THE MAGAZINE FOR COMPUTER APPLICATIONS
Circuit Cellar Online offers articles illustrating creative solutions
and unique applications through complete projects, practical
tutorials, and useful design techniques.

HIGH-TEMPERATURE ELECTRONIC DESIGNS

Part 2: Reducing the Risk
by George Novacek

Start ı Passive Components ı The Future ı The Tradeoff ı Sources and PDF

Last month, I gave you some theory on designs operating at elevated temperatures and the risk involved. At high temperatures, you start getting into an area where regular solder will melt and insulation will disintegrate. Such concerns for high-temperature operation are what Iıd like to discuss this month.

You may find their inner workings fascinating, but unless you work for a semiconductor manufacturer, you have little influence on their specifications and are stuck with what you can buy off the distributorsı shelves. Faced with the need to operate at an elevated temperature, are there suitable devices available on the market, or is there something else you can do?

National Semiconductors produces the LH0021-200, a 1-A power amp in a TO3 metal package rated for ı65ıC to 200ıC operation. Several other big names (Honeywell and International Rectifier, to name just two), produce devices rated well above 125ıC, but some are derated to zero power dissipation at the maximum temperature, so read the specifications carefully. Needless to say, their prices may be staggering. But, donıt lose heart. Youıve already seen that commercial TTL and CMOS devices have been successfully tested at temperatures above their specified limit. You can select devices you will need for a project and characterize them for high-temperature operation yourself.

In his book, High Temperature Electronics, F. Patrick McCluskey gives an example of pushing a common variety IR power MOSFET IRF044 into an operating environment it was never designed for. [1] This device is rated for 60 V and 30 A at a case temperature of 25ıC. Its on-resistance is a 28-milliohm maximum at T_j = 25ıC. The power dissipation is derated linearly to zero at the case temperature T_C = 175ıC. To characterize this MOSFET, the breakdown voltage, leakage current, on-resistance, and V_GTH were measured and plotted over the ambient temperature range from 20ıC to 200ıC. The breakdown voltage measured at 200ıC was 77.5 V, and the on-resistance doubled. The leakage increased from approximately 0.5 nA to 1.1 mA, a whopping six orders of magnitude, but the absolute value of 1.1 mA was still respectable and reasonable to work with. And, the V_GTH decreased from 2.92 V to 1.75 V at 200ıC.

The DC/DC converter shown in Figure 1 with 28-VDC input and 42-VDC/117-W output was built using the above MOSFETs together with similarly characterized additional components. The converter delivered continuous 42 V/117 W with 88.7% efficiency at 200ıC ambient temperature for 1000 h without a glitch, at which point the test was terminated. Diodes D1, D2, D5, and D6 could be the intrinsic MOSFET diodes, but in his book, McCluskey cites a slight improvement in efficiency when using discrete devices. The output voltage of the converter is stepped up using transformer T1, with L1 and C3 not only being the output filter, but also helping to achieve zero-crossing switching of the MOSFETs at the arbitrarily selected 25-kHz switching frequency. To drive home the point that high-temperature operation is not exclusively a semiconductor problem, teflon-coated wires and solderless nickel-plated terminals were used to interconnect the components, which were equipped with hefty heatsinks to minimize the temperature rise of their junctions above ambient.

Characterization of components for high-temperature operation is not the only challenge. No less important is choosing the right circuit topology. The junction, not the ambient temperature is the limiting factor, so besides making sure you can compensate shifts in characteristics caused by temperature changes, you must minimize internal dissipation to keep the junction temperature rise above the ambient temperature as low as possible.

Internal heat dissipation is caused by losses. Remember how I talked about the conduction losses, which are ohmic in nature, proportional to the voltage drop across the junction? To minimize them, devices with low on-resistance operated at the lowest currents possible and minimum leakage must be used. Because leakage increases with temperature, which in turn increases leakage, the danger of thermal runaway always looms in the background.

On the other hand, switching losses are dynamic in nature and result from the simultaneous presence of voltage and current during on and off switching of the semiconductors. To reduce losses, this hard switching has to be changed to soft switching, meaning switching at the point where the voltage and current are both zero. This is achieved by turning the conventional PWM (pulse-width modulated) converters into resonant switches by adding inductors, capacitors, and diodes. The reduction of the switching losses comes at the expense of the peak voltage and current the components must be able to handle. Figure 2 is an example of a common boost converter using an IGBT. The extra C1, L2, and D2 facilitate zero current switching to minimize switching losses. This 28/42-V, 100-W converter was also successfully operated at 200ıC. The test circuit used Unitrodeıs UC1860 controller IC.