Last Modified:
 
email dave@allmon.com

This is the PWM controller. The values chosen represent the best fit from the junk box. The LM2524D can operate well beyond the 31.25 kHz this design uses - in fact I had this one running at 62.5 kHz with no other changes than the capacitor hooked to the Ct pin (pin 7, even though the schematic shows it on the second pin 6<g>). The paralleled capacitors on pin 7 are just because I couldn´t find anything but 0.001 caps. I built it with a pot on Rt, adjusted the frequency many times and measured the efficiency and noise, and then picked 31.25 kHz as the best compromise. I then measured the pot to come up with the value for R8, which turned out to be exactly the same value (8215 Ohms) that the 8.2k resistor in the junk box measured.
 

The transformer is made using a 30mm x 19mm pot core. The core is a little large, but I couldn´t get to anything smaller in the allotted time. The core can handle 75 W in a good design. The MOSFETS are good for 100 Vds and 9.7 A continuous. Under full load they don´t warm up more than 5 degrees, so no heatsink is required. The transorbs at D8 and D9 are required. They clamp the inductive kickback from the transformer primary at 51 V, and thus keep the MOSFETS from melting.  The transformer windings are just 1 turn per volt. No bobbin was used. I kludged a form from a 1/2" dowel and some electrical tape, and then slipped the windings off the rod and into the core. The primary is 20 turns center tapped, made from two parallel 26 ga wires, and is good for a bit over 2 A. In practice, with a 175 mA load on both legs, the input current is 600 mA and remains nearly constant throughout the input voltage range of 10 to 15 V.

The rectifiers are Schottky barrier rectifiers. They are good for 1 A continuous. I bought the inductors at All Electronics (I think) some time back. Anything around 100 uH should work fine. Since this is a push-pull converter, and not a flyback or forward converter, the inductor is just a filter - not an energy storage device. The filter has a resonance at around 1.6 kHz, so the feedback MUST be taken off before the inductor. The supply will oscillate, creating a 1.6 kHz ripple of a couple of hundred millivolts if the feedback is taken after the inductor.

All ceramic capacitors are surface mount, and located on the bottom of the board directly across the leads of the electrolytics. I used 0.1, but would rather have used 1 uF ceramics. The electrolytics are all high frequency models. The Tek showed the 4.7 uF tantalum caps to do little or nothing for the noise performance of the circuit.
 

This is the actual voltage regulator circuit. It is fed by the output of the switcher, which is set to 17.5 volts, and has around 6mV of 31.25 kHz ripple. The pi filter here is the same as the pi filter at the output of the switcher. It is essentially a third order Butterworth with a low pass at 1.6 kHz. The 1000 uF electrolytics at the output are storage capacitors.

My Genesis does not use a fan, so there is no constant draw at 12 volts. It is only used for the shutter. I took advantage of that to create a shaped pulse to the shutter. It is rated for a maximum voltage of 48 volts. I am charging capacitor C11 to 17.5 volts through R2. When the opto-isolator in the camera energizes, this 17.5 volts is dumped into the shutter coil. In about 100 mS the charge is drained off, and the voltage on the capacitor has dropped to a value determined by the drops across D1 and R1. R1 is sized so that the voltage across the shutter coil will drop to around 6 volts. That is enough to hold the shutter open. Total power drawn from the Vbatt supply is lowered by half.

When the shutter is released, the voltage on the capacitor charges both from the Vbatt supply, through R1, and then from the 17.5 volt supply through R2. It takes a bit less than 2 seconds to get a full charge, but less than 100 mS to charge up to Vbatt. In focus mode, the shutter spike will not be as high as when imaging, but it will be the nominal 12 volts it was originally designed to be.