First the "short circuit" is not at all a short circuit, it is a forward voltage drop (Vf) that typically is either .5 volts or 1.0 volts depending upon the diode type (schottky or silicon junction). This forward voltage drop always finds itself in series with the output voltage in all switching topologies. With output voltages going down to the one to two volt ranges, this drop reduces the efficiency to 65% before the rest of the losses are added in. No one can tolerate those losses. At least no one wants to.

Second is the off state in which diodes perform very well as far as leakage in the reverse direction. That is until you reach the reverse breakdown voltage level where all hell breaks loose. Of course the diodes with the .5 volt drop can only operate with about 50 volts in the reverse direction before they break down. With switching circuits, which have high pulsed inputs, the schottky diodes are limited to outputs of 8 volts and below.

Third, the worst of all, is turn off time. The diode we all imagine, is one that is on or off. Well that is only true until you try to change the state from on to off. Diodes that are conducting like to continue conducting and they nastily become a short circuit when you try to turn them off. After a couple of hundred nanoseconds, they generally let go, but by then the damage is done. In these days of high speed switching PFC circuits, a Mosfet is used to turn off a diode. In this circuit the cathode of the diode is strapped to a large capacitor charged to 380 volts. The anode, which was at 380 volts also, is now asked by the FET to stop conducting and go to zero volts. The diode refuses. It turns into a short circuit. Reverse current from the capacitor flows into the FET. This current can be much higher than the original forward current. Finally the diode says uncle. It starts to turn off and then whack it does turn off. This little 200-nanosecond bout between the Mosfet and the diode create as much EMI as the spark coil on the Titanic. This is why most PFC circuits are smaller than the EMI filters required for them. There are also significant losses in the poor Mosfet, which had to endure high current at high voltage during the turn off process.

Now this may sound as if I am taking it personally and maybe I am. When I was an innocent engineer and didn't know how devious diodes could be, I would predict efficiencies and switching frequencies that would make you gulp. Then diode reality entered. Since that time I have been praying for Diode Relief, and now it looks like it is here.

I met with John Hashismoko in his laboratory in Kawasaki, Japan. John is a research engineer for his company, and the products he develops are normally for flat panel displays. Recently though, John has been working on the development of a new diode. I asked John to describe his research.

John: "I had been testing various coatings to be used as part of our flat panel displays when I noticed that a certain formulation had an interesting phenomenon about it. If it were left to cure in a magnetic field it would become a diode. The direction of the current determined by the field polarization. To measure the quality of the diode, I cured it in a variety of structures and measured each ones characteristics. They were excellent. The forward drop was only 5 millivolts and the reverse recovery time was unmeasurable. Of course these were pixel-sized diodes that could only handle currents in the milliamperes. I though that if we could parallel tens of thousands of them we could produce ideal diodes capable of hundreds of amperes."

John went on to describe the techniques he came up with to build the perfect diode. He starts with a flat aluminum plate. He etches the plate with an etching solution, which is sprayed on with a very fine spray at high pressure. The capillary action of the solution is very low causing thousands of miniature beads to form on the aluminum. After cleaning the plate is left with tens of thousands of small holes where the etch had beaded. The diode formula is then poured over the plate and after a minute a squeegee removes all of it but the quantity that settles into the holes. Next comes the curing with both temperature and the magnetic field. When the formula cures it not only becomes a diode, it also rises forming a meniscus above the hole it has settled into. Next John coats the aluminum with a fine non-conductive anodize which does not cover the risen diodes. A second aluminum plate is then placed over the first, making a diode sandwich. The diode meniscus contacts the plate which then becomes the anode with the lower aluminum plate (the pitted one) the cathode.

A diode sandwich of 1 square inch can pass 10 amperes in the forward direction with only a 5-millivolt forward drop. A ten-inch by one-inch sandwich can handle 100 amperes. The reverse recovery time is less than a nanosecond. Reverse voltage depends upon the hole size but typically is 50 volts or higher. To reach a breakdown voltage of 600 volts, John simply puts 12 sandwiches in series. Vf is only 60 millivolts. This also reduces the diode capacitance to a very low number.

I smiled. Wow someone heard my prayers. Finally a diode that has no reverse recovery time, a forward voltage of less than 100 millivolts and it is made for heat sinking. Soon we will be producing 1 volt, 100 ampere power supplies that are 98% efficient. I asked John when he expects to make these diodes available to the industry.

John smiled and said, " This April 1^st."

About the Author

Frank Greenhalgh has been working in power supplies and systems for 38 years. He has many impressive accomplishments and patents. Over the years he has made significant contributions to Trio Laboratories where he held the position of Chief Design Engineer and was then promoted to Vice President.