| Electronics > Projects, Designs, and Technical Stuff |
| Isolated Flyback converter in boost mode |
| (1/2) > >> |
| shangaoren:
Hello everyone, I'm looking to make an isolated power supply for a jbc tip (isolated to not interfere with thermocouple in the tip) I would like to output around 160W at around 30V with an input going from 9 to 25V, it's for a heating element so the stability is not a critical part. As far as i know isolated flyback seems to do the stuff but i'm not an expert in that domain : Is there some already done design that match my needs ? If not what should i choose an all integrated controller or design my own with a microcontroller ? What transformer should i take ? |
| nick_d:
It's really not too easy, I know having spent many hours trying to do it (switched to other projects, will return to it). The best advice I can give is buy a PS matching those specs from China. Although, do look inside before installing it. Otherwise, this is what you do: 1. In Digikey parametric search choose your controller. Many have built-in power MOSFET but I think with your power requirement you may have to use an external MOSFET. 2. Design your transformer, I used a procedure from a book and wrote a Python program to carry it out. It involves basically figuring out how much energy the transformer needs to store and your maximum DC resistance and so on, then choosing a bobbin, core and wire of sufficient capacity. 3. Read datasheet for controller carefully and build one of the examples in it. It will help you calculate resistors etc. 4. Pay particular attention to the feedback network. This is rather tricky as it has an opto-isolator driven in its linear region. There is a part that is used for this, something like a buffered reference with error amplifier or similar. Being clever I tried to make do with a less complicated circuit. Bad! 5. The SMPS IC and the feedback IC will have to be compensated, since there is delay through the loop and without feedback it will oscillate. In your application you can tolerate slow response to step-changes in input voltage or load conditions, so be generous with the compensation. Otherwise it's tricky and perhaps best done by experiment, as I don't know of a good mathematical procedure so far. Still keen? Keep us posted :) cheers, Nick |
| shangaoren:
Thank you for your clear explainations. I need to fast results so i will not use flyback for this iteration but i'm curious, i will try later ;) |
| SiliconWizard:
Your requirements are a bit... much! Especially the input voltage range considering the output voltage and power required. Did you realize that getting 160W output power from a 9V source at say, 80% effiency will draw over 22A! Better have big cables/copper traces/connectors... here. If you don't want to set your house on fire. ;D You can take a look at Murata, RECOM or XPPOWER isolated DC/DC modules for instance. (Eg.: https://power.murata.com/en/products/dc-dc-converters/isolated.html ) You could find something that fits but you may have to settle for a more reasonable input voltage range. Note: I'm assuming a DC input and output voltage here. You didn't clearly state that's what you were after. Was it AC to AC instead? Anyway, why don't you just use a separate mains transformer (or a transformer with two separate secondaries) instead? |
| ocset:
you could use say 2 or 3 parallel boosts to boost your 9V-25V up to 40V, say. Then from there, use a full bridge to get your 30V, 160W. At these low voltages, you could use a couple of bootstrap high side drivers for the hi side fets. Or you could just use a flyback to go from the 40v to 30v, 160w. Use a couple of paralleled schottky diodes in the secondary, or a sync fet. I reckon for your app you could get away with on/off control and just pass the (digital) feedback signal through a digital isolator. As you say, its just a heater so who cares about some all singing and dancing dynamic feedback loop. So it would be in burst mode.......vout above demand......primary side switches away......vout below demand....primary side does not switch. ...youll end up with some hysteretic ripple on the output, but does that matter? |
| Navigation |
| Message Index |
| Next page |