5 Volt Switching (Buck) Regulator

I decided to build this project after experimenting with a cheap buck regulator purchased online from China, which worked OK if run at low currents but if I asked it to deliver more than 1 amp it got very hot. Fitting a heatsink to the regulator IC would have helped, but that was impractical due to it being surface mounted. Some of the other components seemed a bit underrated too. The regulator described here uses the LM2576T and closely follows the design found in the data sheet, so I'm really just showing a method of assembly. At least by building your own you can choose good quality components!

The trusty old 78xx series linear regulator (eg. LM7805) wastes quite a lot of energy as heat, and is increasingly being replaced by the more modern (and efficient) switching regulator, also known as a buck regulator. A switching regulator is basically a high powered square wave oscillator that dumps energy into a coil. When its drive waveform transitions from low to high, a magnetic field builds up around the coil and when the transition goes from high back to low it stops putting energy into the coil and the magnetic field collapses. This collapsing field (or stored energy) causes a voltage to appear across the coil of the opposite polarity. By connecting the negative end of the coil to ground through a reverse biased (Schottky) diode, the positive end continues to supply the load when the regulator is effectively off

Because the switching device (transistor) within the regulator is producing a square wave, it is essentially either fully on or fully off and if it's fully on then it's a short circuit. In theory a pure short circuit cannot have a voltage across it, and if there's no voltage across it then there can be no power dissipated (Power = Volts x Amps so 0 Volts x any amount of Amps = 0 Watts). In the real world though, nothing is perfect and there will always be some losses due to small amounts of resistance with small amounts of voltage across that will produce small amounts of power in the form of small amounts of heat. But with the combination of reduced heat and the regulator being off for some of the time, high efficiency can be achieved

To produce the wanted voltage at the output, the mark to space ratio of the square wave is varied which in turn varies the 'on' and 'off' time of the energy delivered to the coil, and thus the energy produced by the coil when its magnetic field collapses. The resulting pulses are smoothed by the output capacitor and an average voltage is produced. By using feedback resistors within the IC, the correct voltage level can be maintained regardless of the output load. The IC used here is the fixed voltage LM2576T-5 (5 Volt) but variable voltage versions are also available that have external feedback resistors that can be chosen to set the output to whatever voltage you want (within the design limits of the IC of course!)

Matrix board layout (viewed from above)

I usually use veroboard as my preferred method of construction but for this project I chose matrix (or prototype) board instead, mainly because I wasn't sure if the runs of veroboard track would cause problems due to the high frequency switching (capacitance/radiation etc.). The main criteria for the layout is to ensure that the components are located as close to the IC as possible to reduce instability. The circuit shown here was intended to be powered from a solar panel producing 12 Volts and so the capacitor voltage rating was chosen accordingly. I used 35V low ESR types

The final results are pretty good. With a load resistor connected to the output and drawing nearly 2A (with the regulator delivering 5V), the input current is about 1A (running from a 12V car battery). A linear regulator running under the same conditions would be drawing about 2A from the car battery! With the heatsink shown (75mm x 75mm x 2.5 mm aluminium sheet bent at 90) the regulator can supply 1.8 Amps continuously without getting overly hot. I personally wouldn't want to push the circuit much harder than this even though the spec says it is capable. It's always good practice to underrun everything and keep well within limits

Design Notes:

While I was experimenting with this circuit it became apparent that the type of 1000uf capacitor used at the output was quite important, not so much for its smoothing ability but for its ability to stay cool. The first one I tried got quite hot which I was not happy about and so I looked for an alternative with a higher ripple current. The one I used in the final design was the Panasonic EEUFR1V102 (FR series) which has an excellent ripple current rating of 2.6 Amps

The Schottky diode shown in the data sheet is a 1N5822 which is difficult to mount onto the board as its legs are very thick. So instead I used an MBRF1645 which has a higher current rating and legs that can actually go through the holes!

I was also curious about the 100uH choke that I obtained online and so I unwound it to see how it was constructed. I don't know the type number of the ferrite ring, but it's Yellow with an outside diameter of 13mm, an inside diameter of 7.5mm and it has 50 turns of 0.6mm (23 SWG) enamelled copper wire wound onto it. The way the coil is wound takes a bit of explaining. It has 30 turns all the way around its circumference, which with 0.6mm wire just about fits with with no gaps. Then, the windings carry on in the same direction but go back the way they came on top of themslves for another 10 turns, ending up half way round the 30 turn coil where the two tail ends finish opposite what looks like a gap in the winding... well there is! The total length of wire used is just under 1 metre