This is for how to make a full bridge dual transformer switch mode DC-DC power supply for a car amplifier or for any kind of battery use. Input voltage is 10-16V. Input current can reach up to 100A. Output voltage is adjustable between +/-5V and +/-63V, or from 10V to 126V, depending on how the output is connected. Maximum output current is 8A. And directly at the beginning - WARNING! Making a SMPS is hard and difficult and dangerous and definitely not recommended for an inexperienced diyer! You can easily burn your whole car if not careful. You can also electrocute yourself, if not careful. You can vaporise the amplifier connected to this PSU, if not careful. You can blow your car battery, if not careful. Lethal voltages are present in this power supply! You can kill yourself, IF NOT CAREFUL! I will not take any responsibility of any harm or damage or problem of any kind, caused by making or using (or misusing) or connecting this PSU to your car electrical system or any electrical system or battery, be the consequence of it either a direct or indirect cause of this PSU. By making and using this PSU you accept to take the responsibility to all the possible consequences of it. If you do not understand or accept the possible dangers involved, do not build this PSU! This PSU can be dangerously powerful. Great care must be taken when making and connecting and using this PSU! This PSU, if correctly made, connected and used, has been tested to be fully functional. The SMPS topology is full bridge. This topology has some major advantages over any other SMPS topology especially in a low voltage high power circuit, such as a car amplifier power supply, run from an ordinary 12V car battery. You have probably seen several Push-Pull topology PSUs in the internet that are quite commonly used for car amplifiers. Those are good and usable and simple, no doubt... For lower power... However, advantage of the full bridge compared to other topologies is the high utilisation of the SMPS transformer(s) which results in a very high power output and high efficiency. As high power can never be achieved with any other topology as available from a full bridge topology. I measured a maximum power output of the actual prototype of nearly 800W with an efficiency of around 94%. The reason for this is the super simple transformer. It only needs one primary winding and one secondary winding. And because of the fact that there is only one primary winding, it is almost guaranteed that the transformer resets itself between the cycles - no matter how badly you have managed to make the windings around the core - every cycle will go trough the same winding and will reset the core. Therefore it is suitable topology for a beginner, not familiar with winding a SMPS transformer, although the electronics themselves are quite complex - which in turn is the draw back of this topology. In an ordinary push-pull transformer (which has two primary windings) you must be extremely careful that you make the both primary windings precisely similar. If not, there is a possibility for flux walking, which you want to avoid, because that will result in a total disaster - a crash and burn... An other good way of crashing and burning the whole thing is by transformer core saturation. That happens if the transformer is stressed beyond its core material limits. The first stage in the power supply is the PWM control circuit. The PWM controller used here is an SG3525 or KA3525 or UC3525 all interchangeable, so which ever just available to you, you can have... Google for their data sheets, read and learn. Connect the Feed Back (+FB- on the PCB) + to the positive output terminal and - to the negative. Or to which ever rail you want to monitor closely. Adjust output voltage to the desired level with the 1k trimmer. Next stage is the gate drivers. I used FAN7392, but IR2110 or IR2010 are equivalents. Google for data sheets, read and learn. Then the switches. I used IRFP4004 here. But there are a lot of alternatives. What you need and MUST have is an ULTRA LOW ON RESISTANCE MOSFET. The faster it is, the better, but low on resistance is obligatory. IRFP4004 has a Rds(on) of typically 0,00135ohms. The higher the Rds(on), the worse the efficiency. And you do not want that when operating on battery power... Aim at least to 0,002ohm devices. The devices drain to source breakdown voltages do not need to be very high. But look for a minimum of 30Vds device for 12V battery use... The turn-on and turn-off times should be as short as possible. The shorter the better. The longer the switching times, the worse the efficiency, and you do not... IRFP4004 has a total turn-on and turn-off time closer to 1us, hence the 1us delay (dead time) on the controller. Look for faster alternatives if possible. I have also used IRFP2907 in one hi power full bridge SMPSU. Its advantage is the short switching times. Could be a low price option here... The PCB is for a TO-247 package, but there is no reason you couldn't solder a TO-220 pack in place instead. Note, however, the current limitation of the package itself in a TO220... Usually <75A. There is also an input capacitor. The higher its ripple current capability, the better. The capacitor I used has (only) 20A, look for higher... You want as much current here as you can possibly have. Best would be a capacitor bank - several smaller capacitors in parallel. Make one and connect in parallel with the input terminals on the PCB for the best possible results... Aim to 100A ripple current capability! Even more - if possible... And now is the most fun part! The transformers: SMPS is magnetism and magnetism is advanced mathematics, so I am not going in to that. Google for SMPS transformer, read and learn... There is no such a thing as a off the shelf transformer for this project. You must make them yourself - which is fun! You need a ferrite core, or actually two. You could have a toroid core, but an ETD-core is easier to wind, and you are able to get more power out of it because of the short wire lengths needed around it. Remember; high current, short wire, low resistance = performance! You want... Get ETD34 cores with vertical bobbins (coil former). I use Epcos bobbin and core (N87 material), similar core material from other vendors are 3C90 (Ferroxcube), P4 (ACME), P (Mag.Inc.), PC40 (TDK) for example. The wire? Lots of current at the primary side of the transformer... You need thick wire - or so would one think. But in the fact of the matter, a thick wire is exactly what you do NOT want! Then how to transfer a lot of current trough a thin wire? The answer is FOIL! I am not going into the complex concept of skin effect in a wire in high frequencies. Google, read and learn... And now when you know, go get the foil. Lots of current at the primary side i.e. you want only a short wire for the minimum resistance, i.e. make the primary winding first. The foil? From where? And how? I believe EBAY will be your best friend for the copper foil! Look from the Arts and Crafts department. Or from your local hobby shop! Or at the local hardware store... You will be surprised from how many different places you can get the suitable foil for your high frequency power supply purpose. Almost every copper foil I have run into is usable for a electrical transformer. But to be sure, the foil should be at least 99.9% pure copper (ETP-copper), that is C110 class copper - and therefore definitely suitable here... If you want better, try find C102 (99.95%) or even C101 which will be 99.99% pure copper (OFE-copper). If even that will not satisfy your needs for purity and conductivity, the only choice left is SILVER-foil... Now when you have found a source for the foil, the next question is; what thickness? And the answer is 0.125mm, or 0.005" or 5mills. What ever foil you manage to get, cut the foil to the needed width, which is 20.5mm for the ETD34 bobbin. The foil is thin enough to be cut with normal house hold scissors. Be careful, you do not want to leave any kind of sharp edges on the cuts. Next you need to insulate your foil. One layer of KAPTON-tape is enough. Ebay helps again... You need the tape to be just wide enough to be folded around the edges of your foil - hence no sharp corners at the foil - if the foil layers comes to contact with each other it makes a so called shorted turn to the transformer, which completely ruins the whole SMPS - totally! When the tape is folded around the foil edges you will have double insulation in there but still be extremely careful not to break the insulation. Then how many turns? To make life really simple and easy, I made a calculator for this... The calculator will give the number of turns around the core, both for the primary and the secondary windings. This can also be used to make calculations for a push-pull transformer, it needs two identical primary windings, both with the given number of turns. All you need to know is the following data. The Core Area you will find from the core manufacturer data sheets. This is vital information, find it and use it. The given core area is for a ETD34 core. Check from the core material data sheet the maximum flux density (saturation flux density) at the specified frequency. Reduce it to have a good margin to the saturation level (about 30% down from the maximum is OK) - or use 2000gauss as the target flux density if unsure. Input voltage U in = V Output voltage U out = V Frequency f = kHz (40-100kHz) Core area Ac = mm2 (0,10cm2 = 10mm2) Target flux density B = gauss (100mT = 1000gauss) Primary = turns Secondary = turns For maximum efficiency and power, wind the core FULL of wire! FULL! Use many parallel wires to cover all available window area. Allow about 45% of the available window area for the primary and rest for the secondary, and you should be close to the perfection. I chose 5 turns for the primary and 22 turns for the secondary. For the secondary winding I used in the prototype 4 parallel strands of 0.9mm wire. But much better would be to use AWG23 (0.6mm), total of 9 parallel strands. AWG23 is good up to around 50kHz. On top of everything, if you have done a good job with the winding, there should still be room left for the second primary winding. 5 more turns with the foil and connect that in parallel with the first primary. There, transformer ready. Now go make the other one - exactly identical - EXACTLY! Because of the high output voltage, there are not too many to choose from for the rectifier diodes. Best would be Schottky diodes here, but high voltage - high current versions are rare. So pretty much the only choice will be ultra fast recovery diodes. Find some which have the recovery time the shortest possible. MUR1520 to MUR1560 are also OK here. And the smaller the voltage drop, the better... At the output I put a small capacitor bank with LowESR capacitors and a couple of high dV/dt Rifas (400V/uS) to catch the pulses. MKP or preferably even FKP is OK here... That's the PCB. But that's not everything! Remember the stupidly over killed filtering from the previous amplifier articles of mine? Yes - we are still missing that. Because of the high powers this gives and takes. It produces a LOT of HF-noise - A LOT! So at the input you need a big coil with crazy thick cable a couple of times around a toroid core. Make that 6 to 8 rounds at a T200-26 or something similar with your input cable. 6-8mm2 OK for input cable, as long as you can keep it SHORT. And DO NOT FORGET the fuse! Start with a 50A fuse directly at the battery terminal. DO NOT use bigger fuse than 100A, NEVER! Also the output needs filtering. One coil for the (+) side and one for the (-) side. Rated 8A at least and 1mH minimum, but the more the merrier so 10mH or more recommended. Check the loudspeaker crossover coils for this purpose! A coupled inductor would be the best, but really hard to find high current capable ones in the range 1mH to 10mH... If you find one, get one, or a couple. The control circuitry needs a 0.5A fuse. Put the on/off switch for the whole unit in here. The power stage does not need a switch. You can leave that permanently connected to the battery. That's it! BTW: If you find it really really impossible to make that PCB, I may be able to help you. I can make the PCB for you for 10e/pc, shipping not included. Shipping to Europe 5e. For the rest of the planet - ask... Africa, MiddleEast, FarEast - forget it completely - sorry... Transformers are 15e/pc but shipping costs are running up quickly. You can ask but I can not promise you anything... Anyhow, payment by Paypal... Thanks for reading. And happy powering! And BE CAREFUL! Chris.