After going through a turmoil of understanding the exact working principle of DCM operation of flyback converter, I ended up in this amazing explanation. Thank you for such a lucid yet simple explanation. How do we also calculate the core loss and can also verify the BH curve for the same ?
I do believe the ground reference used by the bridge mentioned at 04:54 is the same as the ground reference used by MOSFET. As far as the isolation goes the same is provided by the transformer.
Yes they are the same reference, don't mind the symbol difference, and yes transformer provides isolation which is why I did not put a symbol there but just a negative terminal for an output.
Perhaps not a "deep dive," but still a very good explanation of a simple flyback PS. Worth the 15 minutes. One can spend an entire semester learning switching supplies in detail.
@Zachariah Peterson sir it is perfect for a 15min video, thank you so much. I think the component selection and more details about opto feedback (especially tl431 usage) and snubber in another video would be diving deeper :) and the crucial part EMC filter and some layout tips deserves another video :) what I'm wondering is how do we add current control to this circuit.
IS it really a transformer and not a coupled inductor? asking since i remember you needed an air gap. I tried to build one in the past but the hardest part was the transformer* part since most out there are built on order and figuring out the proper generic part out there was the hardest part, maybe you can expand on this on a later video. Or how to make one from a toroid. Also what is the switching frequency range for flybacks ? buck boosts can go up to a few MHz but I never saw any flyback that was even close , also what would be optimal from a efficiency point ? Does the transformer* limit the switching frequency ? Regarding multi output one i have a few questions: 1 how will the power draw in one affect the other? As in if from the 12v DC rail we draw .1A and then 1A how will this affect the 3v3 rail? the rail with the FB will be ok but i am afraid of voltage saggs on the other one. OR should i size the other one higher and then use a LDO after the diode? 2 can i set a center tap like GND so that i also get negative voltages? 3 If i would desire a few DC rails with the same GND wouldn't it be simpler to make a higher DC single rail and then add a few LDOs or bucks on that DC line? (not sure about cost but i assume it is more expensive at scale)
It's absolutely a coupled inductor in flyback supplies; it's where the name comes from - energy "flies back" - magnetic flux is built up in the core during the primary on-time, then collapsed into the secondary(ies) in the off-time. Each switching cycle transfers a "packet" of energy; the size of this packet is proportional to the inductance and square of the current. Since the time it takes to ramp up the current is proportional to the inductance, it's more beneficial to lower the inductance to increase the current. This does however result in high magnetic flux, which is why flyback cores are either gapped, or a powdered iron core is used (which effectively has a distributed air gap, allowing for higher flux before saturation). You can only decrease the inductance so much before current levels become unmanageable, so between this and other factors such a switching losses, there's a "sweet spot" for the flyback topology, which is why you only really find them in the sub-200W power range; above this, a design using true transformer action wins (eg. forward converter, half/full bridge)
Forgot to add re: multi output. You'd take your feedback from the output rail that needs the tightest regulation. There's nothing stopping you adding linear regulators afterwards, but if you're doing it because you want a really clean output, you'll need careful L-C pre-filtering before the linear regulator, as they aren't very good at filtering out the high frequency components of the switching noise that the flyback will generate.
@@NicBellamy So if i use some of those powdered toroid's i can wind the inductors and not bother cutting into it? NICE But i am still a bit clueless about how to filter out the good ones are the powdered materials from a certain series or is the a material number range? asking since i am a bit lost on mouser
(1) Regarding multiple voltages, they are maintained fairly accurately, so there is mostly no need to add post-conversion and regulation unless you need to maintain better than the typical 5% regulation. You need to understand that, for stability reasons, regulation is done on the most loaded line (with highest current), as that one will be sagging the most under typical loads. Then the other voltages are maintained simply by the ratio of their turns relative to the regulated one. One thing you need to consider here is the inclusion of the rectifying diode voltage in your calculations. Let's say you're regulating the 5V output with its 1A load, and you also have a 15V output with less than 1A. Your turns ratio would be 1:3 between 5V and 15V output. But one small detail is that you also have the average 0.7V forward voltage drop across the lower voltage rectifying diode, and 1.3V across the higher one. For simplicity, we will use 1 turn per volt. Your 5 turns will have to produce 5.7V to give a 5V regulated output. On the higher voltage output, the 15 turns will produce almost exactly 3 times the voltage on the lower output, which is 5.7V × 3 = 17.1V, minus the diode voltage drop of 1.3V under maximum load, which gives 15.8V. If the 5V output is less loaded, diode voltage drop is decreased, let's say to 0.5, so we get 5.5V × 3 = 16.5V, minus 1.3V gives 15.2V. Taking this diode drop into consideration, you could use 14 turns to get closer to 15V. Either way, the voltage change is not too large and most circuits will have no problem with it. You just have to include the diode drop on the regulated rail to get a more accurate estimate of the other rails. You will also have a slight voltage drop on unregulated lines due to coil wire resistance. (2) Yes, you can set a common tap and wind negative voltages in the opposite direction, also turning their rectifying diodes in the opposite direction from the positive voltages. (3) Using LDOs is only recommended when you need more tightly controlled voltages with less noise, but in that case you're still recommended to use mutliple outputs from flyback with voltages only slightly above the regulated ones, to keep losses low. Adding buck converters after a flyback is redundant, because a flyback is already a sort of a buck converter with multiple voltages where you only need additional turns, diodes and capacitors for higher voltages, or a tap for lower voltages. It is possible to use the same wire with taps, so that you only have one continuous coil with multiple taps, but wire thickness needs to be observed for each output's current rating. Buck converters after a flyback only make sense when you need a much more tightly regulated additional voltage but don't worry about the voltage noise much.
Can we use full-bridge rectifier on the output? I never seen that, mostly there is a center-tap coil for secondary and two diodes on both end. I guess because the few turn of copper is much cheaper then two fast switching power diode, but I'm nut sure. Is there any other reason why it is unusual?
I suppose you could but I have never seen it, I would need to think about it. I do recall seeing center-tap coil designs on the secondary side, and to rectify both would need diodes, but I see it more often when you want two output rails or in a full-bridge topology.
I came up with CRT televisions that had HT (frequently on the order of a KV) flyback converters. Could you relate this to your low voltage applications?
If by "reference" you mean the AC equivalent of a negative terminal for return currents then it would be neutral. The protective earth is connection is not meant to be a current carrying conductor so it should not be used as such. It is useful to provide a safety ground using the chassis, or by drawing an earth ground ring around the edge of the board.
After going through a turmoil of understanding the exact working principle of DCM operation of flyback converter, I ended up in this amazing explanation.
Thank you for such a lucid yet simple explanation.
How do we also calculate the core loss and can also verify the BH curve for the same ?
I do believe the ground reference used by the bridge mentioned at 04:54 is the same as the ground reference used by MOSFET.
As far as the isolation goes the same is provided by the transformer.
Yes they are the same reference, don't mind the symbol difference, and yes transformer provides isolation which is why I did not put a symbol there but just a negative terminal for an output.
@@Zachariah-Peterson thanks for the clarification.
Lol at the comments. Thanks for an informative video. I feel this series isn't getting enough love. :)
Perhaps not a "deep dive," but still a very good explanation of a simple flyback PS. Worth the 15 minutes.
One can spend an entire semester learning switching supplies in detail.
About as deep as you're going to get with the 15 minutes we're alloted per video.
@@Zachariah-Peterson Agreed. As I said, worth the time.
@Zachariah Peterson sir it is perfect for a 15min video, thank you so much.
I think the component selection and more details about opto feedback (especially tl431 usage) and snubber in another video would be diving deeper :)
and the crucial part EMC filter and some layout tips deserves another video :)
what I'm wondering is how do we add current control to this circuit.
Wow such a great and informative video ❤❤
IS it really a transformer and not a coupled inductor? asking since i remember you needed an air gap.
I tried to build one in the past but the hardest part was the transformer* part since most out there are built on order and figuring out the proper generic part out there was the hardest part, maybe you can expand on this on a later video. Or how to make one from a toroid.
Also what is the switching frequency range for flybacks ? buck boosts can go up to a few MHz but I never saw any flyback that was even close , also what would be optimal from a efficiency point ? Does the transformer* limit the switching frequency ?
Regarding multi output one i have a few questions:
1 how will the power draw in one affect the other? As in if from the 12v DC rail we draw .1A and then 1A how will this affect the 3v3 rail? the rail with the FB will be ok but i am afraid of voltage saggs on the other one. OR should i size the other one higher and then use a LDO after the diode?
2 can i set a center tap like GND so that i also get negative voltages?
3 If i would desire a few DC rails with the same GND wouldn't it be simpler to make a higher DC single rail and then add a few LDOs or bucks on that DC line? (not sure about cost but i assume it is more expensive at scale)
It's absolutely a coupled inductor in flyback supplies; it's where the name comes from - energy "flies back" - magnetic flux is built up in the core during the primary on-time, then collapsed into the secondary(ies) in the off-time.
Each switching cycle transfers a "packet" of energy; the size of this packet is proportional to the inductance and square of the current. Since the time it takes to ramp up the current is proportional to the inductance, it's more beneficial to lower the inductance to increase the current. This does however result in high magnetic flux, which is why flyback cores are either gapped, or a powdered iron core is used (which effectively has a distributed air gap, allowing for higher flux before saturation).
You can only decrease the inductance so much before current levels become unmanageable, so between this and other factors such a switching losses, there's a "sweet spot" for the flyback topology, which is why you only really find them in the sub-200W power range; above this, a design using true transformer action wins (eg. forward converter, half/full bridge)
Forgot to add re: multi output. You'd take your feedback from the output rail that needs the tightest regulation. There's nothing stopping you adding linear regulators afterwards, but if you're doing it because you want a really clean output, you'll need careful L-C pre-filtering before the linear regulator, as they aren't very good at filtering out the high frequency components of the switching noise that the flyback will generate.
@@NicBellamy THX i was considering a PI filter after the diode anyway + potentially some synchronous rectification
@@NicBellamy So if i use some of those powdered toroid's i can wind the inductors and not bother cutting into it? NICE
But i am still a bit clueless about how to filter out the good ones are the powdered materials from a certain series or is the a material number range? asking since i am a bit lost on mouser
(1) Regarding multiple voltages, they are maintained fairly accurately, so there is mostly no need to add post-conversion and regulation unless you need to maintain better than the typical 5% regulation.
You need to understand that, for stability reasons, regulation is done on the most loaded line (with highest current), as that one will be sagging the most under typical loads.
Then the other voltages are maintained simply by the ratio of their turns relative to the regulated one.
One thing you need to consider here is the inclusion of the rectifying diode voltage in your calculations.
Let's say you're regulating the 5V output with its 1A load, and you also have a 15V output with less than 1A.
Your turns ratio would be 1:3 between 5V and 15V output.
But one small detail is that you also have the average 0.7V forward voltage drop across the lower voltage rectifying diode, and 1.3V across the higher one.
For simplicity, we will use 1 turn per volt.
Your 5 turns will have to produce 5.7V to give a 5V regulated output.
On the higher voltage output, the 15 turns will produce almost exactly 3 times the voltage on the lower output, which is 5.7V × 3 = 17.1V, minus the diode voltage drop of 1.3V under maximum load, which gives 15.8V.
If the 5V output is less loaded, diode voltage drop is decreased, let's say to 0.5, so we get 5.5V × 3 = 16.5V, minus 1.3V gives 15.2V.
Taking this diode drop into consideration, you could use 14 turns to get closer to 15V.
Either way, the voltage change is not too large and most circuits will have no problem with it. You just have to include the diode drop on the regulated rail to get a more accurate estimate of the other rails.
You will also have a slight voltage drop on unregulated lines due to coil wire resistance.
(2) Yes, you can set a common tap and wind negative voltages in the opposite direction, also turning their rectifying diodes in the opposite direction from the positive voltages.
(3) Using LDOs is only recommended when you need more tightly controlled voltages with less noise, but in that case you're still recommended to use mutliple outputs from flyback with voltages only slightly above the regulated ones, to keep losses low.
Adding buck converters after a flyback is redundant, because a flyback is already a sort of a buck converter with multiple voltages where you only need additional turns, diodes and capacitors for higher voltages, or a tap for lower voltages.
It is possible to use the same wire with taps, so that you only have one continuous coil with multiple taps, but wire thickness needs to be observed for each output's current rating.
Buck converters after a flyback only make sense when you need a much more tightly regulated additional voltage but don't worry about the voltage noise much.
Can we use full-bridge rectifier on the output? I never seen that, mostly there is a center-tap coil for secondary and two diodes on both end. I guess because the few turn of copper is much cheaper then two fast switching power diode, but I'm nut sure. Is there any other reason why it is unusual?
I suppose you could but I have never seen it, I would need to think about it. I do recall seeing center-tap coil designs on the secondary side, and to rectify both would need diodes, but I see it more often when you want two output rails or in a full-bridge topology.
@@Zachariah-Peterson Thank you
I came up with CRT televisions that had HT (frequently on the order of a KV) flyback converters. Could you relate this to your low voltage applications?
Thank you Zach, what should the ground reference for AC power be? Neutral? Protective Ground?
If by "reference" you mean the AC equivalent of a negative terminal for return currents then it would be neutral. The protective earth is connection is not meant to be a current carrying conductor so it should not be used as such. It is useful to provide a safety ground using the chassis, or by drawing an earth ground ring around the edge of the board.
شكرآ جزيلا على هذا الفيديو المفيد . تحياتي لك من تركيا.
على الرحب والسعة
Hi,
Kindly do more video about it with schematics and pcb design of flyback convertor.
Geranium diodes?
Yes Germanium diodes can be used instead of Si if you want lower forward voltage.
@@Zachariah-Peterson Yes, for a Germanium diode (though now days I would use a Schottky) but you suggested a Geranium diode. A flowery little critter.
@@michaelardai9703 I don't think I said anything about a flower diode, but maybe the captions did 😆
Thanks
Not even close to deep dive
Well we do what we can with 15 minutes, so....
J4f: triangle is not diode 😂
At least it's pointing the right way 😂
Your bridge rectifier looks incorrect.
It's correct, I just drew the input connections from the side and not from the top.
Doesn't he know how to draw a diode?
Sorry you don't like my hasty diodes. At least you knew it was a diode.