Hello!
I'm attaching the file I made with Claude after putting a reversed engineered schematic (a mess to understand) for it to change the layout without changing the circuit (tidying it up).

The SMPS is model FSP350-60APN from Corsair. I got this SMPS by a ex-colleague that said I could have it if I thought I could fix it.
So far, I couldn't fix it yet, but I'm making progress, and the best part is learning brand new ideas of circuits.

Magnetic amplifiers are something in between linear regulators and switched mode converters (and also in the middle in terms of efficiency), but way more reliable, very resistant to radiation, no moving parts, no transistors (this one has a TL431, but you can do without it if you want and has a low power transistor compared to the 3V3 current).

I simulated the circuit with an estimation of value of the filtering components. The 100 mH is the saturable inductor, it has high inductance but very low current for it to saturate (1 mA). The 5 ohm resistor in series with the inductor is to mitigate ringing between the capacitors and the inductor. It's important to have a good capacitance at the output to prevent oscillations. The 180 nF in series with 1 kΩ are there to lower ripple on the output.
Falstad Circuit Simulator - Magnetic Amplifier circuit of SMPS FSP350-60APN

I found that if you take the emitter of the PNP transistor and put it to the output voltage node, it requires less current to dessaturate the inductor, with seemingly no drawbacks. It goes from ~58 mW avg to ~18 mW avg on the 4.7 ohm resistor.

Also, for those who are unfamiliar (as I was few days ago, even after more than a decade studying electronics):
Explanation of how magnetic amplifiers work

This is the image I found on the web and used Claude to generate the PDF:


TL;DR: this is a post to show something cool that I found and could simulate, no need to help me with something, but if you think you can contribute with our learning, go on, I appreciate.

 

To those who will encounter this thread in the future and to those who are watching it:
Falstad Circuit Simulator does not have a hysteresis effect on any inductor (I discovered 1 hour ago while chatting with Claude), it can saturate but it's always the same value for the same input, so the circuit isn't really working as intended, besides regulating it right. The hysteresis is important because it magnetizes the core with a magnetic retention (the remaining B field when no H field).

The best approach I'd do now is to simulate on LTSpice. I don't feel like I'll do it in the short term, so feel free to simulate and post your results here if you wish, I'd be curious what really is the waveforms and amplitudes (given the values of components).

 

There are a number of different types of "Mag-Amps".
I do not have the knowledge to make SPICE cores that will work. Over my head.
I have used them in TV sets (CRT monitors) to change the Horizontal Size on the fly. Also to adjust the H. Linearity dynamically.
I have used MagAmps; where there are several outputs of a power supply. One output is in the control loop, and the rest need to be regulated in their own loops.
They also adjust the resonant frequency of a LC resonant power supply.
There must be more ways of using them. Now I remember that we used the earth's field to change the saturation points of cores to make solid-state a compass.

 

Magnetic amplifiers are fascinating devices. I found them in a mid 1990s multi output power supply used by a high end workstation.
Even back then, the core material was very expensive. It was made of a thin ribbon of exotic magnetic material wound as a toroid. I believe it was the cost what eventually caused them to be superseded by the rapidly improving Mosfets.

 

I have one question.
Is it possible to use a mag amp to amplify audio without distortion?
I see that the saturation is intrinsically nonlinear, so you can't amplify linearly, right?

 

s it possible to use a mag amp to amplify audio without distortion

An op-amp (without feedback) is not linear either. Op-amps use feedback and high gain to reduce distortion. Almost any power amplifier can be used to make an audio amplifier. It is the feedback that compares the input signal to the output signal and corrects for errors.

 

Magnetic amplifiers are fascinating devices. I found them in a mid 1990s multi output power supply used by a high end workstation.
Even back then, the core material was very expensive. It was made of a thin ribbon of exotic magnetic material wound as a toroid. I believe it was the cost what eventually caused them to be superseded by the rapidly improving Mosfets.

Maybe it was this core?

 

An op-amp (without feedback) is not linear either. Op-amps use feedback and high gain to reduce distortion. Almost any power amplifier can be used to make an audio amplifier. It is the feedback that compares the input signal to the output signal and corrects for errors.

I totally forgot about how negative feedback is so helpful.

 

Not exactly but similar. With a trade name of Square Permalloy.

 

Just came here to explain the solution. Last weekend I found out when I cut the heat-shrink covering the mag amp inductor that the heat-shrink on each inductor leads, one of them melted a bit (probably because of too much current there) and it was just being isolated by the enamel between the 2 leads, basically. I suspect that at that high temperature the enamel was not solid to isolate and it conducted. So the 5 V at somewhat high duty went through the 3V3, causing the filter capacitor to bloat, but not enough to damage it (I tested for capacitance, just the ESR is a bit higher), so the OVP triggered. Nevertheless, I changed it for a good one. After separating the short-circuit, I tested it about 3 times and it worked normally.
Photo to see what I'm talking about. The terminals are made of 3 wires in parallel, so that arrow is the first turn of the mag amp inductor of the lower lead.
This gave me a very good amount and duration of dopamine, as I did for the challenge of doing it, not for money.

 

I examined that circuit and while i saw transformers I did not see what I would call a magnetic amplifier. So what part of the circuits shown is thought to be a "MAG AMP"???

 

The L3 inductor, that's in series to the rectification by BD2 and filtering to the 3V3 output.
I don't have for certain that magnetic amplifier is the correct term for it, but Claude keeps calling it the mag amp. For me it's like the circuit is a mag amp, not just the inductor.

 

The L3 inductor, that's in series to the rectification by BD2 and filtering to the 3V3 output.
I don't have for certain that magnetic amplifier is the correct term for it, but Claude keeps calling it the mag amp. For me it's like the circuit is a mag amp, not just the inductor.

Fundamentally ANY mag-amp must have at least two windings, so that one can vary the magnetization affecting the impedance of the other winding. THAT MEANS that a single inductor can not be a "mag-amp". What is shown would be called a "choke" or a "filter choke", intended to resist rapid changes in the DC current. Almost without exception, an amplifier has separate inputs and outputs. The possible exceptions would be "tunnel diodes", and other negative resistance devices.

It seems that your information source is slightly confused.

 

Fundamentally ANY mag-amp must have at least two windings, so that one can vary the magnetization affecting the impedance of the other winding. THAT MEANS that a single inductor can not be a "mag-amp". What is shown would be called a "choke" or a "filter choke", intended to resist rapid changes in the DC current. Almost without exception, an amplifier has separate inputs and outputs. The possible exceptions would be "tunnel diodes", and other negative resistance devices.

It seems that your information source is slightly confused.

I think you're right in saying it should have 2 windings to be considered a mag amp, but this inductor is not a regular one.
Because Claude said it was a mag amp and because I read on wikipedia "Semiconductor-based solid-state switches have largely superseded them, though recently there has been some regained interest in using mag amps in compact and reliable switching power supplies. PC ATX power supplies often use mag amps for secondary side voltage regulation.", which fits very well what I encontered.
This inductor have high inductance, low saturation current, and square B-H curve, which is suitable to act like a switching element. It passes low current until saturation, then high current until the circuit near it reset its magnetic core.
So I think that a choke or filter doesn't fit the description too.

 

Perhaps it is a saturable choke? I'm guessing.

 

Perhaps it is a saturable choke? I'm guessing.

Choke has this name because it chokes the HF, right?
This one doesn't chokes the HF, instead it blocks when the core is resetted.

 

I have had it explained that not every technical explanation found on the "web" is accurate! In fact, I have observed that some explanations are totally incorrect in every aspect. I have observed that to be correct.
A "choke" coil is used to present ahigh impedance to higher frequencies.
THAT HAS NOTHING AT ALL TO DO WITH MAGNETIC AMPLIFIERS!!!
The CONTROL winding in a MAG-AMP varies the magnetic field in the iron core and thus changes the inductive reactance, and that tends to affect the AC current flow in the controlled winding. By some rather skillful engineering it does not tend to act like a transformer, inducing power ito the control winding. THAT is the part that is confusing, as I see it.

 

There are more than one type of mag. amp.
Here is a "linearity coil" which can be used as a mag. amp. The round wheel is a magnet that rotates by hand. The magnet is so strong it can saturate out the core in the positive or negative direction. At the center the magnet does nothing. The rest of the coil is a "dog bone" core with, usually has heavy Litz wire. One type of mag. amp. uses a core material that has a sharp corner in the saturation curve.

Here is the secondaries of a transformer on a PWM. Lets say output at C5 is regulated by the PWM but the output at C6 is not regulated. It just comes along for the ride and the voltage is unpredictable. We insert a Mag Amp inductor in series with D3. At the power phase of the cycle both L3 and L4 output a positive voltage. C5 gets charged up. But, at zero current the inductance at L7 is very high and very little current gets to C6. The Mag. Amp L7 saturates at (12V x 5uS) After 5uS the core saturates and allows full current to flow toward C6. By turning the wheel magnet, we can set the saturation point from (12V X 20uS) or to (12V X 1uS) So the duty cycle seen by L5 C6 can be adjusted by the hand.

We don't actually adjust by hand but by electronic.

 

OK, I see a "Booster Coil" circuit functioning in a useful manner. Quite interesting, but certainly NOT a mag-amp by the common definition. It does provide a voltage increase by adjusting the timing . There is no control current affecting the boost function as described. "Boost" coils were used in the tube-type horzontal deflection CRT systems and worked quite well. BUT the coils were not amplifiers, they shifted the power flow.

 

Back to post #18, sometimes I would the coil as shown. Left side, green wire is a small number of turns if large wire while red is many turns of small wire for the control winding. A linear supply of 0 to 10mA will set the saturation point. Most of the time I wind the red wire like a coupled transformer but wound in a way to have a large amount of leakage inductance.
On the right side I made transformers like this. Again the green winding is high current and the red is 1000s of turns for control current. The center leg of the core has a large gap so the green wire has low inductance. The outside legs have no gap so the inductance is large. There is no coupling between primary and secondary.


I went looking for different Mag-amp coil I have used. This type I mostly control how they reset so the coil starts out partly saturated. The one on the left, has a control winding. I have never use that one I normally have a heavy wire (maybe 10 turns) that caries the output current and a 100 to 1000 turns of very small wire to force the core into saturation.

I could not find the box today, but I have a collection of toroidal coils where the one winding is wound over the outside of the core and never goes through the center. This uncouples the two windings.

About the cores. One method is to have the core go in/out of saturation every cycle. This adjusts the duty cycle. The core has a sharp saturation knee. The coil is a switch that changes the duty cycle of one winding.

Another method is to adjust the inductance of the coil via an error-amp. The core has a soft knee. The coil is simply adjustable not "switched". In the case of a resonant supply, the frequency is adjusted against a resonant LC to set the output power. A second output has a variable inductor that shifts its resonant frequency to regulate this second output.

I played for a while with a 5-leged core transformer where the control winding shifts the flux (of the power winding) from one set of legs to a different set. In effect I was adding in or subtracting a couple of turns from the transformer.

There are many different ways to use inductors/transformers and controlled saturation to change how a power stage works.