hate to do this - please don't think I am picking on you
but this needs correcting:
The motor coils have a resistance measured in ohms. Ohms law states
that I (amps, current) = Volts/Resistance. You can't just push current
through something. Think of it as water in a pipe. the current is the
amount of water flowing, the volts is how hard the pump is pushing and
the resistance is the diameter of the pipe. If you turn the pump right
down, the current must drop
If the motor has a resistance of 20R and you chuck 12 Volts at it, the
current will be12/20 = 0.6A or 600mA
if you wind the voltage down to say 2V, the resistance hasn't changed
so the current must... it will now be 2/20 = 0.1
The resistance you measure with your multimeter is equivalent to the
off-load resistance - when the thing is under load, the *effective*
resistance will change - more on that later.
not exactly. Back EMF is a high voltage spike generated when the
magnetic field around a coil (motor) collapses and generates a voltage
& current. Motor theory says if you pass a current through a wire
suspended in a magnetic field, you get a kinetic motion attempting to
push the wire against the magnetic field - this motion is captured in
motors and it all goes round. The same rule says that if you move a
wire in a magnetic field, you generate a voltage (and if the circuit
is not open at one or both ends) a current. When you turn the power to
a motor (or any coil) hard-off the magnetic field collapses and as the
lines of flux cut through the coil it is the same effect as if you
were waggling the wire up and down the field, but it is very rapid
and aggresive voltages (damaging ones) get induced in the wiring -
this is why you nearly always see a diode across relay coils driven by
transistors or other elctronic (as opposed to electrical) signals - to
stop these voltages hurting stuff - they can be thousands of volts on
big coils. Used exactly this principle in a HV tesla circuit I made as
a kid... generated big voltages from a nine volt battery but switching
a relay coil on and off and bleeding the HV off through a diode onto a
HV capacitor - it did sting I can tell you. On little 12 volt motors
they might only be in the order of 70 or 80 volts. However ( and here
is the clever part) they are directly proportional to the current
flowing at the moment of collapse. As we all know, electric motors
work best when under load and the more you stall a motor, the harder
it works to turn against you and because the effective resistance (the
inductive resistance not the stuff you measured with a multimeter
above) drops, the current will rise so if the loco is working hard
going up an incline, it will be drawing more current but not going as
fast - by measuring the amount of back EMF (the High voltage spike
made by the collapsing mag field) you can tell how hard the motor is
working and so provide it with a bit more voltage (or effective Volts
if using PWM). It now has more power to overcome the forces trying to
stall it and the current will drop slightly but the proportions will
stay the same and the back EMF will not continue to rise in voltage.
so as the current drops significantly you can wind the voltage back to
maintain the ratio (it's now coming down the other side)
Sorry uncle, but your attempt isn't much better, and thus are urban
No, Back EMF is the voltage generated due to the motor acting as a
i.e. the armature windings are moving in the field of the permanent
magnets and the motor acts as a generator, just like Chris said. You
obviously have the basic theory so why do you then have to try and
complicate things with totally innapropriate theory?
It's nothing to do with collapsing magnetic fields, when you remove
power to the motor the armature keeps moving due to inertia and acts
like a generator.
That's due to the inductance of the relay coil, which tries to keep the
current flowing when it's switched off. The diode acts as a short
circuit for that current.
You may indeed get such a spike from a motor coil, but only one, when
the power is removed. From then on we are talking about Back EMF which
is a totally different mechanism, I'll say it again, the motor is
acting as a generator.
Wrong mechanism. Back EMF occurs because the motor is acting as a
generator and the voltage is proportional to the rotational speed of
the motor. That's how feedback controllers work, by measuring the
*actual* rotation speed of the motor versus the desired speed. To work
properly, the system needs to be tuned to each individual motor. The
best DCC decoders give you full control over the tuning of the Back EMF
to suit the motor in each loco.
It depends what you mean by "work best". Maximum torque, maximum power,
maximum efficiency? Max torque occurs at zero revs. Maximum power at
half the no-load speed of the motor.
Back EMF is being generated by a motor all the time, even when it's
being driven. At lighter loads, and thus higher speed, the Back EMF is
greater. Back EMF has opposite polarity to the drive voltage and
opposes it. It's the lower effective *voltage* that causes lower
current at higher speed. The armature resistance is constant. It's
nothing to do with inductive resistance. At the extreme, if the motor
is totally stalled there is no Back EMF and the total supply voltage is
applied across the motor.
You measure the Back EMF generated by the motor (acting as a generator)
and it gives you a measure of the rotational speed of the motor.
On Tue, 14 Mar 2006 12:09:11 -0500, Christopher A. Lee
erm, yes. it is.
that's why it's pWm - i.e. the width of the pulse is changed. rising
edges (i.e when the power is turned on) happens at a precise frequency
and the power stays on for a variable amount of time and then off for
the rest. the more kick you give it, the longer the on period and
correspondingly the shorter the off period, the frequency of pulses is
always the same just the duty cycle (amount of time on versus amount
of time off) changes.
Typically in a digital system, run the pulse from an output from a
processor having sampled a variable resistor. say that value can vary
between 0 & 255 (8 bit), 0=off BTW... the value you read is how long
you turn the power on for and then you invert that value (so turn all
the 1s to 0s and vice versa) and that gives you how long to turn the
power off... so on a very low setting, say, 1, if your frequency is
200Hz (pulses per second per second), you turn the power on for
1x0.005 seconds and then turn the power off for (255-1=) 254x0.005
seconds. then if you think about half way, you turn the power on for
128x0.005 and off for 127x0.005 - but the sum of the on and off
periods always adds up to 255. The Motor doesn't see the pulses so
much because it takes time to start turning and time to stop so it
tends to smooth out the bumps and the perception is of a constant
power delivered to the motor. This is called "integration" in
I built my own PWM controller (using a PIC) on just this principle and
I could get my farish diesels (and others) to crawl at less than scale
walking speed- lovely for shunters and stuff. Oh, and because you are
always supplying the full 12volts to the motor (albeit in bits), it
overcomes sticky or reluctant beasts - maybe not perfect but much
In practice, 200Hz is a little on the slow side and some motors could
hum a bit. also, I need to do some work on pulse shaping - the rapid
rise in power generates extra heat (because of the inrush current into
the inductance of the motor coils. The magnetic field needs to be set
up and at the rise times we are looking at with a square wave the
energy is quite significant... the only place to dissipate that energy
is in the copper of the coils) if I shaped it to rise slower, the
current is reduced and so the heat - but then you need to do some work
on the timing to get it just right.
On Tue, 14 Mar 2006 12:09:11 -0500, Christopher A. Lee
sorry... forgot to mention... you are talking about Frequency
modulation here because the time is changing but even here, the amount
of time the power spends on is significant. even at 1:1 ratio, you are
only going to get 50% effective power (after integration) and you need
to get the pulse widths right or you get a two speed motor, going and
not-going and nothing in between.
Pulse Width Modulation ... in essence the controller constantly sends
"blips" of full power. It can be done in one of two ways, one way (best)
the blips are all the same size and what's ajusted is teh speed at which
they are are sent by the controller ... low speed a few blips per second,
high speed, lots of blips per second. The other way is to ajust the lengths
of the blips themselves but at a constant rate. Low speed - short blips,
high speed long blips.
The motor thus is only ever fed by full power but it can cause the motor to
make a "put put" sound - and potentially overheat. There's also an issue
with coreless motors ... but I don't have any of those.
Noise and coreless motors are only a problem if the frequency is too
low, hence we have DCC decoders that drive the motor with higher
frequency (usually something > 15KHz).
Your "best" method is actually a form of frequency modulation and more
likely to result in too low a frequency.
Making your own PCBs isn't that hard! For a one-off, you can just get
some etch resist tape and pads from somewhere like Maplin, some ferric
chloride from somewhere like Maplin, and off you go! You don't have to
go the whole hog with photoresist and light boxes and all that palaver -
just slap the tape on the board, dunk in acid, and drill when etched.
OK, there's a little more to it than that, but not much.
Go on, give it a go!
Yup, agreed it's no great stress. It's simply that bit more expense; and if
it ends up working out as much if not more than a comercial product that I
can get off the shelf and not spend an entire weekend putting togther then
it's not worth my while.
On Tue, 14 Mar 2006 15:14:51 +0000, Chris Wilson wrote:
Buy a copy of Roger Penfold's book Practical Electronic Model Railway
Projects, published by Bernard Babani in paperback. He answers your
questions and uses veroboard layouts. Printed circuit boards are a waste
of time other than for bulk production or complex circuit designs, neither
of which apply to your project. And they are less easy to modify.
I've got Rodger Amos's book it may be a little old but everything appears
to make sense.
Well not necessarily, as Nick points out it's easy enough to knock up "one
offs" assuming you have the kit, no different realy to using veroboard
except you tend to need far less bridges and you have far less chances of
No my real question was what's on the market now, IOW if I go in to a shop
and ask for a PWM (or modern equiv) with feedback just what am I going to
be offered and what does it all mean.
many thanks though
if you are doing the board manually I tend to agree with you, but if
you get some PCVB design software (there are some that are free to use
below a certain component count) they can make a much better job of
Problem with strip board is remebering to make sure you put your cuts
in the right places otherwise you can get allsorts of nasties happen.
A computer designed PCB eliminates this.
horse for courses - I use both depending on how lazy I feel and how
big the project is.
Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here.
All logos and trade names are the property of their respective owners.