I found an interesting experiment on the web, which can be done in home. Now let’s take a view of it:
This is an ongoing narrative of attempting to construct a high-power switching power supply to replace the failed power supply in his Heath Warrior amplifier. Since he use the amplifier for experimental (non-Ham) work, it sometimes has to operate at full CW power for several hours at a time. The original plate transformer is unsuitable for that task, so he is attempting to build a really heavy-duty supply to replace the original power supply.
This design concept started when the author inherited a large quantity of well-built Dell computer power supplies. They were rated for 230 watts, and some testing showed that the switching transformers in the supplies could easily handle 250 watts. Hmm… If there was just a way that you could make these things put out HV instead of +5 Volts, you’d be all set. Well, to work!
OK, let’s start with some pictures…
This is an overall view of the test setup. At the far upper right of the picture, you can see the 16 ampere Variac used to help keep the smoke inside the components. At the far upper left of the picture you can see the outlet box on my 2.5 KVA isolation transformer, which helps keep the smoke inside of you and this test equipment. In the top center of the picture is the guts of the switcher system.
Starting from the left of the unit, you can see the set of six transformers which have been taken from the computer power supplies. The green paint on the connections indicates that you have visually and mechanically inspected the solder connections before applying power. To the right of the transformers is a black heat sink, salvaged from a Pentium-II processor. The power switching transistors are located on the back of the heat sink where the wires connect to the transistors. To the right of the P-II heat sink is a fan, also taken from one of the computer power supplies. It is powered by a 9 volt, 450 MA wall-wart purchased at the thrift store for 50 cents. After some experimentation, I found that a spacing of 1 inch from the heat sink gave the best cooling and the most even heat distribution across the heat sink. Closer or further spacing results in uneven cooling. The fan also sucks some air past the heat sink just to the right of the fan – the one with the “DANGER- HIGH VOLTAGE” label on it. That heat sink came from a cast-off Pentium PRO processor, and now does duty as the cooling device for the 30 amp bridge rectifier you can see bolted to the front of the heat sink. Regarding the bridge rectifier – I only use two legs of the bridge, and the current is low enough that I could have used one of the bridge rectifiers from one of the power supplies instead, but I had this one available.
The few parts on the lower left of the right hand white breadboard are part of the base drive circuitry between the power transistors and the driver transformer, which is seen at the bottom right of the left hand white breadboard. Can you guess where the driver transformer came from? Right!!!
The voltage control feedback circuit will sample the +1500 volts at the output and send it to the TL494 comparator inputs. Over current sense will also be provided to shut down the supply in case of a HV load short or serious overload, such as an internal arc in a PA tube. At the bottom right of the picture is you best tool – my calculator. If the math says it will work, it will!
The smoke has seriously escaped from these power transistors! Running the primary DC voltage at about 400 volts instead of 300 (to see what would happen – now I know!!) and the author think he bumped the breadboard and accidentally hit something that caused both transistors to switch on at the same time. This picture also gives you a close up view of the jumper connections for the transformers. Like the warning labels? He scanned one (from a computer supply, of course) and printed a few of them out and stuck ‘em on the various parts.
Ahh! That’s much better! After replacing the switching transistors, He installed a nice heavy-duty terminal strip to make connections easy and positive for testing. The heavy copper wires help keep the transistor leads cool and are easy to unsolder and bend away slightly if it becomes necessary to replace the transistors (again!) Note the cut-off bare wire from the transformer in the lower left of the picture. it went to the transformer that smoked while testing at high power. The transformer has been removed and dissected to determine the cause of failure.
A close-up view of the main filter capacitor bank. It consists of 12 caps, each rated at 680 MFD @ 200 VDC. They are arranged as a two sets of six of these caps in parallel, with both sets in series. The series set is then charged to +/- 150 VDC through half of a 30 Ampere 600 Volt bridge rectifier directly from the power line. The brown Ohmite power resistor is the inrush current limit resistor. The small “night light” lamp is a self-indicating bleeder resistor. There is one across each set of filter capacitors. The final version of the supply will have two of these lamps in series across each set of capacitors for higher reliability.
The breadboard setup after the +17 Volt power supply and the Switcher Driver circuits had been finalized and assembled on perf boards. note the missing transformer from the main prototype setup. At this point, all six of the switcher transformers are out of the circuit and have been replaced by the single hand-wound prototype transformer visible at the far right of the picture. The scope probe is connected to a current sample transformer that generates a signal proportional to the current from the main switching transistors to the transformer. This will eventually be used to perform an instant shutdown of the system in the event of a serious overload.
The completed +17 volt power supply. A 24 VAC @ 450 MA transformer (yellow leads) drives a bridge rectifier (black rectangle) and charged the big blue filter cap to about +30 VDC. An LM317T regulator mounted on the finned heat sink, produces +17 Volts @ 300 MA for the switcher control board. An LED provided a visual indication that the supply is working. He used a fixed resistance voltage divider to set the output of the regulator, so, no adjustment is supplied. It either works right or it’s dead!
This is the completed switcher control board. The IC is the TL454. The large yellow disk caps and the two resistors adjacent to them provide frequency control, in this case, 50 KHz. (The IC divides by two, so the actual square wave output is at 25 KHz.) The output stage of the IC drives the small transformer (370-9041-CO) and was used in one of the computer power supplies to drive the main switcher transformers. He decided not to reinvent the wheel, so to speak, and swiped the circuit for use here. The two large brown capacitors are 2 MFD each, and along with the diodes (partially hidden under the caps) and the two resistors, form the base drive matching circuits needed to connect the transformer secondary windings to the bases of the main switching transistors. The black and blue wires connect to the switching transistors. The green and yellow clip leads provide the +17 Volts to the board. The two white clip leads are for regulation feedback, and are not functional in the present setup. The small beige potentiometer and the adjacent components are the voltage feedback adjustment network.
A quick-and-dirty lash-up of a homebrew switching transformer. He used a pair of ferrite cores from a couple of large LOPT (horizontal output transformers) stacked side by side and scramble wound what I calculated was the right number of turns on the core. Insulation is a few wraps of Cellophane tape! Certainly not recommended for long-term use, but good enough for a quick test. Each core measures about 0.5 x 0.4 inches thick. The exact core material is unknown, but since these transformers originally operated in the same frequency range as he is using, He thought they would probably work well enough for testing. They did.
The last picture! The prototype supply driving the incandescent lamp load. There are a total of 10 – 100 watt lamps brightly illuminated – you do the math. Note that all that power is coming through the home-brew junk-box transformer he wound. The switching waveforms are visible on the oscilloscope, if a bit hard to make out. The lower waveform is the switching transformer primary current, and the upper waveform is the load voltage. It’s AC, since the HV rectifiers have not been installed yet. Switching frequency is 25 KHZ, and the duty cycle is about 98%.
To read more:http://w5jgv.com/hv-ps/index.htm
