Brumbaer / 3D Printing / Revived by a woman's touch


July 2015. Color Wheels, HD and Elastic Resin an Odyssey

Act 1

Troja has fallen, 3D printing commences really well, and I tempt the gods by thinking just that.

Don't know what god is in authority of 3D printing, but he decides to punish me for my thoughts. Wickedly he does not to do so himself, but uses the demi-god  of projectors (the other half is the god of monitors :) )to deliver his wrath in form of one of it's most cruel apparitions - the 9 blink color wheel error.

It only shows after some time so it's like a calm on the Aegean sea, first it looks nice, but with raising temperature everything starts to fall apart.

I have to finish the project in a fortnight, so I can't really spent 10years trying to get home. Sending in the projector would easily take as long (as the fortnight not the 10 years) as it will have to cross the Atlantic and pass customs twice. Even using an express service would most likely leave me without an printer for more than a week.

So I open the projector's 7 screws on the outside and 3 on the inside and check whether the connectors sit firm. They do. So I close the projector 3 screws on the inside and 7 on the outside.

As the problem is temperature related, I do it as Odysseus would do, I get an additional fan, but I can't afford a slave, so I have to use one of those small electrical ones. And when put on the top exhaust grill (a slave wouldn't have fit there anyway) it removes enough hot air to get the print finished.

Odysseus never had Damokles' sword swinging above his head, but I do and like Damokles I don't enjoy it.

So I decide to get a new D912HD and convert it for 3D printing. That night I dream of gods rotfl.

In the morning I learn that the printer is out of production. That's not helpful, but explains the nightly laughter.

Odysseus was well known for his trickiness, and so I mimick him and get a D910HD. It's the sister model to the D912HD and while there might be differences in the power supply, it's most likely that the color wheel is the same.

So I open the projectors' 14 screws on the outside and 6 on the inside and switch the color wheel assemblies (a metal holder with the color wheel and some sensor board counting the revolutions) another two times 5 screws for the assembly and it's cover. The cover seems to be added as an afterthought and so you have to remove an other two times 11 screws to get at it. I wish Odysseus would have  designed the interior of that projector. This could have been made a lot easier with a little thought.

Anyway after the switch and packing up (another 52 screws) everything runs like Odysseus ship when Polyphem threw boulders at it. 

I try to be more clever than Odysseus by not provoking the gods again, shutting down any smug thoughts and so I'm able to finish the project in time.


Act 2

Finally I have some time to consider what to do with the D910. I try to use it as a doorstop, but, because of it's size, I hit it with my feet more than once and that hurts.

Throwing it away wouldn't hurt my feet, but my pride.

I could send it in for repair, but as I have switched color wheel assemblies the warranty will obviously be void.

Still it would be the sensible thing to do. Naturally I do something different. I try to repair it myself.

I'm surprised to find that it is not that difficult to get a color wheel, but than my expectations are met by the fact that the one I need is not on offer.

So I look for one which looks similar, i.e. having six segments with the colors in the same order. But I can not find out whether the sync mark is at the same segment. I reckon that it will not matter as I will basically using it in black/white mode only and can change the color of the grid by software.

Finally I find one available in Berlin and can get it next morning.

I worried about, segments, color, index marks, current and voltage, but what proofs to be a problem is how the cable is connected to the motor coil.

All color wheels use a FFC to connect to the electronics, but some connect the coils directly to the FFC and some use a PCB as an adapter. The Vivitek connects directly, the replacement by PCB. :(

Luckily the FFC and PCB re fastened to the motor with some removable glue and can be pried carefully apart.

So I switch PCB and FFC do a short test with a resistance meter and find three contacts connected. Not really having thought about what to expect that sounds reasonable.

So install it and once switched on the projector immediately reports a color wheel error, not even the lamp gets lit.

Ok it was just a try, nobody did gurantee that the wheels would be compatible - I'm resigned but not surprised.

I remove the color wheel, swicht FFC an PCB, measure the connections and finding all four contacts on the original being connected. Full stop, measuring the contacts of the replacement and see they are all connected as well, I made an error when I soldered the FFC to the replacement. So Full Reverse, solder the FFC to the replacement, check all four connectors to be connected, install the color wheel (we are so well acquainted by now that I might call them oCW and rCW), start a projector calibration, projector switches on, projector lits up - elation - the grid showes in some salmon (cooked not smoked) color - because of the differently placed index mark - waiting for 5 minutes, the image flickers, goes dark and a color wheel error is reported.  **** Insert your favorite swear word here.

So the CW is not at the bottom of the color wheel error. I must say I'm not surprised after knowing the oCW that intimately, I just can't imagaine it doing anything wrong - is that what they call the Stockholm syndrome ?

Ok, remove the FFC and put it back on oCW - again.

Switching the color wheel assembly between projectors solved the problem, the culprit has to be the other component on the assembly, no not the metal chassis, but the sensor PCB.

There is not much on the PCB, a smd reflective interrupter, a connector and some resistors and capacities, the latter in 0603 which means rather small. Usually temperature problems with a PCB are related to bad soldering. So I resolder every contact in use. Put the PCB back on the metal chassis, install the color wheel assembly etc. - you know the drill.

Every time I test I have to fasten unfasten all the screws. I'm now down to 24 seconds - practice matters.

Start projector calibration - looks good

wait for 5 minutes still looks good.

I learned that large amounts of white light heat the assembly up more than the red grid, so I do a dummy test print and ... drums please ... all fine.



Act 3

So what to do with a working D910HD.

The answer seems obvious when you know that I have a 1.1 printer that I use for some jobs, but not as often as I wished, because the print sizes and resolutions of 1.1 and 1.2 printers do not match - or if you read the summary on the contents page.

There are three hardware issues to be solved for using the D910 on the 1.1 printer.

First the focal point shift. 

Second the mounting of the projector.

Third removing the glass from the projector lamp.

On the Titan forum there is a user Kentaro who did an stl for the gasket - many thanks -, which I downloaded. First I thought about cutting the gasket from rubber, but than remember the bottle of Spot-E I always postponed to try out. This is my chance to kill two flies with one stroke and I print the gasket using Spot-E.

I'm prepared to do some test prints to get the exposure right - new resin and all. But I'm denied that, it works out right from the start with the same parameters that I use for Spot-HT.

I added a lot of supports to keep the gasket in place, just to be on the save side.

Later I will find out that I can cut exposure times nearly to half without ill effects and that the additional supports are unnecessary. 

Spot-E is really rubbery and elastic and feels nice. it does not loose elasticity ( and it shouldn't) under UV light as far as first tests show. It seems to cling to the build plate like Paris to Helena. It might be worth to try on large models that warp so much that they come off the build table - if the rubbery consistency doesn't pose no problem. 

Before anybody asks, I do not know whether it is castable, but doubt it.

Installing the gasket was a breeze. Remove the top and front bezel. Remove the black lens ring and you can access the 3 screws holding the lens system. This is where you will put the gasket. You will have to replace those screws with M3x10 countersunk head screws.

If you find me obsessed with screws, keep in mind that I have opened and closed the projector at least 25 times.

To keep live simple, or better to start to make live simpler, I decide to keep the original projector mount and just add an adapter for the D910HD. I have an CNC milling machine, so I cut it from acrylic glass with high enough precision - which isn't really that hard, because the only thing that has to be precise is the size and location of the holes and you can manage that with a drill and a drill rig.

The adapter and the size of the projector will shift the projection more to the front, which isn't a problem.

The 1.1 has less room below the x-axis base plate, so you can not get the projector far enough for 70 um resolution, but 50 um is fine. Also 30 is easily possible, probably 25, I didn't check. As I use only 50 and 30um, I'm content, but if you need 70um you will have to look somewhere else.

The third point is not really worth a comment, except that you have to unfasten 5 and fasten 4 screws, screwssssss, screwssss....

The hardware done only the software is missing.

Printer parameters are selected by the means of a configuration file. So I add 2 configurations one for 30 and one for 50µm. 

The first choice is which driver to use. There is a 1_1_0 and a 1_2_0. It has to be the latter one, because that supports the new projector. Other obvious values to change are print resolution dependent values (for the 30um) and projector resolution values.

That should basically be it, but it isn't the build table movement is not correct. The way will be clear as soon as you realize that the 1_2 version uses half steps, while the 1_1 uses full steps.

So you also have to change the z-resolution and you have to double all step values, which appear in parameter pages and some calibrations.

Because the projection is further to front you will have to change to change the y offset as well.

And that's it copy it to your application (I use a Mac so it has to be copied into the Application Package) and you are ready to go.

The picture shows the additional fan. It's not needed anymore, but where else would I store it ?