Let's go back to first principles. How much Mag force do you need? What are you trying to move, how big & how much does it weigh and on what surface? How far is it from electromagnet to electromagnet? What strength is the magnet attached to the object?

Have you actually built a small testBed?

 

Interesting project. I would review some past designs to see what worked and what worked well.
https://dspace.mit.edu/bitstream/handle/1721.1/17620/54698735-MIT.pdf?sequence=2&isAllowed=y

Claude Shannon's Theseus (1952) One of the first actuation systems that used magnetism to move an object on a table was Claude Shannon's robotic mouse, Theseus, developed in the 1950's [46]. Here, a robotic mouse was equipped with sensors on its whiskers, enabling it to sense contact with one of the walls of the maze, at which point it would rotate ninety degrees and attempt to move forward again. In this manner the mouse constructed a map of the maze and could then run the maze a second time without hitting a wall. This was perhaps one of the first "learning circuits" of its type, and this type of artificial intelligence was the main thrust of Shannon's research with the project. The actuation mechanism was also a novel introduction: electromagnets were mounted on an XY plotter under the surface of the (aluminum) table, and magnetic forces grabbed the metal mouse in order to move it on the table. Multiple electromagnets were used, allowing the system to rotate the mouse on the table. Theseus seems to be the first documented instance of this XY-plotter and electromagnet actuation system, used extensively in later HCI work as well as in commercial toy chess sets.

 

If you look on page 3, you'll find the "Switch Output Peak Current" parameter which 3A specified... or am I completely off? It could be, as I said, I am newbie!

Thanks

Fab.

Post #20 covers some issues. Thew issue is power dissipation. 500 mA with a switch resistance of 15 ohms is 3.75 watts of power. The maximum power for the part is less than 2 watts. The switches in your application probably need an "on resistance" of 1 ohm or less which keeps the power in the switch less than or equal to 0.25 watts. Also, there are ways to make a bidirectional switch with discrete MOSFETs. Perhaps someone can suggest a good circuit?

 

Each magnet may take from 20 to 32v with a max current of 500ma each.

That's going to need a serious amount of enamelled copper wire in each magnet. So 500 magnets are going to be ....... expensive! What is your budget for this project?

 

Look at the test conditions not just the result.

3A for 100nS. At 0.1% duty cycle...
That's 100nS in every 100,000nS...

Not a snowballs chance of being useful...

Oh, I see now... so, what's the "real" max current there?

 

That's going to need a serious amount of enamelled copper wire in each magnet. So 500 magnets are going to be ....... expensive! What is your budget for this project?

A few hundreds $$... yes, pretty expensive. But until I can't figure out the electronics to drive it, no way to move on!

Thanks

 

Let's go back to first principles. How much Mag force do you need? What are you trying to move, how big & how much does it weigh and on what surface? How far is it from electromagnet to electromagnet? What strength is the magnet attached to the object?

Have you actually built a small testBed?

Yes, already built a prototype, magnets are small and powerful is around 50 gauss, and aiming to move small objects. But for the purpose of this discussion, I am trying to figure out the best driving circuit and the important specs of the magnets are their voltage of max 32v with max current of 500ma. Thanks!

 

Interesting project. I would review some past designs to see what worked and what worked well.
file:///home/fred/Downloads/54698735-MIT.pdf

That MIT project is exactly where I got most of my ideas, but.... they used 64 magnets driven by several L293D and shift registers. And here's the trick... I'd like to be able to drive many more magnets than that (over 500), and I think there must be a better way to do it instead than using over 250 L293Ds and over 125 shift registers. Or am I mistaken? Is that the only way to do it??!

 

Can't be done with bi-drectional switches

Here's my take...(diagram 1) Each NMOS/PMOS pair controls flow depending on which way up the polarity is (Forward/Reverse), but there's no easy way to handle the isolation between rows/columns, as a result you need 4 MOSFETs for each coil. You can reduce the number of MOSFETs by using a split supply, then you need only one MOSFET pair per coil, still that's 1000 devices. (diagram 2)

There is a way to reduce complexity though. But it involves modifying the electromagnetics. If we use 2 electromagnets per location, either bifilar wound, or centre tapped then we can revert to a simple X-Y matrix! (diagram 3). To keep costs down run the coils at 1A then need 1/2 the number of turns on each leg.



Split supply


Split coils

 

This is where I would start since I know how to program a PSoC.

The MUXs do not use the “0” output to avoid needing to control the data line.

Basically the master PSoC addresses the 4 PSoCs that address the MUXs that operate the enable line on the bridges also the master controls the forward and reverse lines on the H bridges that are all tied together in parallel. (will probably need buffering)

The PSoCs may be able to do all the switching in hardware, but if not it can be done in software.

The diagram as shown can operate 600 magnets.

Any questions just ask.

 

If you look on page 3, you'll find the "Switch Output Peak Current" parameter which 3A specified... or am I completely off?

Off. Look only at steady-state (continuous) ratings. And, for better long-term reliability, derate things by at least 25%; 50% is better. So a 1 A continuous switch should see no more than 0.5 A.

ak

 

Bell Labs. Without a doubt, the World's foremost private research lab.

Claude Shannon was only one of the many intellectual luminaries which worked and prospered there, including several Nobel Prize winners.

Over 60 years later, the Theseus demonstration is still awesome.

 

Can't be done with bi-drectional switches

Here's my take...(diagram 1) Each NMOS/PMOS pair controls flow depending on which way up the polarity is (Forward/Reverse), but there's no easy way to handle the isolation between rows/columns, as a result you need 4 MOSFETs for each coil. You can reduce the number of MOSFETs by using a split supply, then you need only one MOSFET pair per coil, still that's 1000 devices. (diagram 2)

There is a way to reduce complexity though. But it involves modifying the electromagnetics. If we use 2 electromagnets per location, either bifilar wound, or centre tapped then we can revert to a simple X-Y matrix! (diagram 3). To keep costs down run the coils at 1A then need 1/2 the number of turns on each leg.

View attachment 211236

Split supply
View attachment 211247

Split coils
View attachment 211250

Thank you Irving, I'll study your diagrams and let you know if any questions arise. Appreciated!

 

This is where I would start since I know how to program a PSoC.

The MUXs do not use the “0” output to avoid needing to control the data line.

Basically the master PSoC addresses the 4 PSoCs that address the MUXs that operate the enable line on the bridges also the master controls the forward and reverse lines on the H bridges that are all tied together in parallel. (will probably need buffering)

The PSoCs may be able to do all the switching in hardware, but if not it can be done in software.

The diagram as shown can operate 600 magnets.

Any questions just ask.

View attachment 211258

Wow, thank you very much, that's awesome and I just need to study it. It is my understanding that I'd still need 150 H bridges, right? Will that circuit fire one single magnet at a time or it is capable of firing more than one at a time?

Thanks again!

 

Off. Look only at steady-state (continuous) ratings. And, for better long-term reliability, derate things by at least 25%; 50% is better. So a 1 A continuous switch should see no more than 0.5 A.

ak

Thank you for clarifying. So, what's the max current that switch can support in continuous state? How can I understand that?

Thanks again.

 

The block diagram I posted is based on using 1 bridge per magnet, and can only activate 1 magnet per MUX.

In theory you could activate 40 magnets at a time, so I would imagine the usefulness of that would depend on how you laid them out.

 

The block diagram I posted is based on using 1 bridge per magnet, and can only activate 1 magnet per MUX.

In theory you could activate 40 magnets at a time, so I would imagine the usefulness of that would depend on how you laid them out.

Oh, I see... it looks like then that the way I have done it so far would use pretty much the same number of components (see the first graphic I posted at the start of this thread): for driving 500 magnets I'd use 250 L293D and 125 shift registers.

 

Ok, here ya go reduced component design.

2 Half Bridges
64 Triacs (1024 magnets) Adjust as needed
4 16:1 Analog MUX506s (1024 magnets) Adjust as needed
1 Micro

Simplified version attached with only 25 magnets.

Operation:

The micro produces an output to the mux data line, this determines the turn on mode of the Triac. (your choice) You will want to switch this automatically with the bridge polarity. (one mode is more sensitive than the other) You may end up just placing the data line on either ground or positive if the sensitivity ends up not mattering. (with a resistor)

The micro controls the half bridges to reverse the power polarity.

The micro addresses the MUXs.

The micro enables and disables both the bridges and MUXs.

So…start with all mux outputs and bridges disabled…set an address…set the polarity of the half bridge outputs…enable the MUX switches and the bridges.

To change magnet…disable the MUXs…disable either bridge…change address…re-enable the mux and bridges.

Changing the polarity of the magnets can be done any time by simply reversing both bridges with the MUXs enabled.

This design only operates 1 magnet at a time.

I haven't worked out the suppression system, probably just need snubbers at each coil, maybe none at all.

You can replace the MUXs with shift registers, but I believe the coding would be simpler with the MUXs.

EDIT: Image removed due to faulty design.

 

Ok, here ya go reduced component design.

2 Half Bridges
64 Triacs (1024 magnets) Adjust as needed
4 16:1 Analog MUX506s (1024 magnets) Adjust as needed
1 Micro

Simplified version attached with only 25 magnets.

Operation:

The micro produces an output to the mux data line, this determines the turn on mode of the Triac. (your choice) You will want to switch this automatically with the bridge polarity. (one mode is more sensitive than the other) You may end up just placing the data line on either ground or positive if the sensitivity ends up not mattering. (with a resistor)

The micro controls the half bridges to reverse the power polarity.

The micro addresses the MUXs.

The micro enables and disables both the bridges and MUXs.

So…start with all mux outputs and bridges disabled…set an address…set the polarity of the half bridge outputs…enable the MUX switches and the bridges.

To change magnet…disable the MUXs…disable either bridge…change address…re-enable the mux and bridges.

Changing the polarity of the magnets can be done any time by simply reversing both bridges with the MUXs enabled.

This design only operates 1 magnet at a time.

I haven't worked out the suppression system, probably just need snubbers at each coil, maybe none at all.

You can replace the MUXs with shift registers, but I believe the coding would be simpler with the MUXs.

View attachment 211309

Wow, this is fantastic! Thank you ElectricSpidey! This is much simplified! Now it's time to study your schematics and your listed components... I'll let you know if any questions arise. Thanks again very much

 

One question: What program did you use to draw your schematics? I love it! Thanks again