I decided that it was time to do something about the mess under my table. And by mess I mean the many 230Vac to 12VDC adaptors for my receivers and pc's. The good thing about it was that most of them run on 12VDC...but not all. My Rohde & Schwarz EB-100 receiver runs on 6VDC. I considered using a LM317 to reduce the 12VDC to 6VDC but with the EB-100 using 1A that would result in a 6W loss which is converted to heat in the LM317. I thought this was a nice opportunity to learn more about the switch mode power supply.
The Buck converter was the way to go. It "down-converts" a higher input voltage to a lower output voltage. So I started searching the net for design ideas and talked to coworkers about it. And since most of my colleagues are power engineers I had lots of help from them. On Texas Instruments homepage I found an old Unitrode IC that I could use. The UC3573 is a Buck Pulse Width Modulator Stepdown Voltage Regulator.
I picked a switching frequency of 200 kHz and a ripple current of 10% of the load current. Selecting the right switching frequency is difficult. To low a frequency and the components are large. At low enough switching frequencies you can hear the switch mode. Too high a frequency will result in very high switching losses but the components are smaller but pcb-lay is critical. A good choice for this project was 200 kHz.The Inductor
With the selected values of 200 kHz and 100 mA ripple current the inductor value is 149 µH. Now I needed to find a suitable core material that I could use to make my inductor. At Magnetics I found a toroid that was just right for the job. The 58310 High Flux powder core would have a value of 150 µH with 41 turns of wire. This is the measured value of the 58310 inductor.
To find the right core material for the inductor it is important to do some calculations. The inductor must be able to handle the load current which in this case is 1A. With that current you can't use too thin a wire. And you must also take wireloss into consideration. If you use too thick wire it may not be possible to get the right amount of turns on your core. If you choose a bigger core the number of turns go down but the weight and the physical dimensions of the inductor will go up. It is also important that the core material does not go into saturation.
First the number of turns (N) was found. The inductors value was 150 µH and the AL values was found in the datasheet to be 90 nH. Now the B field could be calculated with the number of turns and the load current. The value of µ r can be found in the datasheet and in this case it is 125. And µ 0 is a constant. Le is the path length of the core. The B field was found to be 0.1136 T which is within limits of the core material. High Flux powder cores from Magnetics can handle up to 1.5 T before going into saturation.
I have also tried with an 58928 core with 2x 27 turns of wire. See the picture above. This is the measured value of the 58928 core. This resulted in a 0.1% drop in efficiency so I will stick to the old 58310 toroid.PWM Controller
The UC3573 was only selected because it was used in the design example I found in the Power Supply Cookbook. Any number of controllers can be used as long as they can operate up to 100% duty cycle. In this converter a 50% duty cycle is used. A special feature of the UC3573 is a Pulse-by-Pulse Current limit function. A maximum current can be selected using a resistor connected between the CS pin and Vcc. If the current exceeds a selected value the MOSFET is switched off. In the efficiency picture shown below it can be seen that my current limit has been set to about 1.5A.Rectifier and MOSFET
About 60% of the powerloss in a Buck converter is due to the switching rectifier. So to reduce the loss a Schottky rectifier with a very low forward voltage drop was used. And about 10% of the powerloss is found in the MOSFET. This is both switching loss and on-resistance. So I used an IRF5305 HEXFET from International Rectifier which offers an on-resistance of 0.06 ohm and has a total gate charge of 63 nC. I was not easy to find a MOSFET that had both low on-resistance and a low gate charge.Capacitors
The output capacitors is responsible for about 7% of the loss. It is important to select capacitors with very low ESR. I used the CH/CV series from AVX. They are designed for switch mode applications and have low ESR and ESL. The specific value for the output capacitors was found with this formula. The output capacitance for this design was 6.2 µF and I used 20 µF. Both my ripple current and voltage is within the limits of my design specifications.
I did not pay much attention to the input capacitors but used some rules of thumb for the selection. The total value at the moment is 220 µF distributed on two 100 µF electrolytic capacitors and two 10 µF tantalum capacitors. The use of tantalum capacitors in the input is not recommended because they are known to short circuit. The two electrolytic capacitors I used has a voltage rating of 100V. By using a high voltage rating the ESR will be low which is alway a good thing.Compensation network
This was by far the most difficult thing of my Buck converter design. First I tried to calculate the correct values but without much success. I ended up with the good old trial and error. If the values are not right the switch mode begins to oscillate. Another result of wrong component values can be "slow" reaction time to fast load changes.The end results
So what did I end up with...!? The goal was a switch mode that could deliver 1A at 6VDC. As can be seen in the picture below the efficiency reached 90% which is pretty good.
The schematics of the Buck converter is very simple. The important thing is to keep leads as short as possible and having a good ground-plane.References