• ### Evaluation board

It seems it's gonna be optimal to freeze that project for a while, but beforehand I decided to make a summary of some sort: that EMF retention technique worked well, making schematics and PCB available is gonna be great for someone willing to experiment with it. Also, previous versions were kinda clumsy, I've learned a little from other stuff I designed recently
That version won't high voltage very much, main focus was onto providing lowest possible resistance near Q2. 40V IRFH7004 were a best choice from what was widely available in my country. Higher voltage means higher efficiency, but hey, most of benchtop power supplies are 0-30V or close to that, should be fine : )

Using solder jumpers you can choice between using two PSU (up to 28V) and one PSU (18V max) arrangements. Also, there are two potentiometers to adjust upper and bottom current limits, which you can disarm in similar fashion, by desoldering JP1 and JP2, to use it with some external controls if you wish (0.5 amps/volt, 0->5VDC with a stock 0.1ohm current-sensing resistor)

Oh, and AD8211 is used there to provide somethat better on-state resistance capabilities, as it provides 20x gain. Anyway, typical electromagnet has a resistance around 15ohms or so. I chose 0.1ohm since lower values are NOT widely available here (that is funny), it's not like I'm a fan of 1W THT resistors or anything , )

• ### Optimizing magnetic scheme

Permanent magnets can trick you into thinking, that 20kg of attraction force is easily achieved. Everyone seen electromagnetic locks, some can do almost 500kg of holding force! However, magnets from electromagnetic locks differ from rare-earth ones, since rare earth magnets can do the same with very long magnetic loops, gangsta!

Achieving something similar without superconductive magnets is quite tricky. So for almost a year I worked onto control circuit that can maintain EMF without converting it into heat very much. And even got some great results, however with current semiconductors it helps 50/50: very impressive overall, not enough for my application. So...

It's obvious for me now, but month ago it truly wasn't : )
There was something that can be done with magnetic "design":

There, poles of an electromagnet are pretty close to poles of a permanent magnet (or steel bar), with that in mind it's closer to electromagnetic locks, while by trying to align magnets, it still has somewhat nice operating range.

How gap affects force? I've made a simple setup with my desktop CNC to sorta map it:

As I use parts from an old disk drive as a linear rail for magnet, it had a very noticeable friction. Also, linkage between scales and a moving magnet wasn't perfect, as it tended to be springy...

But that was enough to get needed data:

• ### BLDC controller bughunting

Time to time work with electronics is deeply confusing.
You may have 3 identical FET drivers connected same way, but still - only specific one burns out. Again and again. I've changed resistance (Rgate), and it seems that situation with heating improved a lot (switching now occurs in a neat way, without confusing oscillations, I was impressed: originally thought that high control currents would make everything better, but it seems that they require a bit more thoughtful layout with damping circuitry)

However, one transistor from six still overheats, low-side. Reason is simple:

• ### BLDC testing rig

Long time no see!
For two months melancholy was my companion, but it seems that season ended x)

And I want to introduce BLDC testing rig, as stated somewhere above. Here it is:

• ### BLDC alpha-test!

I fixed that board from previous update. I wasn't carefull enough and forgot to connect some nodes together, and though MOSFET drivers still die when I power everything with voltages greater than 10V (dV/dT problem?), under that everything works fine. So, I tested it out and it worked out. Hooray! : )

It turned out, that it's pretty complicated to measure maximum power output of a BLDC and it's similarly hard to measure certain effects by an oscilloscope as everything is very noisy. Inappropriately noisy! After a bit of confusion I utilized the fact that motor won't start being underpowered, then - compared modes with and without "advanced" retention

As you can see, it utilized power more effectively with retention. Can't say "how much" in numbers, yet it means that there is a room for further improvement at least! It's great that even first steps in that direction give some noticeable results. That means lesser heat irradiation and higher power density are possible

• ### BLDC controller trials

I had an intention to use amazing PowerPak housings for transistors. And guess what? I managed to fail that twice by assigning things to a wrong pins. Firstly, I messed with a gate and after milling second version I realized that drain and source were swapped. So... I soldered TO-220 instead. Here is a picture of board's better years:

PCB at the bottom would serve as a measurement tool for testing bench, it has 12bit DAC and instrumental amplifier with adjustable gain to measure torque. More of that in next updates as there is some problems with that controller. It works, overall, but "rings" like crazy. Here is a photo with re-soldered FETs and voltage on transistor's gates:

• ### Full-fledged theoretical "How-to"

Hello! Made a site to publish theoretical info in convenient manner

• ### BLDCiing it forward

Next logical step - is to couple theoretical findings with BLDCs, it is fun, educational and even useful at some point! In previous update that was a brushed motor... Pretty unfortunate - it's hard to imagine worse enemy than brushes with that approach.

I started to develop an experimental BLDC controller and imagine my surprise when I realized that circuitry, mentioned in previous posts, fits there ridiculously well! People, familiar with BLDC controllers, would probably notice that there is not much to change. Of course, specific control methods required and some hardware tweaks also, but hey, it's pretty neat : )

Generally speaking - there are high/low side FET's connected to each phase of BLDC coil, therefore there is an opportunity to shortcut coil from GND to GND via that. Or from VCC to VCC. Providing low-resistance path for a current, low voltage drop and so on... (topics from previous posts)

It's pretty pricey to implement analog current-control for each phase (thanks to DACs, low resistance and preferably hall-effect current measuring ICs, logic e.t.c.) - in this iteration, I decided to use old-fashioned way, outsourcing computational power and commands from an external MCU. Not the most elegant solution, neither a reliable one. Let's hope, that it would work without occasional fireworks.

Compromise is to use onboard timer and only ON/OFF signal to switch coils would be sent externally, however it would be a tedious process to make it on one layer board, with all required logic IC's. Simplified everything to a maximum degree possible, with one current sensing IC for all 3 channels:

Had lot of fun tracing that stuff! Aside of queer shape, it should have a damn good resistance and heat dissipation properties. Inductance should be less as well.

Not sure, that I gonna manufacture this one soon, so there is a room for corrections.
I'd like to make traces which go from drivers to a gates wider, as I see it now
(there are interesting 4A source/sink MOSFET drivers)

P.S. And that is our test subject!

• ### DC motors?

This time I've got a more direct reading than current. Pivoting torque.
Theoretical parts:
- Part 1 (how efficiency works while you charge an electromagnetic field)
- Part 2 (why discharge time matters and how it affects heat dissipation)
- Part 3 (that one was half-wrong, but heat dissipation part is likely to be true, read carefully)
- Part 4 (what affects discharge time)
- Part 5 (how to discharge field slowly using MOSFETs)

Current is a great reason to assume that magnetic field is somewhere... there, but not that convincing - I can imagine some unaccounted nuances, there are lots of them, usually. And people try to decrease their amount by doing further research.

It was necessary to prove, that along with current it produces appropriate force, as we use unusual methods to work with an electromagnetic field here, so I made that thing:

P.S. "Power distribution question": I don't know, why it charges field at that rate exactly, previously I thought, that it must be an unused part of U^2/R power, but now I see, that it's kind of different in reality. More complicated? Or opposite : )

• ### Electromagnet discharge guide

Yep! Complete discharge guide!

I don't want to repeat that was there previously anyway, so - links.
Theoretical parts:
- Part 1 (how efficiency works while you charge an electromagnetic field)
- Part 2 (why discharge time matters and how it affects heat dissipation)
- Part 3 (that one was half-wrong, but heat dissipation part is likely to be true, read carefully)
- Part 4 (what affects discharge time)

This time I want to make a finishing pass on discharge topic. It looks like that:

In theoretical Part 4 we came to a conclusion, what voltage drop affects discharge time significantly. Must-read, but simplified - coil tries to produce constant current while discharging and it's a very easy task if voltage drop is minimal. One approach is to use a diode, however, 0.4V is pretty high. And that is where MOSFETs come to play, acting as a low-resistance load.

Zener diode in MOSFETs structure is very helpful, since it can handle current until FET opened completely.
That's why this process has 3 stages and not two.

For example, we have 1A of current, conductive channel provides 10mOhm, U = I*R, voltage drop using this method is only 0.01V, 40 times smaller than what we have on a diode! There is a room to play, in different conditions channel shows different resistances. That's what I've got with a random transistor as proof of concept: