Compensation correction
N-Channel
I would like to correct a previous post about calculating compensation of feedback in Marty Browns book I was following an example for current control mode instead of voltage control mode and in the current control mode the inductor is out of the loop and the slope is -20db not -40db as in voltage control mode therefore all my calculations were wrong along with the statements I made about the example, a lesson learned.😱
N-Channel
I would like to correct a previous post about calculating compensation of feedback in Marty Browns book I was following an example for current control mode instead of voltage control mode and in the current control mode the inductor is out of the loop and the slope is -20db not -40db as in voltage control mode therefore all my calculations were wrong along with the statements I made about the example, a lesson learned.😱
Chas-
I was wondering about that...
After plugging in the values you have shown (BTW, THX for the .pdf) 😀, everything seemed a little out of whack. Your post expalins that.
OK, now for my "feedback" 🙂D) on your schematic:
Basically, I like the overall layout. Short, sweet, and to-the point. I do have a few comments, and because we are using the Web to communicate, we can't express emotions or body language, so some comments might get "lost in translation", so to speak, so please don't take them as criticisms, but rather as questions and observations.
1) For your power levels, I would make C5 (1mF) at least 2.2 mF. Check out Chryssis' design procedure for Cc, the Coupling Cap.
2) Have you simulated the start-up circuit? I realise that it operates for only a second or two, but Brown uses a "highly starved" linear regulator consisting of a TIP50 (400V NPN), 1KW between Vbus and the '50's Collector, with a 440k W resistor srting feeding a 13V zener diode. Also, since the 3525's start-=up current is also its operating current, having over 100mA capacity will help to ensure that the '3525 gets enough power at start-up so it doesn't lock up.
3) What is the inductance of the two output inductors? Are they mutually coupled (to provide good cross regulation)? Also, you could put individual inductors in series with the outputs, between C12 & C13, and C14 & C15, respectively.
4) For C10 (Css), try lowering it to 10 mF. You don't want to have too long a soft-start time. (See comment 1).
5) Where are the Y Caps (Line to Ground & neutral to gnd)? These usually run in the range of 2200pF to 3300pF, and can be scrapped from most older AT & ATX boxes, the ones that actually bother to put the AC filter in. 😉
Other than that, like I said, the design looks pretty good.
Here's a thought: Have you considered Active PFC'ing the front-end. This would put alot lower stress on the input rectifiers, and lower actual power consumed. Just a thought.....
Steve
I was wondering about that...
After plugging in the values you have shown (BTW, THX for the .pdf) 😀, everything seemed a little out of whack. Your post expalins that. OK, now for my "feedback" 🙂D) on your schematic:
Basically, I like the overall layout. Short, sweet, and to-the point. I do have a few comments, and because we are using the Web to communicate, we can't express emotions or body language, so some comments might get "lost in translation", so to speak, so please don't take them as criticisms, but rather as questions and observations.
1) For your power levels, I would make C5 (1mF) at least 2.2 mF. Check out Chryssis' design procedure for Cc, the Coupling Cap.
2) Have you simulated the start-up circuit? I realise that it operates for only a second or two, but Brown uses a "highly starved" linear regulator consisting of a TIP50 (400V NPN), 1KW between Vbus and the '50's Collector, with a 440k W resistor srting feeding a 13V zener diode. Also, since the 3525's start-=up current is also its operating current, having over 100mA capacity will help to ensure that the '3525 gets enough power at start-up so it doesn't lock up.
3) What is the inductance of the two output inductors? Are they mutually coupled (to provide good cross regulation)? Also, you could put individual inductors in series with the outputs, between C12 & C13, and C14 & C15, respectively.
4) For C10 (Css), try lowering it to 10 mF. You don't want to have too long a soft-start time. (See comment 1).
5) Where are the Y Caps (Line to Ground & neutral to gnd)? These usually run in the range of 2200pF to 3300pF, and can be scrapped from most older AT & ATX boxes, the ones that actually bother to put the AC filter in. 😉
Other than that, like I said, the design looks pretty good.
Here's a thought: Have you considered Active PFC'ing the front-end. This would put alot lower stress on the input rectifiers, and lower actual power consumed. Just a thought.....
Steve
Shematic
N-channel
the schematic you are in reference to was posted for you by Luka
I have not posted mine. But rest assured my supply is almost a carbon copy of Brown's except for values of some components and the HGTG12N60A4D IGBT'S, when I have finished all test I will post it for anyone who desires to build one.
Thanks again for all your help and suggestions.
ps in reference to the schematic there seems to be a problem with compensation. In my supply I am changing core set to ETD- 49 and output will be +/- 55 V @20 amps to power a bi-amp system with 300 watt sub,200 watt mid and a 100 watt tweeter . Its a reach for HalfBridge but with these igbt's it should be a small step. About 15 amps collector current. maybe pushing it. I have FullBridge Layout ready .
Chas
N-channel
the schematic you are in reference to was posted for you by Luka
I have not posted mine. But rest assured my supply is almost a carbon copy of Brown's except for values of some components and the HGTG12N60A4D IGBT'S, when I have finished all test I will post it for anyone who desires to build one.
Thanks again for all your help and suggestions.
ps in reference to the schematic there seems to be a problem with compensation. In my supply I am changing core set to ETD- 49 and output will be +/- 55 V @20 amps to power a bi-amp system with 300 watt sub,200 watt mid and a 100 watt tweeter . Its a reach for HalfBridge but with these igbt's it should be a small step. About 15 amps collector current. maybe pushing it. I have FullBridge Layout ready .
Chas
Whoops, my mistake.
Luka, see previous post.
Chas, let's see..... +/-55V @ 20A, = 2.2kW. I would DEFINITELY go with the Full Bridge.
What voltage is the AC mains in the DR? If a split 120/240 like here in the US, I would design it forr 220-240VAC operation. If only 120VAC, then I would use at least a 20A breaker, 30A preferred.
Luka, see previous post.
Chas, let's see..... +/-55V @ 20A, = 2.2kW. I would DEFINITELY go with the Full Bridge.
What voltage is the AC mains in the DR? If a split 120/240 like here in the US, I would design it forr 220-240VAC operation. If only 120VAC, then I would use at least a 20A breaker, 30A preferred.
Poor math skills
N-channel
The mains here are identical to US and I had planed on using 220 V for bridge and as you pointed out I need over 2kW for bulk supply. My first plan was to design a supply for each amp and then I felt redundancy creeping in and I have ETD-49 core sets so I switched gears and I did some quick calculations on my trusty SCALC and thats where the error occured. I have built up a bridge with fets and it presented quite a few problems with gate drive and compensation for light loads, so I fell back to halfbridge and even considered a two switch forward converter until I read most of the posts by EVA so I dropped that for now. I would like to put a PFC in front of bridge but most of the app notes I see and even Pressman is for powers around 300 watts and since I have not been there done that I will not go there. I understand by using current mode control I will elminate most of these problems.
Thanks
chas😱
N-channel
The mains here are identical to US and I had planed on using 220 V for bridge and as you pointed out I need over 2kW for bulk supply. My first plan was to design a supply for each amp and then I felt redundancy creeping in and I have ETD-49 core sets so I switched gears and I did some quick calculations on my trusty SCALC and thats where the error occured. I have built up a bridge with fets and it presented quite a few problems with gate drive and compensation for light loads, so I fell back to halfbridge and even considered a two switch forward converter until I read most of the posts by EVA so I dropped that for now. I would like to put a PFC in front of bridge but most of the app notes I see and even Pressman is for powers around 300 watts and since I have not been there done that I will not go there. I understand by using current mode control I will elminate most of these problems.
Thanks
chas😱
Chas-
Split the one full-bridge into two half-bridges, and sybch the two PWM chips together. Synch'ing the MC33025 is very easy and will eliminate any beat frequencies that will be generated.
See the following datasheet from ONSemi: http://www.onsemi.com/pub/Collateral/MC34025-D.PDF
You will also gain the benefit of separate regulation for each channel.
I am aware of a PFC operating in the kW range, but I don't remember where i saw it. Try googling Kilowatt PFC circuit and see what you get.
If you were to separate the AC input section (filter and rectifier) from the DC-DC section (main filter caps and PWM), and use 500V IGBTs, you would have the option of adding PCF later.
Steve
Split the one full-bridge into two half-bridges, and sybch the two PWM chips together. Synch'ing the MC33025 is very easy and will eliminate any beat frequencies that will be generated.
See the following datasheet from ONSemi: http://www.onsemi.com/pub/Collateral/MC34025-D.PDF
You will also gain the benefit of separate regulation for each channel.
I am aware of a PFC operating in the kW range, but I don't remember where i saw it. Try googling Kilowatt PFC circuit and see what you get.
If you were to separate the AC input section (filter and rectifier) from the DC-DC section (main filter caps and PWM), and use 500V IGBTs, you would have the option of adding PCF later.
Steve
3 great idea's in one post
N-channel
You have made my day. This is the route I will take , I will use all the suggestions and will surf the web to find info on PFC for over a kilo. I should have all parts by wed and then I will stuff my breadboard and start test and fine tunning. When all is ready I will gladly share with others all data and schematics. I am very lucky I have a board house 10 min away and a double sided board will be in my hands two days after I send files with solder mask and silk screen.
Thanks
chas🙂
N-channel
You have made my day. This is the route I will take , I will use all the suggestions and will surf the web to find info on PFC for over a kilo. I should have all parts by wed and then I will stuff my breadboard and start test and fine tunning. When all is ready I will gladly share with others all data and schematics. I am very lucky I have a board house 10 min away and a double sided board will be in my hands two days after I send files with solder mask and silk screen.
Thanks
chas🙂
PFC 1kw
N-channel
If you go to IRF website and check out the PFC chip ir1150 I think you might a solution for all your pfc problems. they also have a nice design program that is a point and click after you enter your values and it produces all components for the design. My understanding this can be used from 85 watts to 4kW. Digi-key has this device in stock for 3.85 in single qty. I have odered 10.
After I browse the app note I plan on doing PCB layout and maybe in two to three weeks have something. You will be surprised at how many components it takes for the design about
20 and most are resistors, the only change I am going to make is use an igbt no mosfet.
chas
N-channel
If you go to IRF website and check out the PFC chip ir1150 I think you might a solution for all your pfc problems. they also have a nice design program that is a point and click after you enter your values and it produces all components for the design. My understanding this can be used from 85 watts to 4kW. Digi-key has this device in stock for 3.85 in single qty. I have odered 10.
After I browse the app note I plan on doing PCB layout and maybe in two to three weeks have something. You will be surprised at how many components it takes for the design about
20 and most are resistors, the only change I am going to make is use an igbt no mosfet.
chas
IRF
Thanks, Chas. I will look at it first thing in the morning. 😱 Very cool having a board house that close. You get to see-touch-feel the finished boards before you even take them home.
Looking forward to seeing your results.
😀 😀 😀
Thanks, Chas. I will look at it first thing in the morning. 😱 Very cool having a board house that close. You get to see-touch-feel the finished boards before you even take them home.
Looking forward to seeing your results.
😀 😀 😀
Lower actual power consumed? PFC lowers effiency by 2-5%.N-Channel said:
Apparent power=effective(actual?) power+reactive power? or how it goes in english terminology??
I dont know about other countries, but at least here in Finland consumers dont pay for reactive power/energy.
3.5kw PFC frontend stuff:
http://www.microsemi.com/micnotes/APT9901.pdf
Also check other http://www.advancedpower.com/ application notes.
Real -v- Apparent -v- Imaginary
mzzj,
I misspoke when I said "Actual power", when I meant to say "Total Power" (Actual + Reactive Power). 🙂 Very interesting App Note.
mzzj,
I misspoke when I said "Actual power", when I meant to say "Total Power" (Actual + Reactive Power). 🙂 Very interesting App Note.
Could you show me some picture of that TIP+resistor to zener stuff, so I can see how to connect it in my design.
Output inductors will be each by itself,inductance will be unknown.I don't caculate it
Why would be 2.2 uF better than 1uF.Is it not enought?
So that is the reason why there is none.
Output inductors will be each by itself,inductance will be unknown.I don't caculate it
Why would be 2.2 uF better than 1uF.Is it not enought?
Since I live in old flat, we don't have ground!
So that is the reason why there is none.
How much would be enought 1uF?
Luka-
Since I don't have a pic to attach, I will attempt to describe it. It is just like a series-pass linear regulator using a resistor and zener diode from Vbus to ground to form a voltage reference. Run two 220k-ohm resistors in series from the rectified +320V bus to the anode of the 13V zener, Next, run the zener's cathode to the 0V of the DC bus. (placing two resistors in series keeps each resistor below its breakdown voltage of about 250V). Next, run two 8.2k-ohm 5W power resistors in series (same reason) from the +320V bus to the collector of the TIP50 NPN. Connect the TIP50's base to the junction between the 13V zener and the lower 220k-ohm resistor. Now, the emitter of the TIP50 goes to the cathode of a 1N4007 diode (to prevent backfeeding). This will then feed the +Vcc and +Vc pins of the SG3525 or MC33025, whichever chip you choose. The auxiliary winding on the main transformer should be calculated to provide about 3V higher than the output of the linear regulator, to ensure that the regulator fully turns off after the power supply reaches steady-state operation.
I don't have the calculations for figuring out the value of the coupling cap (Cc), but when I get home, I will look them up and post them. Basically, for the design example Chryssis gave, (200W, half-bridge, 50kHz, 120/240VAC operation), a value of 1 uF yielded avout 30V across this Cap.
I have seen 1, 2.2, 4.7 and 10uF for the SoftStart Cap, Css. The equation for calculating the start-up time is t(s) = (4.5 • 10e5) Css, for C in Farads. I mostly use 10uF.
I would strongly recommend coupling the output inductors onto a common powdered-iron toroid core, like how it is done for the output of all AT & ATX PSUs. This inductor provides regulation, just like in a buck regulator, and ensures excellent cross-regulation and symmetry for the (+/-) outputs. Putting non-coupled ferrite inductors in between the first two electrolytics and the last two on the output provides noise filtering on the outputs.
You will want to connect the feedback network to the first pair of electrolytics because the additional L-C filter will introduce an extra pole and zero into your compensation equations, making the computations more difficult. Whew! All that in one breath! 😀
Since I don't have a pic to attach, I will attempt to describe it. It is just like a series-pass linear regulator using a resistor and zener diode from Vbus to ground to form a voltage reference. Run two 220k-ohm resistors in series from the rectified +320V bus to the anode of the 13V zener, Next, run the zener's cathode to the 0V of the DC bus. (placing two resistors in series keeps each resistor below its breakdown voltage of about 250V). Next, run two 8.2k-ohm 5W power resistors in series (same reason) from the +320V bus to the collector of the TIP50 NPN. Connect the TIP50's base to the junction between the 13V zener and the lower 220k-ohm resistor. Now, the emitter of the TIP50 goes to the cathode of a 1N4007 diode (to prevent backfeeding). This will then feed the +Vcc and +Vc pins of the SG3525 or MC33025, whichever chip you choose. The auxiliary winding on the main transformer should be calculated to provide about 3V higher than the output of the linear regulator, to ensure that the regulator fully turns off after the power supply reaches steady-state operation.
I don't have the calculations for figuring out the value of the coupling cap (Cc), but when I get home, I will look them up and post them. Basically, for the design example Chryssis gave, (200W, half-bridge, 50kHz, 120/240VAC operation), a value of 1 uF yielded avout 30V across this Cap.
I have seen 1, 2.2, 4.7 and 10uF for the SoftStart Cap, Css. The equation for calculating the start-up time is t(s) = (4.5 • 10e5) Css, for C in Farads. I mostly use 10uF.
I would strongly recommend coupling the output inductors onto a common powdered-iron toroid core, like how it is done for the output of all AT & ATX PSUs. This inductor provides regulation, just like in a buck regulator, and ensures excellent cross-regulation and symmetry for the (+/-) outputs. Putting non-coupled ferrite inductors in between the first two electrolytics and the last two on the output provides noise filtering on the outputs.
You will want to connect the feedback network to the first pair of electrolytics because the additional L-C filter will introduce an extra pole and zero into your compensation equations, making the computations more difficult. Whew! All that in one breath! 😀
Inductor value
Luka
You should consider calculating your inductor value based on load current because in a voltage mode regulator it is the deciding factor for closing the feedback loop and your compensation. In this case you will need a type III compensation circuit in MHOP I have noticed designs without it but they are unregulated.
Based on your schematic I will asuume a load current of 10A and a freq of 50kHz using the formula .5 * Vout *T/Iout I get 75uH inductor will do and since you have +- 37V out you put 40 uh in series with each output before the output caps and like N-Channel says wind them on a common core bifilar and take the outputs from opposite ends for phasing, I have not tried it but the yellow core from a 250 watt ATX should work if not stack two together will still be a small footprint, as far as the number of turns you need an inductance meter or core parameters to calculate the turns.One thing you might want to consider is light loads, the calucation above assumes the inductor is continous, I consider an audio amp as a variable load and at 2 amp the inductor would be about 350 uH or 150uH in each output . Quite a large swing, you can download software from internet to calculate your core size they are free.
This is based on my experience and the methods I use which always result in fine tuning under actual test of supply, yours or others might be different but IMHOP voltage mode control presents some problems that can only be overcome by bench test
chas
Luka
You should consider calculating your inductor value based on load current because in a voltage mode regulator it is the deciding factor for closing the feedback loop and your compensation. In this case you will need a type III compensation circuit in MHOP I have noticed designs without it but they are unregulated.
Based on your schematic I will asuume a load current of 10A and a freq of 50kHz using the formula .5 * Vout *T/Iout I get 75uH inductor will do and since you have +- 37V out you put 40 uh in series with each output before the output caps and like N-Channel says wind them on a common core bifilar and take the outputs from opposite ends for phasing, I have not tried it but the yellow core from a 250 watt ATX should work if not stack two together will still be a small footprint, as far as the number of turns you need an inductance meter or core parameters to calculate the turns.One thing you might want to consider is light loads, the calucation above assumes the inductor is continous, I consider an audio amp as a variable load and at 2 amp the inductor would be about 350 uH or 150uH in each output . Quite a large swing, you can download software from internet to calculate your core size they are free.
This is based on my experience and the methods I use which always result in fine tuning under actual test of supply, yours or others might be different but IMHOP voltage mode control presents some problems that can only be overcome by bench test
chas
Since this will be my first offline smps,I don't want it to be complicated,by caculating no. turns,...
My only goal now is to get it to work with amp.
Then bench test it.
And if you have photo of start-up circuit, could you post it.
How does your supply coming along?
Any pictures?
My only goal now is to get it to work with amp.
Then bench test it.
Which one do you use?
And if you have photo of start-up circuit, could you post it.
How does your supply coming along?
Any pictures?
Guys-
I just thought of something: For the output inductor to be effective, it needs to be physically the same size, or roughly the same size as the main power transformer. So, for a PSU putting out over a kW peak, you might want to go with the following powdered-iron toroid core: T-250-26. Toroid, 2.50" o.d., #26 material. Amidon Associates here in the 'States has #26 material toroid cores of that size. If you're doing something a little smaller, perhaps a T-130-26 might suit you, as it will be smaller, claiming much less real-estate on the pc board. Go to Amidon's website and download their data sheet for all their products and specifications. Or better yet, register for a catalog.
Also, this toroid inductor should be bi-filar wound, as chas1 suggests, because it will provide for cross-regulation between the (+) and (-) buses, keeping them symmetric about the zero volt line. (This is exactly like the BIG YELLOW toroid core in computer PSUs.) The two inductors after the first two output filter caps can be ferrite material, and do NOT share the same core. As I stated in my last post, the coupled inductor is for regulation, and the two un-coupled inductors after it are for filtering.
Hope this clears things up a bit.
I just thought of something: For the output inductor to be effective, it needs to be physically the same size, or roughly the same size as the main power transformer. So, for a PSU putting out over a kW peak, you might want to go with the following powdered-iron toroid core: T-250-26. Toroid, 2.50" o.d., #26 material. Amidon Associates here in the 'States has #26 material toroid cores of that size. If you're doing something a little smaller, perhaps a T-130-26 might suit you, as it will be smaller, claiming much less real-estate on the pc board. Go to Amidon's website and download their data sheet for all their products and specifications. Or better yet, register for a catalog.
Also, this toroid inductor should be bi-filar wound, as chas1 suggests, because it will provide for cross-regulation between the (+) and (-) buses, keeping them symmetric about the zero volt line. (This is exactly like the BIG YELLOW toroid core in computer PSUs.) The two inductors after the first two output filter caps can be ferrite material, and do NOT share the same core. As I stated in my last post, the coupled inductor is for regulation, and the two un-coupled inductors after it are for filtering.
Hope this clears things up a bit.
Software
Luka
I use two Magnetics Inductor design using Powder Cores form the Magnetics website it is very good and it gives a value for the noload inductance and allows you to play with the core sizes and so fourth. The other and my favorite is Mircometals Inductor Design Software april 26, 2005 from thier website and the guys in the application dept will respond quickly to any problems or questions you have plus this software covers swinging choke, pfc inductors and more.
Now I will describe the steps for my design or copy if you prefer:
1. On the web is a German site for designing most SMPS you will use and all you do is input you requirements and it will calculate the values for the transformer and inductor size and recommend cores and bobbins. I use it to get in the ball park.
2. Then I refer to my references and compare the results
3.Pressman has a great chart about the power a core can give
depending on topology used that where I get my core size info.
4. Then I refer to George Chryssis "High Frequency Switching Power Supplies Theory and Design" which I received while attending a seminar of his in 1985. This book is a step by step guide to design of most topologies that interest me.
5.Then I surf the web to see if anyone has a working design that might fit my needs and find out the cost, if I think it is worth while and the spec's are ok I buy it, check and see if it can be modified
easily to fit my needs.
6.I also join a forum such as this one and read all the post and post my questions and ideas and review the reply's.
7. I designed DC motor controller's once upon a time and SMPS
are simular since they are a closed servo system and require compensated feedback loops for solid performance.
8. On a scale of 1 to 10 my intelligence is about a 3 for I can't figure out how to include a JPEG or BMP from my hard drive in my post nor a link to a website, so take all of the above with a grain of salt I am sure there are many member's of this forum that will help you. Most of all be patient you will not come up with a useful
design in a matter of weeks, I have been working on mine for 6 months, finding parts , getting advice, simulating and watching smoke and fire a great destroyer of silicon.
chas1
Luka
I use two Magnetics Inductor design using Powder Cores form the Magnetics website it is very good and it gives a value for the noload inductance and allows you to play with the core sizes and so fourth. The other and my favorite is Mircometals Inductor Design Software april 26, 2005 from thier website and the guys in the application dept will respond quickly to any problems or questions you have plus this software covers swinging choke, pfc inductors and more.
Now I will describe the steps for my design or copy if you prefer:
1. On the web is a German site for designing most SMPS you will use and all you do is input you requirements and it will calculate the values for the transformer and inductor size and recommend cores and bobbins. I use it to get in the ball park.
2. Then I refer to my references and compare the results
3.Pressman has a great chart about the power a core can give
depending on topology used that where I get my core size info.
4. Then I refer to George Chryssis "High Frequency Switching Power Supplies Theory and Design" which I received while attending a seminar of his in 1985. This book is a step by step guide to design of most topologies that interest me.
5.Then I surf the web to see if anyone has a working design that might fit my needs and find out the cost, if I think it is worth while and the spec's are ok I buy it, check and see if it can be modified
easily to fit my needs.
6.I also join a forum such as this one and read all the post and post my questions and ideas and review the reply's.
7. I designed DC motor controller's once upon a time and SMPS
are simular since they are a closed servo system and require compensated feedback loops for solid performance.
8. On a scale of 1 to 10 my intelligence is about a 3 for I can't figure out how to include a JPEG or BMP from my hard drive in my post nor a link to a website, so take all of the above with a grain of salt I am sure there are many member's of this forum that will help you. Most of all be patient you will not come up with a useful
design in a matter of weeks, I have been working on mine for 6 months, finding parts , getting advice, simulating and watching smoke and fire a great destroyer of silicon.
chas1
That's exactly what I would say, hehe, but since you went to seminar of that sort in 1985 when I wasn't even born,you must be lucky coz this kind of thing do not appere in Slovenia yet.
I have done this SMPS with 0% knowlage about SMPS and 3 years later I have working one, JUST NEED TO PUT IT on one board so that it doesn't need anything ealse for it to work.
Three years from 0-60mph, huh? That's impressive. I first became interested in SMPSs in the mid-1980's. I started serious research and experimenting in the early 90's when I built the car amplifier that appeared in Audio Amateur in 1989-90, authored by Mr. Randall Vikan. My version of that amp appeared in the "Showcase" section of the 01/94 Audio Amateur issue.
I am STILL learning things about them today. There are so many different topologies, control methods, frequency ranges, and tricks that SMPS designers and DIYers have and use that convince me that the learning in this topic will never end.
I am STILL learning things about them today. There are so many different topologies, control methods, frequency ranges, and tricks that SMPS designers and DIYers have and use that convince me that the learning in this topic will never end.