...and now for something different


#21

Thanks for the explanation.

FWIW, this is not a DLP printer, it’s MSLA (masked SLA), which means it uses a LCD display, and the vat sits directly on top of it. The light source is provided by 24 UV LED array, distributed directly under the LCD panel, so there’s not much of a “hot spot”, unless one of the LEDs fails.


#22

Ah, Thanks but I do understand the differences in 3D printing techniques, Terminology is often interchangeable for example DLP can be the acronym for digital light projection or direct light printing.

You would be surprised at just how much power deviation there is across an LED array, variables such as the lens on the LED and distance from the screen make huge differences to UV power going through LCD in various areas.

For your reference below, is a 3 Dimensional UV output power distribution graph of a 24 LED 405nm LED Array, this graph is from a high quality array and its distribution is better than many, but the differences in power are easy to see.

Reflectors are often used around the array and can vary from very simple flat sheets to complex curves with vastly changing geometry (often those shapes are 3D printed and then a thin layer of aluminium is put on the plastic by plasma deposition techniques) to try to get the best possible uniformity of power at a certain focal distance from the array. Reflector material is yet another topic, the best ones for 405nm are highly polished aluminium, but in terms of production cost is expensive. A high quality curved surface reflector to give best power uniformity from LED arrays will often (on their own) cost 5X the price of Dudemeister’ whole printer…

An example of the design concept is shown below:


#23

Can you provide more information about the 3d graph you posted?

I’m not an electronics/optical engineer, but I do know how to read a chart, and that one doesn’t look like the chart output of an array, more like the output of a single light source.

This is more in line with what I would expect, without any reflectors or any special distribution aids.

P.S.
“Ah, Thanks but I do understand the differences in 3D printing techniques, Terminology is often interchangeable for example DLP can be the acronym for digital light projection or direct light printing.”
since you brought up acronyms, DLP stands for Digital Light Processing. Never heard of it as Direct Light Printing.


#24

Hay, thanks again for trying to correct me - Will tell my work mates that we have all got it wrong, (we do this as a day job…)

If you do a google search for “power distribution from UV LED array” you will find various power distribution graphs - all are similar and none really are the shape you have drawn (I wish they were - it would make my working life much simpler, but with the laws of physics certain things are difficult to change)

We have the kit here (Radiometer) to actually take the measurements (and we often do as we work through various development projects) We also generate power distribution graphs and the one shown before is VERY typical for a UV LED array. We use Maxima software to generate the graph using the Romberg method. We typically measure the power across a 2mm grid spacing using an automated CNC x/Y travis to move the sensor accurately and repeatedly.

PS: The key word is UV array. They behave in a somewhat different way to visible light sources


#25

There’s no point in comparing this to a Form2, but then again, it costs 1/17 the cost of the Form 2.

So as you can imagine the specs are not in line with the Form 2. It’s build area is in only 118 x 65 x 110, so substantially smaller volume than any form printer. However it’s in line with similar printers costing twice as much if not more.

I’m sure the print times can be reduced. I already found the printer had a “fast” profile, which reduces the layer exposure from 8 to 6 seconds, and there’s the amount of lift and wait in between moves that can be accelerated as well. As I said, I didn’t do any special profiles, or anything, I just loaded the model and hit print.

I’m sure that once I get a bit more into it, I’ll be able to improve on most of the print aspects (speed & quality).


#26

Just so we understand each other, I’m a computer electronics engineer, and been at it since 1995, and while my field doesn’t revolve around optics, light, light distribution and such, I do also understand common sense. The 3D plot you provided shows the light output across a 8cm x8 cm area, which might well be the output from a single source, or maybe a cluster of light sources within that 8x8 area

On the other hand, the graph I posted comes from a white paper on the design of light arrays for uniform light distribution. Here it is in case you want to look at it.


#27

I don’t doubt that you or your workmates may be using that description, but maybe you should look up DLP at Wikipedia:

Or alternately google Digital Light Projection. It all points back to Digital Light Processing and Texas Instruments which invented the technology.

Just saying.


#28

You missed the most important point - led producing UV LIGHT as against led producing visible light.

Just had a quick look at the white paper As you can see from the text quoted below that paper is talking of a Steren 5 white light (visible wavelength and also quite a broad spectrum source)

“Forthepurposeofdemonstration,weassembledalinear array of four LEDs with a LED-to-LED spacing of 1.84 cm. These LEDs (Steren 5ULTRA WHITE) emit white light with a ranging angle of 30° and an angle 12 8.44° m 64.66. From Eq. (13), the optimum LED-to-LED distance is d0 0.189z.A translucent diffuse screen was positioned 9.74 cm from the LED array (optimum panel–target distance is z 9.74 cm, which is measured from the chip image position, coinciding in our LEDs with the encapsulant base.) The light transmitted through the screen was imaged at a charge-coupled device (CCD) camera. The recorded irradiance pattern is shown in Fig. 9(a). Figure 9(b) shows the experimental and simulated irradiance distribution along the center
axis of the array (horizontal direction). The slight disagreement between experimental and modeled data is explained by the fact that the irradiance patterns of assembled LEDs are not equal even among LEDs of the same type.16 Nevertheless, the agreement between theoretical predictions and experimental data confirms that the derived expressions are practical design tools for both a quick estimation (first-order design) and the starting point for exact designs.”


#29

Yeah, the profile of light measured through a diffuse screen with a CCD is quite likely going to be a bit different than light measured with a CNC-positioned radiometric power meter, for a given LED array. Then there’s questions like what’s the intended emission angle on the UV LEDs in the array, is there a big external reflector around them, how well are they binned for emission angle, emission power, are they balanced at all when the array is connected, or are they actively balanced at all by the driver, and so forth… are there any photodiodes or other sensors in there to keep track of the array performance?

Without a repeatably-positioned power sensor, I wonder how difficult it would be to characterize exposure uniformity by doing a long exposure (assuming the uniformity doesn’t vary much with time) against a hard-to-react photochemical mixture, whether that’s a photographic paper, film, or plate, or a photopolymer resin.

Or maybe with the array at constant electrical power, maybe you could try get a measurent proportional to the total light transmitted with a camera/CCD pointed at the mask LCD, and then feed various mask patterns to it. If you mask every pixel, does it block all the light with no leakage? If you mask all the pixels on just one side, do you get 50% of the light or less or more? What about 50% of the pixels at random, or in a checkerboard pattern, or other patterns?


#30

Ike - you are spot on. The diffusion layer has a large effect. Its also a cause of UV energy power loss.

For our own UV power tests of an LCD screen / UV light source, we use an image grid of 2mm diameter circles set to 255,255,255 with a pitch of 4mm between centres in the X/Y plane, its easy then to accurately measure the transmitted light power by moving the sensor head from centre to centre (The head is 1.5mm diameter) its also easy to measure light bleed in the areas that should be “black”

We also have a machined metal mask that uses the same centres so that we can check results of the light source both with and without the LCD screen being present

Normally the array is supplied by a constant current (floating voltage) type PSU.

One factor with this type of arrangement is the losses of power. 56w of input power to the LED array is likely to only give at best 1 mw/cm2 of UV energy at 405nm at the build surface. Most of the other energy is lost in the form of heat, Some of the UV energy is also lost (filtered) by the LCD screen as they are designed to transmit visible light from about 460nm upwards and quite effectively filter light at 405nm.

The heat that is generated is quite an issue with the LCD screen, surface temperatures of approx 80c and greater(depends on crystalline structure of device) causes irreversable damage that reduces the screens life time. We generally advise that the screen should be treated as a consumable, with both the effect of heat and UV energy a typical life time (depends on printer) of 80 to 500 hour max print life - Its not expensive so most people are not worried too much by that.

With regards the masking for UV power level differential across the build plate, that can be controlled by changing the pixel colour, for maximum power transmission a pixel will be set to colour 255.255,255 (White) alter that balance and you let less UV light through that pixel. Which is not so good when you start with relatively low power and then you reduce that further. It does require quite complex algorithms to provide good print definition and uniform power distribution


#31

Interesting. Is it just easier to to get a color LCD screen for the applications these days, or are their other reasons to use that instead of a monochrome display? Could you fine-tune your exposure even further by using non-neutral pixel values, since the red, green, and blue pixels all probably transmit 405 differently at a given 0-255 value?

Anyway, back to the Monoprice printer, that seems to leave open the possibility that you could slap together any old array of loosely similar LEDs, put your LCD screen over it, in the factory, point a camera at the thing and then do some software magic to develop a calibrated bitmap that eats some hopefully-small percentage of the output power but gives you much better exposure uniformity across the screen. Then you could theoretically add/subtract that bitmap from every layer sent to the screen. I wonder whether any such thing actually happens, or whether the printer is just supposed to operate in a regime where it can get away with some unevenness in the illumination.


#32

Ike - Unfortunately the available monochrome screens available are a significantly lower resolution than the same size colour screen - Monochrome is much better at UV transmission (it lets through roughly five times the UV energy than a colour display) so its a great shame that higher resolution monochrome screens are not made for 3D printing

The colour settings values are exactly how the UV power is mapped. The three software packages Dudemeister mentioned do not offer such necessities.

You are quite right in what you say as the possibilities of a mask, unfortunately it would not be possible with a camera, it requires a radiometer to do the uv power mapping and the cost of each calibration would be far greater than the cost of printer!!

The reality is you get what .you pay for, a $300 printer will print but should be regarded as little more than an educational toy, a learning curve, rather than a true tool and work horse that is going to deliver the results you want consistently and accurately each and every time.


#33

Back to the original topic.

So for the past couple of days, I’ve been having some troubles and tribulations with this little bugger, but I think those issues have been sorted now.

First of all, I updated the firmware, and as luck would have it, it created problems. After the update, every print failed in one way or another. Not only that, but noticed that the prints were taking a very long time, so I started watching it print, and noticed that every layer was taking about 40 seconds to print.

for those that are not familiar with DLP/MSLA printer, the first few layers (base layers) have longer exposures (30-40 seconds depending on printer and resin), and subsequent layers have a lot shorter exposures of 6-10 seconds.

In this printer’s case, the base layers were set to 40 seconds, and the rest were supposed to be 7 seconds. After the firmware update, all layers were taking 40 seconds. In fact a print that should have lasted about 5 hours, was still printing the next day some 18 hours later.

Anyway, I re-flashed the firmware, and now it’s working OK. But as I ran out the small sample resin, I bought some Apply Lab Works Modelling Grey for MSLA.

This time the issue is completely different. Even though the exposure rates for this resin are very similar to the resin from Monoprice, it creates such heavy suction to the FEP, that the print fails. It’s virtually impossible to lift the build platform and separate the print from the FEP, even by hand.

After browsing the web for some answers I found a Facebook interest group dedicated to the MP MiniSLA, and it turns out this is a relatively common problem with resins that are more viscous than the MonoPrice resins

What happens, is the bottom FEP itself is touching the LCD screen, and it can create a pretty good vacuum underneath. When the build platform tries to go up, the FEP has no give/flexibility as its stuck to the LCD screen. So the trick is to put a few strips of masking tape around the edge of the LCD with the corners open, so to create a thin separation between the FEP and the screen, and also allow the air to escape/breathe by having those open corners.

I’m printing a model right now, and will post some photos once it’s done.


#34

So here are a a few photos of the print I was running. The first one is the print fresh off the print bed, still wet and on it’s supports.

If you look close, you can see a separation line about 1/4 or 1/3 of the way up, when I paused the print to make sure it was actually attached to the print bed. That pause and lift, left a scar on the print.

The photos below show the model after I removed the supports and cleaned it up a little. The quality and detail are very good for 0.05mm print, and I can go down to 0.025, but this isn’t bad at all. If you look close you can still see a few layer marks here and there, which are due to the coupler between the Z motor and and the threaded rod. I’ll probably replace the coupler they have with a flex coupler I have somewhere around here.

Anyway, here is the output of my MP Mini SLA and Apply Lab Works Grey resin, after only about 4 hours of tweaking during my lunch hour.

BTW, The print time for this model can be seen on the first photo. It took 3:46 hours to complete 996 layers (0.05mm layers).

I arranged the same model in PreForm, at the same angle, and sliced it with default supports. The resulting, model was 990 layers and the print time estimate is 3:31 hours on a Form 1+, and 3:52 minutes on a Form 2 using Grey v3 profile on both.

So the SL Mini is just as fast a a Form printer, a little slower than a Form 1+, and a little fast than a Form 2.

BUT, if I were to pack two of these models on the build platform, the Mini would still take 3:46 minutes, but the Form 2 would take 1 hour longer to print.


#35

That’s got a lot of horrible lines though


#36

Yes, there are a few lines through the print. “a lot of horrible lines” ? I don’t know how you define that but let me put it into perspective. First of all, this is 2" tall model, and some of the lines you are seeing under high magnification are not really immediately apparent when looking at the model. I’ve seen injection molding lines that are A LOT worse.

Secondly, as I previously said, this is work in progress, and I’m aware that there are still a few issues, and I think most can be resolved. Since this printer is at work, and I only get to fiddle with it during my lunch hours, I’ve only put maybe 4-5 hours into it, and that includes tests, and generating different support and trying different settings.

Considering the short amount of time I’ve had with it, I think the results are actually pretty darn good. Sure they’re not up to a Form 2 printer yet, but give me time.

My point is that you CAN get good quality prints with a little work and for little money


#37

You are certainly demonstrating your point that it prints.

The masking tape hack/tweak is pretty interesting, but I wonder if that gap is going to blur the exposure edges of your prints in proportion to the distance. Do you have any interest in quantitative tests for dimensional accuracy/repeatability/material perfformance, and that sort of thing, or are you mostly interested in this printer as a artistic model/miniature/figurine printer where the parts will be painted?

As additional points of comparison, members of this forum may also be interested to hear about tank lifetime, resin changes, large prints, and any other maintenance you find yourself doing in the course of running the printer long term.


#38

Looking at the gun there’s a bunch of lines on there, and then multiple lines along the head and lower on the body. Sure, some injection molded stuff is worse, but you’re not making something injection molded. I’m curious if getting a better coupler for the Z-axis will help with that.


#39

When I bought my Form 1+, I never bought it as a tool for technical models that require utmost accuracy, and I had, I would have been disappointed as the models were never 100% accurate. Even today, my current Form 1+ still has some issues printing very accurate models, it always stretch the prints front to back, and my current setup, is pretty weird (I’ll post more details later when I get home).

I print primarily what you described as artistic models, miniatures or articulated figures, mechs, etc. Almost all my models end up being painted, so I developed my own techniques (or trial and error :wink:) to make sure that interlocking parts fit together. If you are interested in seeing some of the type of work I’ve done, you can look here:

As for the other points you brought up, it’s a little early to tell how long the tank will last or maintenance, as I’ve only had the printer for exactly 1 week.

The tank is made of some sort of heavy nylon type plastic, and has a FEP film on it. Knowing how FEP is supposed to be pretty forgiving in terms of UV wear, I would think it should last quite a while, assuming it doesn’t get physically damaged.

Resin changes seem to be simple enough, pour whatever resin is still in the tank out, then pour in the new resin. When I changed from the original Monoprice Clear to the ApplyLabWorks, I just scraped off as much of the clear as I could, cleaned the sides of the tank, and poured in the ALW Grey. Used a credit card to scrape the bottom of the tank, and mix the resin a little, that was that.

Last but no least, size. As I previously said, this has a pretty small volume, 120x68x150, about half that of the Form 1+. But that’s good enough for the type of small prints I will be doing.

Speaking of prints, last night I set up another print and this morning it was done. This particular one is another Warhammer 40K Sisters of silence print, Saint Celestine. The model was split in 4 parts so it pretty much took up the entire print platform.

Here is the model as it came of the print bed (still wet), and later today, I’ll post some shots of the assembled print.


#40

Saint Celestine model update.

Here is a 6 view around the assembled model. There are a few problems with the model, the bottom feathers of her left wing did not print, probably not enough support, and on the ribbons around her sword, there’s supposed to be a single rose, that just came out like blob.

There are still some supports that need to be cleaned up, as they were very close to the surface of the print, and they got fuse to it. That was my mistake as I put the model in the sun to cure it before I removed the supports. I should know better, but I’ve done this kind of clean up before.

FWIW, I also tried printing this on my Form 1+, and the wings had the same problem as here.

On a side note, this model, was not at the same scale as the the other print (about 36-38mm instead of 48mm), so I’m reprinting it at the right scale on the Form 1+. The re-print is at 0.25mm and it’s going to take 13:08 hours to complete. On Monday, I’ll send the same print to the MP MIni also at 0.25mm so I can do a comparison.