Macro-scale digital mirror device

INSTRUCTABLE: www.instructables.com/Digital...
Applied Science DLP demo: • How Digital Light Proc...
Daniel Rozin: • How This Guy Makes Ama...
0:00 Introduction
0:58 DMD to scale
1:40 Initial testing
2:22 System construction
3:40 Flipping mirrors
My last project focused on building a large-scale version of a CD player. With the same spirit, I decided to build a scaled up version of a Digital Micromirror Device (DMD). The inspiration for this project came from Ben Krasnow's Applied Science channel: How DLPs Work. In this video, Ben provides a great explanation for how Digital Light Projectors work and builds a demonstrator. He mentions the possibility of building a mirror array with solenoids to flip each mirror individually. That idea was always interesting to me, and so I finally decided to try it out.
DMDs usually consist of several hundred thousand mirrors! The device I built has only 25 mirrors... Each mirror can flip between two states using a push-pull solenoid. The pattern on the array is controlled by an Arduino.
This project is also related to the artwork of Daniel Rozin. He builds "mechanical mirrors" out of objects arranged in a 2D array that flip to form an image. Often the array displays an image of the observer in front of it, similar to a mirror.

[@baroque_engineer]

The method to reduce the current draw of the solenoids is to power them through parallel RC network. This is the classic trick that has been used for ages when driving relays. See, a solenoid needs high current only to initially overcome a static friction force. A hold current is typically much less. So what you need to do is to determine which current is sufficient to hold a mirror. You can do it, for example, by connecting it to your lab PSU and gradually reducing voltage/current until the solenoid releases. Then add some 15% to the threshold current and calculate the resistor taking into account the resistance of the solenoid coil itself. Determining the value of the needed capacitor takes some experimentation.

[@melody3741]

So you mean power it then drop the power down to hold it? Like a compound bow, sort of?

[@baroque_engineer]

@@melody3741 Yes. The high-current spike through the capacitor activates the relay/solenoid, and then the holding current flows through a resistor.

[@btmovie905]

I am from India I want to contact you

[@agentslimepunk]

I think I could fall asleep to the sound of flipping mirrors. There's something really relaxing about this

[@make.anything]

Man I love the sound of that thing. Awesome build!

[@frankgutierrez6016]

I work for Texas Instruments. My first 8 years working in the DMD fab.

[@avocadoarms358]

This video deserves mid-rolls, you deserve to be paid by KZbin properly, get on that filming grind and get us a 12-14min video man! Great stuff

[@akkudakkupl]

You can lower the relay current by using PWM. 100% for pull-in, 50% for holding. Or you can have a resistor in series with the coil that has equal resiatance to the coil, have the resistor bypassed by a capacitor so that the time constant of the RC allows for reliable pull-in. Both methods lower the current by half. You can also use a zener in series with the back EMF diode to cut down the release time by around half. Its voltage should be the same as the coil voltage, or two times that for even faster release.

[@jbumstead21]

Thank you for these ideas. Very helpful. Someone else (xenoscientific) suggested that "best solution in terms of speed and circuit size would be to have two driver transistors for each solenoid: use both for activation and one for holding." I am not an expert in circuit design, so I just need some time to understand how speed changes with PWM and RC idea.

[@akkudakkupl]

@@jbumstead21 with PWM you need more code to set it up and PWM outputs for all coils preferably. Two transistor way would work also, but that yould mean having one transistor wired in as you have now, the second one with a resistor in series to cut down on the current. The RC bypass method is more elegant IMHO. Say the coil has 4 Ohms and pulls-in in 2 ms - add a 3.9 Ohm resistor bypassed with 470 uF capacitor in (coil - RC network - transistor). This way the initial current is 1.2 A, then goes down to 600 mA after 2ms. Might even stay pulled in at ~25% current - then try 10 Ohm and 220 uF. Parallel the coil with the back EMF diode as is or add the mentioned 4.7 V zener in series to it to make it release faster.

[@lohikarhu734]

For many solenoids, the hold current is MUCH lower than pull-in current... The inductance of the solenoid coil is much lower when "open", as there is very little core material when the plunger is out... There are fuel injector driver IC's, if you look at how they work, it's essentially the same... It's possible use a big capacitor to provide the start impulse, with probably 1/10 of the pulse current as the " hold" current... Pwm is more elegant, if you have enough I/O pins... If you had I/O pins that you can turn off and on... Pwm through higher value esistors to all drive transistors... I/O pin to smaller resistor to transistor.. Pin in off state, pwm drives xstor Pin at 1 drives xstor on pin at 0 drops gate voltage, xstor off So you have "cheap" pwm, and 3-state drive mechanism... You might be able to use capacitor coupling to the transistors to get hard off and on, but there would still be pwm drive... Just a thought

[@benmakeseverything]

Nice, I love the sound it makes.

[@rhalfik]

Very captivating movement and sound. I wonder what can be done to make the current draw less. Maybe some kind of pneumatic mechanism or with strings, where you only actuate tiny valves. You can also twist permanent magnets like they do on the noga dials. With some mechanical advantage you can make the mechanism very tiny.

[@jensschroder8214]

Control each mirror individually or work with columns and rows? You could switch the rows to plus, work with diodes and coils and switch the columns to minus. Then you have a matrix and because the coils are pulsed they consume less power. A diode in opposite parallel to the coil means that it switches off more slowly.

[@borbelyviktor3057]

Good project. You were crazy enough to assemble it together 😂 Thx for sharing!

[@paranoiia8]

I don't know why but that solenoid sounds and light flicking on wall is weirdly calming... For first few minutes at least 😅

[@ilyass_bouarasse]

Thats so good i like electronics prohects like that

[@chrismofer]

I think any lense like a magnifying glass would work to project an image of the array on the wall. Super cool project and inspiring thanks for sharing :)

[@JonMurray]

So cool man. New subscriber ✌🏻

[@Ziraya0]

You could improve the On Position alignment by adding adjustable stops on two the two corners that move inward when you want it to reflect. This could be as simple as two threaded/threadable holes you can put grub screws in. You really need 3 stops to fully tune the mirrors, but I think it would be easier to redesign the mirror cell than to add a third stop to this design. One alternate design that would help is to have the mirrors push out to turn off, and have the pivot as far to one edge as you can get it. Then you can float the pivot axle on some adjustable blocks and have a stop the solenoid pulls the mirror against in the on position. It would help the power consumption broadly if you had bi-stable solenoids so power is only consumed to cause a flip. Flip dot displays do this using a magnetic memory thing I don't remember well, but also flip dot displays are not suitable to being aligned, given their axis is through the middle of the flag's center of area. The mirrors also have a lot more mass than the flags so I'm not sure the lingering magnetic field will be strong enough to hold them up or out. One basic bistable solenoid design would be to have a magnet on the shaft and another magnet in the unstable center position, with their polarities set to repel. The moving magnet will be repelled away from the center position, as long as the solenoid can push or pull it across it will remain on that side until acted upon. I think you could retrofit the solenoids you have to include this functionality, with a 3D printed center-magnet holder, and a magnet glued to the end of back end of the shaft. Finding magnets of the appropriate strength, at an approachable price, will likely be the hardest part. Truly unfortunate how often sourcing parts is the hard part of a project

[@jbumstead21]

Thanks for providing these ideas for improving the mirror error. A one-time adjustment would be nice if possible. Maybe loctite after making the adjustment with some screw adjustment.

[@danialothman]

flippin awesome

[@Cjx0r]

finally a video for me!

[@irkedoff]

Awesome ❣️

[@CyanTiger]

This design is really cool reminds me of the Nikodama Blinking Eyeballs. That toy used a solenoid to move two half circles to make them "blink". I have clip of an eyeball in my videos (shorts). I was testing how to drive them in a percussion project, not sure I can link to it in YT comments. I would donate a pair if someone could reverse engineer them. If you wanted to keep it simple I have found that Surplus Sales of Nebraska has 24v Deltro Armature Relays for $2.00 ea. )They don't have any armatures attached but have the hardware, making them ripe for mounting something like a mirror on them. Big question is if the 1/32" travel will work to move the mirror? Thanks for taking the time to share and listen.

[@TheHDreality]

That's really awesome! Is it possible to calibrate the mirror position in reverse? i.e. lay out the mirrors on a flat surface like glass and then glue them to the otherwise fully assembled frame in one go so they dry laying flat in the solenoids' resting position

[@jbumstead21]

That's not a bad idea. I can't picture how this would work with the existing submounts, but it there was some rework, then maybe something like this could work.

[@CreeewDeeeTaaay]

I wonder if a flip dot display could be repurposed to make a lower current draw version that’s more easily scalable?

[@fridje]

a mylar reflective layer on each dot of a flipdot display would work great

[@ricktighe191]

The solenoid patterns frequently sound "musical"...nice job.

[@svofski]

Great project! Looks like it could be improved a bit by using first-surface mirrors, those that have reflective layer on the outside.

[@rhalfik]

Find an old combo laser printer. Take out the mirror strips from the laser and scanner assemblies. Chop them up and enjoy your high quality first surface mirrors.

[@svofski]

@@rhalfik hard disk platters are probably still easier to get by, they also make perfect first surface mirrors. If they are aluminium, they're easy to cut in any size pieces.

[@jbumstead21]

I haven't thought of sourcing first-order mirrors this way. Very helpful. But honestly, the mechanical mount is the real problem here, even with first-order mirrors the angle error would be awful. it needs a redesign.

[@stijnbolhuis286]

I Tried to make a similar installation for a theatre decor, ended up with 50 20 cm by 20 cm mirrors, every 5 mirrors were mounted on vertical axis, controlled by one servo. than you have 10 rows of this. eventually I programmed and controlled them in micro-bit a with a coding tool for 8 year olds. i can send you a video if you want to use it as inspiration, although it is not as sophisticated as your mini prototype. but the scale of the installation was quite nice.

[@jbumstead21]

Very cool. Send a link!

[@AristarcoCortes]

Very nice project. I have a question.Have you thought about how to solve the problem of high current flow in the circuit?

[@jbumstead21]

That's for your note. Yes, jimvandamme and xenoscientific on Instructables pointed out that the current draw can be reduced considerably by having the holding current lower than the pull-in current of the solenoids. One solution would be to have two driver transistors for each solenoid: use both for activation and one for holding.

[@failoktemp5232]

you only need high current initially, once the solenoid hits the end stop, you can use PWM with a low pass reduce total used current

[@akkudakkupl]

No need for a lowpass, back EMF diode will allow coil current to flow even when the transistor is off.

[@jeinnerabdel]

This is some transformer's stuff level! Pretty dope!

[@pileofstuff]

That's pretty cool.Too bad it's such a colossal (and expensive) job to scale it up.

[@danamulter]

huh. that was kinda cool.

[@tejeshwvardhan1165]

use surface mirror! generally mirror coating is done on back of mirror! back coating mirror add refraction to reflection!

[@HeathLedgersChemist]

What is sodder?

[@DrQuadrivium]

Does this have any real practical purpose other than just being a relatively expensive toy?

[@Hukkinen]

:) Make (half) disco ball.. Should it rotate or not.. Hmm...

[@SwordFreakPower]

Bowling lanes are another example of the lengths americans go to to avoid using SI units.

[@glitter_fart]

do nano scale

[@Teczito]

Now do bad apple!

[@alexanderdemidovich4372]

Looks like Barbarella's computer.

[@jbumstead21]

Wow, needed to look that one up, but yes. yes, it does.

[@alexanderdemidovich4372]

@@jbumstead21 Something similar was in the game Chronicles of Riddick On PS2. The game menu was organized in a similar way. Only in the form of a spinning cube. But I have warmer emotions associated with Barbarella. :)

[@ophello]

Isn’t the purpose of this idea to have a reflection change? Also it would have been better to use coils on a board, and use magnetic mirrors which slightly deflect when the coil is energized.

[@neatt3815]

Make a bigger one and play Tetris on it 😁

[@winout96]

"Four bowling lanes" Let me guess: You are from the USA!

[@PredragMrvic]

Vidim da piše "Ogledalo" to ne može biti slučajno! :)

[@jbumstead21]

Ne mogu vjerovati da ste vidjeli!

[@tswellersalzer1850]

Stop the Music.

[@zoeyzhang9866]

Creative project! Wondering if any custom PCBs, or 3D printed parts may help for any upcoming projects? Would love to sponsor free services and reach any collab together! (PCBWay zoey)