Duration: 2:20

22 views

1

DC-DC converter chips generally have 2 types of control loops: 1, peak voltage and 2, current sensing; The peak voltage limits the output voltage to a celling or fixed voltage level. The current sensing is done so by a small shunt resistor and this tiny voltage is amplified using an Op-Amp and the output is combined with a ramp generator to create a cycle to cycle duty cycle value. At 500kHz, this is done so 500000 times in a second. If a higher current is sensed and higher voltage is detected, this will either lower the duty cycle or resets the ramp generator. Both these 2 control loops work together to regulate the output voltage; easy said than done. The stability of such circuit requires good function transfer responsiveness by the so-called Error Amplifier Compensation circuit, which in this case is an internal one done so via the internal mosfet's source terminal.

Trying to fool/bypass such feature by grounding this input works, but the system is less stable with a non-matching inductor for the load. Ideally, the duty cycle drops if a picked voltage has been reached, but unfortunately due to unknown reason (probably due to the driver not being designed for a high voltage MOSFET here or maybe the so-called soft start feature limits the initial duty cycle to a lower percentage), the voltage doesn't step up more than 165v regardless. 500kHz is proven to be too high for low power high voltage converter. A 4.7uH inductor is needed, but this also implies a huge current source, which most 5v power supply can't handle. By switching at lower frequency, a bigger inductance inductor can be used and so the input current can be reduced. 47uH is about ideal, and so it's been used on many of my projects. To reduce switching noise, a 470uH can also be used, but at this value, to reduce DC loss, the inductor will be much larger in size. so again, 47uH is about ideal in terms of switching frequency, switching current (5v power supply), physical size, and last but not least, cost.

PS. This is pro7, an antique in term of prototype boards, but while it's considered a failed prototype, a lot has been learned from it to make it worth the time and money.

PSS. I can be very wrong. The measured results are as correct and I can manage. How the results are interpreted is a matter of opinions rather than facts. Without using more sophisticated measuring equipment, it's just guess.