Any radio controlled machine with motors has potential problems. If you understand the problems, you can minimise them and this page is aimed at helping you. However - the problem can appear very complicated. Hopefully this page is written in plain language and will enlighten you rather than confuse you.
The block diagram shows the electrical system of a typical radio controlled machine, such as a robot. You may like to open this in another window so you can refer to it throughout this page.
At one end we have a radio receiver, whose input is a very small signal - maybe 2 microvolts at 40MHz. At the other end we have a motor whose input signal is a PWM waveform of maybe 36v peak, or about 18v rms of 20kHz.
So the input signal voltage is some 9,000,000 times smaller than the output, but at 2,000 times the frequency. You will realise from this that there is some potential for the output to interfere with the input!
4QD controllers are however well-behaved and do not generate a significant amount of interference. The same cannot be said for motors however! A motor contains a commutator and brushes. This arrangement is switching the motor current maybe eight times per revolution of the motor armature. A typical motor can revolve 3,000 times per second. So the current is being switched at a frequency that maybe around 24kHz. The motor current can be as high as 100 amps (or more).
Now any electrical switching arrangement tends to arc and spark. Sparks generate noise. The noise from a spark is broad-band and can have a very high radio frequency component, extending to hundreds, even thousands of megahertz.
You may recall that the original experiments with radio transmission were made using a spark transmitter and with this a signal was sent across the Atlantic! The wining in your robot is not going to be as good an aerial as was Marconi's original, but the moral should be clear.
So you need to make sure the motor, in particular its commutator and brush gear is in the best possible condition to minimise arcing. You also need to take steps to suppress and noise that is generated and clearly this needs to be done as close to the noise source as is possible.
Then you need to attend to the transmission paths so that this motor noise is not easily fed back either into the controller or into the receiver. Transmission paths are partially along wiring and partially by radiation, so screening should be considered.
Although the controller is a lot better behaved than the motor, and contains internal suppression, you should still be aware of noise transmission paths from controller to radio receiver - these will conduct not only controller noise but also motor noise.
You also need to attend to the aerial so that reception of wanted signal is maximised and unwanted noise is minimised.
This motor capacitor is also needed if you are using relays to reverse the controller as it helps quench any arcing which may occur if the motor is reversed fast.
Noise, at least the noise which is a nuisance in the type of machine we are discussing, travels by four paths:
All of these paths need to be considered and optimised to get best results.
It's possible to go 'overboard' and fit ferrite cores to the motor wires, but this is not often done and is probably not worth doing.
Any electrical signal gets radiated as a 'radio' transmission. The motor wires are carrying high currents, switched at 20,000 times per second. This background is also overlaid with any noise generated by the brushes. Clearly, you should reduce radiation from thee wires as much as practicable and the best way to do this is to keep them as short as possible and to make area of the open loop between then as small as possible.
In the first diagram the motor wires are widely separated and there is a large area of 'loop' between then, indicated by the gray area. This loop will act as an aerial, radiating any interference.
So run the two wires (M+ and M1) close to each other as shown in the second diagram where the area of the loop is very much reduced. If the wires are of a length that makes it possible, twist them together.
The motor body is generally a good screen. However there is coupling between it and the guts of the motor, so there will be a noise signal present on it. Exactly what you should do with this will depend on the design of the machine, but it will generally be screwed to the metal chassis and you have little choice of other action.
Any two wires running parallel to each other for any distance interact because of mutual inductance and capacitance. So consider which wires are likely to be 'hot', i.e. are connected to noise sources, and which are likely to be sensitive, i.e. connected to inputs where noise might get in and cause problems.
'Hot' wires will generally be motor-to-controller and controller-to-battery wires, and sensitive wires will generally be those associated with the radio receiver and the interface.
Keep these two groups of wires as far as possible from each other. If they must be in close proximity, you may need to consider using screened cable for the sensitive ones.
The controller is a lot more predictable and better behaved than the motor but it would be foolish to not minimise the radiation from it. The noise is not random, being generated by the PWM which is closely controlled, so it is a lot easier to study and to understand. Transmission mechanisms are of course exactly the same as for the motor and the steps you have taken already to minimise motor noise will also minimise controller noise from the motor wires: these are 'doing all the work' so you've already done a lot.
Exactly the same considerations apply to the battery wiring as apply to the motor. Keep it as short as reasonably possible, run positive and negative together to reduce any radiation loop area and twist the wires together if the length allows this.
Measures to reduce battery loop area also reduce the battery loop impedance - which (if excessive) can cause the controller's capacitors to overheat. See PWM speed control on 4QD-TEC site.
The signal transmitted from a typical radio control transmitter is well polarised and is not directional. Polarised means that it tends to radiate in a flat plane. So it tends to radiate in a sphere from the transmitter and the signal energy is therefore reduced as the square of the distance. At a certain distance from the transmitter, the signal is so know that the receiver cannot separate wanted signal from noise. Doubling the signal (or halving the noise) will only increase effective range by 40% therefore.
The signal received from any particular transmitter is a strong function of the receiving aerial and a good aerial can alter this factor by orders of magnitude. So any measures to improve range will depend far more heavily on having a good aerial than on any other factor.
Only once the receiving aerial is as good as possible is it worth worrying about noise and if you have followed the hints above - you've already done a lot. But any electrical system is 'noisy'. There will come a point, with increasing distance, where this noise will cause signal to be lost.
The other factor is that an aerial of the type used in a radio control receiver is not a single 'pole' - any electrical signal works between two points and the strength of the signal received depends on the signal voltage between the input from the aerial and the effective earth point that the electronics is comparing the aerial input with. So the aerial is in effect one half only of the overall aerial. The other half being 'virtual' but a function of the wiring, metalwork and chassis of the vehicle. So in effect you have a dipole aerial with a (potentially noisy) motor and controller (and battery) hanging on one end. If that is significantly noisy it injects the radio noise back into the effective earth of the aerial, where it appears to the receiver as a difference between earth and aerial signals and the poor old receiver cannot tell it apart from the signal coming in the aerial!
Exactly what effect this 'lump' (motor, battery and controller) has is of course scarcely predictable. But it it's well-constructed and suppressed, the effect should be minimal. So is a well constructed machine, you don't need any isolation in the earth. However if the machine is not good, then you want to prevent any noise signal travelling back along the earth wire to the receiver.
4QD's DCI uses potentiometers mechanically controlled by the radio control servos. With this system there is no electrical path and you have complete isolation.
With a digital interface, unless you specifically prevent it, there is an electrical signal path, as shown in the block diagram above, where the 'virtual' second half of the dipole is shown dotted. The earth connections through the individual boards are shown in green.
The best isolation device, should isolation be required, is an opto coupler. There are two places where opto couplers (aka opto-isolators) can be used, indicated by the two pecked yellow lines B1 and B2.
4QD have opto isolated versions of both our single channel and our dual channel digital interfaces and the optos in both are in the B1 position. This has several advantages:
From battery negative through controller 1 to C1b, through the multicore cable to C1a which is (and really should be) joined in the interface to C2a. Thence back through controller 2 to the battery. There is a page, complementary to this, on Machine wiring - Good and Bad practise which explains that earth loops are generally undesirable!
In this instance, avoiding this loop is possibly likely to cause more problems than it causes. However there are circumstances when it can cause a problem:
In the event of a motor fault, both tracks could, of course blow, so two fuses are required.