This invention relates to the control of LEDs.

The replacement of Cold Cathode Fluorescent Lamp (CCFL) by Electro- Luminescent Diodes is becoming more common due to the lower cost of LED solutions as well as superior performance in terms of contrast.

In typical modern LED illuminated displays (TV, notebook), several LED strings are used to cover the entire display surface. These strings are placed either on the edge of the display (edge lit) or directly behind the display (direct lit).

The typical block diagram of a LED backlight driver (with a boost converter) is shown in Fig 1 .

A DC/DC converter 10 is built to provide the high side voltage of the LED string (V ₀ U _T ) whereas a current source 12 supplies the LED with the appropriate current (L _ED )- The voltage at the low side of the LED string 14 (V _FB ) is controlled by a feedback loop ensuring sufficient headroom for the current source while minimizing the power dissipation across this current source (better efficiency). If V _FB is below a reference voltage V _REF , the DC/DC converter output power is increased to increase V ₀ UT and thus V _F B-

In order to control the brightness of the LED, a pulse width modulation of the LED current is used. This modifies the average value of the current and thus the brightness of the LED.

Figure 2 shows the LED current pulse width modulation, and shows the PWM signal as the top plot and the average current as the bottom plot.

Current products on the market use two different basic configurations:

(i) One dedicated power conversion stage (AC or DC/DC boost or buck or fly-back) and a feedback loop per LED string.

(ii) One common power conversion stage and a feedback loop for several LED strings. The solution (i) is best optimized in terms of efficiency as well as heat dissipation as the dedicated DC/DC boost will adjust V ₀ UT to get the lowest V _F B for each and every LED string. This approach is shown in Figure 3, which essentially comprises multiple circuits of Figure 1 in parallel. The disadvantage is the cost due to a higher number of used components as well as the board area occupation.

For cost reduction reasons, the solution (ii) with one DC/DC boost and feedback loop for several LED strings can be preferred. The lowest V _F B is extracted and used to control V ₀ UT- This solution is for example described US2009230874, from which Figure 4 is taken.

The drawback of the solution (ii) is twofold:

(i) Heat dissipation

The LED string forward voltage mismatch can cause severe heat dissipation for the current regulators. As the system will adjust V ₀ UT for the channel with the highest LED forward voltage drop, V _F B(VT) for the other channels will be higher than the minimum regulated value due to LED string mismatch.

For explanation a system is assumed with a minimum set-point for V _F B(VT) of 0.7V, a nominal LED string forward voltage of 80V and 3% mismatch maximum between strings, 4 LED strings to be driven with 300mA current per string and 50% duty cycle PWM.

The IC for this system will have to dissipate 0.5 ^* 0.3A ^* (0.7V + 3 ^* (0.7V +

80V ^* 3/100)) = 1 .5W in worst case mismatch instead of 0.42W in case there is no mismatch in the LED string forward voltage.

This heat limitation limits the number of LED strings that can be supplied by one power converter and/or increases the number of IC to be used due to the heat dissipation capability of available packages. All of this increases the overall cost of the application.

Figure 5 shows a generic circuit diagram, for a single controller 50 controlling a set of strings 51 of LEDs. A feedback block 52 is shown generically. In the prior art method this is a minimum selector as explained above.

The example figures above are inserted into Figure 5 to show the effect of the LED string forward voltage mismatch.

(ii) Speed of operation As the DC/DC clock speed is limited to a few hundred of kHz for practical reasons (component size, signal speed, switching losses), the loop speed is limited. In case independent dimming signals are used (called 2D or local dimming), switching from a low forward voltage string to a high forward voltage string can cause V _F B of the latter string to be below the operating limit of its current source. The LED current of the latter string may then be well below the desired value.

As a practical example, consider that in Figure 5 only the channel 3 is on. The forward voltage Vf is 80V*97%=77.6V for this string and the feedback control loop will react such that V _F B3 is 0.7V (no other string is on). V ₀ UT will then be adjusted to 78.3V (0.7V+77.6V). If now channel 2 is switched on, knowing that the forward voltage of this string is 80V, there is V _FB 2 = 78.3V - 80V = -1 .7V. The current source of the channel 2 can not deliver any current under this condition before the loop increases the DC/DC power transfer to get V ₀ UT reaching 80.7V.

This can take a long time due to the slow speed of the DC/DC converter and no light will be emitted by the string 2 during this time.

The invention is directed to these two problems.

According to a first aspect of the invention, there is provided an LED driver for driving a plurality of strings of LEDs, comprising:

a controller for controlling a voltage or a current flow applied to one end of each of the strings of LEDs;

a plurality of threshold circuits, wherein each threshold circuit is for comparing the voltage at the other end of a respective string of LEDs with a respective threshold voltage; and

a combiner for combining the threshold circuit outputs thereby to generate a signal for controlling the controller.

This aspect of the invention provides a shared controller between multiple strings of LEDs. This reduces the amount of circuitry required. In order to improve the efficiency of the system, individual threshold circuits are provided so that the voltage control function takes into account the individual requirements of the different strings of LEDs, and does not simply react to the lowest voltage present. The threshold circuits enable voltage boosting to be delayed until the voltage boosting really is required to maintain the desired voltage levels elsewhere in the circuit.

Each threshold voltage for example comprises the voltage at the other end of the string of LEDs in response to the same voltage (V ₀ UT) applied to the one end of the string of LEDs. Thus, the threshold voltages take into account the differences in response of the different strings to a given drive voltage.

The driver can further comprise:

a current balancing arrangement for each string of LEDs, wherein the other end of the string of LEDs is connected to the current balancing arrangement through first and second paths, the first path comprising a heat dissipater,

wherein the current balancing arrangement is part of an integrated circuit, the heat dissipater is external to the integrated circuit, and the current balancing arrangement is adapted to control the splitting of the current through the LED string between the first and second paths.

This arrangement enables heat dissipation to take place off the integrated circuit, and in a controlled manner, so that the generation of the required integrated circuit biasing voltage is assured, and excess current can then be dissipated.

According to a second aspect of the invention, there is provided

an LED driver for driving a plurality of strings of LEDs, comprising:

a controller for controlling a voltage or a current flow applied to one end of each of the strings of LEDs in dependence on voltages measured at the other ends of the strings of LEDs;

a respective current controller for each string of LEDs for controlling the current in each string of LEDs independently;

a current balancing arrangement for each string of LEDs, wherein the other end of the string of LEDs is connected to the current balancing arrangement (88) through first (80) and second paths, the first path comprising a heat dissipater (RHEAT),

wherein the current balancing arrangement is part of an integrated circuit, the heat dissipater is external to the integrated circuit, and the current balancing arrangement is adapted to control the splitting of the current through the LED string between the first and second paths.

This provides the external heat dissipation aspect outlined above.

The invention in its two aspects provides solutions to solve the two above mentioned issues. The invention offers solutions having only one power conversion stage (AC or DC/DC buck or boost or fly-back) supplying several LED strings, each of which can be driven independently (for example by a dedicated independent PWM dimming signal) without facing the current source saturation as well as the high heat dissipation due to LED string voltage mismatch.

The benefit is total system cost reduction by removing the constraint on

LED string matching and using fewer chips. The two aspects above can be employed independently to each other. The heat reduction/dissipation concept can be implemented without having the LED string forward threshold voltage setting and vice-versa. The invention provides an approach which gives advanced dynamic headroom control for a multi LED-string driver.

The current balancing arrangement can comprise a feedback path for maintaining constant a biasing voltage for the integrated circuit. This assures the integrated circuit biasing. Energy is then only dissipated in the external resistor if this is possible while maintaining the biasing voltage. Thus, the headroom is controlled dynamically.

The heat dissipater can comprise:

a single resistor; or

a parallel set of resistors each having an associated control transistor; or a series set of resistors each having an associated control transistor.

The use of multiple resistors enables the control transistors to pass between their linear and saturated modes of operation in a staggered manner, and this can improve the efficiency of the system. In this case of a set of resistors, the associated control transistors are each between a terminal of the associated resistor and a common terminal (VISRC), and the current balancing arrangement is adapted to control a first transistor in the second path and the set of control transistors. Thus, both paths are actively controlled. The first aspect of the invention also provides a method of controlling a plurality of strings of LEDs, comprising:

controlling a voltage or a current flow applied to one end of each of the strings of LEDs;

comparing the voltage at the other end of a respective string of LEDs with a respective threshold voltage; and

combining the threshold comparisons and controlling the controller based on the combination.

Determining the threshold voltages can be by:

applying an increasing voltage to the one end of the strings of LEDs;

monitoring the voltages at the other end of the strings of LEDs;

when the voltage at the other end of all of the strings of LEDs has reached a minimum threshold, storing the voltages at the other end of all of the strings as the individual threshold voltages for the different strings of LEDs.

The second aspect of the invention also provides a method of controlling a plurality of strings of LEDs, comprising:

controlling a voltage or a current flow applied to one end of each of the strings of LEDs in dependence on voltages measured at the other ends of the strings of LEDs;

coupling the other end of each string of LEDs to a respective current balancing arrangement which is part of an integrated circuit, through first and second paths, the first path comprising a heat dissipater which is external to the integrated circuit; and

controlling the splitting of the current through each LED string between the first and second paths.

The invention also provides an LED circuit comprising the driver of the invention and a plurality of strings of LEDs. Each string preferably comprises a plurality of series-connected LEDs, but a string should be understood as being able to include a single LED.

Examples of the invention will now be described in detail with reference to the accompanying drawings, in which: Figure 1 shows a typical block diagram of a LED backlight driver with a boost converter;

Figure 2 shows the LED current pulse width modulation,

Figure 3 shows a known design with one dedicated power conversion stage and a feedback loop per LED string;

Figure 4 shows a known design with one common power conversion stage and a feedback loop for several LED strings;

Figure 5 shows an arrangement with one common power conversion stage and a feedback loop for several LED strings and which can be controlled in accordance with the invention;

Figure 6 shows an example of selector circuit of the invention which can be used in the arrangement of Figure 5;

Figure 7 shows a threshold voltage setting process of the invention;

Figure 8 is used to explain a heat dissipation reduction concept of the invention;

Figure 9 shows a first implementation of the circuit of Figure 8;

Figure 10 shows simulation results of the heat dissipation concept implementation;

Figure 1 1 shows the power dissipation for three values of heat dissipater; Figure 12 shows a modification to the circuit of Figure 9;

Figure 13 shows the resulting LED current accuracy improvement based on the approach of Figure 12.

Figure 14 shows a further modification to the circuit of Figure 9 with multiple parallel resistive paths;

Figure 15 shows simulation results for the multi-path implementation of

Figure 14;

Figure 16 shows a further modification to the circuit of Figure 9 with multiple series resistive paths;

Figure 17 shows simulation results for the multi-path implementation of Figure 16. The invention provides a driver design in which a single controller can be used to drive multiple stings of LEDs, but in one aspect with dynamic threshold voltage setting so that the individual characteristics of the LED strings can be taken into account in the voltage control loop. In another aspect, excess energy is dissipated off-chip in a dedicated heat dissipater, and the routing of current to the heat dissipater is controlled dynamically such that a desired integrated circuit biasing remains stable.

The invention can be applied to the circuit of Figure 5, which shows a system with 4 LED strings and 4 independent dimming PWM signals, one per channel. The 4 strings are supplied by a unique DC/DC boost stage 50. The concept is valid for any number of LED strings and type of power conversion stage.

The first aspect of the invention can be applied by using a more intelligent feedback system in the feedback unit 52. In particular, the feedback unit 52 can be implemented as a selector circuit for setting suitable threshold values for each LED string.

Figure 6 shows an example of the selector circuit 52 of the invention, and shows how two LED strings are processed, for simplicity.

As shown in Figure 6, the single boost controller 50 receives signals from the two strings, each of which has a separate threshold circuit 62,64. Each threshold circuit 62,64 supplies a threshold value to a comparator C1 , C2, which also receives the feedback voltage V _F B from the LED string.

LED string forward voltage detection

In order to ensure that there is no distortion in the current shape due to mismatch between the feedback voltages V _F B when switching strings, this mismatch must be included in the V _F B threshold voltage used for the feedback loop.

First a charging phase is launched when the power converter is activated. All the LED channels are forced to be on at the same time and having the feedback loop regulating on the lowest V _FB . During this phase, called phase 1 , the V _FB voltage of each string will be compared with a common minimum threshold Vthjnin. The multiplexers (MUX) of Figure 6 set with Vth_n (with n, the string number) to Vthjnin.

Thanks to the energy transfer by the DC/DC boost converter, V ₀ UT increases continuously and consequently the V _F B voltage of all channels. As soon as the lowest V _F B exceeds the minimum desired value (Vthjnin), i.e. the output of all comparators C_n has been set HIGH, the DC/DC is stopped (stop DC/DC is set HIGH) and the threshold voltage setting phase occurs (phase 2).

During this phase, the actual V _F B voltage value at the feedback pin of each string is captured and stored. Figure 6 shows this capture and storage operation by the use of a sample and hold circuit "S/H" but other implementations are possible such as the use of an A D converter along with a digital register.

At that time, the V _F B voltage of the string with the highest forward voltage drop (Vfjnax) is equal to Vthjnin. For the other strings the V _FB voltage is equal to Vthjnin + (Vfjnax - Vf_n), with Vf_n their own forward voltage. The V _FB threshold voltage includes now the relative mismatch between the strings.

After this second phase, the IC will enter the normal operation phase. The multiplexers MUX set their threshold levels Vth_n to the stored V _FB voltage value which will be used for the control of V ₀ UT-

This threshold voltage setting process is shown in Figure 7.

As explained above, after power up in step 70, the process comprises generating an output voltage in step 71 , then comparing the feedback voltage V _FB of each LED string with the minimum threshold Vthjnin (step 72). While the output voltage rises, the comparisons continue (feedback path 74) until the value of V _FB for all strings has reach the Vthjnin value. The set of V _FB values for all strings is then stored in step 76 and the comparator thresholds are set accordingly, in step 78.

In this way, individual reference values for V _FB are found for each LED string.

With this method, it is ensured that V ₀ UT will always be at the minimum value while ensuring sufficient voltage headroom for proper operation of all current sources at any time. The DC/DC loop time constant does not need to be shorter than the LED current pulses to avoid LED current distortion. Heat dissipation reduction

The second issue to solve is high heat dissipation within the chip due to the LED string forward voltage mismatch. Assuming a system with a minimum V _F B threshold voltage of 0.7V, a nominal LED string forward voltage of 80V and 3% mismatch maximum between strings. If 4 LED strings are to be driven with 300mA current per string at 50% duty cycle PWM (see Figure 5) for all of them. The IC will have to dissipate 0.5 ^* (0.7 ^* 0.3 + 3 ^* (0.7 + 80 ^* 3/100) ^* 0.3) = 1 .5W instead of 0.42W in case there is no mismatch in the LED string forward voltage.

Figure 8(a) shows the heat dissipation reduction concept. The LED string is now fed with two parallel current paths. The first path 80 contains an external power dissipation element RHEAT between the LED string 82 and the IC (the components below the line 84). In case of low V _F B voltage, this path may not be able carry the full LED current while ensuring that the IC has sufficient voltage headroom to operate. The second path 86, linking directly the IC to the LED string, provides the remaining current.

The IC contains a current balancing element 88 as well as a current generation and control element 90. The current balancing element 88 is aimed at driving the maximum possible current in the first path 80 while ensuring sufficient voltage headroom for the current generation and control element 90. It will also divert the remaining current into the second path. The current generation and control element 90 supplies the required LED current and performs the dimming operation based on a dimming control signal.

In this way, the required current is provided, and the voltage headroom is ensured by the second path 86. The required heat dissipation takes place outside the IC.

There are different possible implementations of the current balancing and current generation and control. Figure 8(b) shows an example of block diagram.

A feedback loop manages the split of the generated LED current into 11 (second path 86) and I2 (first path 80) in order to maintain the voltage at the node VisRc constant. The higher the V _FB voltage value, the higher the 12 current value. It means that when V _FB increases, the power dissipation will be shifted to the external resistor R _HEAT instead of the IC.

The current balancing element 88 has a comparator in a feedback path so that the voltage V _I S _R C is constant, by selectively controlling transmission gates which select between the currents 11 and I2. This voltage is in this example a biasing voltage for the IC.

These transmission gates are shown in more detail in Figure 9.

Figure 9(a) is a standard implementation with feedback control of the voltage V _I S _R C but no heat dissipation circuit (i.e. no current splitting control).

Figure 9(b) shows the implementation with the current balancing heat reduction circuit 88 of Figure 8. The current balancing circuit comprises an amplifier A1 in a feedback path to set the voltage V| _SRC , and two transistors M1 and M2. The current balance is controlled while maintaining the voltage at the VISRC node equal to a constant value V _S AT- If V _F B increases, the loop will increase I2 in order to maintain V _I S _R C constant. V _SAT will be chosen close to the minimum headroom voltage required by the current source in order to minimize the heat dissipation.

The proper setting of this implementation requires that the transistor M2 is always in the linear region, meaning that it acts as a resistor. The size of M2 must be chosen as high as possible (minimizing its resistance R _D SON) to take full benefits of the heat reduction effect. The optimum value of R _HEAT depends on the maximum V _F B voltage as well as the current LED-

Assuming RDSON_M2 is small as compared to RHEAT and knowing that VISRC equal to V _SAT thanks to the regulation loop, this gives:

RHEAT

I\ = ILED - ^VFB - ^{VSA T}

RHEAT Eq. 2 V _FB must always be higher than V _SAT to ensure proper operation of the current source. This will be ensured by that feedback loop with the power conversion stage with Vthjmin set to a value higher than V _SAT (see Figure 6).

The power dissipation (P_wohr) inside the IC without the heat reduction circuit is:

P _ wohr = ILED * VFB _Eq 3

With the heat reduction circuit, part of the power dissipation is done with the external resistor R _HEAT (I ₂ ^2* R _HEAT )- Then the power dissipation (P_whr) with heat reduction in this case is

P_ whr = ILED * VFB - 12 ² * RHEAT P _V kr = ILED > VFB - ^(VFB - ^{VSA T)}

RHEAT Eq. 4

As shown by Eq. 1 , Eq. 2 and Eq. 4, when V _FB is low (close to V _SAT ), most of the current is balanced to the 11 branch and in that case the power dissipation is similar with or without the heat reduction implementation.

With the heat reduction concept, when V _FB increases, I2 will increase accordingly and the power dissipation inside the IC is reduced ( Eq. 4). The higher the V _FB increase, the higher the IC power dissipation reduction. This has for effect to maintain almost constant the power dissipation inside the IC.

As I2 can not exceed L _ED , if V _FB continues to increase above the limit ILED ^* RHEAT+VSAT the power dissipation inside the IC increases again. The power dissipation reduction reaches its maximum value which is R _HEAT ^* I _LED ² - The table below shows the power dissipation reduction formulae as a function of V _F B-

The derivative of Eq. 4 gives

dP_ whr _ILED VFB - VSAT)

dVFB RHEAT aP_ **r ILED ' RHEAT

dVFB 2

VFB = ^ILED * ^RHEAT ₊ VSAT

² Eq. 6

By combining Eq. 4 and Eq. 6, the maximum power dissipation in the region where V _FB ≤ILED ^* RHEAT+VSAT is

P whr max = ILED * VSAT + RHEAT * U £ Ly

- - 2 Eq. 7

This maximum power dissipation is reached when 11 = I2 = I _LED /2 and its value increases with RHEAT-

Figure 10 shows simulation results of the heat dissipation concept implementation.

The left hand side graph shows a comparison of the heat dissipation without (plot 100) and with (plot 102) the heat dissipation concept with 12Ω R _HEAT , 250 mA LED and 0.33V V _S AT- The calculated P_whr_max is 0.27W at 2.1V V _FB from Eq. 7.

Due to the R _D SON of M2, the simulated value is slightly shifted (0.32W at 2.35V). The V _FB limit for maximum heat dissipation reduction is calculated to be at 3.33V VFB (see table above) as compared to 3.9V from the simulation (impact of M2 R _D SON)- The simulated maximum heat reduction is 12 x 0.25 ² = 0.75W as calculated.

The right hand side graph shows the current through M1 (plot 104), M2 (plot 106) and the total LED current (plot 108). Including the M1 and M2 R _DSON , the minimum V _FB for the current source to operate correctly is 0.9V (I _LED reaches its nominal value).

Figure 1 1 shows the power dissipation for 3 values of R _HEAT : 9Ω (plot 1 10), 12Ω (plot 1 12) and 15 Ω (plot 1 14) with 250mA l _LED . A trade-off is to be found between the extension of the region where the IC power dissipation is maintained constant (high value of RHEAT) and the maximum IC power dissipation (low value of RHEAT) in this region.

As explained above, the transistor M2 should be operated in the linear region. However, the heat reduction feedback loop in the example above is controlling only the transistor M1 . For the V _F B voltage above V _S AT + V _D SSAT_M2 + ILED*RHEAT, the transistor M2 is no longer in the linear region and the entire LED current is going through M2. M1 is off and the loop is no longer controlling V| _S RC- VisRc is then equal to VBIAS - V _G S_M2, where V _G S_M2 is the gate-to-source voltage of M2. This can change depending on the temperature and process variation. Due to the high amount of LED current and low VSAT value (necessary to minimize the IC heat dissipation), the output impedance of the current source is quite low (few tens of Ohms). The variation of V| _S RC will induce variation of the LED current.

Thus, to improve the LED current accuracy for the whole range of V _F B, the gate of M2 can also be controlled. This can be done with a double control feedback loop as depicted in Figure 1 2.

The gate of transistor M1 is controlled by an operational transconductance amplifier 1 20. The output is connected to the gate of transistor M 1 , and through resistor R1 to the gate of transistor M2. The gate of transistor M2 is also connected to the bias voltage VBIAS through resistor R2. The current out of the OTA goes through R1 and R2 and thus will control the gate of both M 1 and M2.

The loop gain for M2 is to be made smaller than M 1 (R2 < R1 ) to take full benefits of the heat reduction. A capacitor C1 is in parallel with resistor R1 and bypasses at high frequency the pole made with R1 and the M1 gate capacitance, in order to improve the loop stability.

The result of LED current accuracy improvement is shown in Figure 1 3.

The left plots show the currents of M 1 and M2. Plot 1 30 is the current through M 1 without the double control loop and plot 1 32 is the current through M 1 with the double control loop. Plot 1 34 is the current through M2 without the double control loop and plot 1 36 is the current through M2 with the double control loop. The effect on the LED current is shown in the right plot, where 138 is the LED current without the double control loop and plot 139 is the LED current with the double control loop.

As indicated by Figure 1 1 and the above formulas, increasing RHEAT allows to extend the power dissipation rejection zone but with the price of an increased maximum power dissipation (P_whr_max). In order to have both benefits, i.e. lower P_whr_max and large power dissipation zone, a variable resistor RHEAT can be used which is low value for low to medium V _F B (decreasing P_whr_max) and high for high V _F B (extension of power rejection zone).

For this purpose, several resistive paths (R-paths) 140 in parallel can be used as shown in Figure 14. Each of the R-paths 140 uses an increasing value of RHEAT- Figurer 14 shows four external heat dissipation resistors RHEATI to RHEAT4- For low values of V _F B, all the transistors of the R-paths are in the linear region and thus the equivalent RHEAT value is equivalent to all RHEAT values in parallel. For increasing V _F B, the different transistors of the R-path are shifting progressively into the saturation region and fewer paths have transistor in linear region. The effect is an increasing equivalent value of RHEAT for increasing V _FB . The more R-paths, the flatter the power dissipation curves.

Figure 15 shows simulation results for the multi R-path implementation of Figure 14. For low V _FB , MRP1 , MRP2, MRP3 and MRP4 are in linear region, the equivalent RHEAT is RHEATI // RH EAT2 // RHEAT3 // RHEAT4 (the four resistors in parallel). When V _FB increases, MRP1 switches off followed by MPR2 and then by MRP3 (this is shown in the middle graph) leading to a progressive increase of the equivalent RHEAT- The left graph of Figure 15 shows the IC power dissipation for different V _FB values. Plot 150 is without the heat dissipation concept, plot 152 is for a single R- path and plot 154 is for the 4 R-paths implementation.

RHEAT in the case of a single R-path is 12Ω and for the 4 R-path, RHEATI = 3/6 ^* 4 ^* 12 Ω , RHEAT2 = 4/6 ^* 4 ^* 12 Ω , RHEATS = 5/6 ^* 4 ^* 12 Ω , R _HE AT4 = 4 ^* 12 Ω.

The middle graph in Figure 15 shows the current passing through the direct path (11 ) - plot 156 - and through the resistors (together forming I2) - plots R1 to R4 - as a function of V _FB and for the four R-paths in the 4-path implementation. The right graph in Figure 15 again shows the current passing through the direct path (11 ) - plot 156 - and through the single resistor (I2) - plot 157 -as a function of VFB and for a single R-path implementation.

A variant with heat dissipating resistors placed in series as arrangement 160 is also possible as shown in Figure 16. In this configuration, RHEATI , RHEAT2, RHEAT3 and RHEAT4 are set equal to 12 Ω IA. Figure 17 shows the same plots as in Figure 15. The left graph adds the power dissipation curve 155 for the series resistor arrangement. The middle graph is the same as in Figure 15, and the right graph shows the current contributions from the resistors in the series arrangement (for comparison with the middle graph).

The invention enables is a significant cost reduction of the total solution. The reduction of the heat dissipation within the IC will allow reducing the number of used components, ICs as well as discrete components.

Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measured cannot be used to advantage. Any reference signs in the claims should not be construed as limiting the scope.