Magnetic amplifier made from common 12 volt transformers and magnetic cores

[eneuro]

Hello,
I've found sometime ago magnetic amplifier and magnetic switches used in military applications in fighter aircrafts
On magnetic amplifers in aitcraft application

What I'm interested in now in tree-phase application project
Modeling of a three-phase application of a magnetic amplifier

Here it is interesting article about Magnetic amplifier made from common 12 volt transformers
It looks like 9V battery and 1k potentiometer does the trick using common transformers 


However, in this schematics from this article I'm not sure whether voltage created in controll windings (it looks like primary with higher number of turns is used? ) will cancel each other 


Idea behind using two cores in magnetic switch is to cancel this load inducted voltage in controll wires , so everything is ok at above magnetic amplifier circuit?
I DO NOT see there exchanged wires in those two transformers used  on primary or secondary

Below it is correct schematics with two cores from professional papers linked at the begining of this thread:


Did you have expierience with magnetic amplifiers and could take a look at this 12V transformers schematics to verify this, please?
My guess is this schematics is wrong and danger high voltage could apear in DC controll circuit 
Maybe those rectifier diodes do some trick with higher gains, but I think those transformers should be connected to cancel induced voltage in DC controll loop... 

[T3sl4co1l]

Looks right to me.  The dot phasing is same-same and the winding connection is same-opposite.

[Andy Watson]

I DO NOT see there exchanged wires in those two transformers used  on primary or secondary

Why should they be swapped? The two transformers are performing the same action for different parts of the cycle - as determined by the diodes.

[eneuro]

Thank you for hints.

Why should they be swapped?

In magnetic amplifier with reactance coils in series they are swapped and it does not matter if we swap transformers/toroids controll wires or load wires

So, in this case they have to be swapped and it is easy to understand why induced voltage in controll wires will cancel when two cores are used in this case.

When magnetic amplifier with reactance coils in parallel (with or without diodes) are used it looks like they are connected without swapping and this configuration is more tricky 
I did not notice  it before, so its time to follow AC curents paths and try to simulate this circuit with diodes in brain again and understand what is going on when reactance coils are in parallel 
 

[IanB]

However, in this schematics from this article I'm not sure whether voltage created in controll windings (it looks like primary with higher number of turns is used? ) will cancel each other 


Idea behind using two cores in magnetic switch is to cancel this load inducted voltage in controll wires , so everything is ok at above magnetic amplifier circuit?
I DO NOT see there exchanged wires in those two transformers used  on primary or secondary

Let's consider the diagram above and call the windings on the left the control windings, and the windings on the right the load windings.

Now the DC control current, being DC, can only flow through the control windings in one direction. Therefore the load current must flow in the same direction through both load windings also, so that the magnetic fields have the same polarity. This is what the diodes are for.

First, consider the case when there is no control current. The active load transformer has a high self-inductance and a high impedance, and little current flows in the load. However, a significant voltage may be induced in the control winding. However, the other transformer also has a high impedance and so it will tend to block and absorb this voltage, being in series with the first.

Second, consider the case when there is a large control current. The active load transformer will have a saturated core, a low self-inductance and a low impedance, so a large current can flow in the load. However, due to the low inductance there will be little voltage induced in the control winding.

Either way, the battery and control potentiometer are relatively safe.

[Zepnat]

Wow fascinating. Trying hard to understand the principles of operation but sleep depravation isn't helping.  Where might something like this be used, would the average person (me!) have unknowingly used something with this inside?

[tom66]

A magamp regulator is found inside many computer ATX power supplies. It is used to regulate the 3.3V rail.

Most ATX power supplies have a regulated 5V rail, and a quasi-regulated 12V rail. The 12V rail is regulated via a multi-winding inductor as about 1V variation overall is tolerable (11.4V to 12.6V.) But the 3.3V is usually regulated by a magamp given its smaller tolerance of 300mV (3.15V to 3.45V.)

The main transformer produces an AC square wave, and a driver circuit is used to control the saturation of an inductor to regulate the output voltage. With this, power transistor isn't necessary to switch the full load current (only needs to switch a fraction of it, for the magamp control voltage) and the magamp inductor is cheaper than a proper buck converter inductor (since it is designed to saturate, whereas a buck converter never wants a saturated inductor) the circuit costs less than using a buck converter.

[ignator]

Quote from: Zepnat on Today at 04:05:47 PM

Wow fascinating. Trying hard to understand the principles of operation but sleep depravation isn't helping.  Where might something like this be used, would the average person (me!) have unknowingly used something with this inside?

These were used in avionics for analog autopilots, and many other equipments back in the day with TUBES/VALVES, before transistors and OPAMPS became common. Aircraft used 26VAC 400Hz for synchros/resolvers and such that needed amplifiers in the AC realm.
This predates me (avionics design engineer) by about 15 years but the old guys (30 years ago) cut their teeth on them at the company I worked for.

[IanB]

It's something I had never heard about before this thread, but now I know of it I'm definitely going to try it out. I think I can use one  transformer instead of two, with a full wave bridge on the load side, and I'm wondering if I can use a snubber diode on the control side to catch the voltage spikes?

[johnwa]

I had a play around with these circuits a while ago, it was fun being able to control a 50W halogen lamp using nothing more than an LDR! Though I can't remember the exact details of the phasing I used.

For the 'self biasing' configuration, you want positive feedback action, so the load current reinforces the control current. I think that this will only happen for one polarity of the input signal - the other will result in negative feedback, and reduced sensitivity. The output polarity is the same in both cases, so the amplifier effectively responds to the absolute value of the input, though with differing gain depending on the polarity of the input.

Note that this is a separate effect to the relative phasing of the transformers though. I will have to think about it a bit more, but I believe that what IanB wrote is basically right.

On another note, does anyone have any thoughts on an appropriate parameter or procedure for evaluating the performance of these magnetic amplifiers? I did attempt to do some power gain measurements on the circuit that I built up, but realised that the results were somewhat meaningless, depending on how you accounted for the DC bias power, and the power transferred back from the load circuit to the input.

[mzzj]

A magamp regulator is found inside many computer ATX power supplies. It is used to regulate the 3.3V rail.

Most ATX power supplies have a regulated 5V rail, and a quasi-regulated 12V rail. The 12V rail is regulated via a multi-winding inductor as about 1V variation overall is tolerable (11.4V to 12.6V.) But the 3.3V is usually regulated by a magamp given its smaller tolerance of 300mV (3.15V to 3.45V.)

The main transformer produces an AC square wave, and a driver circuit is used to control the saturation of an inductor to regulate the output voltage. With this, power transistor isn't necessary to switch the full load current (only needs to switch a fraction of it, for the magamp control voltage) and the magamp inductor is cheaper than a proper buck converter inductor (since it is designed to saturate, whereas a buck converter never wants a saturated inductor) the circuit costs less than using a buck converter.

Maybe so 15 years ago but not anymore. 12v rail has been the regulated work-horse since year 2003 ATX12V 2.0 or so.
Nowadays most? atx power supplies also use invidually regulated power supply rails instead of magamps.

[eneuro]

The main transformer produces an AC square wave, and a driver circuit is used to control the saturation of an inductor to regulate the output voltage.

It is interesting that changeing  the saturation we can change conductivity in continous way without disconnecting load from mains if I understand it well.

Here it is another article about Magnetic Amplifiers: How They Work, What They Do dated to July 1960 
This magamp schematics from these days shows exactly how transformer wires are connected with rectifier diodes and load in AC circuit to form full-wave magnetic amplifier.


I think it is worth spend a while and test this using  transformer with known core hysteresis and maybe AC sinus signal generator at higher frequency lets say 10kHz with fast Shotky rectifiers-than  much less power turns are needed I guess and it might work, while now it is clear how these two power coils and controll coil in the middle are connected   

Update: I've found nice (iron?) E core transformer from microwave oven teardown 

It has 10.5cm2 outer core area and 21cm2 in the middle.
It looks like there is about 144 turns @ its primary 230VAC coil (2.3 Ohm).
If I remove high voltage coil by cutting welder tracks at the bottom than I could easy add another power coil and controll in the middle  for testing.

Just wondering what magnetic relative permeability u/u0: 200-1000 of this microwave transformer core (iron?) might be for calculation?

BTW: Started read this excelent book "Magnetic Amplifiers - Principles and Applications - 1960" by Paul Mali and it explains a lot.
It is time to  and  those transformer magamps 

However, it is always worth to see and hear nice looking girl in electronics lab which seams to know howto not to destroy oscilloscope 

[eneuro]

Update: I've found nice (iron?) E core transformer from microwave oven teardown 

It has 10.5cm2 outer core area and 21cm2 in the middle.
It looks like there is about 144 turns @ its primary 230VAC coil (2.3 Ohm).
If I remove high voltage coil by cutting welder tracks at the bottom than I could easy add another power coil and controll in the middle  for testing.

Ok, it is time to try... saturate this microwave oven transformer core using oryginal 3 turns secondary with 230VAC primary connected in series with load 


Note: I've disconnected HV secondary wire from transformer core and it was the first thing I've made-of course ensured HV cap is discharged-it wasn't powered up before.

Be carefull and remove HV secondary, unless you know what are you doing 

Those low voltage 3 turns are made with ~0.5mm wire, but probably it is not copper, I guess. In microwave oven it was used to heat up magnetron filnament.
As I remember it were 3VAC on those 3 turns
I have some idea to make something like synchronized in time with 230VAC mains saturating this transformer core using 2 phase H-bridge driver and 12VDC with current limit.

How do you think-how many DC amperes could be needed to saturate this trafo with those 3 turns?   

BTW: I can remove HV secondary, so will be ablle to make more than 3 turns if needed, but I wouldn't like to go above 24VAC and use two transformers in this "magamp" experiment.

[eneuro]

Yep, it is time to remove this scary HV secondary windings and put there in place a few 12VAC low current PSUs as well as  2 coils designed to saturate this thing with the help of two mosfets and PC817 in antiparallel to detect mains polarity 

Something similar was done in this DIy spot welder to limit current in welder tarsformer:
http://www.mike-worth.com/2013/07/31/adding-a-current-limiter-to-my-mot-welder/
However, current is limited there by changing inductor core gap on the bottom of cut microwave oven transformer (increasing gap-moving up-will lower down inductance and increase  welder current in one of welder transformers)


So, it will be interesting if I will be able to do something similar not by changing inductor core gap, but synchronized saturation of its core.
Probably using two separate a few turn coils (for each polarity saturation) and two mosfets will be easier than messing with H-bridge there, than 2 MPU pins PWM will control amount of DC current in those coils based on detected mains polarity, so I should be able saturate this core in each AC mains +/- period with low voltage and decent current.
Lets see what happends in real experiment 

[eneuro]

Lets see what happends in real experiment 

Everything was fine and I was able to remove HV secondary windings without need to cut MOT at the bottom welder paths


Unfortunatelly, wanted also cut bloody insulation paper close to primary winding above magnetic shunts and.... at this side was hiden slightly inside core and angle grinder slightly touched primary windings and of course cut less than 1mm, but few at the top of wires are damaged 


Probably only Mac Gaywer skills could repair this MOT primary windings 
But maybe you have some hints, howto reconnect those top two layers wires?
I haven't got such thin copper pipes to make bridges, but while this primary coil will not be moved and some space above available there is a little chance some patent could save this nice MOT primary 

[Zero999]

Have you measured the continuity of the damage primary? If it's still in the 100mOhm range then you could still use it, as long as you use it at a tiny fraction of the rated current.

[eneuro]

Have you measured the continuity of the damage primary?

Nope, there is no connection between terminals from half of this damaged windings-a few top were cut.

Anyway, will try to find such small copper pipes with 1mm inside diameter, while I need them for other project too and make bridges.
Another option is add small copper wires and solder together with reinforced thin copper using spot welder, but need to adjust current while 1mm copper wires disapear at longer spot welder pulses 

It will be used in prototype and I do not expect higher currents than 10A, so maybe will be able "glue" lower layer wires at 5mm bridges and above them another layer 10mm.
There is still decent window above for other experimantal windings.
If it fails, than will cut bottom and remove this coil and I'll have nice E core for custom magamp  shown below and above, but it is classic one so here again 

[SeanB]

That wire is copper plated aluminium, and is not going to solder at all. Best is to simply get another donor microwave and try again, as you will not be able to fix this, just rewind with new wire.

[eneuro]

That wire is copper plated aluminium, and is not going to solder at all.

Yep, it is copper plated aluminium and any attempt to press it harder destroys it 

Anyway, identified those 13-14 damaged wires and it doesn't look good.


However, realized that I don't need transformer with exact numbers of oryginal turns, while it will be used mainly as saturable inductor, so there is some hope, that I will be able take those identified damaged wires back and connect with another endm than try to use spot welder to melt joined aluminium wires which should be possible, since copper wires at such gage basicly disapeared at spot welder maximum power with noise similar to gun shot 
This way I'll loose those few turns (13) but it is still only maybe ~10%, so no problem-only minimum current whill be slightly higher than if this primary coil were oryginal 

We'll see if I will be able to melt those aluminium wires joints to create strong connected alumium wires.

[eneuro]

It could be epic MOT primary repair too, while managed to move all cut wires back from its power terminals and 30% connections found 


However, have to find a way to weld those bloody coppered aluminium wires, since it looks like the best idea try to spot them together when connection order puzle will be solved.


Holy crap! Do not turn on microwave ovens- take them apart 

[jolshefsky]

However, have to find a way to weld those bloody coppered aluminium wires, since it looks like the best idea try to spot them together when connection order puzle will be solved.

Can you use crimps?

[paulie]

I loooooove Jeri Ellsworth. Jeri Ellsworth is my favorite girl. I want to marry Jeri Ellsworth when I grow up.

[eneuro]

I loooooove Jeri Ellsworth.

Circuit Girl 

[eneuro]

Can you use crimps?

Probably this is the only option, while no way attempt to spot weld those coppered aluminium wires leads to nowhere and thinking about something like this PCB but with 5mm copper vias, put two wires srew them and reinforce with copper wires and solder everything to bottom PCb copper-of course aluminium will not solder, but it will hold those wires in PCb, than by adding hot glue or epoxy will insulate everything.

I need 8 such holders on one pCB with additional 9th free if needed do the same trick in the future during experiments



Removed few useless bloody aluminium wires, no longer needed since do not creates any loops no more:


Epic reconnected wires now shows decent resistance lowered from 2.3 Ohm to 2.1 Ohm, so it makes sense   


However, it was difficult operation, while one bloody aluminium wire cracked close to windings when tried to push it from the coil (its is very weak mechanicaly) and hopefully after removing a few other cut wires was able catch it and unwind back 

It is never easy, but it looks like I will be able experiment with magamp using this inductor or add classic magamp two other coils using of course copper wires and this crappy aluminium coil will be used as control wires to saturate magamp 

[TerraHertz]

Can you use crimps?

Probably this is the only option, while no way attempt to spot weld those coppered aluminium wires leads to nowhere and thinking about something like this PCB but with 5mm copper vias, put two wires srew them and reinforce with copper wires and solder everything to bottom PCb copper-of course aluminium will not solder, but it will hold those wires in PCb, than by adding hot glue or epoxy will insulate everything.

What on Earth?! Attempting to splice aluminum wires?
Why are you putting so much effort into a lost-cause damaged MOT? Just go find another one. Or ten. Every single pile of discarded junk anywhere will have multiple old microwave ovens. (OK, exaggeration, but not by much.) I made myself stop picking them up, as enough is enough.
And next time, when angle grinding a transformer, shove in something solid as a protection on the winding you want to keep. Iron is good, as you can see by the sparks when you hit it. Or ceramic.