105 replies to this topic

[Old_Hag]

I hope you aren't discouraged at all by the lack of comments, I think it's just because you're at a level with programming and engineering only a couple people on this site can understand. But once you finish this, and if it works as planned, I think it'll be without a doubt the greatest technical achievement ever on this site.

[pkiknex25]

Yes, this is so far above everyone's head that nobody can really comment.

[Connector1o1]

It seems as though good old rubber bands are too old fashioned and boring, right?
But really, I must say this is quite the project to take on, I agree that this will be one of the greatest technical achievements on this site and I think this build is underrated. It's so interesting to read! I never really commented but I always read every update, I apologize. I can't wait to see how the aluminum car parts turn out!

[ThisIsGabe]

Thanks a lot guys!
I'm going to be hopefully getting the milling all done Tuesday! I would have been able to do it this last week but the mechanical engineering sophomores had a milling lab and reserved all the mills :/
On a happier note, I've got the code written and compiled! 1024 lines of arduino code got done at 2am friday xD All that needs done is writing a function to send and receive SPI information from the DRV8303. However as it stands it can be used, it works, and I just need to actually energize the board!
Metal for the launch track arrived Thursday, and whats kind of nice is that it sort of works without any changes with an actual knex car! I just need to chop some metal off the center bar then I'm going to JB Weld them all together.
I also need to make the electromagnets and the drive motors, but they shouldn't be too hard to do.
Home stretch!

[ThisIsGabe]

*cough*
Well milling takes a tad longer than I thought it did..
Tuesday I learned how to mill (hooray!), and made the aluminum plate into a bunch of aluminum rectangles.
Today I made the rough rectangles into very precise rectangles! Now next Monday and Tuesday I've got about 8 hours of milling time reserved and I hope to get the cuts done by the end of the day.
BUT
I put the magnets on one of the 3d printed pieces and tried pushing one of the pieces of aluminum past it.....and when you hold it ~2mm away, you can definitely feel the force of the eddy currents! This is before even putting the other half of the magnets on, which should greatly increase the magnetic field (since the field of one half is mostly looped back on itself, and a second half will cause the field to go across anything placed between them).
So at worst I've got a very elaborate eddy current braking system!
I'm not going to put pictures here just yet as currently the aluminum is just rectangles that define the outside of the components

[ThisIsGabe]

Wheeeeeee I'm getting there! 

I'm getting milling done as quickly as possible, but I had the sudden realization that I'd need to mill out the track somewhat as well, so I'm currently machining out the track. I've broken the launch into 4 12" sections, and have 3 other sections for the station, block brakes and final brakes. Two of them have been machined down to ~2/100" of the ideal width, the other parts I hope to finish tomorrow. The track is in three parts, an aluminum rectangle with two side channels cut in, and two rails the correct width. JB Weld is going to be used as a fastener because it'll work

I need to ask my dean where in the engineering building I can set up the coaster, as I need the engineering building's power, hoping I get a chance to do that tomorrow.

To Do:

Finish milling track & Glue parts together

Mill wheel holder

Assemble track

Wind electromagnets

Assemble LSMs

 

Try to do:

Brakes

Drive tires 

[Rollerfreak987]

Sounds like a very interesting project... Would you mind posting some pictures, I'd really love to see what it looks like as of now .

[ThisIsGabe]

Sounds like a very interesting project... Would you mind posting some pictures, I'd really love to see what it looks like as of now .

Here's the only pic of the track I have currently,  http://www.sscoaster...ex-test-car-r28

And here's the magnet holder http://www.sscoaster...lsm-car-wip-r12

[ThisIsGabe]

Ok, really long winded update!

I got a place to set up in my school! Yay!

Track is done, I shot a time lapse which should go up soon.

The custom LSM car doesn't work! BOO!

This caused a minor (ok major) panic attack because oh my lord all this time and effort for nothing BUT

I decided to try mounting a single magnet under the car, and IT WORKED. Not only is the poor SS car that I screwed the mounting plate into able to handle the track, but I wound a single coil and it moved the car with a reasonable amount of power on just 5 volts!

https://www.youtube....h?v=0FDtHI_elNY

I got the PLC totally wired up to sensors and panels and all that fun stuff!

I had a minor problem of none of the proxys worked as planned, but after sleeping on the problem, discovered that a 1k resistor and a transistor made everything good. No idea how, but it all works now hip hop hooray.

 

ALL THAT REMAINS:

Wind the rest of the LSM coils

Final reprogramming of PLC

Testing and Programming of Supercell LSM Driver

IAAPA

 

One of the bonuses of using the K'nex car is I can use the knex track to hold the magnets rather than trying to make the metal track work, which is sweet! 

But the downside is I've spent at least 100 hours designing and 40 odd hours machining the custom car and metal track and now it's just a pretty decoration. >:I

Live and learn I s'pose

 

I'm coming in rather hot, as I really really REALLY want to have this finished by IAAPA. TBH I'll probably finish it Saturday or Sunday night, and spend the next week just crashed because I'm staying up waaaay later than I should working on this. Tonight is actually going to be the first time since last monday that I've gone to bed the same day I woke up

The engineering building is really creepy at 3am.

[ThisIsGabe]

Alright. Bittersweet news.

Something's going wrong with my drivers, because I can't get the coils to actually switch. I jury rigged a system that I felt would replace the driver board, but it didn't work.

Tonight or tomorrow I shall try changing the wiring scheme, using three mosfets as low side switches, and just having 12v on the high side.

[ThisIsGabe]

Alright. 

I've now realized that as it stands, the LSM system doesn't work. 

HOWEVER:

This does not mean that this is the end

I've learned a great deal about what should and should not be done, and come up with several optimizations that will be implemented in the next version!

• Magnetic sensors: Photointerrupters were cheap, but they added complexity, and didn't work very well with the PLC. Using a reed switch, I can remove all of the complex wiring that the photointerrupters required, and they can work with both the arduino and PLC without any additional circuitry
• LSM Coils: In the next version, I will make the coils independently driven. Instead of using three phases which means if one coil is shorted, all of them don't work. Independent driving will make detecting faulty coils very simple.
• Assembly/design process: I made a major goof by trying to design and implement everything at once. This resulted in me learning that the custom car wouldn't work as I was trying to put everything together. In the next attempt, the first step will be designing and building the car, then building a short test track to assemble magnets, then building the whole coaster around it.
• Keep it simple: I could have probably cut a few more corners, that made the assembly a lot more complicated. 

I'll be moving away from campus for a year on co-op, and while I'm away I'll probably take a few months break from the project, but I'll be working on the new design for the drive!

I'll take lots of pictures before teardown

See you later!~

Gabe

[TheSUCKCrew]

I do not recommend using reed switches for any project that involves magnetic fields. They get very unpredictable.

 

Other than that, I'm just kinda sad to see you're changing your plans to this extend. I wouldn't like to see you change this so that the coils are driven independently.

There should be a better way to test for faulty coils, right?

At least you're not giving up yet, though. My suggestions to get this project off the ground, would be to start with prototypes that are a lot simpler and use more forgiving techniques (like crafting the magnethousings with common materials and a glue-gun, instead of starting with CNC stuff right away. Lots of mistakes can be avoided that way.

 

Anyways, I'll be looking forward to the updates in the distant future.

 

Greetings,

Bart

Edited by TheSUCKCrew, 12 December 2015 - 09:27 PM.

[ThisIsGabe]

Thanks Bart!

I had really only been meaning to use reed switches for the "out on the track" proximity sensors, because they won't be by the coils, and they'll just be talking to the PLC. Or I'll look and see if there's wider photosensors.

I am somewhat curious as to why you don't recommend driving coils independently. What I've been thinking of is ~20 MOSFET H-bridges that are still driven externally, so that I could still treat it as a 3 phase motor, but it's easier to spot trouble coils. I'd been noticing that in order to get it to work, I'd basically have to have all of the coils pretty near identical, and any coil that was even slightly off would cause the overall flow of current to change. Individual driving would take care of it, as I could at worst fire coils one after the other.

I'm going to be going about it in a different order next time, because you were right. Trying to cram in everything at once, under a deadline was a very bad idea. What I want to do is make a car (or mod a K'nex car), then make just a straight section of test track, and get that working before starting on controls and a full circuit.

[ThisIsGabe]

Guess who's back...

back again...

Yesterday I finished the design of my H^4 H-Bridge PCB. I spent a few weeks planning these so that they would be more efficient than the old system. Because the wiring that I had to use with the old controller was quite frankly awful, I redid how the whole thing works together.

http://dirtypcbs.com...d260747876ef61d take a look! (BTW, if for some reason you want a copy of this board, wait a few weeks. Once I verify it actually works as well as I want, I'll post it to their store)

This board is designed to drive 4 coils (or even some parallel/series coils). Each H-Bridge has its own current sense resistor, and 741 amplifier to let an Arduino read the current. The big difference with this board is the communication system. On the middle top and middle bottom, I've put Ethernet ports, which will let me link the H-Bridge board to an arduino shield by simply plugging common Ethernet cables in. I made this to pump at least 4A constant per channel, although with heatsinks I could probably put a lot more through especially in pulses.

Now just cause they're so neat, I'm gonna do a brief plug~

Check out dirtypcbs.com - I'm getting ~10 copies of this board for $25. For comparison, the single copy of my last board cost $53. I'll do a more thorough review once the boards come in, but at least I do love the pricing!

[ThisIsGabe]

Still waiting on the shipment of boards (gotta love cheap Chinese shipping) - they've been shipping since last Tuesday and I'm impatient xD

While I wait, I've gotten some more planning done for the launch system as a whole. This month's section is position detection!

Now, there's several ways that I could detect the position of the car as it launches, and they've all got their ups and downs. Laser reflection is expensive, reed switches won't work around the coils, but photogates did work fairly well - the problem was the width that I had.

To combat this, I'm planning another Dirty PCB. What I'm going to do is have one "double use" board, which will be able to hold either infrared LEDs or infrared phototransistors. Then I'm going to position them so that the whole car passes through the beams, rather than trying to make something pass through a slot (which was extremely difficult). The sensors will connect to a secondary arduino mega through Ethernet cables (which really simplifies wiring), and then the two arduinos will be able to communicate with each other. I want to get boards made and ordered by next weekend, so I can begin working on designing big beefy new electromagnets.

[dgohstand]

I guess this is a dead thread but I really am curious as too what happened with this project. I was going to attempt to build a LSM launch so I looked around and this seems to be the only thread with someone trying to build this. So I went through the whole thread and boy was there alot of info! I didn't realize how complicated it actually is but I will still try it. Anyways, I was wondering if you're still working on this? 

 

Sorry again about posting on a dead thread but I couldn't stop myself. 

[ThisIsGabe]

Actually I probably should have posted more stuff but I was trying to wait until I had motion (which im hoping will be by next week).

I made the H bridge boards, assembled and tested and originally they were shorted due to a symbols error, but by blind luck I had swapped N and P channel fets, and they had the same pinout. So the boards are working, and can pass current.

I have new electromagnet cores, in a more sensible position. Rather than making passthrough I'm putting a magnet on the bottom of an existing car. I've already got the car mounted, and a test track set up. 

Coils are going to be wound on a 3d printed coil form. This was 3d printed by shapeways and turned out amazing. Tolerances were just right and i should be able to fit LOTS of turns of heavy gauge wire. All the parts are in my apartment but due to some school related travel, I am not.

Hopefully during the week I'll be wrapping the first coil, and next weekend I should be able to try to make the car shoot off a single coil. If it works, I'll have 7 more coils  printed and have a launch section to work with.

Don't worry, this project isn't going to die, I've spent too much money and time at this point to give up now xD

[Caps]

This was definitely interesting to read through! 

 

I've been working on my own as well. As a proof of concept I threw something together just to see how it would work (With only 4 magnets; more are gonna be added)

 

 

My trains:

http://i.imgur.com/5oC2pLT.jpg

http://i.imgur.com/dh2lDcI.jpg

 

Control panel I'm working on: 

http://i.imgur.com/GhXKNKJ.jpg

 

The main switching box:

http://i.imgur.com/hKCTkBX.jpg

 

The biggest issue will be latency over USB but I've got plans to use predictive timings based on some interpolation (lagrangian) of the sensor data so we'll see how that goes

 

 

Edit: I uploaded another video that shows it better

 

Edited by Caps, 06 October 2016 - 03:11 PM.

[dgohstand]

Wow, I like what you've built. You should start your own thread on your launch though. Also a description of how your launch works would be nice. But based on what I saw it seems to work well considering there are only a few magnets. Keep up the great work! 

[BGTKing]

Cool work Caps! Welcome to the site and we look forward to seeing how your LSM progresses! 

 

PS does the turkey pot pie give you extra power?