Current Surges in DC Motor
The steady-state armature current depends on the small difference between the back e.m.f. E and the applied voltage V. In a converter-fed drive it is vital that the current is kept within safe bounds, otherwise the thyristors or transistors (which have very limited overcurrent capacity) will be destroyed. In order to prevent the current from exceeding its rated value we cannot afford to let V and E differ by more than IR, where I is the rated current.
It would be unacceptable, for example, to attempt to bring all but the smallest of d.c. motors up to speed simply by switching on rated voltage. For example, if rated voltage of a d.c. motor is 500 V, and the armature resistance is 1 Ω. At standstill, the back e.m.f. is zero, and hence the initial current would be 500 /1 = 500 A, or 25 times rated current! This would destroy the thyristors in the supply converter (and/or blow the fuses).
Clearly the initial voltage we must apply is much less than 500 V; and if we want to limit the current to rated value (20 A in the example) the voltage needed will be 20 x 1, i.e. only 20 V.
As the speed picks up, the back e.m.f. rises, and to maintain the full current V must also be ramped up so that the difference between V and E remains constant at 20 V.
Of course, the motor will not accelerate nearly so rapidly when the current is kept in check as it would if we had switched on full voltage, and allowed the current to do as it pleased. But this is the price we must pay in order to protect the converter.
Similar current-surge difficulties occur if the load on the motor is suddenly increased, because this will result in the motor slowing down, with a consequent fall in E.
In a sense we welcome the fall in E because this brings about the increase in current needed to supply the extra load, but of course we only want the current to rise to its rated value; beyond that point we must be ready to reduce V, to prevent an excessive current.
The solution to the problem of overcurrents lies in providing closed-loop current-limiting as an integral feature of the motor/drive package. The motor current is sensed, and the voltage V is automatically adjusted so that rated current is either never exceeded or is allowed to reach perhaps twice rated value for a few seconds.
Dynamic Behaviour and Time-Constants
The use of the terms ‘surge’ and ‘sudden’ in the discussion above would have doubtless created the impression that changes in the motor current or speed can take place instantaneously, whereas in fact a finite time is always necessary to effect changes in both.
If the current changes, then so does the stored energy in the armature inductance; and if speed changes, so does the rotary kinetic energy stored in the inertia. For either of these changes to take place in zero time it would be necessary for there to be a pulse of infinite power, which is clearly impossible.
The theoretical treatment of the transient dynamics of the d.c. machine is easier than for any other type of electric motor but is nevertheless beyond our scope. However it is worth summarising the principal features of the dynamic behaviour, and highlighting the fact that all the transient changes that occur are determined by only two time-constants.
The first (and most important from the user’s viewpoint) is the electromechanical time-constant, which governs the way the speed settles to a new level following a disturbance such as a change in armature voltage or load torque.
The second is the electrical (or armature) time-constant, which is usually much shorter and governs the rate of change of armature current immediately following a change in armature voltage.
When the motor is running, there are two ‘inputs’ that we can change suddenly, namely the applied voltage and the load torque. When either of these is changed, the motor enters a transient period before settling to its new steady state.
It turns out that if we ignore the armature inductance (i.e. we take the armature time-constant to be zero), the transient period is characterised by first-order exponential responses in the speed and current. This assumption is valid for all but the very largest motors.
For example, if we suddenly increased the armature voltage of a frictionless and unloaded motor from V1 to V2, its speed and current would vary as shown in Figure 1.
There is an immediate increase in the current (note that we have ignored the inductance), reflecting the fact that the applied voltage is suddenly more than the back e.m.f.; the increased current produces more torque and hence the motor accelerates; the rising speed is accompanied by an increase in back e.m.f., so the current begins to fall; and the process continues until a new steady speed is reached corresponding to the new voltage.
In this particular case, the steady-state current is zero because we have assumed that there is no friction or load torque, but the shape of the dynamic response would be the same if there had been an initial load, or if we had suddenly changed the load.
The expression describing the current as a function of time (t) is:
The expression for the change in speed is similar, the time dependence again featuring the exponential transient term e-t/τ.
The significance of the time-constant (t) is shown in Figure 1. If the initial gradient of the current–time graph is projected, it intersects the final value after one time-constant. In theory, it takes an infinite time for the response to settle, but in practice the transient is usually regarded as over after about four or five time-constants.
We note that the transient response is very satisfactory: as soon as the voltage is increased the current immediately increases to provide more torque and begin the acceleration, but the accelerating torque is reduced progressively to ensure that the new target speed is approached smoothly.
Happily, because the system is first-order, there is no suggestion of an oscillatory response with overshoots. Analysis yields the relationship between the time-constant and the motor/system parameters as
τ = RJ/k2 ……(equation 2)
Where R is the armature resistance, J is the total rotary inertia of motor plus load, and k is the motor constant.
The appropriateness of the term ‘electromechanical time-constant’ should be clear from equation (2), because τ depends on the electrical parameters (R and k) and the mechanical parameter, J.
The fact that if the inertia was doubled, the time-constant would double and transients would take twice as long is perhaps to be expected, but the influence of the motor parameters R and k is probably not so obvious.
The electrical or armature time-constant is defined in the usual way for a series L, R circuit, i.e.
τa = L/R ……(equation 3)
If we were to hold the rotor of a d.c. motor stationary and apply a step voltage V to the armature, the current would climb exponentially to a final value of V/R with a time-constant τ a.
If we always applied pure d.c. voltage to the motor, we would probably want τa to be as short as possible, so that there was no delay in the build-up of current when the voltage is changed.
But given that most motors are fed with voltage waveforms which are far from smooth, we are actually rather pleased to find that because of the inductance and associated time-constant, the current waveforms (and hence the torque) are smoother than the voltage waveform. So the unavoidable presence of armature inductance turns out (in most cases) to be a blessing in disguise.
So far we have looked at the two time-constants as if they were unrelated in the influence they have on the current. We began with the electromechanical time-constant, assuming that the armature time-constant was zero, and saw that the dominant influence on the current during the transient was the motional e.m.f.
We then examined the current when the rotor was stationary (so that the motional e.m.f. is zero), and saw that the growth or decay of current is governed by the armature inductance, manifested via the armature time-constant.
In reality, both time-constants influence the current simultaneously, and the picture is more complicated than we have implied, as the system is in fact a second-order one.
However, the good news is that for most motors, and for most purposes, we can take advantage of the fact that the armature time-constant is much shorter than the electromechanical time-constant.
This allows us to approximate the behaviour by decoupling the relatively fast ‘electrical transients’ in the armature circuit from the much slower ‘electromechanical transients’ which are apparent to the user. From the latter’s point of view, only the electromechanical transient is likely to be of interest.