by Antony Chazapis
During a discussion with fellow radio amateurs, someone brought up the subject of linear power supplies. My friend Kostas, SV1GSU, said he had a half-finished one. I offered to help him complete it, so I could see the layout and the construction process in practice. I have read a lot on linear power supplies and always wanted to build one, but was kept off by the cost of the large input transformer, which may exceed the price of a new switching device. Kostas said that back then, when he had ordered the transformer for his supply, he had two of them made and if I helped him finish his project, he would give me the other one to start mine! Too much motivation.
As I got the power supply home, I realised it was mainly a bunch of components. This was an abandoned project, started over 10 years ago, when Kostas – a now avid DIYer – had little experience. Aside the dust and corrosion, it had some first-time mistakes. The output transistors were not properly attached to their heatsink and the enclosure’s bottom plate was bent from the enormous weight of the transformer. Moreover, I thought that the wooden box meant to guide the air flow was not necessary, the bridge rectifier needed better cooling and the 2N3055 output transistors should be replaced with the more powerful 2N3772 devices, to better handle the output current (also used by Giorgos, SV1CDY, in his own build). The bridge rectifier, regulator board, and all cables were in bad shape.
First step was to break down the existing construction to its components and decide what to keep. The heatsink was actually two parts. They were separated and cleaned along with the enclosure, meters, and everything else that did not end up in the trash can. Then, before putting the power supply back together, the bottom plate was straightened and reinforced with a piece of MDF. Reconstruction started from the AC side: the power connector, fuse, switch and transformer. I drilled the front panel to mount the switch and covered the fans with protective grills. Power up produced AC voltage at the other end of the transformer.
Next step, the bridge rectifier and capacitors. Both required new mounts. I attached the bridge rectifier to a new heatsink and fabricated two simple capacitor mounts from hose clamps and L-shaped mounting brackets, with one side cut to half length. I connected the transformer to the bridge using screw terminals, powered up, and got nearly 32V DC at the capacitors. Note that the transformer’s secondary is rated at 24V.
Four new 2N3772 output transistors and one 2N3055 were screwed on the old heatsinks using silicone pads for both insulation and better thermal transfer. The two heatsinks were bolted on a strip of L-shaped aluminium, in order to be able to mount them as one module on the bottom plate. The 2N3055 was wired to drive the 2N3772s, new, 10W output balancing resistors were added, and all necessary cables were constructed and attached. For regulation, I made up an LM7815-based circuit, on a small piece of perforated board. A pair of capacitors at input and output (10uF and 100nF at both ends), and a red LED to indicate it’s working. Examples of using an LM78xx device to regulate high current power supplies can be found here and here. I placed the output module in the enclosure, connected the regulator board, powered up, and to my surprise – it’s always a surprise, isn’t it? – got 13.8V at the output! That would be 15V minus about 0.6V for each pass through the driver or an output transistor.
As a finishing touch, I put a bleeder resistor across the output, to discharge the capacitors when powered off. The selected value of 100 Ohm produces a nice effect on the voltage meter, that gradually drops down to zero, a second after turning off the switch.
The supply was given back to Kostas, who promised to take good care of it and make it even better, starting from adding a soft-start circuit to limit the inrush current. Linear power supplies are relatively simple in design, but – as I found out – demanding in the construction process. The most difficult part of building one is to shape the enclosure to accommodate the necessary switches, dials and connectors, and mount all components properly. That’s why no two linear power supplies are the same. Each bears the unique signature of its builder. Overall – as every construction project, but this one even more – a very rewarding and educating experience.
Now, a giant input transformer sits on my desk and is waiting for its turn…