Minimising power loss with dynamic power control
For example, a digital-to-analogue converter (DAC) needs to supply up to 20 mA to a user-defined load in the 100Ω to 1 kΩ range. In this case the minimum supply voltage must be 20V. The maximum power supplied by the DAC is V × I = 20V × 20 mA = 0.4W. If a 1-kΩ load is used, all of the power is consumed by the load, resulting in no lost power. A 100-Ω load consumes only 0.04W, so 0.36W is wasted or dissipated on chip. In some cases, a 0-Ω load is a valid condition, resulting in all power being dissipated on chip.
In a 64-pin LFCSP package, the maximum ambient temperature cannot exceed 125°C. With four channels each dissipating 0.4W, the total power dissipated is 1.6W. The thermal impedance of a 64-LFCSP package is 28ºC/W. In the previous example, the temperature rise is PD × θJA = 1.6W × 28°C/W = 44.8°C. Therefore the maximum safe ambient temperature is only 80.2°C. Heat sinks can be added to overcome this problem, but this may not be viable due to the required space and cost.
Dynamic power control (DPC) directly addresses this issue. A dc-to-dc converter boosts a 5-V supply to create a 7.5-V to 29.5-V supply. This boosted supply powers the DAC current output driver, which delivers the required power to the load. With a 0-Ω load, the output of the DC/DC converter is 7.5V, its lowest value. The maximum power dissipated in the DAC is only 7.5V × 20 mA = 0.15W, saving 0.25W compared to the original solution.
With DPC, the maximum power dissipated by four channels (each short-circuited to ground) is 0.6W. The temperature rise is PD × θJA = 0.6W × 28°C/W = 16.8°C. Therefore the maximum safe operating temperature increases to 108.2°C. DPC provides the most benefit in systems having a wide undefined load range, high channel density, and high temperatures that leave little room for large power losses.
The AD5755 four-channel 16bit digital-to-analogue converter provides voltage and current outputs for programmable logic controllers (PLC), distributed control systems (DCS), and other industrial process-control applications. Dynamic power control regulates the voltage on the output driver, minimising power dissipation with low-value load resistors—and easing thermal management. Each channel can be configured to provide:
Voltage output, with 0V to 5V, 0V to 10V, ±5V, or ±10V full-scale range and ±0.04% total unadjusted error (TUE);
Current output, with 0 mA to 20 mA, 4 mA to 20 mA, or 0 mA to 24 mA full-scale range and ±0.05% TUE.
Offset and gain can be individually programmed for each channel. The devices can be used with the on-chip 5 V, ±5-ppm/°C reference or an external reference. Available in a 9 × 9 × 0.85-mm 64-lead LFCSP package, it is specified from –40°C to +105°C and priced at $13.65 (1000).
When the current output is enabled, VDS of the output FET is sensed. This voltage controls the MOSFET in the power control block to regulate VBOOST, which in turn controls VDS as determined by the output current requirement. With the MOSFET switched on, the inductor charges to a value determined by the difference in the actual value of VDS and the required value. When switched off, the inductor discharges into the capacitor and VBOOST pin. This process repeats on each clock cycle. There is one dc-to-dc converter per channel.
About the author
David Rice isan Applications Engineer at Analog Devices, Limerick, Ireland. He holds a Masters in Engineering in Embedded Systems from Cork Institute of Technology, Ireland.
To download the PDF version of this article, click here.