Power supply

Lets build a power supply, target 50V 10A, challenge, most parts come from the scrapyard, some parts had to be purchased, other parts are from hammarkets.

  • Frame without a top plate and a backplate : scrapyard
  • Toroidal transformer 1KW which I estimated from its size because there is no documentation,  input 2 times 110V, output 2 times 43V and 18V (unused), powerswitch : scrapyard
  • A 100 Volt, 22000 micro farad capacitor : hammarket
  • Netzteil + other capacitors, rectifier, 2N3055’s, power diodes : Reichelt
  • Two 220V fans : hammarket
  • A high voltage LM317HVK in TO-3 case: Ebay (UK)
  • 10 mm Pertinax plate, scrapyard
  • Nuts and bolts, 1.5mm2 copper wire, hinges,  aluminum/plastic piping : all comes from the hardware store
  • Top / Back plate 2mm aluminum including the work to drill two 76mm diameter openings for the fans: metal workshop in the city.
  • All heat sinks come from the scrapyard or hammarkets. Just find something with a rating of about 1 to 2C / Watt.

The circuit is not that difficult, nothing fancy here, not drawn is a power switch and a 10S 385K varistor over the power plug:


The blue netzteil from Reichelt is a delay switch that you install before the transformer, it reduces the transient current caused by induction of the toroidal transformer when you turn it on. Essentially the blue block is a series resistor that is short circuited after 60 msec. The image below is my test setup for any unknown transformer that I get, it is a variac where the primary has a manual protection against run-up induction currents. In that case the resistor is a lamp in series with the primary of the variac which you short circuit with a push button automatic fuse (called a Stotz, this is the 1960 version). If you don’t use the Stotz then the transient induction current goes in the net, usually it blows the fuse on the meter case in the hallway which you want to avoid. The transient of a 1Kw toroidal transformer does the same, so this is why you need the blue netzteil from Reichelt which is a modern solution for the problem.


The 100 Ohm bleeder resistor over the 22000 micro Farad initial capacitor is on a separate heat sink (maybe it is a bit overrated), a 0.05 Ohm series resistor is mounted to the bottom plate of the case behind the first capacitor, it is there to allow reduce a transient caused by charging and discharging a large capacitor, that is, within 5 seconds it is stabilized at the required 60V (approx). Rapidly discharging any capacitor over 10 micro Farad at 60V is an issue, if you want to see sparks then start here. So the bleeder resistor is a must for safety.

The LM317HVK regulator is a TO-3 version that can the used up to 60V, don’t get them in China, buy them here (thus second hand on eBay). The LM317HVK with the rectifier go together on a separate heat sink, and the 2N3055 transistors are as well on a separate heat-sink. Both diodes are 10Amp 1kV, and all capacitors are rated at 100 Volt. The measured ripple reduction is better than 10mV with a load of 2Amp or 100W. I will still try to find a solution to test this supply under higher loads.

The weight of this supply is mostly caused by the transformer, the top plate has two hinges and is closed with six Parker screws. Also, you need air circulation because of the heat dissipated in the power supply. All wiring is 1.5 mm2 copper, clamp screws on 10mm pertinax slightly elevating above the base plate. Also the transformer has a 8 mm isolation from the case to reduce stray volatage on the case, it is mounted with two screw threads to the base plate. Pertinax is not used in modern designs because it is hygroscopic, epoxy resin is the preferred solution. But this is the reason why you find pertinax on scrapyards.
The blue block is the Netzteil from Reichelt, Toroidal transformer with two 43V secondaries in parallel, the high amp regulator,  a steel LM317HVK on the same heatsink, top center is the 22000 micro Farad capacitor, to the right the bleeder resistor on a separate heat sink, on the third heat sink, the largest, you find both 2N3055 transistors, it is a current amplifier.
The way it looks when the case is closed.

The ripple voltage to current ratio for this circuit follows from the equation:

du = I / C * dt;  Vrms = du/2*0.7

where dt is 10 msec for a 50 Hz powergrid, Vrms to be measured at the input pin of the LM317 and the common collector of the 2N3055, it will become 0.8 Volt at a load of 5 Amp and it scales linearly. The power dissipated by the 2N3055’s is the difference between 60V and 50V times the current used, so at 10Amp a total of 100Watt is dissipated while 500W is provided to the load. The measured ripple at the output of the transformer is a lot less than Vrms because of the LM317 regulator.

Open points to be improved or tested in this experimental design are:

  • What is the temperature doing at high loads, for this I need to build (or find) a regulated load to test it up to higher currents. Without the fans in a open case the heat sink of the bleeder resistor goes up to 90C which is still acceptable although it is too hot to touch, the bleeder resistor can handle up to 50 Watt and it is dissipating 36W. With the fans in a closed case the measured temperature at the backside of the heatsink stays under 40C at a room temperature of 18C. The other heat sinks don’t really get warm at a 2Amp load.
  • Don’t try to short circuit this power supply, even with a fuse it will kill both 2N3055’s. PH0BAS (Bas) has a solution for me and this will be the next improvement.


Performance specs of this power supply:

  • V1 measured after the 0.05 Ohm resistor is depending on the power provided by the supply, the relation is V1 = -0.925 A + 57.5 Volt, at 10 Amperes we end up at approximately 48 Volt.
  • The ripple measured at 55V after the 0.05 Ohm resistor is 0.5 V rms at 3 Ampere
  • The ripple measured after the 2N3055’s is 10 mVolt at 50 Volt and at 3 Amperes, so the regulator is doing what it is supposed to do.
  • The thermal resistance of the small heat sink on the north east side is 1.88 C/w, probably 1 C/W is the thermal resistance of the large heat sink. More than 100 W dissipation in the 2n3055’s looks like a no go.

This looks like a 500W lab power supply, don’t think this transformer can handle a lot more. Added a fuse holder before the 0.05 Ohm resistor, so this is what we currently have:


Still to do

  • add a current limiter
  • add a pot meter / and or selector to the front panel

Last update: 17-Nov-2019 6:45

3 thoughts on “Power supply

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.