| Electronics > Projects, Designs, and Technical Stuff |
| Low power isolated DC/DC converter |
| << < (2/2) |
| Zero999:
I consider 200mW of idle power dissipation to be quite bad. I suppose in the applications where I've used these DC:DC converters, quiescent power wasn't a consideration, so I didn't even bother checking the specification on the data sheet. Out of the data sheets you've linked to in your first post, the LT3001 seems like the it's the most suitable for your application, because it's optimised to lower supply voltages. The LT8300 will work, but the higher voltage rating of the internal MOSFET, means it also has a much higher on resistance, so is less suited for low voltage operation. I wonder how well it would work with a centre tapped transformer? There are also lots of unregulated transformer driver ICs available, but then you'll want something to regulate the secondary voltage. Linear regulators are probably out of the question, as you want low power dissipation, so a transformer with two secondary windings, each powering a buck regulator, could be used. https://www.analog.com/media/en/technical-documentation/data-sheets/3439fs.pdf |
| Kleinstein:
One point to watch with ready made DC/DC bricks is capacitive coupling and common mode signal injection. Some of the cheaper ones are really bad in this respect (probably due to using a really tiny transformer). The injected common mode signal is very hard to filter in a really isolated application. One can combine the push pull configuration with an additional series inductor (like in the maxim picture) on the secondary side. This way the PWM ratio can also be use for voltage regulation (a little like a buck converter, using the same switches). The old style TL0494 can work this way - chances are there are more modern lower power versions available. For low power loss chances are it gets better with a relatively low frequency and a slightly larger core, as this allows for more spacing / isolation. If the primary side uses USB anyway, it could be possible to power the converter from the USB side instead of batteries. As USB tends to be noisy anyway, another switched mode converter from an extern 6-12 V to 5 V if need may not make things much worse. |
| SiliconWizard:
You can also take a look at XP Power products. Not bad stuff. Just a note. The 6V min input voltage will be a bit on the low side to find a suitable converter with a +/-15V output. https://www.xppower.com/Products/DC-DC-Converters |
| jaromir:
I spent another evening with this. At first I made another transformer for flyback. The same wire, core and bobbin as for previous one, primary is 30 turns, secondary 60 turns. I built flyback with BC337-40 as before and results were not exactly convincing. The circuit worked, but efficiency was around 60%. I noticed the same thing as mentioned in my first post about push-pull converter - the emitter current continued to flow after base voltage decreased to zero - see https://snipboard.io/dZzK8Y.jpg After replacing BC337 with MOSFET IRLD120 this delay disappeared. I knew BC337 isn't stellar transistor, with not exactly great switching times, but I really underestimated it. With new transistor in place I was able to achieve higher efficiency. With 23,5V on output, feeding two parallel 4k7 resistors - that is 235mW of output power - I had consumption of 30,5mA from 10V input (305mW), meaning efficiency 77%, being a bit lower at 6V. I call it acceptable result, though I'm sure there is a room for improvement. Efficiency doesn't include feedback and control, though. The transformer has a few pF of capacity between primary and secondary. With better choice of transistor I returned back to push-pull circuit. I was able to get better efficiency than before (around 75%). After employing rectifier with accumulation choke I was able to achieve quite wide and more linear output control range, but since flyback has simpler transformer and overall circuit, I'm going the simpler way. Now I'll wait for LT8300 and LT3001 to arrive, but this simple flyback sounds like good backup plan. Thanks all for comments. |
| T3sl4co1l:
What was your circuit? Remember that base charge needs to be removed, just as gate charge needs to be removed from the MOSFET. Usually a voltage divider is used, to get a logic-level input while providing a discharge path for the base. A speed-up cap can be placed in parallel with the top divider resistor to yank some charge out immediately. A delay of 100ns should be easily achieved this way. :-+ Tim |
| Navigation |
| Message Index |
| Previous page |