A sneak peek!

[allthatwhichis]

What he is doing here is not the same as the color correction board. The CCB is for making the modulation speeds of the different lasers closer together so they start and stop at the same time (or appear to).

Here, he is taking a small sample of the beam of each laser and comparing it against the modulation signal from the DAC to try and solve the inconsistant linearity of the laser output... I think...

[buffo]

What DZ is doing is slowing down the faster lasers, flying an f16 like a sluggish 747, which will work.

Actually, that was what his FIRST project (the color correction board) was for - to slow down the modulation response of the fastest laser to match the slower ones.

But this new board is designed to be installed after the color correction board, and it acts much faster. It uses optical feedback to smooth out non-linearities in the power output of the laser. Believe it or not, it can nearly eliminate jellybeaning! So yeah, this is a totally different board.

Since the the environment of the lasers are well controlled (temp mainly) there are no significant outside disturbances so an extra feedback loop is not needed.

This is generally true for direct injection lasers, which are quite stableas a rule. However, for DPSS lasers (blue lasers in particular, but also some economy-priced greens), there is a great deal of power variation in the output, even with a constant modulation input. These non-linearities manifest as "static" in a scanned line. Bill Benner coined the term "jellybeaning", because the scanned line looks like a bunch of jellybeans stacked end-to-end, rather than a nice smooth unbroken line. And until now, there hasn't been a good solution to eliminate this defect.

the laser's internal feedback control (current, temperature) make it a predictable system so is can be regarded as a 'black box' with a known mapping from modulation input to light output.

In a perfect world, this would be true. However, real-world tests of many lasers has shown that the response curve of power output to modulation input is *not* linear, and indeed is not even consistent over time, for the vast majority of DPSS lasers available. (At least, for the lasers that are priced within reach of the average hobbyist!)

OK, so a couple of questions from someone who still has the big "?" over their head ():
1. In "less-electronically-oriented" layman's terms, what is the end-result functional difference between this particular circuit, and the color correction board?

The color correction board ensures that all your lasers come on at the same time and turn off at the same time.

This new board will smooth out the slope of the power output to modulation input curve so that it is nearly linear. That is, 2.5 volts on the modulation line = 50% of max output, 3.75 volts = 75% output, etc. It will also ensure that the power output will not drift. That is, if you have 2.5 volts on the modulation line, you will *ALWAYS* have 50% output, even after the laser "warms up". Finally, it will get rid of most (if not all) of the jellybeaning described above.

2. Would this be applicable in a "less common" projector setup, such as RGBV?

It will work in any projector, but the most improvement will be seen in projectors that use DPSS lasers, since these lasers display the most non-linearity. Also, for the RGBV example, you have 4 separate color channels, so you'd need two of these boards. (As far as I'm aware, DZ is designing these boards with only 3 channels.)

What he is doing here is not the same as the color correction board. The CCB is for making the modulation speeds of the different lasers closer together so they start and stop at the same time (or appear to).

Here, he is taking a small sample of the beam of each laser and comparing it against the modulation signal from the DAC to try and solve the inconsistant linearity of the laser output...

Yes - exactly! What Aaron said...

Adam

[Stuka]

Sounds like I need this, too -

So many accessories, so little time to get the frickin' rig built...

[DZ]

Wow allthat, very nice explaination!

But this new board is designed to be installed after the color correction board, and it acts much faster. It uses optical feedback to smooth out non-linearities in the power output of the laser. Believe it or not, it can nearly eliminate jellybeaning! So yeah, this is a totally different board.

Well, I wont say that I can eliminate jellybeaning with this. I've been able to do some neat things with various circuits so far, however, being able to eliminate jellybeaning was only at a low modulation speed or no modulation at all. For instance, the laser media test pattern, I was able to clean up nearly perfect. Ultimately, it's limited by the bandwidth of the modulation input on the laser. Red has been flawless, blue has been very challenging, I think I'm pretty close though.

[300EVIL]

Very nice David! Definitely going to pick one of these up as well. You just have a little beam pickoff on each laser head going to a photodiode,,, correct? Pics?

[heroic]

Red has been flawless, blue has been very challenging, I think I'm pretty close though.

Did I mention that I managed to blow up another diode working on my DSP-based version of this? Sometimes I can be depressingly stupid. I had a bug in my code, no light, and so I thought, well, maybe I need to twiddle the driver settings (the signal was scoping out ok, so it should have been fine). It wasn't fine. It wasn't fine at all. :/

[FourDee]

I think that most of the budget blue dpss lasers like cni and laser-wave (I can only say for sure about CNI, haven't tried laser-wave) are lacking performance because of not-so-good drivers.

I know for a fact for example that CNI drivers can only drive a peltier single endedly so they can only cool ...
I've been working on an alternate driver for my CNI blue, but it would be difficult to distribute since you have to change the internal wiring of the head ... (the ntc's and the tec share a common gnd pin)

I have seen however that CNI now has an OEM line of lasers. These are very small laser heads with very attractive prices that are sold without drivers


example :

http://cnilaser.com/PDF/OEM-H-473.pdf

[Zoof]

DZ, I think you are doing great work. With the optical feedback you make the laser's response linear, and proportional to 0 to 5 V. The best results are probably to be had with a HW solution but I have managed to come a very long way with a SW solution.

As I was developing my software I found that the one variable that gave me the largest increase in blue performance was the idle modulation input. When blanked my blue still gets a >0 modulation input, just below lasing threshold. It greatly increases performance in exactly the same way that DZ shows in the pics at the beginning of the thread.

Finally each laser's modulation is corrected with 4 parameters:
maximum, minimum, blanked, and linearity (similar to gamma curve).

These parameters allow excellent compensation for chinese dpss problems. With these corrections I can produce a grid with a white to black color gradient without losing white balance for any intermediate gray value! Will show when I have the pics.

However, the problems with this SW solution w.r.t. a H.W. solution are twofold:
1) cannot insert it an any existing laser control SW-HW solution.
2) timing of the lasers can only be matched on a point time interval basis, not arbitrarily as with HW.

So I see the need for a HW solution but SW works for me.
BTW this SW solution is feedforward (still no need for feedback, but the parameters have to be known).

Anyone wanting to SEE what I talk about, take a look at the 3D demo software and play with the R,G,B, tabs' settings.

[DZ]

You just have a little beam pickoff on each laser head going to a photodiode,,, correct? Pics?

Wouldn't that be a pain, having to setup 3 different detectors. I found a different way using 1 detector and "1" photodiode. Was going to be kinda the big secret, but what the heck, I'm sure my secret's safe with you! Ignore the tin foil, was out of shielded cable at the time and since it's working just fine with the tin foil, I haven't bothered to change it.

Did I mention that I managed to blow up another diode working on my DSP-based version of this?

Sorry to hear, I did have green in the mix on this, however after a few hours it stopped working. I'm still not sure what the cause of failure is for the green, it was a bit of a franken-laser. Unknown laser hacked to work on a lasever driver. I'll have to take a closer look at it, maybe something just disconnected on it.

[/quote]I have seen however that CNI now has an OEM line of lasers. These are very small laser heads with very attractive prices that are sold without drivers[/quote]

That's interesting, how attractive are the prices? Some cheap blue's matched to some drlava drivers would be nice.

Zoof, I'm sure many of these problems could be resolved with software, but not all of them! From seeing what the circuit has to do to keep the output stable, I just don't see how software could match it. Delaying and linearity to a certain degree could be cleaned up a bit. Does your software run on the QM2K? I'd like to give it a try.

[daedal]

Wouldn't that be a pain, having to setup 3 different detectors. I found a different way using 1 detector and "1" photodiode. Was going to be kinda the big secret, but what the heck, I'm sure my secret's safe with you! Ignore the tin foil, was out of shielded cable at the time and since it's working just fine with the tin foil, I haven't bothered to change it.

...

You got me thinking again...

Wouldn't compensating for a single "combined" color ruin the colors? Unless of course you are using a triple 'line' PD and the optic is splitting the three colors into the three separate PD's. That would be brilliant IMO, and I can see form your pictures that you have 5 wires into the PD

Nice one David... this is absolutely amazing! If you need a second test rig (nearby, i.e. shorter drive) let me know

--DDL