The present invention relates to an electronic switch comprising a controlled semiconductor device controlled by current delivered from control cir¬ cuitry, and overcurrent protection circuitry, where the control circuitry is galvanically separated from the controlled semiconductor device by means of a transformer, and where the power for the overcurrent protection circuitry is supplied from said trans¬ former. The present invention originates in the desire to use electronic switches for controlling equipment such as electromagnetic valves and fans drawing air over heat exchangers in refrigeration systems and cooling plants. Frequently the valves are operated using pulse width modulation. The duty cycle for the control of the valves is relatively long, i.e. sev¬ eral seconds. Electronic switches are switches, which, as op¬ posed to electro-mechanical switches, do not make and break electrical contacts, but instead control the current through a semiconductor device, such as a transistor. Such switches are well known in the art and have many advantages, such as being reliable over many switching cycles and having short on-off switch¬ ing times. Because of these advantages such elec¬ tronic switches are useful for modulation, such as the pulse width modulation mentioned above. Depending on the specific purpose of the pulse width modulation the duty cycle thereof may vary from microseconds to the above-mentioned several seconds. Compared to controlled electro-mechanical switches, i.e. relays, electronic switches do however suffer from the drawback that they do not, unless ap¬ propriate measures are taken, provide any galvanic separation of the control circuit and switched cir¬ cuit . In the prior art it has been found advantageous to use a transformer to provide the galvanic separa¬ tion. Examples of such prior art electronic switches are found in DE-U-93 16 157, DE-U-296 17 892, US-B- 6,462,603, US-A-5, 168, 182 , and JP-A-57-131125. Moreover, separating the control circuit from the switching circuitry provides the desired advan¬ tage that the voltage that may be switched by the switching devices becomes independent from voltage with which the control circuitry is operated. This mean that the control circuit may be operated at low voltages commonly used for digital circuitry, e.g. 5 V or less, while voltages substantially higher e.g. in the range from 24 V to 600 V may be switched by the switching devices. Also, these switched voltages may float with respect to the control circuitry, al¬ lowing the electronic switch to be inserted almost anywhere in the electric circuit that has to be con¬ trolled, e.g. in a phase between the supply and the load without the switch having a terminal on ground or zero potential. This again makes it difficult to provide elec¬ tronic overcurrent protection such as overload pro¬ tection or short circuit protection, because the switched voltage cannot be used as supply for the electronic overcurrent protection. Essentially the voltage over the terminals of the electronic switch will be zero during the switching cycles. Fuses are normally to slow to be of any use in protecting the switching devices, and it will be no¬ ticed that even though some of the above prior art documents have built in fuses, these fuses do not provide protection against overload or short circuit of the switching devices. Instead these fuses protect against faults in the overall switching circuitry, but they do not protect the switching devices or the switching circuit itself against overcurrents from overloads or short circuits. JP-A-57-131125 mentioned above does, however, have a basic overcurrent protection. In one embodi¬ ment the overcurrent protection uses an SCR to draw down the basis of a transistor constituting the semi- conductor switch. The SCR is in turn controlled by the potential over a measuring resistor in series with the emitter of the semiconductor switch in the load circuit. Hence, if the current through the meas¬ uring resistor exceeds a given value, the SCR turns on and switches off the semiconductor switch. This overcurrent protection, however, does not have any selectivity between various overcurrents, such as short circuits, overloads, or in-rush currents. US-A-4363064 and DE-A-4007539 describe con- trolled electronic switches with overcurrent sensing. In US-A-4363064 the sensing circuitry is supplied via a transformer from the control circuitry as part of the drive signal . The supply for the control cir¬ cuitry, however, is taken from the main circuit to be interrupted, and is thus not independent thereof. In both US-A-4363064 and DE-A-4007539, if an overcurrent is detected, a signal is transmitted back through the transformer to the control circuitry. Transmitting the signal back through the trans¬ former to the control circuitry, in order to allow the control circuitry to interrupt the control sig¬ nal, and thus cutting off the overcurrent, is however complicated and involves a time delay. Moreover, if selectivity between various overcurrents such as short circuits, overloads, and in-rush currents, is to be achieved, the signal transmitted must contain sufficient information to distinguish these, thus in- creasing the complexity of the signal and the trans¬ mission thereof. It is the object of the invention to provide an electronic switch according to the opening paragraph wherein the switching devices have a better and more versatile protection against overcurrents such as overloads or short circuits. The invention achieves this object by providing an electronic switch according to the opening para¬ graph, characterised in that the overcurrent protec- tion circuitry being adapted to respond autonomously and selectively to overcurrents and so as to control the turning off of the controlled semiconductor de¬ vice in dependence of whether the overcurrent is a short circuit or an overload. Making the protection circuitry autonomous ob¬ viates any need of transmitting signals back to the control circuitry. Instead, the overcurrent protec¬ tion circuitry may immediately itself turn off the controlled semiconductor switch in case the overcur- rent is a harmful overcurrent such as a short circuit or an overload. Moreover, as the overcurrent protec¬ tion circuitry is adapted to respond selectively be¬ tween overloads and short circuits, it is achieved that the overcurrent protection circuitry autono¬ mously does not effect unnecessary interruptions of the load. At the same time any need for transmission of information regarding the nature of the overcur- rent back to the control circuitry, because the se¬ lectivity is implemented directly in the overcurrent protection circuitry. According to a further preferred embodiment, the overcurrent protecting circuitry comprises means for automatic reset after a time delay after a re¬ sponse to an overcurrent. Thus if the overcurrent condition does not persist, the overcurrent protec¬ tion circuitry will again allow the controlled semi¬ conductor device to conduct. Preferably the time de- lay depends on the magnitude of the overcurrent to which the overcurrent protection circuit responded. According to a further preferred embodiment, the overcurrent protection circuitry is adapted to not to respond to in-rush currents. Thereby the autonomous overcurrent protection circuitry becomes even less prone to effect unnecessary interruptions of the load. Also, this is achieved without the ne¬ cessity of transmitting any information back to the control circuitry. According to another preferred embodiment, the overcurrent protection circuitry comprises a compara¬ tor, where said transformer provides the power supply for the comparator, and the output of the comparator controls the controlled semiconductor device. Thereby, under normal circumstances the control signal transferred via the transformer, is used to switch on and off the comparator, the output of which consequently controls the controlled semiconductor device, switching it on when the comparator with the associated overcurrent protection circuit is on, and switching the controlled semiconductor device off when there is no supply to the comparator, the com- parator thus not providing any output signal . Thus by using the comparator the overcurrent protection cir¬ cuitry becomes autonomous, i.e. it is able to turn off the controlled semiconductor device independently of the control circuitry. According to yet another preferred embodiment the overcurrent protection circuitry comprises a com¬ parator in conjunction with an RCD network. This al¬ lows for the desired selectivity. Thus, according to still a further preferred embodiment the RCD network is adapted to provide both overload protection and short circuit protection. This can be realised in a simple fashion with appro¬ priate arrangements of diodes in the branches of the RCD network. In particular said overload protection and said short circuit protection are provided by charging one and the same capacitor through respec¬ tive branches of the RCD network to a predetermined trigger voltage of the comparator. With one and the same capacitor the respective time constants involved will be determined by the resistors in the branches. The invention will now be described in greater detail based on an exemplary embodiment and with ref¬ erence to the drawings. In the drawings, fig. 1 illustrates a block diagram of the elec- tronic switch according to the invention, fig. 2 illustrates a schematic of the elec¬ tronic circuitry of the electronic switch according to the invention, fig. 3 illustrates a the time-current protec¬ tion response of the electronic switch. In fig. 1 is shown a block diagram of a switch¬ ing device incorporating an electronic switch accord- ing to the invention. The active element of the elec¬ tronic switch is a semiconductor switch 1, such as an Insulated Gate Bipolar Transistor. The semiconductor switch controls the current through a load 2 in a load circuit. In order to use the electronic switch for both AC and DC a bridge rectifier 3 is inserted in the circuit load circuit. A current sensing device 4 is connected in series with the semiconductor switch 1. The semiconductor switch is controlled in¬ directly by supervisor circuitry 5, such as overcur- rent protection circuitry based on the current meas¬ ured by the current sensing device 4 , connected thereto. The supervisor circuitry 5 is supplied by a gate driver supply 6 which draws its own supply cur¬ rent from a transformer 7 providing galvanic separa- tion between the logical control circuitry 8 control¬ ling a high frequency signal from an oscillator 9. Typically the various parts of the electronic switch would operate at different voltages. The logi¬ cal circuitry 8 and the oscillator 9 would normally operate at 5 V or less, the gate driver and supervi¬ sor at 12 V. The load circuitry could be any desired voltage between e.g. 12 V and 230 V or more and could be DC or AC, e.g. 50 Hz or 60 Hz mains AC. The oscillator produces a high frequency pulse train, with a frequency of e.g. 4 MHz, which is fed to an input of a NOR gate in the logical control cir¬ cuitry 8 with two inputs. The NOR gate may be one among a plurality integrated in the same unit and fed by the same pulse train. More specifically there would be one NOR gate for each pair of switch termi¬ nals of the unit, i.e. one NOR gate for each elec¬ tronic switch. To the other input of the NOR gate a control signal, such as a pulse width modulation sig¬ nal, can be applied. When the control signal is ap¬ plied to the NOR gate the output will also produce a 4 MHz pulse train on the line 10 in dependence of the state of the control signal. Thus, in accordance with the control signal to the NOR gate an intermittent pulse train is produced on the line 10. This inter¬ mittent pulse train is fed to the primary side of a transformer 7. The transformer 7 provides galvanic separation, so that the logical control circuitry thus far described is separated from the supervisor circuitry 5 and the load circuit that follows. Gal¬ vanic separation gives a reinforced insulation, fre¬ quently demanded by authorities. Reference is now made to fig. 2 where the blocks 1, 3, 4, 5, 6 and 7 are illustrated in greater detail . The 4 MHz, 5 V pulse train described with ref¬ erence to fig. 1 is received on the line 10 and fed to the primary side of the transformer Tl via a ca- pacitor Cl. On the secondary side the pulse train, as passed by the transformer, i.e. distorted and pref¬ erably with an increased voltage in the range of 12 V, is rectified using diodes Dl and D2 and capacitors C2 and C3. The voltage is stabilised using a Zener diode DZl, thereby producing a supply voltage Vcc of e.g. nominally 12 V, as long as the control signal applied to the NOR gate is high. The use of a NOR gate is merely an example, and other ways of realis- ing this logical control circuitry are well within the skilled person's reach. As an alternative to us¬ ing logical gates a microprocessor could be used for producing the intermittent pulse trains directly to the transformer, whereby the logical gates can be omitted. In this respect, each of the individual electronic switches of the unit could have each own associated microprocessor, or signals from different outputs from a single microprocessor could be used to control a respective electronic switch. As an alter¬ native to a dedicated oscillator the clock signal of a microprocessor being part of the unit could be used. Also, regarding the rectification it should be noted, that the skilled person would realise that other types of rectifiers could be used, e.g. a stan¬ dard bridge rectifier with four diodes and, prefera¬ bly, one smoothing capacitor, as indicated in the block diagram of fig. 1. Though the rectified signal could be used di- rectly to control the semiconductor switch Z2 , this is not the case with the present invention. Instead, according to the invention the supply voltage Vcc is used to supply the supervisor circuitry 5. According to a preferred embodiment the super- visor circuitry 5 comprises an operational amplifier Ul coupled as a comparator. The gate of the semicon¬ ductor switch is coupled to he output of the compara¬ tor. The operational amplifier is supplied from Vcc. Consequently the supervisor circuitry 5 gets no supply current and will not operate when no pulses are received on the line 10, i.e. in the intermis¬ sions in the intermittent pulse train from the logi¬ cal control circuitry of fig. 1. The output will therefore not produce any output voltage to switch the semiconductor switch Z2 on. The semiconductor switch Z2 will thus be switched of in accordance with an off state of the control signal to the logical control circuitry 8. When on the other hand pulses of the pulse train are received on line 10, i.e. in accordance with an on state of the control signal to the logical control circuitry, the supervisor circuitry receives the supply voltage Vcc. The supervisor circuitry will thus be energized and operate, as will be described below. The operational amplifier Ul of the supervisor circuitry is coupled as a comparator, in particular as a Schmitt trigger. In normal operation the output of the compara¬ tor will be high, i.e. follow the supply voltage Vcc. Consequently, the semiconductor switch device Z2 will be on as long as the control signal applied to the input of the NOR gate of the logical control cir¬ cuitry 8 of fig. 1 is high. During normal operation the semiconductor switch Z2 may thus be switched on and off using the control signal to the logical con¬ trol circuitry of fig. 1. Consequently, the current through a load cir¬ cuit, e.g. a power line supplying a fan or a an elec¬ tromagnetic valve may be switched. Since the semicon¬ ductor switch Z2 is controlled by current supplied from the transformer Tl, which provides galvanic separation, it floats with respect to the voltage on the switched power line. The electronic switch may thus be used anywhere in the circuit to be switched, be it connected to the phase, connected to the neu- tral or even connected between two loads. Also, it may be inserted independently of polarity of the load to be switched, i.e. it is not important which termi¬ nal is connected to a phase or to the DC side of a circuit. This removes the risk of incorrect wiring. As already mentioned a bridge rectifier Dl is interposed in the load circuit, in order to be able to switch both DC and AC. When the semiconductor switch Z2 is on, the switched current passes through a first branch of the bridge rectifier Dl, through the semiconductor switch Z2, through a current sens¬ ing resistor R8, and through a second branch of the bridge rectifier. Returning now to the current sensing resistor R8. Any current through the load circuit will give rise to at voltage drop over R8. The amount of volt¬ age drop will depend on the magnitude of the current in the load circuit. The current sensing resistor R8 forms part of the supervisor circuitry 5, which protects the load circuit against overload and short circuits. In par¬ ticular it forms part of an RCD network, i.e. a net¬ work of resistors, a capacitor and diodes, coupled to the negative input of the operational amplifier Ul. The capacitor C4 is coupled to the negative in¬ put of the operational amplifier Ul via a resistor R3. Depending on the current through the current sensing resistor R8, and the voltage drop that fol¬ lows therefrom, the capacitor is charged through the resistors R2, R4 and diodes D4, D4, D5. When the voltage over the capacitor C4 reaches a given value, the output of the operational ampli- fier switches to zero. Thus the semiconductor switch Z2 turns off and interrupts the load circuit. As mentioned the charging of the capacitor C4 depends on the current through the current sensing resistor R8. The purpose is to provide selectivity, so that the supervisor circuit does not indiscrimi¬ nately effects the interruption of the load upon any overcurrent . Some transient overcurrents are fully acceptable, e.g. those transient overcurrents which occur when an inductive load such as a relay, an electromagnetic valve, or an electric motor of a fan is switched on. Such transient currents, commonly known as in-rush currents, may exceed the current of an overload, but are acceptable because they are transient, i.e. of short duration. Hence, the over- current protection should not respond to them. Overloads occur over longer time spans and the overcurrent protection should respond to them before they are critical, i.e. before the electronic switch, the load or other parts of the circuitry is damaged. Short circuits currents, which have a very high mag¬ nitude should be detected and responded to immedi¬ ately. More specifically the circuitry responds in the following way to overcurrents. When under normal load conditions the supervi¬ sor circuitry is active and the transistor is on, es¬ sentially all the current of the load circuit flows through the current sensing resistor R8, and gives rise to a voltage over R8. The resistance of R8 is relatively low, e.g. 0.5 Ω. The current that flows through the other branches of the network R2-C4, R4- D3-C4, D6-D5-D4-C4 and D6-D5-R7 can be neglected, be- cause they all have substantially higher impedance values. A typical value for R2 would be 1 MΩ, and the diodes could be considered non-conducting. If the current in the load circuit increases, e.g. as a result of an overload, the voltage over R8 increases proportionally. Eventually with increased voltage the diode D3 becomes conductive and current flows through D3 and R4, thereby charging C4. If suf¬ ficient current flows long enough to charge C4 to a predetermined voltage, depending on the supply volt¬ age Vcc and the voltage divider formed by Rl and R6, the comparator is triggered because the voltage on the negative input of the operational amplifier Ul exceeds the voltage on the positive input of the op- erational amplifier. In that case the output goes low and the semiconductor switch Z2 switches the current in the load circuit off. It should be noted that some hysteresis in the switching exists due to the voltage division between R6 and the positive feedback resis- tor R5. The load circuit is thus protected against overloads. Subsequently the capacitor C4 is slowly dis¬ charged through R2 and R8. The voltage over C4 drops and eventually makes it possible for the output of the operational amplifier to go high to Vcc again. If the overload persists, the procedure will repeat itself. In case of a short circuit the current in the load circuit increases rapidly to currents, which the semiconductor switch Z2 can only endure for a few mi¬ croseconds. In that case the voltage over R8 increases rap¬ idly until it reaches a voltage where the diodes D5, O6 start conducting and allow the current to also flow in R7. The resistance value of R7 is larger than that of R8, e.g. thousand times bigger. Typical val¬ ues would be R8 = 0.5 Ω and R7 = 470 Ω. With further increased current, D4 would start conducting, thereby charging the capacitor C4. Since the capacitor C4 is charged solely through the diodes D4, D5, D6 with no other current limiting resistances than R7 in the circuit, the predetermined voltage is reached very quickly and the output of the opera¬ tional amplifier Ul goes low and the semiconductor switch switches the current in the load circuit off. In fig. 3 the curve illustrates the current versus the response time. It can be seen that in the normal working area below Iovi no action will ever be taken. In the overload area above Iovi and up to IShort response times from a few milliseconds to approxi¬ mately 50 milliseconds depending on the magnitude of the overload current are achieved. For the short cir- cuit area above IShort quick responses are achieved. The response time for a short circuit current would normally be in the microsecond area, i.e. less one millisecond. Also in this case, the capacitor C4 is subse- quently slowly discharged through R2 and R8. The voltage over C4 drops and eventually makes it possi¬ ble for the output of the operational amplifier to go high to Vcc again. If the short circuit persists, the procedure will repeat itself. It should be noted that the circuitry as de¬ scribed only protects against short circuits or over¬ load conditions. It does not itself take any action to remove the cause for such a condition or alerting to the existence thereof. The addition of such cir¬ cuitry is not considered relevant for the invention as such, and would under any circumstances lie within the reach of the skilled person. Thus with the arrangement described above an electronic switch with overcurrent protection and galvanic separation between the logical control cir¬ cuitry has been provided. In particular an electronic switch, which is suitable for inductive loads is pro¬ vided, because the overcurrent protection circuitry, will not respond to transient overcurrents which oc¬ cur when such inductive loads are switched. Also the electronic switch does not have any minimum holding current, as e.g. switches using Tri- acs do, meaning that it can switch loads with very low currents. With the galvanic separation of the logical control circuitry and the electronic switch the elec- tronic switching device may be in used in the exact same manner as traditional normally open relay, thereby eliminating the consequences of incorrect wiring. Though the above description has been given for a single electronic switch it should be noted that several of the electronic switches illustrated in fig. 1 could be integrated in one and the same unit using the same common oscillator 9.