Using a 12 V battery to power an array of electromagnets drawing ~800 A, a switch is necessary to change polarity in the EMs. It seems that the best option is an H-Bridge using four IGBTs which would likely require building from components to deal with heat dissipation (at around 10 kW). My question is, why are 4 transistors necessary? That is, is it possible simply to construct 2 circuits to the same EM with ends A and B such that in one circuit, [+] goes to A and [-] goes to B, while in the other, [-] goes to A and [+] goes to B; switching one to the other such that the IGBT in one is 'open' while the other is 'closed' and vice versa. If as I suspect this is not possible, because for example one circuit interferes with the other, then how would that situation differ from the case using an H-Bridge configuration? By the way, only low frequency switching of 2 to 4 Hz is required.

 

What you are discribing sounds like an H bridge.

 

I agree with Brevor. You are describing what an H-bridge does.
Why do you want to reverse the electromagnet current? A magnet will exert the same force on soft ferromagnetic objects regardless of polarity.
Switching 800A means you will have some humungous voltage spikes to contend with.

 

What I'm imagining though requires only 2 transistor switches; on the assumption, possibly incorrect, that when one circuit is switched on, no current will proceed back through the other circuit due to the presence of its own transistor. I understand the sense of the H-Bridge, but am not learned enough--yet!--to understand why in this situation, of EM polarity switching, transistors are necessary on both sides of the load.

The reason for the switch is that I've constructed an apparatus--a generator of some sort-- which requires a rotating magnetic field to be elaborated in a total of 12 EMs (fixed as a stator; each EM positioned at the mid-point of the 12 edges of a cubic structure, and directed at the centre of that cube) capable of driving a magnetic core functioning as the rotor. The idea is that each EM first attracts one side of the magnetic core, and upon switching of EM polarity then repels it. The apparatus is definitely capable of doing this; provided these polarities can be successively switched, which requires transistors.

A copper winding is attached to this rotating core in which, because it is continuously subject to the operation of a varying magnetic field arising in the relation of the rotating core and the fixed stator EMs, a current should be generated; and the general aim of the experiment is to ascertain the nature and properties of such a current, if any.

I would rather not have to build an H-Bridge myself--since one with the requisite specifications is problematic to operate due to heat--, and this is the main reason I seek your assistance; for which incidentally I am so far most grateful.

 

So your jumping right into 800 amps as an experiment? If this is an experimental project, how big will the finished one be?

How you're describing your "apparatus" sounds eerily like a BLDC motor.

 

What you require is an H-bridge unless you have a bipolar power supply available which allows use of a "half bridge."

At 12 volts and 800 A, MOSFETs are much better suited than IGBTs. It is still far from trivial to handle currents of that magnitude.

 

aim of the experiment is to ascertain the nature and properties of such a current, if any.

If the coil is spinning around, apparently unconstrained in 3 dimensions, how will you support the core and measure the coil current?

The idea is that each EM first attracts one side of the magnetic core, and upon switching of EM polarity then repels it.

That will require very rapid polarity switching during the brief period (a few millisecs?) that the side is within the range of the field of the EM. Switching devices to handle that will need to be very robust and likely very expensive.

 

I agree with the others that you need an H-bridge configuration. Classic case.

Question: How fast are you planning to switch this thing and can you tolerate a dead time in between switching polarities? One problem with this sort of thing is that no switch is instantaneous. It can be challenging to avoid any chance, however brief, of both sides of the H-bridge "on" at the same time, ie. a short of biblical proportions. Purpose-built H-bridge driver ICs are obviously designed with this in mind.

Since your first H-bridge project is 4 orders of magnitude larger than a hobbyist project, I'd look very seriously for commerical products designed to operate at your specifications. A DIY solution sounds super dangerous.

 

A typical example is scrap yard magnet cranes where the current of each magnet is ~100A.
A reverse shot is required for demag purposes to remove residual.
Typically this is done rather crudely by modern standards with a reversing contactor set.
Max.

 

What I'm imagining though requires only 2 transistor switches; on the assumption, possibly incorrect, that when one circuit is switched on, no current will proceed back through the other circuit due to the presence of its own transistor.

This is one of those things that seems reasonable spoken out loud into the air, but becomes obvious when drawn.

Try actually drawing a schematic of your two-transistor solution, then imagine the current flow for each combination of switch states. The problems will become obvious quickly.

 

Apologies for the delay.

So your jumping right into 800 amps as an experiment? If this is an experimental project, how big will the finished one be?

How you're describing your "apparatus" sounds eerily like a BLDC motor.

Yes, it is set up like a motor; but the aim is to generate a current in the winding, which is a continuous progressive lap-type affair, mounted on the rotor; which contains a permanent magnetic core. That core is configured as a helix--effectively a propeller (a helix contained in a sphere)--, and is driven, along with the winding mounted on it, by the rotating magnetic field elaborated in the EMs. The reason for the high amperage is only the extremely low resistance of the EM coils required for sufficient field strength; and while this is only a prototype--with all sorts of technical problems, to do with heat in those coils etcetc.--, the eventual device would operate at a comparable scale--since the aim is only to generate a type of current, not huge amounts of it.

If the coil is spinning around, apparently unconstrained in 3 dimensions, how will you support the core and measure the coil current?

That will require very rapid polarity switching during the brief period (a few millisecs?) that the side is within the range of the field of the EM. Switching devices to handle that will need to be very robust and likely very expensive.

The rotor/armature is mounted on a vertical axle; and yes, the switching needs to be as seamless as possible. So far, my information is that due to the problem of heat, any existing H-Bridge module capable of handling the current will not be able to dissipate heat--which seems odd. Perhaps you can clarify that for me; surely, a switch designed to handle 800 A and 10s of kW of power is capable of accommodating heat.

This is one of those things that seems reasonable spoken out loud into the air, but becomes obvious when drawn.

Try actually drawing a schematic of your two-transistor solution, then imagine the current flow for each combination of switch states. The problems will become obvious quickly.

This is my question: does current feedback from one circuit to the other? Why are 4 transistors required in an H-Bridge when 2 would seem sufficient; unless one of each pair were required to prevent such short-circuiting? I've tried to find the answer to this, but am simply not well-educated enough in the intricacies of transistor function--yet--to understand why they wouldn't function as one-way channels.

A typical example is scrap yard magnet cranes where the current of each magnet is ~100A.
A reverse shot is required for demag purposes to remove residual.
Typically this is done rather crudely by modern standards with a reversing contactor set.
Max.

Max! Nice to see you. Crude is fine, as long as it will work. Can such a device be made to alternate polarity rapidly; and at a rate of 2 to 4 Hz?

I agree with the others that you need an H-bridge configuration. Classic case.

Question: How fast are you planning to switch this thing and can you tolerate a dead time in between switching polarities? One problem with this sort of thing is that no switch is instantaneous. It can be challenging to avoid any chance, however brief, of both sides of the H-bridge "on" at the same time, ie. a short of biblical proportions. Purpose-built H-bridge driver ICs are obviously designed with this in mind.

Since your first H-bridge project is 4 orders of magnitude larger than a hobbyist project, I'd look very seriously for commerical products designed to operate at your specifications. A DIY solution sounds super dangerous.

I certainly would prefer to find a commercially available product (capable of switching instantaneously, though at low frequency only: 2 to 4 Hz). The reply to 'Shortbus' summarises things..

 

my information is that due to the problem of heat, any existing H-Bridge module capable of handling the current will not be able to dissipate heat--which seems odd.

Your information is incorrect. Water-cooled and forced-air cooled systems are in use, especially for industrial applications.

 

What you require is an H-bridge unless you have a bipolar power supply available which allows use of a "half bridge."

At 12 volts and 800 A, MOSFETs are much better suited than IGBTs. It is still far from trivial to handle currents of that magnitude.

This seems to be a point of contention. I had originally thought MOFSETs would be best, but was confused by the suspicion that bipolar transistors are preferable to MOFSETs at high voltage and current; and the fact that IGBTs seem better suited to low frequency switching. What do you think of Max's idea about reversing contactors?

 

Why are 4 transistors required in an H-Bridge when 2 would seem sufficient

Suppose you have a single-polarity power supply with terminals marked '+' and '-'. Consider a coil with terminals A and B. For current to flow in one direction you connect A to + and B to -. That requires two switches (transistors) to close. To reverse the current, those switches must open and you connect A to - and B to +. That requires two further switches, since the original two can pass current in one direction only.

 

Your information is incorrect. Water-cooled and forced-air cooled systems are in use, especially for industrial applications.

Suppose you have a single-polarity power supply with terminals marked '+' and '-'. Consider a coil with terminals A and B. For current to flow in one direction you connect A to + and B to -. That requires two switches (transistors) to close. To reverse the current, those switches must open and you connect A to - and B to +. That requires two further switches, since the original two can pass current in one direction only.

Yes, it seems the part I don't understand is the reason, if it's a circuit, 2 switches are necessary to open/close it, as if the switch were just an ordinary light switch. Clearly there is something about the use of transistors as switches which I don't yet understand. I'll take your word for it though, and commence the search for an existing H-Bridge module--MOFSETS or IGBTs??--with appropriate flyback diodes, cooling etc.. I'm slightly incapacitated physically which restricts such a search to wonderful portals like this. 'Max's' suggestion about a 'reversing contactor set' sounds feasible, and I'll look at that too. In retrospect, I might have designed the EMs to require less current. Oh well.

 

This is my question: does current feedback from one circuit to the other? Why are 4 transistors required in an H-Bridge when 2 would seem sufficient; unless one of each pair were required to prevent such short-circuiting? I've tried to find the answer to this, but am simply not well-educated enough in the intricacies of transistor function--yet--to understand why they wouldn't function as one-way channels.

Sorry, but it's hard for me to describe these things in words without some context. If you want to share a schematic of how you think two switches will get it done, I'll be happy to explain the problems in the context of that circuit. You can draw the circuit with simple SPST switches instead of transistors in order to simplify things if you want.

I can't think of any plausible two-switch solution, so I don't have a starting point from which to discuss why it won't work. If you can draw a circuit to show us how you think two switches will work, we can talk through the problems. Or, if you want to start with a diagram of an existing h-bridge circuit and tell us which two switches you think can be omitted, we can talk from there.

As I said before, I think once you actually try to draw it, you'll realize that it doesn't add up.

The only two switch solid state solution I'm aware of would require a bipolar power supply (as mentioned by ebp in post 6.)

 

Apologies for the delay.


Yes, it is set up like a motor; but the aim is to generate a current in the winding, which is a continuous progressive lap-type affair, mounted on the rotor; which contains a permanent magnetic core. That core is configured as a helix--effectively a propeller (a helix contained in a sphere)--, and is driven, along with the winding mounted on it, by the rotating magnetic field elaborated in the EMs. The reason for the high amperage is only the extremely low resistance of the EM coils required for sufficient field strength; and while this is only a prototype--with all sorts of technical problems, to do with heat in those coils etcetc.--, the eventual device would operate at a comparable scale--since the aim is only to generate a type of current, not huge amounts of it.

You're making even less sense as you try to explain things. You want to have electromagnets generate a current in permanent magnets??? A helix contained in a sphere??? How old are you, may we ask that? Or what grade in school are you in. Before spending money on this "experiment" you may want to spend some time reading and studying how things work.

I also don't think you have any real idea how high a current 800Amps is, at any voltage. Houses here in the USA are usually only supplied with around 200Amps at 240 volts. I have no idea where you would even find a 800A 12V DC supply for your 'experiment'. Or for that matter there is no wire size chart that goes that high of DC amps, but thewire would be huge if you could find some that would work. But do applaud you for thinking big from the get go.

 

'Max's' suggestion about a 'reversing contactor set' sounds feasible, and I'll look at that too. .

Incidentally these are required to have some kind of contact arc quenching in place such as P.M. or other means.
Max.

 

What cross-section copper bar are you presently using/planning for the EMs?
I think you should re-design your EMs to have a lot more turns passing a lot less current to achieve the same field strength.
Even a heavy duty 12V car battery will not like having to pass 800A for more than a few seconds at a time.

 

What kind of inductance are you talking about in these electromagnets?

If it's more than a very small amount, when you try to switch that current quickly, you will discover that one of the quicker ways to die is open-circuiting an energized electromagnet.

If your current is 800 A and you are cycling at 4 Hz, then even 100 mH of total inductance will require 640 V in order to ramp linearly.