A heavy duty 13.8V power supply is a fine thing to have in the shack, but unless you acquire one secondhand, is an expensive little beastie to buy. This means building one should be considered, not only for the cost savings, but also because you can brag about it on air to your mates. Of course, careful consideration must be given to the properties of the completed supply, and after talking to a few of my friends who have built their own and fallen into all the traps, here are the printable ones : RF proof, easy to make, commonly available parts used, but above all CHEAP.

Well, last things first. Breaking down the construction costs of a heavy duty regulated supply, they are in order:

Dealing with these in turn, we can reduce the cost greatly by rewinding a microwave transformer (about $A5 total), scrounging old computer grade electrolytics (lots around), and designing the electronics to be so RF proof that a wooden case can be used - yes, that's right - wooden! If you are really stuck for a dollar, then good supply regulation and overload protection also allow all metering to be deleted. Finally the wooden case allows 1/4 inch bolts and washers to be substituted for expensive terminals. If you can't put the whole thing together for less than $A50 then frankly you don't even qualify for the junior scroungers league.

Moving on to the other points, manufacture is easy as no etched PCB is used. The PCB is simply made by using a hacksaw to cut through the copper overlay on the PCB material breaking it up into separate pads. Details are given in the drawings.

Keeping the supply RF proof is another matter entirely. During development, several designs were tested, based around such chips as the 723 regulator, the 3140 op amp. and a 7912 three terminal regulator with bypass transistors. In all cases, the high gain of the control amplifier forced the use of a PCB with a ground plane to which everything was heavily bypassed. This technique limits RF interference and also prevents motorboating and high frequency instability (a common problem in high current circuits such as power supplies and audio amplifiers) as the ground plane acts as both an RF shield and a single point ground.

However, for home construction, the use of a double sided PCB is undesirable and anyway, the performance of each of these circuits is totally over the top. After all, ham rigs powered from 13.8 volt are designed for use in a car where the supply voltage wanders all over the place. Two volts of variation is quite typical. Regulator circuits which hold the output voltage constant within a few millivolt for all conditions of load are simply not required. It is much more important that the output voltage is free of noise and ripple, and the published design does this very well. Noise and ripple are well under 5 millivolt peak to peak, and output regulation (no load to full load) is around 200 millivolt. A simple control circuit is used without overall feedback and the result is a cheap, very stable design. RF proofing is provided by physically earthing the heatsink, and also by using it as a ground plane. The collectors of the TIP3055s are physically earthed to the heatsink (no micas), and so a good section of the circuit is actually at earth potential. Two other advantages are easy assembly and excellent heatsinking.

General Comments

The circuit of this power supply can be broken into two parts.

The details published provide a transformer and rectifier structure capable of providing 8 amps DC continuously (and short current peaks up to 20 amps). This is sufficient to adequately power SSB transceivers with 100W PEP outputs, BUT WILL NOT power a transceiver providing a continuous carrier of greater than about 40 watts.e.g. AM, FM, continuous key down morse, single tone SSB testing etc. Demanding a continuous output of more than 8 amps will result in the transformer secondary overheating, with a possible fire risk. The reason we can get away with a supply with an 8 amp continuous rating is simply that speech is very "peaky" data, and so SSB has the odd high power peak but a very low average power level (usually about 20 -30% of peak value). It is on this basis that transformer and heatsink sizes are usually selected for domestic hi-fi equipment.

The winding resistances of the transformer have been very carefully chosen to avoid excessive current peaks which will cause failure of the 35 amp bridge rectifier. If you wish to use a transformer with higher current ratings to make a continuously rated 20 amp supply, you MUST use more heavily rated diodes with peak repetitive surge current ratings of around 800 amp.

The second part of the circuit (filter caps and regulator) will supply 20 amps continuously and can be used unchanged in a heavier duty supply. The transformer, rectifiers, and regulator circuit all generate large amounts of heat, and fan cooling MUST BE PROVIDED for safe and satisfactory operation.

How It Works

The first section of the circuit is the transformer, rectifier, filter capacitors and bleed resistor which turns the incoming 240 volt AC into roughly smoothed DC. At first glance, this is a simple circuit and so the operation is rarely discussed anywhere in detail. However, being a very high current supply, this really is a different can of worms and needs to be completely understood, if for no other reason than to prevent rectifiers failing and electrolytics either overheating or exploding. Those of you who are already planning to increase the output current above 20 amps should read the next section very carefully as you won't find it in any common text.

The circuit operates by topping up the charge stored in the electrolytic capacitor every half cycle via the rectifier. Under load, it is desirable that the AC ripple voltage existing across the filter capacitor is kept low and for this to occur, the recharging must occur in a very short time just before the peak of the cycle. If the ripple allowed is 10%, then very roughly the recharging must occur in around 10% of the half cycle length. If the average current delivered by the supply is 20 amps, then the average current during the recharging period must be around 200 amps. This huge current peak must be tolerated by both the rectifier and filter capacitor. Another way of looking at this problem is to regard the charging current spike as 20 amps of DC together with a large AC current superimposed on top of it. This AC component flows through the filter capacitor internal loss resistance causing large amounts of heat to be generated. For these reasons, filter capacitors used in high current supplies have three published ratings, capacitance, maximum DC operating voltage and RMS ripple current rating. If reference is made to the famous Schade curves for rectifiers (see references), for the ripple percentages used in this design (10%) the relationship between the RMS AC ripple current and DC output current is around 2.5 This means that the filter capacitors used here must have an RMS ripple current rating of a least 50 amps. Capacitors with these sorts of ratings are physically large, to provide the big surface area necessary to get rid of the internally generated heat. I would recommend one computer grade 100,000 microfarad 25 volt aluminium electrolytic around 140mm long by 75mm dia. or 2 @ 47,000 microfarad 25 volt aluminium electrolytics around 105mm long by 75mm dia. Smaller capacitors must not be used unless you have specifications which clearly show that they have ripple current ratings at 40 deg. C of at least 50 amps for a single unit or 25 amps for each of 2 units.

An even worse situation occurs at switch-on of the supply as the electrolytics are fully discharged and represent a short circuit. If this should happen at the peak of the cycle, enormous peak currents flow and the principal thing that limits the peak current is the winding resistance of the transformer primary and secondary. (It is not surprising that the house lights blink!) The 35 amp bridge used in this design has a single surge rating of 475 amps and in order not to exceed this rating, a particular wire gauge has been selected for the rewinding of the transformer secondary (2.5 square mm). Under no circumstances should this be varied.

So much for the operation of the simple part of the circuit. The next bit is the constant current source (TIP41C) and 6.8 volt zeners. This part of the circuit reduces the ripple existing across the filter capacitors by around 70 db to produce a clean stable reference voltage of 14.5 volt. Current flowing through the 1K5 resistor forward biases the two 1N4004s producing an almost ripple free voltage of 1.4 volt across the base emitter junction and 15 ohm emitter resistor of the TIP41. Thus 0.7 volt exists across the 15 ohm resistor, setting the collector current of this transistor to about 50 ma. Most of this current flows through the two zeners, further reducing ripple and producing the 14.5 volt reference potential (and yes the zeners are 6.8 volt but this is measured at a test current of 5ma, not the 50ma used here).

So the power supply output voltage is 13.8 volt, due 0.7 volt being lost across the base emitter junction of the TIP2955.

The last part of the circuit consisting of a TIP2955 and five TIP3055s is really just a big compound emitter follower. At very low output currents (less than 7 ma), the only transistor supplying the output is the TIP2955. This is because there is insufficient voltage existing across the 100 ohm collector resistor to turn on the TIP3055s. However, once this limit is exceeded, the TIP3055s progressively turn on, supplying whatever current is required. The five emitter resistor sets of 0.22 and 1 ohm simply ensure that the total output current is equally shared by each of the 3055s. At a current of 4 amps through each 3055, or 20 amps total, 0.7 volt exists across each emitter resistor combination, turning on the BC548 which then starts to shut down the constant current source. This limits the maximum output current to 20 amps. By the way, don't try to cut out any of the 3055s. If you check the specs, you'll discover that the maximum current a 3055 can handle with 20 volt across it (output short-circuit) is 4.5 amps.

Making The Transformer

You can use a professionally wound transformer which produces 15V RMS at the rated output current, or you can roll your own.

Click here for rewinding details

Assembling The Supply

The first thing to do is gather all your bits together and design and make your box. Remember that a fan is mandatory (you can get a good quiet one from an old computer power supply) because the iron in a microwave transformer is flogged to death to keep both costs and weight to a minimum. Unventilated, they get very hot after about 30 minutes.

I made my case using 19mm chipboard for the base, 5mm thick 3-ply for the front and rear panel, and 3mm masonite for the lid top and sides. The front and rear panels of the case were drilled to accommodate switches, meters, fuse holders, etc. Then 12mm square timber was nailed and glued around everything but the bottom edge of these panels to provide a timber frame for the lid retaining screws. The completed panels were then nailed and glued to the base. The lid of the box was assembled using 12mm square timber at the junction of each of the panels. Everything was both nailed and glued for strength. A pattern of air holes was included at the front of both of the lid side panels to ensure that good ventilation was obtained.

All of the components on the heatsink were then assembled (see diagram). Finally, all components (case bottom, heatsink, transformer, electros, front and rear panel bits, etc. etc.,) were married together to produce a unit ready for final wiring and testing.

Wiring and Testing

Simple enough really - use the left over 7 x 0.69mm wire for all the high current wiring (see the heatsink diagram) and thin plastic covered multi strand wire for all the rest. Wire up the transformer, rectifier, filter cap, and bleed resistor first and test the assembly. Watch your rectifier and electrolytic capacitor polarities like a hawk. If things go wrong, they will do so in a big way. Next, complete the voltage reference circuitry and test that (14.5 volt across the zeners). Last, add the super emitter follower and test the completed supply. A 60 watt headlamp bulb makes an excellent load. Testing the current limit is not easy and involves laying your hands on a 0.5 ohm 300 watt resistor. Do not just short the supply terminals and hope. If the current limit does not work the damage will be awesome. With the 0.5 ohm in circuit, 27 amp will flow if the limit is not working and the output voltage will be 13.8 volt. If the current limit is working, the terminal voltage will be around 10 volt and the current around 21 amp.

BRIEFLY connect the 0.5 ohm and see what happens. Do not leave the output shorted for longer than 60 seconds. Remember that the power supply design has been optimized for SSB operation and designed to supply an average drain of 8 amps.

I made my 0.5 ohm resistor from nichrome wire reclaimed from an old bar radiator and immersed it in a bucket of water. Steel wire of around 1mm dia obtainable from your local hardware shop for picture hanging could also be pressed into service.

Adding More Muscle

The supply can be relatively easily extended in capacity - here are the stops. First, throw away the mickey mouse 35 amp bridge and replace it with some heavy stud mounted diodes, e.g. BYX52s or similar which have peak forward current ratings of 800 amps or more. These will need to be mounted on a decent sized heatsink. Alternatively, 2 **IDENTICAL** 35 amp bridges can be paralleled. Use a transformer from a 1kW microwave and rewind the secondary using heavy multistrand wire with an area of around 4 to 6 square mm. The outer sheath of old RG-8 or RG-213 coax. with its 200 or so conductors flattened, stretched and covered with paper masking tape is ideal. Add filter capacitors as necessary to get the appropriate capacitance (50,000uF per 10 amps) and ripple current rating (25 amps of ripple current rating at 40 degrees C for every 10 amp of DC output). Add TIP3055s at the rate of one 3055 for every 4 amp additional output. Drop the 15 ohm emitter resistor in the constant current source to 12 ohm (30 amp). Beef up the heatsinks to suit your application.

Have fun and try not to liberate the magic smoke which makes all electronics work.

References and Further Reading
A Final Comment

Note that a wooden case does not prevent radiation of the stray magnetic field surrounding the transformer. This can induce mains hum in microphone leads etc so keep your supply and tranceiver at least 1 metre apart...