Page 5 of 7

Re: lab electromagnet from scratch

Posted: Fri May 11, 2018 12:37 am
by Rich Feldman
Chris, thanks for the electrical conductor price-per-pound data. Its importance is one of the main learnings and teachings of our projects.

Now an update on the three-inch magnet. Filling voids using TIG torch, in my inexperienced hands, was not worth the trouble. Minor "pits" remain after both bars were finished to the target length, with ends flat and square, within about 0.001". (As we pretend that 1" x 4" shapes can be called square.) Then it was only a matter of drilling a few holes, and tapping four of them, before a declaration that yoke assembly has begun.

Ever wonder about teenage amateur scientists driven by birthday deadlines? I'm facing the horrible spectre of five years elapsed since original post about this project. Here's one statement from early on:
This magnet project is to:
* get my hands dirty
* test my purported knowledge of E & M, engineering, and practical scrounging
* explore the low-cost low-power corner of magnets producing 1 tesla in 1 inch air gap.

Here's the post which presents the starter plates. Sort of like the stone, or nail, with which to begin making a tasty and wholesome soup. Also the first scale drawing. viewtopic.php?f=15&t=8600#p59448

As mentioned more than once before, and as the Mullins family well knows, electric power requirement is not very sensitive to pole area. But it goes up as the square of air gap length, and as the square of flux density B. And it goes down in direct proportion to conductor mass going up, if the average turn length doesn't change.

The scale drawings below illustrate that principle, and the 2015 evolution of my design. Grid is 1", just like in that early drawing. In all cases the cross-sectional area of flux paths in the yoke is slightly more than that in the cylindrical pole pieces. All coils have the same winding area and current density, and generate the same flux density in air gap. All pole pieces are 7" long, a consequence of my coil conductor choice.
The 6" magnet is 20" x 20" on the outside, and to my eye has an ordinary aspect ratio.
The 3" magnet in the middle uses 22% as much steel but 71% as much conductor and power (same as ratio of average turn length).
The 3" magnet on the right is what's being assembled. The original flat-finished plates go on the sides instead of the ends. That makes the air gap length infinitely adjustable, without any precision turning or boring. Predicted and measured steel weight is 56 lbs for the four central parts, and forecast to be 102 lbs when side plates are included.

Re: lab electromagnet from scratch

Posted: Tue May 15, 2018 1:58 am
by Chris Mullins

Having an adjustable pole gap would be really useful. That feature was way beyond my construction abilities. You mentioned another temporary winding, at 100V and 100A to get 1 Tesla with a 1" gap. Even with much less current on a temporary winding, if you can adjust the gap continuously you could experiment with saturation limits. Are you still planning to use that aluminum foil tape for the final winding?

Re: lab electromagnet from scratch

Posted: Wed May 16, 2018 7:02 am
by Rich Feldman
Yes, Chris, I want to use the aluminum strip coils. The material will need some work to make connections to the inside, and to avoid any shorted turns on the exposed edges. Any rewinding onto different spools will need a fixture, I think, and a way to clean up damaged edges on the way through. Facing a sort of deadline, and not much spare time, I might shoot for 1 tesla-inch using lesser coil(s) and lots more power. Duty cycle could be 2 seconds on and 2 hours off. Must avoid throw-away work just to meet the demo date. :-)

Here's the first re-assembly after degreasing and painting the yoke parts.
Joints between horizontal and vertical bars are clamped with 1/4-20 threaded rods. Just as tight as they would be with two 1/4-20 machine screws at the same torque, like the pole piece connections, but without thick-steel drilling and blind-hole tapping.
At the bottom we see a low-profile caster cart, presently on a flat concrete block while glue dries.

The upper pole piece and top plate are secured with an identical clamp set. Presently the gap is set to be 1 inch.
Without further ado, it's my pleasure to present:
Next step is to do some magnetizing. Another high priority is making some accessory parts for aligning the upper pole, and placing the upper clamp at the right height. Clamp rod tensions are matched by sound, like bicycle wheel spokes. If the upper clamp had a quick release feature, this contraption might make an effective mousetrap. :-)

Re: lab electromagnet from scratch

Posted: Thu May 17, 2018 5:52 pm
by Rich Feldman
My confidence was shaken to the core the other day, during casual magnetization experiments.
An unexpected phenomenon, which I think is from flux leakage, immediately reared its ugly head. The pole farther from the coil is _much_ weaker than the nearer pole. I bet the factor is worse in this slender design than in cores with stubby poles and stubbier gaps, like the Mullins'. Adding a coil around the upper pole won't improve the "loss" factor, it will just superimpose a strong pole on top and weak pole on the bottom.

Not explainable by the one-dimensional magnetic circuit model, indicated by blue lines in the scale drawings above, and in this signed one from a 2013 post:
3in_magnet.JPG (35.56 KiB) Viewed 6048 times
Never previously considered by me.

No time for discussion now. Anyone else want to chime in, or point to a magnet design book? We can measure flux any place we can encircle with a wire. We can simulate using FEMM, but its restriction to axisymmetric geometries limits the realism in this case. Who has access to a real 3D FEA simulator for magnetics, or even for static heat flow or electric conduction?

If the problem is what I fear, then three remedies come to mind. Use more ampere turns, change the thick steel design, or magnetically insulate the flux leakage paths with sheets of superconductor. :-)

Re: lab electromagnet from scratch

Posted: Fri May 18, 2018 3:48 am
by Richard Hull
Simple variable gap systems of your design are fine, but all systems of your type design have coils over each round pole section and much thicker flux path pieces equal to or of greater cross sectional area than the round poles of carefully chosen high permeability steels. It all depends on the final use for the magnets. Hysteresis loses in such special steel is not an issue in DC magnets, of course.

Richard Hull

Re: lab electromagnet from scratch

Posted: Fri May 18, 2018 7:37 am
by John Futter
I agree with Richard your flux return path steel is way too thin. I meant to photogragh several of our big magnets @ work but ran out of time.
These are made here in NZ by who make 95% of the magnets used in the semiconductor industry world wide and a great deal of the research type magnets as well. Click on the video at the top of the page to see one being assembled (a bit quicker than you are doing). note the massive flux return paths they use to get max field in the gap
I've seen their factory in Auckland and some syncnotron magnets they were making at the time 20 tons each and dozens of them. I also saw some magnets just out of the factory in test that had Danfysik labels on them
They use special steel that is made for them using Buckley systems steel recipe by a steel mill in Australia it starts to saturate at just under 2 tesla. They say that most normal steels start saturating around 1 Tesla.

Re: lab electromagnet from scratch

Posted: Fri May 18, 2018 6:04 pm
by Chris Mullins

Can you power the coil with an adjustable, constant current source, and measure the field strength at the face of each pole? When I only had one coil, I had around a 3 to 5% difference at each pole face, across a 1.42" gap (e.g. I'd read 144.9 mT on the bottom pole face (the one with the coil), and 137.7 mT on the top pole face). That spread also varied with measurement position. In the center of the pole, there was almost no variation, while I'd have up to 5% near the pole edge. The spread also varied a bit with overall field strength. I attributed the variation from center to edge of the pole to the asymmetry of only having one coil (one big motivator for finishing the second coil, apart from doubling the max strength).

I used 3" thick steel due to two design requirements:
1. The total cross-sectional area of the frame should be at least equal to the pole piece area, like Richard said above. Since the flux divides into two directions (into left and right sides of the frame), the area of each section can be half the pole piece area. I had 8" pole pieces (from the design requirements for the cyclotron), or 50 sq. inches. So the frame has to be roughly 25 sq. inches, otherwise it will saturate before the pole.

2. The frame width should be roughly the same as the pole diameter. This was more a rule of thumb I picked up, and although there are several examples where it's not the case, many working cyclotron magnets roughly my size were shaped like that (e.g. this one: from the Rutgers cyclotron), so it's an aspect ratio that works. In mine, that puts the frame at 8" wide, and the thickness at 3.125". I used 8"x3" = 24 square inches, close enough for me.

Even with that, my magnet efficiency is around 90% at 450 mT, and 80% at 820 mT across the 1.42" gap, so it's already starting to drop off.

If you're close to saturating that steel in the thin section, you might find it works closer to expectations if you lower the current. If you can make some measurements at low currents, and see how the performance varies as you crank it up, that might help in understanding what's happening.

I do plan on modelling my magnet, and in another thread Scott Moroch mentioned some tools to do that. The Rutgers and Houghton cyclotron folks have used those (some of which are free) and posted the info online, so we could start from their models and modify them to fit our magnets (which are pretty similar).

Re: lab electromagnet from scratch

Posted: Fri May 18, 2018 8:03 pm
by Richard Hull
Chris, Your measurement differential is a total non-issue. You seem to think it signifcant. So very much has to do with the critical placement of the sensor used to measure the strength. Equally important is the absolute flawlessness of the steel pole itself. Are there voids or imperfect alloying regions in the pole piece? Are there small imperfections in the face or numerous tiny bubble voids near the pole face measurement area? Most likely, however, the increased flux path to the other pole piece and any joints between the coil and the other pole piece would be the causative agent.

You are talking the difference of 1377 Gauss versus 1449 Gauss or a mere 72 Gauss differential. This is a pitiable differential which can be caused by you or the flux path very easily. Detailed magnetic theory and the issues involved with permanent or DC electromagnets are one of the most understudied areas by technical people forcing them to make general assumptions related to magnetic measurement and magnetic circuitry. That's right, a magnetic circuit! Look at the ohms laws for electrical circuitry versus the magnetic circuit laws.....They are, in effect, identical. (An exercise left to the student)

There is a slight difference in that in electricity we look at resistance in ohms and bypass the term conductance, its reciprocal. In magnetics, we look at permeability,(magnetic conductance), and not its reciprocal, magnetic resistance, reluctance, stated in "Rels". The Rel is virtually unknown, some consider it archaic, however electrical conductance in "mhos" is more commonly known and is still used.

Measuring magnetic field strength is like measuring neutrons. Both are less commonly followed paths for most technical people and just because you a have a neutron counter or a magnetic field strength meter, does not confer on you the ability to understand what you are measuring or how to apply and interpret information given by the instrument. A much deeper understanding of what you are dealing with at the material and theoretical level is needed.

I equate this with a person with a GM counter seeing a radioactive source read a 58 mrem/hr rate and another person with a different GM counter finding only a 20 mrem/hr rate from the exact, same source! Assuming each counter to have been recently calibrated by a national standards referenced facility, why the differential? Are they both right or both wrong or is only one reading really correct?? (Another exercise left to the student)

When I use the "exercise left to the student" phrase, it is a clue that we assume you to be inquisitive enough to be come a student and seek out references as self-directed learning would force a naturally inquisitive person to do in order to find out causative agents and , thereby, "learn".

In the end, I more or less blame the manner in which magnetics is dealt with, If the Rel and magnetic resistance were emphasized in magnetic circuits we would see more wire,(electrical), in the metal magnetic circuit and more joints,(electrical), in the magnetic circuit which would cause a diminution of current(magnetic flux,Gauss), at the pole face. We are more trained to see and recognize resistance in circuits, where magnetic engineers are more on the look out for keeping the most permeability, (conductance), in their circuits.

Reluctance is discussed here and its relationship to electrical ohm's law.

Richard Hull

Re: lab electromagnet from scratch

Posted: Fri May 18, 2018 9:17 pm
by Peter Schmelcher
I believe you should split the electromagnetic, half around the top and half around the bottom. This should change the magnetic circuit potential in the support arms to zero opposite the air gap, the reduced coil diameter requires less wire so you will get more amp turns from your spool.

My fading memory also recalls an advantage to actually covering the air gap with the windings making the flux lines more constrained within the air gap.

Just my 2 cents

Re: lab electromagnet from scratch

Posted: Fri May 18, 2018 11:13 pm
by Chris Mullins

No doubt I still have much to learn!

I was thinking it could be significant since my difference was much smaller (0.5% or less) once I had both coils in place, compared to a single coil on one pole.