Controlling a high power electromagnet (for eddy current brake)

[TomcM]

Hi,
Would you kindly help a student out of his depth (I have very limited experience in power electronics)?

Current situation (picture in attachment): 220V AC -> variac (0-240V AC) -> Graetz bridge (to rectify) -> the coil. Magnetic field is created in the airgap and rotation of the aluminum disc is stopped/slowed. Resistance of the coil is 380 Ohm. I borrowed an oscilloscope and I remeber that the voltage oscillations on the coil were rather small, so the L/R ratio should be high, but I don't have the inductance value

End goal: Create a circuit to control the eddy current brake. Input signal: 0 - 5V from microcontroller, Power supply: 220V AC, Output: 0 - 0.6A DC current in the coil of the electromagnet. The steady state of output should be able to change from 0 TO 0.6A or any other target value within miliseconds. The changes should occur approximately every second.

My current idea: make a classic AC dimmer (zero cross detector circuit, optocoupler, triac), then rectify the output with a Graetz bridge and connect it to the coil (my inductive load).

Questions:

• Is the base of my idea a good way to approach the problem?
• If yes: Should I be worried about filtering the rectified phase cut signal, before connecting it to the load?
• Are there any commercialy availible circuits that solve this exact problem already?
• Any other tips?

Thank you for your time and knowledge, I appreciate your help.
TomcM

[T3sl4co1l]

Yes, that is fine.  Some voltage drops can be saved by making the bridge out of SCRs, but that will do for starters.

I would be shocked if the L/R time constant is small compared to mains frequency; anything with enough winding resistance should have plenty of inductance to keep the current flowing between cycles.  Conversely, control varies on the same scale, so don't expect sharp on/off behavior.  On and off a second should be fine.

Tim

[Terry Bites]

So you need 0-220V DC for the coil. The coil will heat up and its DC reistance will change. So do you want a constant current supply?

[NiHaoMike]

Just rectify mains and use PWM to control the current. Easy way would be to use an isolated gate drive and MOSFET, with a fast diode across the coil. Use a linear Hall sensor placed in the middle of the core for feedback. Try to find an EE in power electronics to help you.

For a one off application, it might be cheaper (and safer) to just buy a high voltage bench power supply than investing in the time to develop a custom circuit. Or if it's trivial to remove the coil, replace it with one that operates at a lower voltage and higher current. Probably 24V would be a good balance since 24V power supplies are commonly available for 3D printers. A brushed motor driver can then be used to control the current.

[TomcM]

So you need 0-220V DC for the coil. The coil will heat up and its DC reistance will change. So do you want a constant current supply?

Yes, ideally the current remains constant, since the braking force is proportional to current^2. Since there is already 132 W of power loss at room temperature, the coil heats up pretty quickly. Should I be looking into adding some feedback somewhere, or creating a constant current source? Problem is, I only know how to do that in low power circuits and I imagine that the practices in high power circuits would be quite different.

[TomcM]

For a one off application, it might be cheaper (and safer) to just buy a high voltage bench power supply than investing in the time to develop a custom circuit. Or if it's trivial to remove the coil, replace it with one that operates at a lower voltage and higher current. Probably 24V would be a good balance since 24V power supplies are commonly available for 3D printers. A brushed motor driver can then be used to control the current.

Borrowing a programmable bench power supply from the university might actually be the best idea.

The coil was a custom job and cannot be changed. It has 5500 turns and a 0.22mm diameter copper wire was used. Since the braking force is proportional to current^2 * turns^2 , this was the only viable setup that I could have had made.

[T3sl4co1l]

The main difficulty is isolation, as the coil is at mains voltage.  If you don't mind putting an isolation transformer in there, you could do that, and then have a common ground side that's easy to work with; a microcontroller can read the current and feed back to set the phase angle or whatever, using a PID loop.  Or if you don't mind that the MCU is riding on mains voltage, you can put it there directly; this can be a bit of a pain for development though, plus whatever kind of interfacing you're looking for (digital control e.g. serial port? just a plain on/off switch and potentiometer?).

For example, your typical plug-in power monitor (Kill-a-Watt and clones) uses an MCU riding on mains voltage, reading the load voltage and current, and calculating the stats displayed.  This works out because its interface is well insulated, just plastic buttons and an LCD, real simple.  If you wanted to, say, log values from a similar thing though -- you'd need some kind of port connection, and you can't just attach a serial port to mains voltage, you need isolation.  Digital isolators are a heck of a lot easier than isolation transformers, so that'd be a typical solution in that case.  But if you need to have a programmer attached to it as well (as for development), you'll at least need an isolator for that interface as well; which can be about as simple as a serial port isolator, but it might not be as standard as a serial port, so you might have to cook up something by hand, say.  At which point just using an isolation transformer might be more attractive.  So that's all that drives such a choice, really.

There are other ways -- you can isolate the sense signal itself, say with a Hall effect sensor (though these aren't great at low currents -- they aren't very sensitive, generally), or with an analog-digital converter and digital isolator.  (Pure analog isolators happen to be rather tricky to make, so it's usually actually better to convert it first.)  Or even sense the magnetic field directly (something a Hall effect sensor is adept at -- the advantage is all the turns on the magnetic core, versus sampling the current through a tiny (usually one-turn hairpin) winding).  Though sensing field has the downside of also being sensitive to the effect of the brake disc itself, the directionality of magnetic fields, and needing calibration.

Tim

[Terry Bites]

What levels of flux do you expect. You are going to need to measure it to validate your project? In the 10's of mT a TLE5501 might be just the job.
Any control loop wil have its time constant, so will the solenoid. Making fast changes to the flux will be compicated. Inductors are equvalent to mechanical mass. A PID cohntroller is great if you can grapple with the tuning issues.
I'd start by defining the range of currents, he flux you hope th achieve and the di/dt maxima.

BTW variacs do not provide any isolation from the mains- even when set to zero. Thats a problem that needs fixing from the outset.

If its only used intermitently then a stack of lead acid batteries or Lipos and a PWM controller will do. You can still get zapped though.  The investment in designing a controlled recifier with big old mains transformer will cost you much more and is a big undertaking leading to tears, recrimmination, the loss of your good eye and cross generational blood fueds. It will likely be dumped in landfill after your project.

[Kleinstein]

With 220 V and 0.6 A the would give out some 130 W of heat. With the ripple current and power factor the loss may be a little different, but not much.
So unless there is good cooling (not sure a fan would be sufficient) or only short time use, one should consider using less current / voltage or maybe change the coil (more copper).

One can extimate the magentic field from current * turns / (core gap + core length / µ).
This would be something on the order of 500 kA/m and thus quite some field.

[NiHaoMike]

The coil was a custom job and cannot be changed. It has 5500 turns and a 0.22mm diameter copper wire was used. Since the braking force is proportional to current^2 * turns^2 , this was the only viable setup that I could have had made.

Is the core one piece or can it be easily disassembled and the coil removed? If it can be disassembled, it looks like you could 3D print a bigger coil form (using a heat resistant plastic like PETG or ASA) and wind a new coil with thicker wire. In particular, using as much of the opening in the core (leaving only enough to give clearance from the disk) would help minimize resistive losses.