Laser Projector (Red, Green, Purple) "Laser 3D Party Light" from Spencer's

[james]

Take a look at the TLO74 as a replacement for the LM324.

[tribble]

Nice tests... couple questions. The missing white, are you sure that's not lazy green again? What if you changed the colors to have red or blue be the color to turn on for white? Also, regarding the weird handling of the blank vectors... does it handle a black (non-blank) vector differently than a blank one?

Pretty sure. First, note that the color is identical to the previous color. Second, note that by adjusting the vertices as described (plus some additional complexity I haven't blathered on about here yet) the white returns. See the pic up in post 236.

I didn't think to test the black vector, so I'm not sure.

I have an alpha of the tool that takes a normal format 0 section and produces the spencerized file working pretty well, but it will probably be a few days before I have time to do some before/after pix of various frames.

[DogP]

Take a look at the TLO74 as a replacement for the LM324.

I've got some TL084s laying around that I plan to try, which I think are pretty similar to the TL074s.

DogP

[james]

See if this gets you anywhere. I have no way to test it!

http://laserboy.org/forum/index.php?topic=553.msg6364

James.

[adventurevision]

Just an FYI, our local Spencer's has this for $129.99 now.

My mail-order projector arrived broken so my wife swapped it out for me tonight and she noticed the sale price. YMMV.

A big thank you goes to DogP for posting about this over at the KLOV forums. Thank you Sir.

[DogP]

So, I got a bunch of tests done on this.

On the DAC stuff, I looked at the output of the DAC and first stage opamp, and there doesn't seem to be any problem (no quality difference that I could see).

DAC out at speed 10:


DAC out at speed 40:


Regarding the laser diode color drive timing... the colors relative to each other are nearly identical (red is first, then green 300ns later, then blue 600ns after). Below are pics of the color drive on the scope for turning on a white vector (these kinds of tests are the few times a 4 channel scope would be useful at home... usually I don't have enough hands for 4 channels ).




The laser timing relative to the vectors is really screwy, as we've been discussing. Below are details from the tests I ran on the scope. The first pattern starts with 10 blank points at the origin, then goes 4 points red, back to the origin for 10 more blank points, and so on from 1-4 in the pic (finishes with 10 blank points at the origin).


Here's at speed 20:


And speed 10:


This is just screwy... so yes, it delays the laser signal by 1 sample... which might actually be reasonable if they're trying to sync the galvos and lasers. The problem here is that it stretches that DAC sample where the laser turns on (the yellow trace should step at a constant rate). It looks like going from blank to color, it waits one sample, plus some time, and then continues (the time after the laser turns on is about one normal sample). Now look at speed 10 (note the time scale change)... this looks VERY similar to speed 20, except the amount of time it waits is basically double (like it's telling it to wait in sample time, not actual time).

The other test I ran is this. This starts with 10 blank points at the origin, then goes to the upper right with a red vector and a purple vector (checks the single vector before blank timing). Then back to the origin for 10 blank points and to the lower right for a green vector and a 3-point red vector (checks a multi point vector before blank timing). Then it goes blank and slews to the lower left corner and back to the origin (checks laser to blank timing).




This shows more views of the screwy behavior. Blue never turns on (single vector before the blank), but red stays on as it should (blue should have turned on at the top hump on the left ramp, and turned off at the same time as red).

I turned on persistence on my scope so you could see green and red overlaid on the same screen... green is the very dim red trace right before red on the right pulse. So, you can see that one color to the next transitions just fine (after the DAC stalls for green, it switches to red in a regular DAC period). BUT, notice that the blank vector is also stalled for red to turn off (the top, middle, bottom steps on the right should have been evenly spaced).

So... hopefully this info helps understand what's going on under the hood a little bit better.

On the amp side of things, I swapped the LM324s in the X side for MC33079s, and the TDA2030A for an LM1875, and it works great. I don't notice any major improvement, though I took "before" pictures, so I'll have to take some "after" pictures and compare side by side to check. The opamp that didn't work was the TSH24 (not sure exactly why... didn't really investigate). I have some TL084s and TDA2050s as well, but I think the parts I'm using now should perform better than those.

I also had an idea about getting the green to drive quicker... and it sorta works. The header for the green has the green signal and ground... when the signal is low, it turns on. So, I thought maybe I could add a pot across the green drive header, and bias it so the green diode is right at the edge of being visible, so it wouldn't take much for the drive signal to turn the diode fully on. I found that ~660 ohms was good. The problem is that it's VERY sensitive (a couple 10s of ohms is the difference between green lines everywhere and no benefit)... and I'll bet that it's temperature dependent as well. So, it'd probably need some sort of active control to reliably be useful. Not sure that it's worth it... but it was just something that I wanted to look into.

Pretty sure. First, note that the color is identical to the previous color. Second, note that by adjusting the vertices as described (plus some additional complexity I haven't blathered on about here yet) the white returns. See the pic up in post 236.

I didn't think to test the black vector, so I'm not sure.

Yep, after doing some more tests (above), I'm fairly certain you're right. About the black vector... I tried it, and the weird timing stretch stuff went away, but it doesn't seem to know how to handle it (lasers were on when they shouldn't be and stuff). I didn't really investigate, though with the screwy handling of special points, it'd be nice if we could trick it into drawing regular points, but just leaving the lasers off. It might be worth another look (maybe instead of black, see what it does with really dim points), or maybe I should verify that there wasn't a palette issue or some other stupid error. Without all the goofy delays, we should get smoother graphics/higher frame rates/etc.

See if this gets you anywhere. I have no way to test it!

http://laserboy.org/forum/index.php?topic=553.msg6364

Cool... I'll try to compile it when I get a chance. Or if you post the binary, I'm sure one of us would try it out sooner.

Just an FYI, our local Spencer's has this for $129.99 now.

My mail-order projector arrived broken so my wife swapped it out for me tonight and she noticed the sale price. YMMV.

A big thank you goes to DogP for posting about this over at the KLOV forums. Thank you Sir.

Cool... thanks for the heads up! I might have to pick up a spare at that price. BTW, what got broken on the one you ordered?

DogP

[tribble]

So, I got a bunch of tests done on this.

Excellent work, DogP, top notch! If we meet at a LEM I need to buy you a libation of your choice.

So, the MCU inserts, what, 3 or 4 vertex periods of dead time to turn on the laser between blanks and lit vectors? Am I reading that right?

I guess it's worth a look to see what happens when we just change the palette index for 'blank' vectors to black... turns out the black palette entry is number 64. I'll modify my spencerizer code to see what happens if I replace the blanked vertexes with black when I get a chance. When I get the rest of my build environment set up, I'll try James' code too

Does the laser driver chip have any sort of bias adjustment on it? I have noticed that the green gets much faster when it gets warm, but it is still nowhere near as fast as the direct diodes. I don't see putting a 520 in this box, though!

[tribble]

OH, and another thing... the ILDA test pattern looks really neat on the scope! But, I notice that the transitions between points are along a 'grid' and not so much on the diagonal... is the MCU outputting X, then Y, instead of both simultaneously?

[adventurevision]

Cool... thanks for the heads up! I might have to pick up a spare at that price. BTW, what got broken on the one you ordered?

DogP

Well... right from the get-go it would only draw a horizontal line. Thinking something could have jiggled loose in shipping I ended up taking it apart to see and I have to say... I think either this unit was messed up right from the factory or someone else got their hands into it and then really messed it up.

The violet laser looks to have been replaced and then spliced in very poorly. A bunch of the hot glue globs were broken(meaning someone was messing around). The wiring was very messy(when compared to the replacement projector I got last night thanks to the wifey!). There was a screw half way out of the module that is responsible for horizontal deflection. There was a very fine wire that was broken on the PCB of this module. And solder was touched up in many places for the circuit that lead to this module.

It was really fugly... too fugly.

BTW, when the son got home last night I had the laser running in the family room and he flipped for all the Nintendo graphics in that package you put together... really awesome of you to share! Thanks again.

Now I'll let you smart people get back to being smart while I just sit here trying to follow along scratching my... head.

[tribble]

Also, regarding the weird handling of the blank vectors... does it handle a black (non-blank) vector differently than a blank one?

DogP, it was a GREAT idea... but it doesn't work. :-)

After some testing, it looks like my copy of the Spencer projector draws color index 64 (black) as WHITE. Also index 65. And index 255. I didn't try them all, but I suspect there is no "lit black" option in this showcard. Also, other than index 64, there is no color in the default palette that does not have at least one channel at or above 128.

And wow... for $130... might be worth it just for the parts. I mean, how much do a pair of 15K galvos cost on ebay? Consider you get a set of lasers as well, and you have parts for a nice Lissajous demo.