What's the normal low-load working temperature for an SMPS transformer?

[HomeExperiement]

Hi!

I have question about high frequency transformers inside switch mode power supplies (SMPS). Is it normal for them to become hot under no load or low load? For example if power supply is rated for 144A at 13.8V and I only pull from it 2A at 9.6V, it normal that transformer becomes about 50C just in about 10 minutes or should it stay at room temperature at that low load? The size of transformer is roughly 5x5x5 cm. Or does it mean that something is wrong with transformer? Also, when load is less than 1A transformer starts to make high pitch (around 10-12 kz I think, can't measure it) whistling noise - is this normal or is it sign of low quality cheap PWM controller? When I increase the load the noise fades out - or what it seems like it just goes outside of human hearing range.

I am asking about this temperature thing because on feature list temperature controlled fan on/off was advertised. But if 2A (if you have patience to wait then even 1A) load can make fan to start because high temperature threshold is reached then what's the point of that feature? Well of course from china all kinds of scam could come. Therefore I would like to know if such temperature at such load simply means that I've got bad unit or is it physically impossible to design a high frequency transformer that is around 5x5x5 cm in size and that stays cool under 2A load?

Below is also a product image. If this power supply at such size (29x13x7cm) were well designed what do you think how much current/power would it be realistic to draw from it without need for fan to start spinning (lets say the threshold is 50C for both transformer and heatsink and whatever gets there first starts the fan)?

 

[Baluncore]

Running a 144 A * 13.8 V = 2 kW supply on a low load will be inefficient. It takes energy to run a switching converter, even when there is no load.

I²R losses in the transformer winding will be proportional to load current, but the core losses will be decided by switching frequency.

The audio sound from the SMPS at very low load may be single isolated cycles of operation, before it shuts down due to output over-voltage, then restarts. I expect it has poor regulation at low load.

You do not specify ambient air temperature. Heat loss is proportional to temperature difference.

 

[DaveE]

Of course the real answer requires detailed knowledge of the circuit design. However, a reasonable rule of thumb for transformer design is that the core losses should be roughly equal to the full load copper losses. As Baluncore said the core losses are pretty constant for a fixed frequency SMPS, the copper losses are proportional to I². So I wouldn't worry too much.

If we assume that temperature rise is proportional to power loss then you could expect about twice the temperature rise as at full load vs. light load. It's actually not linear because cores and copper are less efficient at elevated temperature, so I would guess 250-300% hotter (above ambient) at full load.

Added cooling for SMPS is always a good idea, as is derating. Reliability and lifetime are a strong function of temperature, and SMPS manufacturers are under a lot of cost pressure not to add fans, bigger heatsinks, larger transformers, etc. Read the datasheet carefully about the environment the PS is specified for and/or the cooling you are supposed to provide. There is a lot of "specsmanship" at play in these products.

Having worked in this area for many years, I can confidently say that anyone that thinks you should buy a 100W power supply to power a 100W load lacks hard earned experience. Cooling and derating are your friends, even though they cost extra.

 

[HomeExperiement]

You do not specify ambient air temperature

23C

In other words if SMPS is designed for 2KW and I only draw from it 2A then transformer getting 50C is normal due to inefficiency? From the other hand I have heard that some PC PSUs can handle up to 65% of load in 0 RPM mode. Do they simply have better low load regulation?

Just trying to understand what is the purpose of advertising feature such as temperature temperature based fan off and on control? Is this just some scam or is there any point in it? I mean if at lower loads it gets hot due inefficiency and at higher loads get hot due to load then basically they could already sell units with fan always on instead.

 

[essenmein]

The transformer is only part of the equation. A transformer getting warm even with no load is fine. 50C is nothing. Even a mains connected 50/60Hz transformer will get warm even if there is nothing connected to the secondary due to the hysteresis loss in the core material.

Your fan may not come on at all if under no load, while the transformer is getting a bit warm, all the other components, main switching bridge, rectifier diodes (or fets if synch rect) etc are experiencing very little loss at this point. As you load the supply down, all those other components will start adding losses to the equation, then things start really warming up and the fan kicks in.

If you want to check the no load power consumption, measure input current/power with nothing connected on the output, as others have noted this will always be non zero, you have to energize your control circuit, provide gate drive energy, energize the transformer etc etc etc.

 

[HomeExperiement]

Your fan may not come on at all if under no load, while the transformer is getting a bit warm, all the other components, main switching bridge, rectifier diodes (or fets if synch rect) etc are experiencing very little loss at this point. As you load the supply down, all those other components will start adding losses to the equation, then things start really warming up and the fan kicks in.

That happens. However it comes on already at low low. I am just trying to figure out how much load would it be realistic to expect SMPS like specified in OP handle before fan starts. As I have understood it's design flow if for 2kw unit 20W is already enough to need the fan but if it had not any major design flaws at what point would you expect fan to start?

 

[essenmein]

That happens. However it comes on already at low low. I am just trying to figure out how much load would it be realistic to expect SMPS like specified in OP handle before fan starts. As I have understood it's design flow if for 2kw unit 20W is already enough to need the fan but if it had not any major design flaws at what point would you expect fan to start?

How long is a piece of wood?

Even two seemingly identical power supplies (ie visually similar) can perform dramatically differently.

It could just be that the fan threshold was poorly chosen, or its a "feature" because they copied a different design but did not modify it to suit a new application.

Then if the power supply is not experiencing any natural convection, you would expect the fan to either cycle on and off, or if its a smarter controller, run at a minimal speed to remove the losses.

 

[Baluncore]

How can we know that, if instead of 23C it was operated at 18C, that it would not operate the fan ?

We are in an engineering data vacuum here. We have no model number, data sheet, topology or circuit diagram.

I would consider a 2 kW maximum rated supply would be optimised to operate between 1 kW and 1.5 kW. Operating such a supply at less than 1% might be considered abuse.

How long is a piece of wood?

A piece of wood is longer than a short piece of string, but shorter than a parliamentary term.

 

[alan123hk]

For example if power supply is rated for 144A at 13.8V and I only pull from it 2A at 9.6V, it normal that transformer becomes about 50C just in about 10 minutes or should it stay at room temperature at that low load?

If I were you, I wouldn't worry too early.
Let's make a very rough guess, assuming that the SMPU has 95% efficiency at 2KW full load, and the transformer core loss accounts for 1/3 of the total loss, then the total loss = 2000*(1/0.95-1) = 105W, Core loss = 105/3 = 35W.

Assuming again that the core loss at low power is approximately equal to full power, then at 35W power consumption, for a 5x5x5cm transformer, a temperature increase of 27(50-23)°C will not surprise me.

I am asking about this temperature thing because on feature list temperature controlled fan on/off was advertised. But if 2A (if you have patience to wait then even 1A) load can make fan to start because high temperature threshold is reached then what's the point of that feature?

It is difficult to comment on the start-up of the fan at 50°C. Without knowing the design details, we cannot challenge the cooling fan control strategy. I personally think that 50°C may not be too low. Did you remove the cover of the SMPS to measure the temperature of the transformer? In this case, the convection conditions may be different from that enclosed inside the chassis, or is it possible that the fan has intelligent control function, and is intentionally started in advance to prevent overheating inside the transformer or other components?

 

[artis]

if the transformer is 50c at light load be sure it will get hotter when powering near full load capacity , I would say that is normal, after all the transformer is physically small it doesn't have the best heat convection.
I built some smps myself time ago and one was a rather powerful one so when I used it near maximum load the transformer was hot enough so that when I touched it I almost burnt my finger.Whether a smps needs fan or doesn't and when the fan should switch on depends on the specific psu in question, they are not all the same, some are tightly packed in a small volume so need fan at all times otherwise electrolytic capacitors get too hot, other don't need forced ventilation at all, like the one I built I had more space available and used larger transformer core etc so I don't have any forced ventilation at all even at full power.can you give us the model of the smps or something? By the picture it seems like a weird psu there seem to be many transistors etc.

 

[HomeExperiement]

It could just be that the fan threshold was poorly chosen, or its a "feature" because they copied a different design but did not modify it to suit a new application.

What would be the geed enough threshold for transformer? I mean some manufacturers have it set to 65C instead of 50C. But is it still healthy?

I personally think that 50°C may not be too low.

Or what do you think is 50C just right?

or is it possible that the fan has intelligent control function, and is intentionally started in advance to prevent overheating inside the transformer or other components?

Yes, it is.

Did you remove the cover of the SMPS to measure the temperature of the transformer?

Yes, but shouldn't have much effect because the case already is of metal which conducts heat well and has ventilation holes in it.

We are in an engineering data vacuum here. We have no model number, data sheet, topology or circuit diagram.

can you give us the model of the smps or something? By the picture it seems like a weird psu there seem to be many transistors etc.

Without knowing the design details, we cannot challenge the cooling fan control strategy.

I don't think model details would be of much help here.
https://www.aliexpress.com/item/4000328528983.html?spm=a2g0s.9042311.0.0.19f84c4dVuoDDx
If you want I can make photo gallery of inside photos it and link it here.

like the one I built I had more space available and used larger transformer core etc so I don't have any forced ventilation at all even at full power.

If I want quiet power supply with output voltage adjustable in entire range, is building it myself only option? I can't seem to find such models at all. And if I want good quality output that has stable voltage under different loads and that has low ripple, do I require oscilloscope for that or is good modeling on the paper also good enough? And how did you solve PWM management? Did you use arduino to switch mosfets on and off or did you use some specialized chip?

 

[Baluncore]

I don't think model details would be of much help here.

I have just read the new review of the product.

AliExpress titles and descriptions include irrelevant key words associated with similar products that increase the item hit-rate in a search. You need to be aware of that as it leads directly to confusion, and misrepresentation of the item being listed. It also makes it difficult to search for an item you need, because of the huge number of irrelevant search results.

In my experience, when it comes to disputes over faulty goods, or misrepresentation of a product by the Seller, AliExpress stands behind the Chinese seller, while eBay and Paypal protect the Western Buyer.

So, if I find what looks like a good device on AliExpress, I then search for the equivalent device on eBay, then buy it from the eBay supplier whenever possible, even if the initial price is higher.

If you cannot get a return on the supply misrepresented by the seller, you could replace the noisy fan with a larger and quieter external fan that runs at a lower speed, from an old computer supply, or from eBay.

 

[alan123hk]

If I want quiet power supply with output voltage adjustable in entire range, is building it myself only option? I can't seem to find such models at all. And if I want good quality output that has stable voltage under different loads and that has low ripple, do I require oscilloscope for that or is good modeling on the paper also good enough? And how did you solve PWM management? Did you use arduino to switch mosfets on and off or did you use some specialized chip?

If you really try to design and build a 2000W SMPS yourself, I will admire your ambition and passion for perfection. I personally may not do this unless it is necessary and there are no alternatives. I do think that designing and developing SMPS is a very complex and difficult task. In addition to basic electronic circuit theory and analysis, there are also many EMC issues and safety issues of electricity need to consider, so it is definitely not suitable for someone like me who is not fully trained and inexperienced in this area of expertise.

Anyway, the link below is a free online switch-mode power supply, circuit and transformer design, calculation and simulation software. This seems to be a valuable reference for those who want to learn SMPS design but It also shows the complexity of such a task.

https://www.poweresim.com/index.jsp?pc=24766

 

[artis]

@HomeExperiement Well I guess it wouldn't be the best idea to recommend to you to make your own SMPS supply especially for such high power.
If you really wanted to understand better how they work start by reading the fundamentals and maybe get yourself a 12v battery and then you could indeed make a small transformer that is switched by two mosfets which could be driven by an arduino but making a 2kW smps would require powerful mosfets and good circuitry and mains power so I don't recommend that you unless you are ready to become and electrical engineer.

I've built mine smps simply because I do have the fanaticism and it's a hobby for me , otherwise I would have just bought one because let me tell you something from experience, in 99% cases if you aren't well experienced in a field it's always cheaper to either pay a professional or buy a plug&play device than to make one yourself, even if the parts will be cheaper the time and risk still doesn't pay off.As for the temperature of the transformer I say don't sweat about it , unlike semiconductor devices , copper wires and ferrite blocks are much more forigiving to temperature , I'd say unless the transformer get's over say 90+ degrees celsius don't worry. nothing can happen to it , the only thing that can break in a transformer is the enamel insulation of the copper wires, those insulations differ but generally they can take over 100 degrees celsius and start to melt closer to 150. at least the good ones, so running your on say 80 degrees won't do any harm.
Maybe the electrolytic capacitor will get hotter and dry out faster.

When I run my smps at full power for longer periods the ferrite traffo gets hot too, I haven't measured exactly but I can't hold my finger on the core for long.
And yes almost all smps use specially designed drive circuits with IC's not arduino or other small computers. In my circuit I have SG3525 IIRC which just adjusts things like duty cycle and frequency etc, then the output is fed into another IC in my case IR2113 which then drives the switching devices (IGBT in my case) which then drive the transformer primary which is a half bridge circuitry in my case, but as I said that took me some few years of understanding before I got myself a working model.By the way what are you doing with 2kW of power?

 

[zoki85]

Such transformers are quite resilient to the heat stress. Usually there's more danger to other SMPS components in vicinity of it, to fail or malfunction, if the transformer gets too hot.

 

[artis]

exactly so a good practice would be to put filter electrolytics away from such heat sources as far as possible

 

[HomeExperiement]

If you really try to design and build a 2000W SMPS yourself, I will admire your ambition and passion for perfection. I personally may not do this unless it is necessary and there are no alternatives. I do think that designing and developing SMPS is a very complex and difficult task. In addition to basic electronic circuit theory and analysis, there are also many EMC issues and safety issues of electricity need to consider, so it is definitely not suitable for someone like me who is not fully trained and inexperienced in this area of expertise.

Anyway, the link below is a free online switch-mode power supply, circuit and transformer design, calculation and simulation software. This seems to be a valuable reference for those who want to learn SMPS design but It also shows the complexity of such a task.

No I would prefer not to do it. Just asked to get some more knowledge on this field. One thing is complexity, but other thing is that finding all the parts itself would already take like yer or so. But from the other hand if I want to get unit that can continiously supply 600w without any noise at all then I can't find any such option. I hard that some people buy waterproof units if they want passive cooling but those waterproof units don't have output voltage regulation. As I have not had any I don't know if its possible or how easy it is to modify it to be adjustable in range 0-100% or how well such attempts turn out.
an other idea was that if I could find 20x100w power supplies that are 0-100% adjustable I could conect these in parallel because small ones are typically fanless. But how possible would it be to control them all from the same potentiometer? I mean desolder the factory originals and then connect cables from all 20 units to one potentiometer. An other question would be inrush current. If typical for such size is let's say 30A then 20x30A inrush would blow my wall socket black on the moment I connect? So that idea doesn't sound very good either. But would still be great to hear some more professional thoughts - even if it's just for educational purposes.

As for the temperature of the transformer I say don't sweat about it , unlike semiconductor devices , copper wires and ferrite blocks are much more forigiving to temperature , I'd say unless the transformer get's over say 90+ degrees celsius don't worry. nothing can happen to it , the only thing that can break in a transformer is the enamel insulation of the copper wires, those insulations differ but generally they can take over 100 degrees celsius and start to melt closer to 150. at least the good ones, so running your on say 80 degrees won't do any harm.
Maybe the electrolytic capacitor will get hotter and dry out faster.

Such transformers are quite resilient to the heat stress. Usually there's more danger to other SMPS components in vicinity of it, to fail or malfunction, if the transformer gets too hot.

Then if they don't accept return shipping I could maybe try playing around with some resistors to make temperature sensing chip read let's say 40% lower values from sensors that are in transformer? I guess It shouldn't be too complex task.
But if capacitors dry out, do they pop with loud bang?

When I run my smps at full power for longer periods the ferrite traffo gets hot too, I haven't measured exactly but I can't hold my finger on the core for long.
And yes almost all smps use specially designed drive circuits with IC's not arduino or other small computers. In my circuit I have SG3525 IIRC which just adjusts things like duty cycle and frequency etc, then the output is fed into another IC in my case IR2113 which then drives the switching devices (IGBT in my case) which then drive the transformer primary which is a half bridge circuitry in my case, but as I said that took me some few years of understanding before I got myself a working model.

Saw video from youtube where one guy was using this. Just thought that it might be easier in a sense that you can write your own custom logic there as much as you want. But not sure about other aspects if it would make life easier or not.

By the way what are you doing with 2kW of power?

2Kw not really. Waned to have about 600W continuous and since it advertised smart temperature controlled fan the seller said that at 2kw I won't start till load hits 600w. And since price difference between 1kw and 2kw wasn't that big then I did not see problem with it. Thought that maybe if in future I want to do something that is power hungry then I would have 2kw available - but nothing such planned yet. Since I heard that there are 0 RPM mode PC PSUs that can handle 65% load without starting the fan,getting 600W from 2000W SMPS seemed something that might work.

exactly so a good practice would be to put filter electrolytics away from such heat sources as far as possible

Would 3cm be already good enough?

 

[zoki85]

But if capacitors dry out, do they pop with loud bang?

Electrolytic caps vaporizing its' electrolyte may die silently or with spectacular explosion

 

[Baluncore]

But if capacitors dry out, do they pop with loud bang?

Electrolytic capacitors have a K pattern scored in the exposed end. When they dry out, that end swells before perforating and venting. Look for bulged ends and replace them with high temperature rated components of the same voltage and capacitance.

If fan noise is a problem, fit a bigger slower fan, change the case if needed. Change the orientation of the supply so cooling air can flow up through the case.

 

[artis]

or just get yourself a thermal sensing probe like there comes on various multimeters and see how hot the transformer gets under load, then decide whether you need cooling at all.capacitors usually dry out slowly in elevated temperatures like 50/60 etc degrees celsius , in order for them to explode because of temperature the temperature has to be really high and the capacitors have to be under power.
I once had such a thing , I played some video games on my older pc, it had the Nvidia gtx8800 card in it (famous for it's extreme heat) and the card was hot and apparently the heat was so large it went upwards by convection and passed through the pc power supply fan through the supply and out , then as I was playing the pc at one moment there was a large boom and the room went dark.
took me a moment to realize what had happened, the caps blew up in the power supply input and shorted the AC mains which blew my fuses.
So even with all modern caps having venting structures built into them , never say never

 

[alan123hk]

an other idea was that if I could find 20x100w power supplies that are 0-100% adjustable I could conect these in parallel because small ones are typically fanless. But how possible would it be to control them all from the same potentiometer?

For ordinary users, it doesn't look easy and simple to connect two or more SMPS in parallel by thenselves, unless the SMPS already have a built-in design that allows users to connect through standard procedures.

Because there are multiple SMPS connected in parallel, and each SMPS has its own feedback control to provide stable input-output relationship, I would concern the stability of the whole system in addition to the issues mentioned in the links below.

https://www.frei.de/assets/templates/frei/Dateien/Dateien/PDF/Stromversorgungen/en/Technische_Beschreibung_Parallelschaltung_ENGL.pdf
https://www.digikey.hk/en/articles/techzone/2016/sep/properly-configure-parallel-power-supplies

An other question would be inrush current. If typical for such size is let's say 30A then 20x30A inrush would blow my wall socket black on the moment I connect?

Yes, inrush current can be a problem, unless each power supply in parallel can also keep its ratio of inrush current to normal current low.

 

[HomeExperiement]

If fan noise is a problem, fit a bigger slower fan, change the case if needed. Change the orientation of the supply so cooling air can flow up through the case.

Would it help if I glued heatsinks on transformer's frame? From the one had protective layer of tape isolates some of heat too so efficiency would be reduced. Also I am not sure how much magnetic fields in transformer like it when there is heatsink attached to it.
The best would be 0 RPM mode. Does anyone happen to know anyone who makes SMPS what are fully passive cooled and that can adjusted in entire range?

or just get yourself a thermal sensing probe like there comes on various multimeters and see how hot the transformer gets under load, then decide whether you need cooling at all.

I have IR thermomoter. With emissivity of 95 I got 60C at the bottom of transformer. With probe I got 50 on top and about 55-56 on bottom - I suspect that since probe is of metal then the other side that was not touching the transformer was giving heat away. I didnt dare to stick it into transformer or press it too hard against it because I was afraid to damage it.

I was playing the pc at one moment there was a large boom and the .

What would be best way to protect against it (it = room went dark)? One way that I know is to use some kind of ballast - like connect it in series with lightbulb and then you have current limited power supply. But the negative side is that it wouldn't allow me to test it's behavior under 1kw load for example. And also it causes slight voltage drop. and thirdly it consumes extra power in addition to unit that's being ballasted. Is there some better way to construct ballasts? I mean like something that would give full 230V till 2kw and then cut power as soon as it reaches 2001W and that would be fast enough that short circuit wouldn't make room dark?

Because there are multiple SMPS connected in parallel, and each SMPS has its own feedback control to provide stable input-output relationship, I would concern the stability of the whole system in addition to the issues mentioned in the links below.

Even if they're all the exact same model and all have exact same value pot for voltage regulation?

Yes, inrush current can be a problem, unless each power supply in parallel can also keep its ratio of inrush current to normal current low.

Could probably fix it with correct thermistor in series with SMPS bank I guess? I just looked up one 300W unit. It's interesting that 300W also has 50A inrush current just like 2kw has.

 

[artis]

forget the inrush current that is mostly defined by the input filter capacitors.
also forget about capacitors blowing up, what I said was just an example as to what can happen if a electrolytic capacitor is used and overheated when it's powered. I believe you don't have any such major heat sources close to them so you'll be fine.In theory you can take a heatsink and attach it to the top of the ferrite core it won't do any damage , in fact I attached my transformer to the heatsink as well , it helps to take away some of the heat made in the core.
In theory you can parallel multiple smps outputs if they are of the same type Although in practice it might not be the best idea, because all smps have filter capacitors in their output to make smooth dc , now those filter capacitors get charged up much like the ones on the mains voltage side upon turning on the device, the inrush current taken from the smps transformer is proportional to the capacitance in the output.
not all smps will turn on at the exact same time so some of the smps output will be more loaded than others, if this load is beyond the load the smps can take it will either shut off or blow a switching transistor.

, you did not answer what you need the power for ? that information can give folks here better shot at an advice.

 

[alan123hk]

Even if they're all the exact same model and all have exact same value pot for voltage regulation?

Do you really believe that all the output voltages of parallel SMPS can be set exactly same by providing the same pot value?

I believe this does not work at all, and even if you individually adjust the pot of each SMPS to have exactly the same output voltage, I still doubt the effectiveness and reliability of the system.

In my opinion, putting more than one converters having regulated output voltages in parallel is definitely not a good idea in principle unless current sharing control (droop method) is applied.

In practice, it is easier to connect converters with regulated output voltage in series than in parallel. In parallel operation, a better choice is current mode converter.

The link below shows a brand of high-performance DC power supplies that can be programmed for constant-voltage mode, constant-current mode or constant-power mode operation. It also allows users to connect power supplies in parallel using a "master / slave" daisy-Chain wiring configuration.

https://www.powerandtest.com/power/dc-power-supplies/asterion-dc

 

[HomeExperiement]

Contacted the seller. He forwarded my problem to factory engineer. Engineer said I could try to unplug the fan because the unit should be able to handle up to 600W load (of 2kW) with passive cooling only. Turned out that at 80W (11.7V*6.8A but probably transformer outputs 24V or something because the the max it outputs is a little over 15V so transformer has to be more than that which means that load on transformer side is probably half of the amps at output) load it took about 30-40 min till transformer hit 75C which triggered over temperature protection which shut off the unit. Just curious to know if it's normal for transformers to get hot that easy?

, you did not answer what you need the power for ? that information can give folks here better shot at an advice.

For self mad LED light setup. that needs to be quiet but that consume up to 600W at full brightness. Occasionally might use it for other projects that need power in that voltage range. Needs to be voltage adjustable in entire range.

 

[Baluncore]

...Turned out that at 80W (11.7V*6.8A but probably transformer outputs 24V or something because the the max it outputs is a little over 15V so transformer has to be more than that which means that load on transformer side is probably half of the amps at output)...

You do not understand the way a switching supply regulates the output voltage.
There will be no linear regulator on the output of that switching supply. In the earlier photo I see what looks like five Schottky power diodes and five capacitors as the output. Output regulation is done on the input side of the transformer by adjusting the duty cycle.

The 230 VAC input is first rectified to 325 VDC. That is switched at about 50 kHz with variable duty cycle into the primary of the transformer. The output from the transformer is rectified directly into the output capacitors.

If the regulation feedback is adjusted to produce 12 VDC out, then an 800 watt load will require an average 66.7 amp output current. The transformer input current will only average about 800 W / 300 V = 2.7 amp.

There will be transformer core losses, but the major source of inefficiency and heat will be from the output rectifier diodes which will drop about 1 volt * 67 amp = 67 watt. That is why they are mounted on the heatsink. For lower voltages and higher current outputs, switching supplies are more likely to use mosfets as synchronous rectifiers at the output in place of Schottky diodes. I cannot read the part numbers on the rectifiers pictured earlier.

Some w = I²R power will also be lost in the resistance of the transformer secondary. The problem with the output rectifiers will be the conduction angle needs to remain small for good regulation, so secondary current pulses must be much larger than average. I²R in the transformer secondary circuit then dominates resistive losses at full power.

 

[HomeExperiement]

The 230 VAC input is first rectified to 325 VDC. That is switched at about 50 kHz with variable duty cycle into the primary of the transformer. The output from the transformer is rectified directly into the output capacitors.

Do you man that DC is is sent to transformer? Or what makes this DC to AC again in order to send it to transformer?

The output from the transformer is rectified directly into the output capacitors.

What do you mean exactly? As I understand, it is already rectified, but then it gets rectified again?

If the regulation feedback is adjusted to produce 12 VDC out, then an 800 watt load will require an average 66.7 amp output current. The transformer input current will only average about 800 W / 300 V = 2.7 amp.

The input is always constant? It does not change with different output voltage adjusting?

Also I created this picture with part names for more info.

 

[Baluncore]

The 230 VAC input is rectified into the big capacitors (10), at 325 VDC.

The controller (12) drives transistors (4,5,6,7), to alternately switch that DC into 325 VAC at 50 kHz in the transformer (14) primary.

The transformer (14) low voltage secondary, at 50 kHz, is rectified by diodes (8,8,8,8), to make the low voltage DC output on capacitors (9,9,9,9,9).

The op-amp chip (13) senses the actual output DC voltage, it sends signals to the regulator (12), which adjusts the on time of the alternating input transistors (4,5,6,7) to control energy flow and so stabilise the output voltage.

https://en.wikipedia.org/wiki/Switched-mode_power_supply