Circuit Corner - Issue 10
In this issue, I expand on Issue 9's coverage of current sense transformers with some practical advice. I also present some results from one of my more involved design projects: a simple, inexpensive boost regulator.
Practical Current Sense Transformers
This is easily fixed, however. Readers familiar with solid state charge pumps will recognize the current steering design. On every negative-going pulse, the capacitor takes on a value equal to the inverse of the voltage of the negative-going pulse, thus tying it to 0V.
The first circuit works well for sense transformers with a high sensitivity, where the voltage drop from a diode does not substantially effect the reading. Schottkey diodes, with their lower voltage drops, are very useful in this arrangement.
The second circuit uses a "superdiode" arrangement to eliminate the voltage offset. This circuit is exceptionally precise, but at the cost of an opamp.
Simple Boost/Flyback Converter
Before I had a chance to publish that circuit, however, my research continued. I had been tinkering with a two-transistor arrangement for motor PWM and LED control. I had almost abandoned the project, as it was plauged with interactions between the various components, but while looking for a circuit to boostrap a large converter that I have been working on, I struck gold.
By eliminating the timing loop entirely and replacing it with an off-time loop driven from the flyback phase, I had managed to get the two transistors to reliably oscillate, turning off when the overcurrent limit tripped, being held off by the flyback effect of the inductor, and turning back on when the flyback effect ceased.
While that circuit was indeed quite useful for the task at hand, it had no voltage regulation. It would put out a constant peak current into the output, regardless of how much power was being drawn. While this is fine for some applications, like motor PWM or LED driving, when used to fill a capacitor, it would fill it until the point of catastrophic failure. It occurred to me that I could add a suppression signal by wiring a zener diode from the existing summing junction on the turn-off transistor to the output voltage. The resulting circuit worked, providing something resembling a regulated supply from two transistors.
That circuit also had problems. The regulation, such as it was, operated in a "bang-bang" mode, turning either fully on or fully off, causing substantial voltage ripple. Added to that was the disadvantage of using a zener diode: it requires a different diode for every voltage output, and zener diodes tend to only exist in a few specific sizes in the average junk-box. I lacked a diode of the size I needed, and that was enough reason to keep going.
Serendipity kicked in at this point. I decided to try using a standard one transistor voltage regulator cell instead of the zener diode. What I had forgotten to anticipate was that the softer characteristic of the regulation would apply a DC bias to the summing junction. This had the effect of lowering the peak current as the voltage rose, and led to very smooth regulation during operation. The resulting circuit is what I present here.
The two leftmost transistors form a sort of astable multivibrator, where the inductor sets the time constant. The middle transistor acts as the main switch for the regulator, while the left one handles the turn-off conditions. The PNP transistor forms a voltage regulator, with a voltage roughly set by 0.5V/1k*(100k+1k). In this case, the output voltage is roughly 51V. Changing the 100k resistor can yield a variety of other voltages, although this is likely the highest voltage that can be safely achieved with these components. The circuit runs in a self-tuned intrinsic discontinuous conduction mode, and so long as the inductor is rated for the peak current, it is free of saturation effects.
The 100k resistor sets voltage and the 10 ohm resistor sets maximum current (1.5V/10ohm=150mA). I ran it from a 9V battery, bypassed with a .1uF capacitor.
I built mine using a pair of 2N2222s and a BC557B. I had first tried using a BC547B for the left transistor, but it performed badly, so I swapped a 2N2222 in place. As is often the case when I do these things, I really should have paid more attention to the symptom, which cropped up again some time later. As a result, I have updated the diagram to incorporate a slight correction in the positioning of the summing junction diode. The new positioning allows the shutdown transistor base to be rapidly drained by the inductor recoil while also preventing a nasty glitch that would hold the shutdown transistor on through the entire cycle, thus effectively shutting down the converter.
This circuit can also be adapted to function as a flyback, simply by adding a flyback winding to the inductor, and feeding it into a capacitor through a diode. Turns ratio determines the flyback voltage versus the boost output voltage.
