CorelDraw 18

 

CorelDraw 18

Cool! Thank you!

One question: how fast do you think will be in milliseconds (or nanoseconds) to switch the routing from a magnet to another? And the polarity? Trying to understand how flexible this system would be to fast switch between magnets.

Thanks again

 

That is a difficult question to answer, it would depend on several different factors such as...

Speed of the micro.
How good you are at the coding.
How much you can get done with hardware in the micro vs. software.
Whether you use polling or make use of interrupts.

I'm still pretty new at using micros, so I'm sure somebody with more experience could answer much better.

 

That is a difficult question to answer, it would depend on several different factors such as...

Speed of the micro.
How good you are at the coding.
How much you can get done with hardware in the micro vs. software.
Whether you use polling or make use of interrupts.

I'm still pretty new at using micros, so I'm sure somebody with more experience could answer much better.

Fair enough for me. I shouldn't have problems with coding, I have been coding for 35 years, but I am weak in electronics. I'll experiment with your design and move from there.

I really appreciated your help, your design is super!

All the best,
Fab.

 

You're welcome.

Keep us updated.

 

Ok, here ya go reduced component design.

Don't think so. First mentioned in post #13, there are multiple sneak paths around most of the coils. Add reference designators and pin numbers to each coil and I will describe one.

Unless I'm having a senior moment.

ak

 

Ouch, I can't believe I missed that. You are correct the circuit won't work.

So I guess it was me having the senior moment.

Sorry, fablau

 

Easily done.

This challenge is a bit of a conundrum. There is no obvious easy solution - even if you use the matrix solution to reduce the steering logic, you still end up with one coil one driver, whether that be a full-bridge, or a split supply and a half-bridge. The least component option is the x-y matrix and dual coils.

 

One question: how fast do you think will be in milliseconds (or nanoseconds) to switch the routing from a magnet to another?

It's not just the routing you have to consider. The inductance of any coil delays current build-up, so will limit how fast you can reach full magnet strength.

 

Switching in software won't be more than a few mS. Depending on the coil/magnetics situation, likely to be 10s of mS.

 

Ouch, I can't believe I missed that. You are correct the circuit won't work.

So I guess it was me having the senior moment.

Sorry, fablau

I still don't see the problem though... Your schematics looks good to me... what am I missing? Please, help a super-junior-baby here to understand. Thanks!

 

Normal inductors are not *semi* conductors. They conduct (and in your case, require) electron flow with equal facility in both directions. In the #38 schematic, close the two bottom-right TRIACs (bottom of the right column, right end of the bottom row). That is, replace them with straight lines. Now follow the current path from the left-side Half Bridge Output, through the TRIAC, diagonally up-and-right through the coil, and through the column TRIAC.

BUT ...

Current from the column TRIAC can go up to the next coil, up-and-right through it, left to the next coil, down-and-left through it, down to the next coil, up-and-right through it, and out the row TRIAC. There are many other similar paths around the intended coil.

X-Y matrix drive works when driving an array of LEDs because an LED is a semi-conductor (as long as the reverse voltage is not too high). Same for transistor base-emitter junctions, zener diodes, etc. If your system had incandescent light bulbs, then full-wave-rectifying the incoming AC and putting a diode in series with each bulb would work. But your electro-magnets need the full sine wavevoltage and current waveforms, and no cutsie-switching trick will get around that.

ak

 

Normal inductors are not *semi* conductors. They conduct (and in your case, require) electron flow with equal facility in both directions. In the #38 schematic, close the two bottom-right TRIACs (bottom of the right column, right end of the bottom row). That is, replace them with straight lines. Now follow the current path from the left-side Half Bridge Output, through the TRIAC, diagonally up-and-right through the coil, and through the column TRIAC.

BUT ...

Current from the column TRIAC can go up to the next coil, up-and-right through it, left to the next coil, down-and-left through it, down to the next coil, up-and-right through it, and out the row TRIAC. There are many other similar paths around the intended coil.

X-Y matrix drive works when driving an array of LEDs because an LED is a semi-conductor (as long as the reverse voltage is not too high). Same for transistor base-emitter junctions, zener diodes, etc. If your system had incandescent light bulbs, then full-wave-rectifying the incoming AC and putting a diode in series with each bulb would work. But your electro-magnets need the full sine wavevoltage and current waveforms, and no cutsie-switching trick will get around that.

ak

Thank you for your clear explanation, now I see where the problem resides... just wondering: what if I make the coils "becoming" semiconductors by adding diodes at their inputs in some way?

 

Yea, I have to apologize again, that was a really stupid mistake.

There could be a solution, but is has its own issues.

1. The coil supply voltage range would be limited.
2. It adds components. (of course)
3. I can't test it for lack of the proper LTSpice models. (and actual components)

But basically it uses the fact that the "sneak around current" has to go through multiple coils to complete a path.

So using this fact and placing something in series with each coil that doesn't conduct at the lower voltage you can keep those coils from seeing current.

Originally I was thinking DIAC, but there aren't any available that suit the need. (and of course that would have been too easy)

So now I'm thinking Triac with 2 back to back zeners at the gate, but I have no way to test this, and i doubt you would want to add the extra components anyway.

So basically I need to quit while I'm behind, and again...sorry.

 

Yea, I have to apologize again, that was a really stupid mistake.

There could be a solution, but is has its own issues.

1. The coil supply voltage range would be limited.
2. It adds components. (of course)
3. I can't test it for lack of the proper LTSpice models. (and actual components)

But basically it uses the fact that the "sneak around current" has to go through multiple coils to complete a path.

So using this fact and placing something in series with each coil that doesn't conduct at the lower voltage you can keep those coils from seeing current.

Originally I was thinking DIAC, but there aren't any available that suit the need. (and of course that would have been too easy)

So now I'm thinking Triac with 2 back to back zeners at the gate, but I have no way to test this, and i doubt you would want to add the extra components anyway.

So basically I need to quit while I'm behind, and again...sorry.

No need to apologize! I really appreciated your suggestion and I still think your design must work with some modification.

I like the idea to stop the current to flow at low voltage, and maybe I can find a solution to that. I don't mind to add more simple and cheap components if needed. What I don't like is the idea to use 250 L293Ds and 125 shift registers, plus extra stuff to control over 500 PWM points with 500 magnets, when that's not actually needed.

Thanks again

 

I would like to dial it back to a theoretical level. I use LTSpice to depict the idea. The scheme deals with the extra paths using diodes. The entire array is powered with a reversible polarity power source (V1) to allow for reversing polarity of the selected magnet. The extra paths are blocked with a pair of diodes at each magnet. To deal with the diodes, two switches are used for each column (for example, S1 and S2). One is used when the power is one polarity and the other is used when the power is the opposite polarity. All the switches are bidirectional. V2 through V7 depict control signals used to turn the switches on and off. These are TBD for now. I don't know how to deal with the voltage created by the magnets when turned off. Perhaps someone can suggest how to deal with these. Is this a start?

 

Call me crazy or stupid but I think that layout would work even without 2 column switches.

Of course I could just be having another senior moment.

 

I would like to dial it back to a theoretical level. I use LTSpice to depict the idea. The scheme deals with the extra paths using diodes. The entire array is powered with a reversible polarity power source (V1) to allow for reversing polarity of the selected magnet. The extra paths are blocked with a pair of diodes at each magnet. To deal with the diodes, two switches are used for each column (for example, S1 and S2). One is used when the power is one polarity and the other is used when the power is the opposite polarity. All the switches are bidirectional. V2 through V7 depict control signals used to turn the switches on and off. These are TBD for now. I don't know how to deal with the voltage created by the magnets when turned off. Perhaps someone can suggest how to deal with these. Is this a start?
View attachment 211389

Call me crazy or stupid but I think that layout would work even without 2 column switches.

Of course I could just be having another senior moment.

Yes, it works Analog Ground! The double switches are actually needed for reverse polarity, according to the polarity use either switch. Tested with iCircuit and works!

See the attached image below, I just put regular manual switches in place of Triacs, but it seems to work. Now time to see how complex becomes by adding more magnets... but I guess the trick is to add 2 diodes to each magnet and then the correct number of Triacs accordingly... do we all agree with that?

 

Call me crazy or stupid but I think that layout would work even without 2 column switches.

Of course I could just be having another senior moment.

The pesky extra paths are tricky. I think the separate column switches are needed to allow the diodes to block the extra paths for the dual polarity power supply.

 

Bonus goodie: it looks like it is possible to fire more magnets at the same time!