??? 03/07/08 07:56 Read: times Msg Score: +1 +1 Informative |
#151986 - Some principles learned the hard way Responding to: ???'s previous message |
A flyback converter in this power class can be somewhat demanding to design and implement.
I have done several such converters the smallest ones having only a few watts of output power and the biggest for linear HF amplifier in kilowatt class. The suggestion for using a mosfet is very good. However, select the mosfet carefully. There may exist higher voltage spikes than the 100 volts. I would use a 500 volt type here. Same applies to the current, in Your case a safe value would be in the neghbourhood of 10 amps (continuous). You do not need any heatsink if the ON resistance of the FET is low. This is particularly true for high current types. You will need a good drive for the mosfet gate. If the FET is of high voltage/ high current type then there will be a large capacitance on the grid. It can be several nanofarads - which is large because You want to switch the thing on and off rapidly. If the switch is not fast You will end up generating heat. As a matter in fact when the circuit is working properly the charge/purge of the grid capacitor may generate more heat than the actual work current. Connect a 2..22 ohm resistor in series with the grid and then clamp the other (drive) end of this resistor to supply lines with shottky diodes. This is because the PCB will generate extra oscillations because of the fast transients. It is not nice to have the FET switching back on after just switching off because of these oscillations. The schottky diodes "eat" the oscillations. Layout is VERY critical here. There are some very good IC:s to take care of the PWM as well as the MOSFET drive. If You don't need variable output voltage then a dedicated IC is propably the best bet here. Those are simple to use and there are a lot of experience on how to use and how not to use them. One example of such IC is UC3825. Although this controller is for push-pull topology it runs just fine in flyback configuration. If You are using a flyback topology with an 'E' shaped transformer then You should consider leaving an air-gap between the halves. This will make the inductance lower but it will make the thing harder to saturate. If You end up saturating the thing then You are basically generating heat somewhere as the energy is not transferred to the output. And guess what - it is usually not the FET that heats up - it is the transformer ... fried several of them and those smell VERY bad ;) The flyback or "booster" diode should ofcourse be one of the fast recovery types. For lower voltages a schottky diode is OK but in Your case it might be hard to find a type that survives the voltage. Note that this diode is going to generate some heat - it may be the only component requiring a heatsink. Pay close attention to PCB layout. Make all connections as short as possible and use separate power and signal ground planes. It is NOT a sin to use jumper wires as those conduct better than the copper on the PCB. Sometimes You can get very good results by soldering a thick copper wire on top of existing PCB strip. It is a lot better solution than using tin alone. |
Topic | Author | Date |
flyback boost | 01/01/70 00:00 | |
Isolated? | 01/01/70 00:00 | |
Isolated flyback ? | 01/01/70 00:00 | |
Hence the name? | 01/01/70 00:00 | |
Yep | 01/01/70 00:00 | |
The same circuit | 01/01/70 00:00 | |
Clarification | 01/01/70 00:00 | |
Design Approach | 01/01/70 00:00 | |
Using Iron Powder core | 01/01/70 00:00 | |
The frequency depends ... | 01/01/70 00:00 | |
Faraday had something to do with this.. | 01/01/70 00:00 | |
Stored energy | 01/01/70 00:00 | |
Curious: what if? | 01/01/70 00:00 | |
Some principles learned the hard way | 01/01/70 00:00 | |
My observation.... | 01/01/70 00:00 |