This application claims the benefit of Israeli Patent Application Number 188497 filed on Dec. 30, 2007, which is hereby incorporated by reference herein.
This invention relates to power supplies.
In many applications it is desirable to connect multiple DC power supplies in parallel. This may be done to increase the total current available or for redundancy. One particular application with which the present invention relates to is the powering of low voltage tungsten halogen or xenon lamps via a low voltage DC supply. To this end, it is known to fix a DC supply rail comprising positive and negative DC supply lines to or near the ceiling and then to mount one or more lamps to the supply rail across the supply lines. Commonly, the lamps can be moved along the supply rail so as to allow them to be directed in a preferred location.
In such a configuration, the number of lamps of any given power rating that be can be connected to the supply rail is limited by the power rating of the supply. This means that once the maximum power rating of the supply is exceeded, additional lamps can be connected to the supply rail only by coupling one or more additional power supplies to the supply rail. In practice this is fraught with difficulties.
Another known approach to power sharing and disclosed in US 2005/083640 to Weidmüller seeks to have the total load current split evenly among the supplies in parallel. For example, if four supplies are used to deliver 20 A they seek to have each supply loaded to 5 A in order that each supply will be operating properly. If a supply is required to go from no load to full load instantaneously (which is the case when a supply delivering almost all of the load current fails), it may go into current limiting. Proper load sharing also means the operating life will be maximized (the MTBF is longer at 80% of full load than at 100%).
Proper load sharing can only be accomplished when the output voltage of the supplies are at the same level at the point where they are commoned. This means that voltage drop in the wiring must also be taken into account. For example, if the terminals of one supply are used as the common point (i.e. two supplies are connected in parallel by daisy-chaining the output terminals and the load is connected directly to the terminals of one supply), then the voltage drop in the wires between the two supplies may affect the load sharing. An imbalance of as little as 50-75 mV can lead to the supply with the highest output voltage delivering virtually all the load current. If the output current rating is not sufficient for such a load current, the power supply will shutdown because of over-current or over-temperature. Maintaining a zero imbalance condition is very difficult—temperature fluctuation, component tolerances, and power supply location (i.e. wire lengths) are some of the factors that can influence the output voltage.
Often in applications involving parallel power supplies diodes are used to prevent a supply with a low output voltage from drawing current from a supply with a higher output voltage. This approach does not improve the load sharing situation and also introduces a voltage drop as well as additional heat dissipation. For example, a typical diode with a 0.7V drop used with a 10 A power supply would have to be rated for 7 W of power dissipation. This means a heat sink is required and the heat dissipation may affect other devices in the control circuitry. Schottky diodes offer a lower voltage drop and thus less heat dissipation, but still do not eliminate the problems.
To overcome these problems, Weidmüller proposes two solutions: (i) diode modules with pre-calibration of power supplies to ±50 millivolt and (ii) load sharing similar to the Unitrode design shown in
In the first approach, the diode modules enable multiple power supplies connected in parallel to adjust their outputs to maintain zero current imbalance. It is apparent that such an approach requires additional circuitry in each power supply and this adds to the expense and the bulk of the power supply, to such an extent that the power supply becomes both prohibitively expensive and bulky for low voltage lamp applications. It should be borne in mind that low voltage lamps do not need constant DC for their proper operation and this facilitates the use of low cost power supplies. Hence, the solution proposed by Weidmüller is hardly practical for such applications.
Even apart from this, such a solution requires pre-calibration of the power supplies to with ±50 mV and preferably requires that the parallel connection be effected as close as possible to the load, thus effectively militating against distributed loads along the complete span of the supply rail. In most lamp rail or track applications it is usually more convenient to connect multiple power supplies at different locations across the supply rail, for example at opposite ends.
DC power supplies that are designed to operate from the mains AC power supply include a diode bridge rectifier. Consequently, the connection of multiple DC supplies in parallel is equivalent to the circuit described above.
The full resonance converter shown in
Conventional full resonance converters do not lend themselves to parallel connection for the reasons described above. Specifically, the peak voltage across each of the resonant capacitors C3 and C4 is not limited. As the load increases the effective resistance of the voltage transformer decreases and this can lead to capacitors C3 and C4 to a state of over-voltage, whose magnitude is a function of:
This means that there is no effective limit to the load sourced by one of the resonant capacitors since if the resulting capacitor voltage increases beyond the peak supply voltage, the voltage across the other resonant capacitor will simply go negative, so that the sum of the capacitor voltages remains equal to the peak input voltage. Consequently, if two such converters are connected in parallel across a load that is actually larger than the power rating of one of the converters, there is no intrinsic mechanism to stop one converter from attempting to supply the full load. Of course, over-current and temperature protection may, and typically will, be provided but this merely stops the converter from working altogether and then the same problem is repeated in respect of the second converter, with the end result that all the converters will be shut down and no power will be applied to the load at all.
Pavlosky devotes much space to a comparison of the circuits reproduced in
Pavlosky also devotes much space to use of the converters shown in
It would clearly be desirable to provide a method and circuit that allows multiple converters to be connected in a power sharing arrangement without being prone to these drawbacks.
According to an aspect of the invention, there is provided a method for sharing power to an electrical DC load that exceeds a capacity of a first DC power supply between the first DC power supply and a second DC power supply such that in combination all of the power supplies are able to provide rated load power, said method comprising:
According to another aspect of the invention, there is provided a lamp illumination system comprising:
a pair of DC supply rails configured for connecting multiple lamps thereto; and
at least two spatially distributed power supplies connected to the DC supply rails, each of the power supplies comprising a partial resonance converter being operable in clamping mode so as to prevent the respective power supply from attempting to source a load that exceeds a nominal power rating of the power supply within a predetermined accuracy.
In order to understand the invention and to see how it may be carried out in practice, some embodiments will now be described, by way of non-limiting example only, with reference to the accompanying drawings, in which:
In the following description, components that are identical to those already described with reference to
The resulting voltage pulses are stepped-down by a transformer TR, so as to produce at a secondary thereof low voltage AC that is rectified by a pair of FET transistors F1 and F2 driven by a FET driver prior to feeding a low voltage DC load. Respective resonant capacitors Cr1 and Cr2 are connected across the switches and operate as dc-link capacitors which serve as a voltage divider for the half-bridge inverter. The transformer TR has a primary winding (Lp) that is coupled via a current transformer CT in series with the inductor Lr to the common junction between the transorbs T1 and T2. The transformer TR has an auto-secondary winding comprising a pair of windings (Lsec) each having a first end connected at a common junction and a respective second end connected to the drain terminals of the respective FETs F1 and F2. A secondary of the current transformer CT is coupled to the FET driver, which is responsive to zero current flowing through the primary winding (Lp) for switching the FETs F1 and F2. The source terminals of the FETs F1 and F2 are commonly connected to the negative voltage supply terminal of the load to the load via an LC filter comprising Lf and Cf, the positive voltage supply terminal being derived from the mid-point of the voltage transformer secondary winding.
The resonance frequency is defined by the resonant capacitors Cr1 and Cr2 in conjunction with the leakage inductance Ls of the transformer TR. Clamping diodes D1 and D2 are connected across the resonant capacitors Cr1 and Cr2 and serve to clamp the output voltage of the series resonance converter by preventing the build up of negative voltage across the resonant capacitors.
The current drive transformer generates high frequency current pulses that are in phase with the high current primary and serve to feed low voltage gate signals to the respective FETs in anti-phase via the respective auto-secondary windings of the step down voltage transformer TR. The voltage transformer operates as a self-oscillator whose output is substantially sinusoidal and reduces switch losses.
The diodes D1 and D2 serve to clamp the output voltage of the series resonance converter by preventing the build up of negative voltage across the resonant capacitors Cr1 and Cr2. As a result the output voltage of the series resonance converter can never go higher than the network voltage and the maximum power is limited to:
where T is the resonance period.
Having described the circuit topology, we will now explain its operation. When operating at low power, the voltage at the junction of the capacitors Cr1 and Cr2 will be equal to half the supply voltage. If we work at full power, the voltage across each output capacitor is equal to the peak supply voltage. As a result if we now connect say two power supplies in parallel across the DC supply rails 31a and 31b as shown in
It is seen from
In order to understand operation of the system, consider a first situation where in
In practice, the accuracy is a function of the tolerance of the circuit components.
The applicant has found that two power supplies having components of 5% tolerance and operating at within 5% of the resonance frequency, will achieve full load distribution by one of the power supplies providing half the full load power+15% and the other will provide the shortfall equal to half the full load power−15%. Between 50% and full power, one unit takes full power+15% and second unit only takes the rest. Below 50% only one power supply is needed.
If the load is more than doubled, then additional power supplies will need to be connected to the supply rails and all but one will typically operate in clamped mode, any shortfall being taken up by the remaining power supply, which will not be clamped.
Number | Date | Country | Kind |
---|---|---|---|
188497 | Dec 2007 | IL | national |