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

[james]

Are you doing something different than the typical correction amp for the sound card DAC?

[DogP]

Are you doing something different than the typical correction amp for the sound card DAC?

I'm not sure... for X/Y, I'm just using an op amp to go from 2.25V bias unipolar to 0V bias bipolar (by referencing to Vref), with variable gain (up to ~4.5x gain). For R, G, and B, I'm just using a comparator against Vref for output (since I don't have an analog laser driver anyway, and the open collector output of the LM339 is convenient on this board). Then I'll probably switch the X and Ys when the USB cable is detected (either analog switch, or maybe just an audio relay), and feed the signals into the "focus" knob, so I can still control the scale from the knob on the back.

DogP

[james]

Well, that will work.

The LaserBoy Correction Amp is designed to be a fits-all solution. It has its own negative voltage regulator to set a reference to sum with the DC offset. It also has as much as 10X on the gain. Some USB cards like the USB-SND8 only have a DC offset of about +1.2V and they don't have quite as much of a peak to peak output.

James.

[DogP]

Yeah, I'm just looking for a quick simple solution to work w/ the CM6206, which is nice enough to provide the Vref, and has ~3Vpp output. I don't want too much gain though... supposedly the amp wants +/-5V signal, though it looks like at max size, it's got about 15Vpp going into it. But I'd rather not push it too much.

DogP

[DogP]

Okay... I had some fun today.

The power supply arrived... SUPER fast for $3 USPS mail from west coast to east coast. It's pretty small, but before I started putting it in a case, I needed to check out the performance. I crimped my cables, connected it directly to the amp board (JST XHP-3 housing - http://www.digikey.com/product-detai...219-ND/1651017 and http://www.digikey.com/product-detai...35-1-ND/527370 pins).




I turned it on, and qualitatively speaking, DEFINITE improvement (straighter lines, sharper corners, better circle, etc).
Before:




After:




Large scan angles had an even bigger improvement. Look at the ugly overshoot on the corners at 25 degrees
Before:


After:


Of course this could use a proper tune (shouldn't have overshoot like that), but I didn't want to change anything until I could compare before/after.

Quantitatively speaking, the new supply is good, but not perfect. Here are pics of the voltage on the scope:
+15V (~750mVpp, but only about +/-375mV)


-15V (~600mVpp, but only about +/-300mV)


It looks like the load regulation isn't perfect (but within the +/- 3% spec). The biggest thing though, is that the voltages are evenly distributed around +/- 15V. So, the regulation loop is doing its job (and a couple large caps might help this... I should have room in the project box for them). On the stock switcher, it is almost entirely voltage sag, like it's not even trying to compensate. Also, notice that -15V is better than +15V now (I assume the negative rail draws less current).

All the scope pics have been with a scan angle of 13 degrees... I imagine the larger performance gap at 25 degrees is because the stock switcher performs even worse at full deflection. I haven't measured the current draw on this yet either, but I'll do that when I get a chance

So, to sum it up... IMO this supply is well worth the $17 (shipped) upgrade. But, there are probably even better performing supplies out there. Not sure how much performance is left to squeeze out of this by just upgrading the power supply though. Maybe I'll try this with a couple quality lab supplies some day.


I also modified the USB sound card.


It was a pretty basic configuration... the DAC output went through an AC coupling cap, a series resistor, and into the amp, then out of the amp through another AC coupling cap to the header that the wires attached to. I originally just jumpered the AC coupling caps, since the amp should have had unity gain, but there was some offset, so I just decided to jumper the input resistors, pull the amps, and jumper from in to out. It's small, but not unworkable (0402 Rs and Cs). I put some kapton tape under where the large caps were, just to make sure my jumpers didn't touch anything going under them. You could connect right to the pins of the QFP and pull the first cap too, but that header looked nice, and should allow a sturdy connection.

I was planning to build the circuit with whatever stuff I had laying around, but changed my mind. I've got enough other things to do, and need to order some stuff from Digikey anyway, so I'm gonna order the parts I need to make the circuit how I want it.

Also, is there a collection of test wave files out there? Something to make sure my 5.1ch card is working correctly, and maybe an ILDA pattern (I know I could output it from Laserboy, but known working ones are always nice )? I kinda remember seeing a download for some wave files for initial adjustment.


And I was hoping to build the HFD adjustment section tonight, but got busy with other stuff... hopefully tomorrow. If that works, I'll try properly tuning the whole thing.

DogP

[tribble]

DogP, that's great work on the power supply! I looked at the crimpers for those pins... I can't justify paying $500 for a crimping tool. Any of them. One of them is over $1000. For that I'd buy a whole new chijector from golden star, pray it cleared customs, and crimp with pliers. Do those spikes correlate with the corners of the test pattern? If so I would be very interested to see how much some fat caps clean that up. I think I posted a link to some 30,000uF ones above. This is just like building audio amps... for storage caps... go big or go home. ;-)

Your overshoot will be cleaned up with a little additional LFD on the X axis. Note the spiral into the home position, though, on the pic with the better power supply... overshoot on the overshoot correction. I think that's what the HFD is supposed to clean up.

[DogP]

DogP, that's great work on the power supply! I looked at the crimpers for those pins... I can't justify paying $500 for a crimping tool. Any of them. One of them is over $1000. For that I'd buy a whole new chijector from golden star, pray it cleared customs, and crimp with pliers. Do those spikes correlate with the corners of the test pattern? If so I would be very interested to see how much some fat caps clean that up. I think I posted a link to some 30,000uF ones above. This is just like building audio amps... for storage caps... go big or go home. ;-)

Your overshoot will be cleaned up with a little additional LFD on the X axis. Note the spiral into the home position, though, on the pic with the better power supply... overshoot on the overshoot correction. I think that's what the HFD is supposed to clean up.

Yeah, the real crimpers are ridiculous. At work we have the real Molex crimpers for pins (a different one specific for each pin series, @ about $300 each)... they're really nice, but totally not worth the cost. At home, I just use my cheap-o Radio Shack D-Sub pin crimpers: http://www.radioshack.com/product/in...uctId=2103683# . They're not the best, but at $10 and locally available, they're well worth it. They work for a lot of pins, though of course some fit better than others. If it doesn't crimp well, or if the pin doesn't fit in the crimper (like the large pins I used on the PS side), I just use pliars and bend the fingers over the wire and add a dab of solder.

I'm not sure where the spikes actually occur, but they're at 11Hz on your red ILDA test file at speed 20. I'm guessing it's the start of each frame (it's probably running at ~11FPS). I've got a 10000uF cap right next to me (spare from an arcade game repair)... I'll try hooking it up and see if it helps. I'm happy to see that this supply is actually regulating though. There's no way you're gonna hold up a sagging power supply with a cap like this, but it should help buy the feedback loop some time to compensate.

I think before I do any tuning, I need to turn off blanking on the test files. Like Steve mentioned, it tells you more about what's going on, but more importantly, I think, is that we don't get the goofy handling between blank and non-blank vectors. And I don't think we want to add any sort of optimization to these files (i.e. the spencerizer), or we kill some of the test purpose.

DogP

[tribble]

I think before I do any tuning, I need to turn off blanking on the test files. Like Steve mentioned, it tells you more about what's going on, but more importantly, I think, is that we don't get the goofy handling between blank and non-blank vectors. And I don't think we want to add any sort of optimization to these files (i.e. the spencerizer), or we kill some of the test purpose.

Pretty sure you can do the unblanking with minimal side effects in LB; k menu, then a. Agreed on the optimization.

Thanks for the link to the RS crimpers; that sounds like a winner.

I don't think I can retune faster than 20 until I solve the "wandering hotspot" problem. I was going to try the 8 degree thing, but if I can't ramp up the speed in the showcard, that's kind of pointless. I recall you saw it in "larger" files, but not small files using the built-in repeat. The test files I'm using are not all that big, certainly not what would be used for a song-length 'show,' so if it can't be worked out, speed 20 is it. Or maybe 21, 22... I'm not sure where the limit is.

I'm thinking more and more of trying to put an ILDA pigtail on this puppy.

[mixedgas]

Dog, trust me on this, tune at 8 or 10 degrees some time in the future. The smaller the angle, the faster you can go.

Steve

[DogP]

I don't think I can retune faster than 20 until I solve the "wandering hotspot" problem. I was going to try the 8 degree thing, but if I can't ramp up the speed in the showcard, that's kind of pointless. I recall you saw it in "larger" files, but not small files using the built-in repeat. The test files I'm using are not all that big, certainly not what would be used for a song-length 'show,' so if it can't be worked out, speed 20 is it. Or maybe 21, 22... I'm not sure where the limit is.

On mine, 20 looks fine all the time... 21-23 look fine, except there is one dot on the first frame after loading. 24+ have multiple dots, progressively getting worse the higher you go. Speed 20 is actually ~16.67KPPS (60us/sample). So, IMO, I'd tune for speed 20. According to the 20KPPS ebay auction, the galvos are supposed to be good for 20K at 20 degrees, and 31K at 8 degrees (maybe it's exaggerated, I dunno).

Dog, trust me on this, tune at 8 or 10 degrees some time in the future. The smaller the angle, the faster you can go.

Yeah, but 8 degrees isn't really useful for me. I use this in my basement, where there really isn't much room to shoot it from far away... where I'm projecting it, I can only be 6 ft. away. A 10"x10" projection isn't very impressive. I was surprised though... when I ran the unblanked file, it actually looked pretty good at 8 degrees @ speed 20. That's probably where they tuned it. Of course with blanking, their software screws it up. :-P

I had some other fun today, but it's too late... I'll post pics and stuff tomorrow.

DogP