GB2143342A - Device and means for controlling air conditioning plants in buildings - Google Patents

Abstract

A control device, for an air-conditioning plant having a central heater or cooler and several units 1 in different rooms, which takes as a reference the room in which, from time to time, the ambient temperature differs the most from a nominal pre-chosen value unique to that room on account of the heat load at that particular moment, and varies the power output of the heater or cooler as a predetermined function of that temperature difference. <IMAGE>

Description

SPECIFICATION Thermostatically controlled mixer This invention relates to a thermostatically controlled mixer valve for mixing hot and cold water, for example in showers.

Thermostatic mixer valves are known in which a temperature sensor in the valve output provides an electircal signal which is compared with a presettable reference signal, the difference between these two signals being used as a control signal for a motor driving the valve mechanism. Figure 1 illustrates in block diagram form a motor controller 1 for a motor 2 which drives, via a gear box 3, a mixer valve 4. The valve has inlets for cold water at temperature TK and hot water at temperature Tw.

The valve outlet provides water at temperature TM.

Temperature sensor 5 measures the outlet temperature and derives a signal TM. A comparator receives a "set value" signal Afrom a manual control 6 together with the feedback signal B from sensor 5.

The difference signal from the comparator is the control signal for the motor controller 1.

A major drawback to such a single loop system as shown in Figure 1 is that it can only detect and compensate for changes affecting the control loop after they have taken place, and these are reflected in temperature changes which are "loaded" with the temperature sensor time constant. Other factors affecting the performance of the system include gearbox play (hysteresis), motor speed, calcification of the valve elements etc. Known designs presuppose constant linear transfer functions of all elements which in practice is not the case. Electric motors have a load dependent starting voltage, as Figure 2(a) shows.Typically an electric motor will not begin to operate under even a light load until a threshold voltage F (-e) is reached, the motorthen having an angular velocity co proportional to change of the applied voltage U. A gearbox hysteresis, much of which is due to play or backlash in the gears, is shown in Figure 2rub). The temperature sensor has a dead period due to the insulation provided by its encapsulation, and in addition it usually has an inherent non-linear temperature-time function, as shown in Figure 2(c).

The present invention seeks to provide a low cost thermostatically controlled mixer which avoids or minimises the foregoing disadvantages.

According to the present invention there is provided a thermostatically controlled mixer valve arrangement wherein a motor controlled valve is responsive to a control signal being the difference between a pre-set temperature value and outlet temperature value derived via a first feedback loop from the valve outlet to the motor control, characterised in that the mixer valve arrangement further includes one or more secondary feedback loops wherein a secondary control signal(s) is derived from mechanical or electromechanical dynamic parameters of the valve arrangement, said secondary control signal(s) being applied to a secondary control device for the mixer arrangement.

In order that the invention may be more clearly understood embodiments thereof will now be described with reference to Figures 3-13 of the accompanying drawings, in which: Figure 3 illustrates a mixer arrangement having a secondary feedback loop incorporating a valve position sensor, Figure 4 illustrates a mixer arrangement having a secondary feedback loop incorporating a valve velocity sensor, Figures 5 & 5a illustrate a mixer arrangement having a secondary feedback loop incorporating a motor speed sensor, Figure 6 illustrates a modification of the mixer arrangement of Figure 5 in which the secondary feedback loop further incorporates an integrating circuit, Figure 7 illustrates a mixer arrangement with a feedback loop incorporating a motor speed sensor and a temperature controller, Figure 8 illustrates a modification of the arrangement shown in Figure 7, Figures 9 & 9a illustrate a mixer arrangement incorporating a lead controllerto simplify the arrangement of Figure 8, Figure 10 illustrates a mixer arrangement incorporating a feedback loop with a signal proportional to the rate of change of the temperature TM being superimposed on the velocity control signal, via an additional control circuit, Figure 1 7 illustrates a modification for a mixer to allow a sudden change in the "set" value to be imposed on the thermostat without overshoot, Figure 12 illustrates a modification to a mixer to eliminate system oscillations, and Figures 13 & 13a illustrate the incorporation of a "lag" circuit into the mixer control to compensate for variations in the starting voltage of the valve motor.

Referring to Figure 3, a mixer arrangement comprises a temperature controller 1 which controls a motor 2 driving, via a gearbox 3, a mixing valve 4.

The mixing valve 4 has a hot water inlet Tw and a cold water inlet TK. The outlet TM provides water at a temperature tM dependent on the mixing ratio and the inlet temperatures. The temperature tM of the outlet is determined by a temperature sensor 5 and this provides a signal tM which is compared with the set temperature Ts from a set value control 6. The difference e between Ts and TM is the input to the temperature controller 1. The temperature controller 1 processes the difference signal e according to a control law preferably Proportional Integral Differentiation (PID) and feeds a control signal to a valve position controller 8. This control function may be implemented by a microprocessor.

In order to overcome some of the drawbacks in single loop systems as previously referred to, secondary feedback loops are incorporated. The simplest form of secondary loop is one in which a position sensor 7 determines the actual position q of the mixer valve 4, e.g. at the outlet of the gearbox 3, and provides a position signal qwhich can be compared with a desired position signal q set in comparator 8a.

The error signal is fed to a position controller 8 which in turn controls the motor 2. The position sensor7 may be no more than a d.c. fed rotary potentiometer directly coupled to the gearbox valve drive train or via a low play, i.e. backlash free, gear. For a translatory position an inductive position sensor can be used, whilst if the drive is via a stepping motor the built-in pulse counter can act as a position sensor. In the arrangement shown in Figure 3 the position feedback signal q is, in effect, compared with the temperature difference signal e acting as a quasi "set position" signal. The result is that the position controller 8 gives an output voltage to the motor 2 to drive the mixer valve until e = q.If the position controller is a proportional device then a suitable choice of gain factor will ensurethatthe valve is adjusted at maximum speed without overshoot, and that uncertainties resulting from gearbox play are virtually eliminated. The voltage supplied to the drive circuit is thus always greater by the gain factor of the controller 8 than would have been the case if only the single temperature feedback loop had been active. Thus mechanical loads in the valve are more quickly overcome and cannot significantly affect the outlet temperature.Electronic position sensing also makes end step switches superfluous whereas an electric motor in a single loop arrangement acts as an integrating element, which would lead to stability problems if a further phase lag element were included in the control loop, the double feedback loop arrangement of Figure 3 does not appear as an integrator to the outside observer.

The temperature controller 1 with subordinate position control could therefore be a proportional integral (PI) controller. In contrast to a simple P controller it is able to remove even the smallest control deviation. This is particularly applicable when the valve becomes difficult to drive owing to calcite deposits which would lead to increasing control deviation in the case of a simple proportional (P) controller.

An alternative arrangement to that of Figure 3 is one where the secondary loop uses a valve velocity sensor 10, Figure 4. Typically the sensor may be a tachogenerator producing an electrical signal proportional to the velocity Vat which the valve is moved. Similar two the position sensing arrangement, the velocity signal V is taken to a comparator 9a where it is compared with the output of the temperature controller acting to produce a "set velocity" signal V set. The error signal is applied to a velocity controller 9 in an analogous manner to the arrangement of Figure 3. The arrangement of Figure 4 has similar advantages to that of Figure 3 relative to drive train hysteresis and valve calcification. However, the arrangement of Figure 4 does look like an integrator therefore the temperature controller 1 cannot be implemented as a PI controller.The same behaviour as with the PI temperature controller and position feedback can be obtained, however, if now the temperature controller is implemented as a proportional differential (PD) controller. But there is a disadvantage in that the starting or threshold voltage of the motor can lead to a permanent control error. Furthermore this implementation cannot replace end switches.

Considerable reduction in technological effort of the previous scheme can be obtained by deriving the velocity from an electronic circuit 11 using the motor current, Figure 5. The angular velocity of the motor can be obtained from two voltage measurements at the input and output of the motor with interposition of a known measuring resistance. This implementation is especially cost effective since it avoids mechanical measuring elements. The implementation can be made even more effective by adding an integrated circuit 12 to the velocity feedback circuit, consisting of an op amp plus a few standard components, Figure 6. This turns the velocity signal into a position signal. However, the output of the integrator 12 may drift to the rail voltage as a result of systematic errors which can occur in the calculation of the angular velocity.This can be avoided by adding a negative feedback circuit to the integrator by means of a resistance 12a so chosen that the output of the integrator declines to zero at zero input with a time constant which is large compared with the time constant of the control process. This arrangement is now again equivalent to a position feedback circuit. It has however the disadvantage that the gearbox with its hysteresis lies outside the feedback loop. The disadvantage can be compensated by replacement of the costly electromechanical position sensor by a cheap electronic circuit which has the further advantage that no mechanical wear occurs as it would have done with a position sensor.

If the temperature controller 1 is implemented in the form of a PI controller, a further simplification is possible, Figure 7. The PI circuit can be constructed with separate proportional and integrating parts 13, 14, 15. The integrator 13 can be constructed in the same manner as the integrator in the velocity feedback loop described previously. If now the signal from the velocity calculator 11 is not taken via an additional integrating element 12 but via that belonging to the temperature controller 13 we have again a position control loop but using fewer electronic elements, Figure 7. The problem of integ ratordrift is now also removed because any tendency to drift is self-compensated by the feedback of the temperature signal.

Praxis shows that the last named arrangement, Figure 7, satisfactorily solves the task provided the velocity feedback signal is sufficiently free from noise.

In the case of d.c. motors only high quality components can guarantee this. Collector noise in cheaper motors can cause deterioration in the control process. The previously described arrangement can be improved to avoid the effect of collector noise. This is accomplished by splitting up the return loop of the velocity signal between the velocity calculator 11 and the feed input to the temperature controller 1, FigureS. The velocity signal is taken to a comparison point 9a interposed between the velocity controller 9 and the temperature controller 1 while the signal taken back to the temperature controller is taken from a branch point just before the comparison point 9a. Finally, a first order delay element 16 is placed between the new branch point and the temperature controller 1 whose transfer function is given by F(S) = V1(1 + TS). Here V is the amplification factor of the control element 8 and T the mechanical time constant of the motor. In the arrangement of Figure 8, the elements 9, 9a, 2 and 11 again form an internal control loop for the valve velocity as described for Figure 5. Making the gain of the proportional controller 9 sufficiently large ensures that the valve velocity can follow any change in the "set-velocity" signal almost immediately. This together with the delay element 16 causes the signal feedback to the point 13a to be almost equivalent in its time behaviour to the corresponding signal of Figure 7. Therefore, the arrangement of Figure 8 is equivalent to that of Figure 7 with the added advantage that cheaper motors (with higher collector noise) can now be used.

The transfer function G(S) of the elements 13 to 16 is formally equivalent to a conventional lead circuit provided the mechanical time constant of the motor T iS sufficiently small. This gives a possibility of greatly simplifying the circuit of Figure 8 while retaining all its control properties by replacing the temperature controller 1 together with the delay element 16 and the feedback to 1 3a by a conventional lead element 9a consisting of a few components.

Such an arrangement is, in its good control behaviour, similar to the initially discussed PI controller with subsidiary position sensing and control.

Although here the effects of drive train hysteresis and motor start-up voltage are not compensated, it has the advantage of cheapness.

The basic idea of control improvement by subsidiary circuits can be extended by assigning these to more than one internal dynamic parameter. Thus it is possible to add a feedback loop with a signal proportional to Tto the arrangement of Figure 5 which is superimposed on the velocity feedback loop.

This would normally require a separate measuring sensor but it is possible to derive Tfrom the valve velocity v if it is assumed that t corresponds to the valve velocity, itself delayed by the mixing process and running times. To implement this (Figure 10), a branch point is inserted into the velocity feedback loop between the velocity calculator 11 and the comparison point 9a from which the signal is taken to an electronic "TM calculator" represented by a first order delay circuit whose time constant is chosen to correspond to the delay between valve displacement and temperature change. Thus the T calculator outputs a signal proportional tot. The inner loop is closed by the tM controller and a comparison point 19a.Elements 1,9 and 19 are (in a preferred version) implemented as P controllers.

This arrangement again is equivalent to the lead arrangements already discussed in relation to Figures 8 and 9.

Certain further flaws in the various elements of the thermostat are - according to the invention - compensated by the introduction of various circuits into the main control loop.

A perturbation in the physical parameters affecting control path like a sudden pressure drop can only be sensed via the temperature sensor "loaded" with its time constant, whereas a sudden change in set value can be externally imposed on the thermostat without delay. This means that a thermostat optimised for internal perturbation exhibits strong overshoot during sudden changes in the set value while a thermostat optimised for good response to changes in the set value reacts too slowly to changes in the physical parameters. This situation is dealt with in accordance with the invention by following the set value block 6 with a delay circuit 21, Figure 11, whose time constant is made closely similar to that of the temperature sensor. This causes a thermostat optimised for outside disturbances to react to changes in set values without overshoot.

Practical considerations make it difficult for the time constant of the temperature sensor to fall below a certain minimum value. However, cheaper, slower temperature sensors can be employed if the delayed temperature signal can afterwards be so processed that it appears like an undelayed signal to the controller. This is accomplished by following the temperature sensor with an electronic lead circuit 23, Figure 11. In practice an effective time constant of temperature sensing of 0.1 second can be obtained if a temperature sensor of time constant T = 0.5 seconds is used.

Another problem is mixing noise due to imperfect mixing of the two water streams in the mixer valve.

This difficulty can be removed by incorporation of a low pass filter 24 into the temperature feedback loop. A 10 Hz filter is the most suitable means of attenuating the noise without any considerable increase in the effective time constant of temperature measurement.

In general a mixer thermostat cannot be characterised as a pure delay element. In addition to the delay constant Ta dead time to exists mainly caused by the running time effect of the valve movement and the arrival of the mixed water at the temperature sensor. The impact of the dead time can be described by an additional term exp (-sto) in the transfer function L(s) of the entire open control loop which causes a phase delay which increases with frequency at constant absolute L(s). This influences system stability in a decisive manner and in general tends to enhance the tendancytowards system oscillations. The effect can be reduced by reducing the system gain but only at the expense of slower system response and greater residual control errors.

The problem is solved by incorporating into the main circuit a further lead element 26 (Figure 12) constructed as shown in Figure 9a. Its constants can be so chosen that the phase delay is compensated such that the original gain value can be preserved.

It has already been pointed out that a control circuit with a lead controller and subordinate velocity controller cannot compensate for the deleterious effect of the motor starting (threshold) voltage (c.f.

Figure 2a) as can the more expensive PI controller.

This manifests itself in practice by a greater residual error e=.

However, the effect of the lead controller can be compensated for by increasing the value of the transfer function of the temperature controller H(s) selectively by inserting a lag circuit 28 before the temperature controller 1 as shown in Figure 13. By correct choice of the components of the lag circuit one can ensure that the transfer function is en hanced at low frequencies without affecting the stability of the crucial frequency w, (at which the overall systems transfer function becomes unity).

The residual error due to the lead controller can thus be reduced by the lag element by suitable compromise design. The starting threshold voltage of the motor increases with time because the motor load increases with increasing calcification of the valve.

The addition of the lag element 28 contributes to retaining the good control properties of the thermostat over extended periods of usage.

Thus the measures of Figure 13 allow the cost effective lead controller of Figure 9 to exhibit a performance which approaches closely to the excellent ones of the more costly PI controller. In similar manner it allows the implementation of Figure 10 also to approximate closely to the performance of the PI controller.

In summary the present invention gives rise to three specially advantageous systems:-the PI con trollerwith subsidiary position control of Figure 3 together with the measures described in Figures 11 and 12 form the ideal system. The lead controller of Figure 9 and the T controller of Figure 10 with subsidiary velocity control also incorporating the measures of Figures 11 and 12 form systems whose performance is somewhat weaker but whose cost is lower.

The above principles apply to other processes where a physical parameter is measured by a sensor "loaded" with a delay and dead time and where the permitted control times lie within similar orders of magnitude as the time constants (delay and dead) characterising the sensor.

Claims (23)

1. Athermostaticallycontrolled mixer valve arrangement wherein a power driven valve is responsive to a control signal being the difference between a pre-settemperature value and outlet temperature value derived via a first feedback loop from the valve outlet to the power control, characterised in that the mixer valve arrangement further includes one or more secondary feedback loops wherein a secondary control signal(s) is derived from mechanical or electromechanical dynamic pa rameters of the valve arrangement, said secondary control signal(s) being applied to a secondary control device for the mixer arrangement.

2. A mixer valve arrangement according to claim 1 characterised in that the secondary feedback loop comprises valve position sensing means yielding an electrical signal for measuring the control position of the mixer valve, said electrical signal being applied to a comparison means for comparison with a pre-set signal value derived from the main control circuit, the difference between the electrical signal and the pre-set signal being applied to the secondary control device for the mixer arrangement.

3. A mixer valve arrangement according to claim 1 characterised in that the secondary feedback loop comprises valve velocity sensing means yielding an electrica I signal for measuring the velocity of the mixer valve, said electrical signal being applied to a comparison means for comparison with a pre-set signal value derived from the main control circuit, the difference between the electrical signal and the pre-set signal being applied to the secondary control device for the mixer arrangement.

4. A mixer valve arrangement according to claim 3 characterised in that the valve velocity sensing means comprises an electrical circuit whose input parameters are the voltage dropped across the valve prime mover including a series measuring resistor, and the voltage dropped across the series measuring resistor alone measured at the same time.

5. A mixer valve arrangement according to claim 2 characterised in that the valve position sensing means comprises a d.c. fed rotary potentiometer mechanically coupled to the valve orto a gearbox intermediate the motor and the valve.

6. A mixer valve arrangement according to claim 2 characterised in that the valve position sensing means comprises an inductive position sensor coupled to the valve.

7. A mixer valve arrangement according to claim 2 characterised in that the motor is a stepping motor and the valve position sensing means comprises a pulse counter arranged to count pulses applied to the stepping motor.

8. A mixer valve arrangement according to claim 2 characterised in that the element measuring the valve velocity is comprised of an electrical network whose output parameters are two voltages one of which is dropped across the motor and a measuring resistance placed in series with it and the other which is dropped across the series measuring resistance only.

9. A mixer valve arrangement according to claim 3 or 4 characterised in that the motor control is a PI controller and that the integrating element of the PI controller is used for integration of the valve velocity signal.

10. A mixer valve arrangement according to claim 2 characterised in that a subsidiary valve velocity control is provided and that the electrical signal proportional to the valve position is derived by electronic integration of the set value signal for the valve velocity.

11. A mixer valve arrangement according to any preceding claim characterised in that the motor control incorporates a lead control circuit.

12. A mixer valve arrangement according to claim 3 or 4 characterised in that the arrangement incorporates an additional circuit element yielding an electrical signal which is a function of the rate of change of the outlet temperature and a subsidiary feedback control loop whereby the rate of change signal is compared with a pre-set signal generated by the first feedback loop and the difference between the rate of change signal and the pre-set signal is imposed on the loop controlling the valve velocity.

13. A mixer valve arrangement according to claim 12 characterised in that the additional circuit element yielding a rate of change signal is provided by an electric circuit whose input signal is the signal proportional to the valve velocity.

14. A mixer valve arrangement according to any preceding claim characterised in that the arrange ment includes compensation means for decoupling of the response of the controller to changes in set values from the response towards changes brought about by external factors.

15. A mixer valve arrangement according to any preceding claim characterised in that the arrangement includes compensation means for processing sensor generated measurement signals and for smoothing said signals and compensating for time lags associated with the temperature sensor.

16. A mixer valve arrangement according to any preceding claim characterised in that the arrangement includes compensation means for nonlinearities in the transfer functions of the feedback control loop in respect of dead times and motor starting or threshold voltages.

17. A mixer valve arrangement according to claim 14 characterised in that the pre-set temperature value is applied via an electronic network functioning as a lag circuit which delays the setvalue signal so as to prevent overshoot of the outlet temperature as a result of the response of the motor controller to changes in the set value of the temperature and that thereby the response of the controller to changes of set values of the temperature is optionally separated from its responses towards other factors.

18. A mixer valve arrangement according to claim 15 characterised in that the output of the temperature sensor in the first feedback loop is applied via a lead circuit the parameters of which are so adjusted that the effective time constant of the series arrangement consisting of the sensor and the lead circuit is smaller than that of the temperature sensor alone.

19. A mixer valve arrangement according to any preceding claim characterised in that the output of the temperature sensor is low pass filtered to suppress noise imposed on the temperature signal due to imperfect mixing or other causes.

20. A mixer valve arrangement according to any preceding claim characterised in that the control signal for the motor controlled valve is applied via an electronic circuit of the lead type whose parameters are adjusted to compensate for a phase delay in the transfer function of the first feedback loop caused by dead times occurring in the feedback loop elements.

21. A mixer valve arrangement according to any preceding claim characterised in that the control signal forthe motor controlled valve is applied via an electronic lag circuit whose parameters are adjusted to increase the gain of the control loop at low temperatures to reduce residual control errors.

22. A mixer valve arrangement substantially as described with reference to Figures 3-13a of the accompanying drawings.

23. An electronic water mixing thermostat with robust control behaviour not sensitive to parameter changes along the control path, characterised in that the main control circuit for controlling the temperature of the water mix is supplemented by one or more auxiliary control circuits for the inner dynamic parameters of the control path.