Soon starting Open source low cost RGB laser dac. Need input

[macpod]

Starting after the holidays I'm going to start making an open source USB rgb laser dac board.

If I were making it only for myself I would make it have these features and only these features:
-usb
-Small
-low cost
-only up to 64kpps speeds
-12 bit XY channels (differential)
-3 analog laser (single ended, non-differential) unsure of resolution
-possibly have an interlock pin
-breakout pins if convenient.. but probably not 0.1" headers since those take up lots of board space and make routing more difficult on 2-layer boards.
-mounting holes
-no ilda connector

Chances are if I showed this board with these features to a forum like this I would get a ton of "why didn't you ad / if only this board had [insert feature here]" responses which can be depressing and drop my motivation since many of the feature requests would be simple to implement or at least make possible at a later point in time via breakouts.


SO! I would like to solicit feedback from you folks to see what features you would want or deem necessary in a usb dac intended to be stuffed in a box along with the lasers such as:
-Minium/maximum ilda rates
-how many color channels should it have?
-should the color channels be differential? single vs bipolar?
-what other channels/connectors should it have (i.e. shutter, focus, digital turboencabulator pwm thrust bearing autoaligner input, etc)?
-What resolution should the color channels be? Before I started the lasershark I read around on these forums and saw folks requesting 12bit color channels. When I showed the Lasershark board here and a few other places I had many pm or email me and ask why it supported 12 vs 8 which surprised me. Is there really no desire for more colors?! Is 8 deemed to be good enough? It would significantly cut costs is why I ask.

[mixedgas]

I dont have time to type this, so this is a crummy response with no citations.

The demand for 12 bit color channels is because when color manipulations are done in software, it "eats" bits. 12 is not really needed, nor can the eye see more then about 64 colors at a time. The problem is you can't do a 8 bit multiply on a 8 bit number and expect to have a lot of energy left in the laser beam.

The demand for 12 bits is coming from some one who wants more. However they have not studied the lit on eye color response in laser projectors. Your eye (At any given time) sees about 32 colors if your normal, and perhaps 64 colors if your a "genetic" artist. When your dimming and fading the image is where the problems start.
Bad jerky fades show up. The eye is lousy on color, but great on seeing shades. In fact maybe the central 33% of your vision is color vision, the brain computes the rest from grey scale sensors and motion cues. But it sees changes in overall image intensity VERY well.


Say you have a 8 bit dac and you do a matrix based axis rotate on a vector image in a 8 bit processor. Your effective resolution when done is 5-6 bits at best. The image can look Jaggy if your rotate slowly. The issue is similar, but WORSE, with color.

Early show systems got around this by using a DAC to drive the Reference input current of another DAC.

Its not realistic from the eye point of view. I'd dare anyone to see the difference at the mid to high end of the palette of a 1 bit change at 9 bits. At the low end of intensity, bit counts matter. One best selling laser software has the worse color palette (lots of stairsteps) I have ever seen on its low and midrange products.

4 outputs from 10 bit DACs for color is more realistic. Its not hard to have 4 different diode laser wavelengths in a projector. Ie red green blue and 405 nanometer violet.
A few people have the hardware for this or have 4 or 8 channel PCAOMs* for color. (I have 8 color channels to feed if I want)

Another alternative would be to have a intensity channel DAC on the board that drives the DAC reference input. You'd need a more expensive color DAC to do this.
So you have classical RGB + I, and if the software does not support I, then the I dac is driven to full high by default by your board.

BTW, TTL (On/OFF) color is dead and should DIE. Analog drivers cost the same to build as a good TTL driver. The zero to five volt signal, if buffered through a good opamp with 100 ohm or less output impedance, should drive a TTL line just fine. Especially if there is a weak 10K pulldown on the output for those really cheesy Asian TTL projectors that do not comply with even the TTL standard voltages.

Analog color with three channels should be 256x256x256 = ~16 million possible colors. You cant possibly see that many, but you need good fades in a show.

Even if you do have the extra resolution or intensity channel, probably only LSX Software and Wave files will support it NOW. Extra channels = VERY GOOD IDEA, extra resolution = nice but future software needed.

Its your call, you may need a jumper on the board to force 8 bit x RGB color when needed by certain software. You may be able to drive a bit of a revolution by doing a I channel in hardware to drive the RGB hardware. It is a proven way out of this. PM me if you have questions on how to do that.

*Polychromatic Acousto Optic Modulator, what was used with ion lasers before these lousy RGB only projectors came out.
I can have two reds, a yellow, two greens and three blues in my projector.

Steve

[macpod]

Thank you Steve,

I think I've started typing a reply to your post about 3-4 times now, but decided not to hit the submit button. Really sorry for the slow response

Your color explanation and The RGB+I setup makes total sense to me. Originally to save cost I was thinking of using a 2-channel 12-bit dac for XY and a {TBD NUMBER GOES HERE} channel {TBD NUMBER GOES HERE}bit dac for the colors.

I could just as easily leave the 4-channel 12-bit dac I'm using on the current Lasershark on the PCB and use one of the extra two channels on it for the Intensity. How would these specs sound:

• usb
• Small
• low cost
• only up to 64kpps speeds
• 12 bit XY channels (differential) Tuneable via on-board pots.
• 12bits of Intensity control (not broken out, this connects to the analog laser dac for scaling). If SW doesn't support it, it is maxed.
• 12bit channel for ???? thoughts on use?
• 4 analog laser channels (single ended, non-differential) 8 bits, but possibly slightly higher if price is neglegible on components.
• 5v TTL Interlock A (output) and B (input)
• breakout pins if convenient.. but probably not 0.1" headers since those take up lots of board space and make routing more difficult on 2-layer boards.
• mounting holes
• no ilda connector
• Serial number

I'm not really interested in dmx stuff, but are there any other important connectors/channels that should be added too?

[mixedgas]

Built in Serial Number for LSX?
I'll get back to you.

Steve

[macpod]

Built in serial number will be there. I'm using the GUID of the microprocessor. I'm actually programming this for the Lasershark board as we speak

[macpod]

Built in serial number will be there. I'm using the GUID of the microprocessor. I'm actually programming this for the Lasershark board as we speak

Serial number stuff is done. 128bit GUID I also wrote some more documentation on how to setup the lasershark stuff.. now to get back to the new version...


For the Lasershark pro, I'm going to use either the LPC1347 or one of the LPC175x series. I'm thinking the latter since they have 1024byte usb ISO transfers vs the limited 512byte transfers of the LPC134x series. It's also faster, but a downside is it is more expensive and larger.

The X and Y channel design won't change much from the Lasershark as I haven't heard complaints about that. I am considering ditching the pots for digital pots and auto-calibration setup via a adc, but the more I think about that, the more I think it's just overkill.

[als1980]

No such thing as overkill I just got my board in the mail today and your build quality is amazing. Thanks again for letting get one of, these guys from you. I will be hooking up to test in the next few days. I'll will post some pics of everything a happy customer is playing with
Thanks again! Keep up the great work!

[shrad]

as for low intensity resolution, wouldn't a logarithmic curve improve things? I don't know if this is supported in any software, but it should improve things

[macpod]

No such thing as overkill I just got my board in the mail today and your build quality is amazing. Thanks again for letting get one of, these guys from you. I will be hooking up to test in the next few days. I'll will post some pics of everything a happy customer is playing with
Thanks again! Keep up the great work!

I'm glad it got there quick. Let me know if you have any questions. If you have them, then others probably will too.. and those would be helpful to know for documentation purposes

[colouredmirrorball]

as for low intensity resolution, wouldn't a logarithmic curve improve things? I don't know if this is supported in any software, but it should improve things

In LSX you can edit the linearity curve of each colour (+ master intensity).

This is a very interesting project, keep up the good work!