4 diodes better than 1. DIY Array on the cheap

[kernelpanic]

[quote author=Hemlock Mike link=1229985350/20#29 date=1230085590]You probably will see some mixing of the feedback current with all the tabs common.  I'd use mylar insulator pads and heatsink grease.   Common ground will be through the LD's.   The tab is the refference in each circuit.   (I think)  

Mike

The center tab is Vout:[/quote]

Sharing Vout negates the use of multiple drivers. The circuit would act like one driver with a current output equal to the sum of the individual drivers. But at the ~100-150mA that the PHR diodes run at, you don't need a heatsink at all unless your supply voltage is much larger than the diode's running voltage. So if you are driving from 12V or more, you'll want to heatsink. But if your supply is 9V, then each LM317 is only dissipating around 1/3 to 1/2 a Watt. (~3V x 100-150mA) That is about 1/20 of it's capacity. I usually only heatsink these if I'm running more than 500mA.


hope this helps,
kernelpanic

 

[JLSE]

Yeah , thats what I figured. Wasnt sure though, thought there may be a trick other than isolating, but it only made sense that it would couple the adj from all four. I wanted to etch a beefy driver just in case I do this with reds.  But for now, heat wont be an issue, and I will arrange the board to accept one large single H/S just in case.


Thanks, all the same

 

[randomlugia]

Hmmm... Couldn't you use use a second lens to keep them in one beam? Example:

 

[suiraM]

Simple instructions for joining laser beams:

Get an aluminium plate, and mount a diffraction grating on it. A wooden block or similar should probably do fine, depending on the grating. The pro ones can be supplied with mounts. For cheap types, you can glue it to a piece of glass with optic glue and sandwich the glass into a cut in the wooden block. The grating should be parallell to the far edge of the plate.

Put one 2x4" on each side edge, and shine a BR laser through the grating from the far side (if transmissive) or near side (if reflective). Draw the beam paths on the plate with a felt tip pen or something like that. With the right sort of grating, you should see at least 4 beams, probably far more. Mark off the points where the beam needs to be deflected in order for each beam to end up at the 2x4"s at regularly spaced intervals, with no beam paths crossing each other. Draw them all the way to the 2x4". Turn off the BR laser.

Find your compass, and halve the angle between the beam and deflection, then construct a line that is perpendicular to the halved angle at the precise point of deflection. Set up one mirror at each point of deflection. Remove the 2x4"s from the sides. If you get, say, 8 beams, and only have 4 lasers, then place mirrors or beamstops at the end of the deflected lines that you will not be using.

Mount one Meredith module (the block type) for each deflected line, so that the laser output will be perfectly parallell with the line in question. You will want to use such a module when you're drawing the lines if the grating is biaxial, to get the right height, but it shouldn't be needed if you're using a regular single-axis grating. Obviously, these modules should be screwed onto the bottom plate, and if you want thermal control, you can use a torque wrench and some arctic silver epoxy to clamp the peltiér plates between the modules and bottom plate, which will give a good coupling. TEC will improve the quality of the output beam, particularly if you can set the temperatures individually. Again, not necessary.

If the grating is reflective, the output beam will be on the near side of the plate. If it is transmissive, the output beam will be on the far side. The beam will be parallell to the grating. You will probably see some scattered light from higher orders when you power up the rig, but most of the input power will be directed into the first order, which will be the output beam. At least, that's the theory. Alignment is somewhat critical to the output beam quality, and not all diodes respond well to being mounted in this way. With a bit of luck, though, you should be able to get a collimated output beam that has a reasonably narrow linewidth, and if your angles are carefully set up, the output should behave as a single beam for the distances you're interested in, without too much fuss with the optics later on. The same cannot be said for the rig you've already got, although the divergence can be smaller in your setup if you get perfect alignment.

With this suggestion, you're realizing a filled aperture. If you use a reflective grating, it will also serve as a Littrow external cavity diode laser configuration, which will give you a very narrow linewidth by locking the lasers to each other. In both cases, it may be beneficial to place a microscope slide glass in the path of the output beam, parallell to the grating, but I'm not sure how close the tolerances are, so you might not want to commit to that. The point being to lock the lasers from both sides of the optical axis by providing about 4% optical feedback.

Depending on the grating, you can probably merge anything from 3 to 17 beams with a similar technique, but the output divergence depends on how closely matched their wavelengths are. For a regular grating of the sort hobbyists are likely to use, the spread shouldn't be so large as to cause a problem. With a good ruled grating, you would need to use temperature, current and polarization control to lock them tightly (but then you would probably also be using them for something where that would be the least of your worries).

Absent a grating, you can cheat with a DVD, as they are grooved closely enough for this wavelength, but the alignment will be trickier. Note that DVDs are reflective gratings, not transmissive ones, so please take good care of your eyes by protecting them from the (many) reflections.

Hope this helps.

NB¹ • If you're using the same technique for a different wavelength, you have to use that wavelength when drawing the lines, or else it won't work.

 

[JLSE]

Hmmm... Couldn't you use use a second lens to keep them in one beam? Example:

Unfortunately, its easier said than done. When 4 seperate beams enter the optics, you will always get 4 out. I believe the idea is to live with this problem, but get the beams closer together in paralell. So essentially you still have 4 beams in the final output, but they are so close together, they behave as one.

The cube used in the setup (pending) somehow eliminates the difficulty in bringing them together, but id imagine they still remain as 4 seperate beams tightly paralelled.

To see this behavior up close, take two diodes with raw output, and no focusing lens on the end of the modules. Place them close together and power them up. Now put a larger collimating lens infront of the both of them. Voila! lol, two seperate beams

They will always remain as seperate beams, travelling on their own paths. Ive tried many optics in the past to try and achieve this. Ive tried the above with two modules side by side, focused them down together, placed an expander in the path and them collimate again after the expander. Guess what, two beams again

If there is one thing ive learned in my time trying different things, is that to get more power, mainly by combining, things become very tricky. Combining of different wavelengths is much easier than this, and is a good start to begin understanding.


@ suiraM-

That is an interesting idea, though it sounds equally as involved as the route ive chosen. Do you by any chance have a crude sketch or diagram to aid and illustrate the proceedure? I'd like to give this one a try as well, but would expect the losses to be a bit more than my current arrangement.

I had thought of trying a diffraction gratting, but hadnt put much time in the idea. The problem with BR that I also see for this, would be the absorbtion of the gratting material itself, but may be wrong.

I cant picture this arangement in my head, but if you can post a pic, im sure things will become much clearer.

As usual, all input is welcome ;D



And Merry Christmas to all

 

[suiraM]

Let's see if this drawing explains it better.

I've drawn it for four beams in two orders, based on a reflective grating.

The boxes on the sides are the Meredith block modules (easy to mount to a plate). The smaller boxes around the centre are first surface mirrors, except the middle one (a microscope slide glass). The remaining large box at the top is the reflective grating, and you can think of the paper surface as the mounting plate. Beam paths are shown in red. The red arrow shows the direction of the output beam. To find out what the diffraction angles are, so that you'll know where to put the first surface mirrors, and at what angle, you must use a laser of the exact same wavelength in the opposite direction of the output beam.

If you have a photodiode with a tiny surface area, such as those found inside laser diodes, then you can use that to fine tune the alignments of the mirrors, since the output will be at its maximum when each laser is deflected at precisely the correct angle. You can't possibly eyeball that to any extent. Also, note that while I have shown the drawing with side mounted lasers firing toward the centre axis, and the beam deflection as done entirely by the mirrors, you may find it easier to place the lasers differently. A lot of mirrors work best with 45° or 90° angle of deflection, and that also gives a wider tuning range. Probably not as sensitive, either.

I'll make a better drawing for you to illustrate a related concept, in case you happen to have easy access to reflective gratings.

This one should give a good idea, though...

 

[JLSE]

Let's see if this drawing explains it better.

I've drawn it for four beams in two orders, based on a reflective grating.

The boxes on the sides are the Meredith block modules (easy to mount to a plate). The smaller boxes around the centre are first surface mirrors, except the middle one (a microscope slide glass). The remaining large box at the top is the reflective grating, and you can think of the paper surface as the mounting plate. Beam paths are shown in red. The red arrow shows the direction of the output beam. To find out what the diffraction angles are, so that you'll know where to put the first surface mirrors, and at what angle, you must use a laser of the exact same wavelength in the opposite direction of the output beam.

If you have a photodiode with a tiny surface area, such as those found inside laser diodes, then you can use that to fine tune the alignments of the mirrors, since the output will be at its maximum when each laser is deflected at precisely the correct angle. You can't possibly eyeball that to any extent. Also, note that while I have shown the drawing with side mounted lasers firing toward the centre axis, and the beam deflection as done entirely by the mirrors, you may find it easier to place the lasers differently. A lot of mirrors work best with 45° or 90° angle of deflection, and that also gives a wider tuning range. Probably not as sensitive, either.

I'll make a better drawing for you to illustrate a related concept, in case you happen to have easy access to reflective gratings.

This one should give a good idea, though...

Ahhh, now that makes sense! Some things are hard to picture, but that sums up the concept quite well.. ;D

Have you tried this configuration? And if so, what kind of losses would be expected? And no that im questioning it, but is this of your design or is there a tutorial out there with the results using this setup?

Im going to complete the first setup, then move on to other methods of beam combining, and see what the better way is. If the grating proves to be a simpler method, there will be persons on this forum who may wish to try. Though if the losses are reasonable, I predict many combining DIY's is LPF's future.

I will be trading a nice BR laser to another member for a 2in. RGB cube, so this project should progress nicely, if not just give me more ideas


Thanks again for the great input

 

[suiraM]

You're welcome.

It's a common technique, called "filled aperture" beam joining.

I haven't played too much with it yet, because of the high cost of good gratings, particularly after the insane customs fees up here. No serious supplier is going to mark their shipment as a gift, so I end up with a 100% effective surcharge. And Thorlabs wants the price of a plane ticket to ship me some simple sockets! Since I'm using this stuff for other reasons than hobbyist use, a low quality grating just isn't worth it.

The expected losses depend on the grating. A decent reflective grating could probably use 70% of the output from each diode. Transmissive type gratings, no idea, but I hear some of them have better overall efficiency, so it might be worth a shot. You can get an idea of the concept using a cheap grating, and upgrade to a better one if you're happy with the results. Edmund Optics can probably help you select the best one for your use.

Be sure to keep us updated on how it works out, if you do try it.

(Beamshot? )

I think you can also use this technique to merge several wavelengths. Essentially, one grating should be able to cover a full octave (say, from 400nm to 800nm) of the spectrum by mapping the correct diffraction angles for each wavelength involved. How much power you can merge at a given wavelength will depend on the number of diffraction orders you get (two orders in the drawing I provided). There's always m=±1 at least, so you will always be able to merge at least two beams per wavelength with a single grating. To get the best possible beam merging, you can even lock the wavelengths with additional gratings used in a so-called Littrow configuration, or the modified version that tunes the mirror angle instead of the grating angle. If you have some temperature control, or at least a very good driver and heatsink, then you should be able to set all the diodes of each type to the same central wavelength. If the mirrors are precisely aligned, that can give a damn impressive beam quality. No tuning is needed for the greens, since they are usually single crystal types pumped with a single mode diode.

If you need more power, you can use a reflective "star pattern" diffraction grating. Just suspend it rigidly over an aluminium plate (be careful with those reflections!) and then mark off all the "stars". Make a note of the beam angles, and drill holes in the plate that match the angles of those beams. Insert AixiZ heads into the holes, adjusting them for the best alignment, then weld or epoxy them in place at those angles. The net result will be a very large amount of power in a single output beam. If you are worried about the beam quality, a spherical lens can focus that beam into a multimode fiber. The output will be circular, although the lens setup will be more expensive.

I'm considering doing something along those lines for a multi-watt shortwave laser rig...

Just have to see if my inquiry is taken seriously or not, and whether the price of crystals will compete with diodes.

 

[viroy]

I've been thinking about this for a while and may have a solution:
Imagine a mirrored cone about 1" tall, at the top it comes to a very fine point.
The idea is to bounce multiple lasers in 360 degrees at the tip of the mirrored cone.
They would reflect at 90 degrees, and if they all were on the tip just right, you should be able to maintain a very tight beam out. from there all beams should be perfectly parallel and easily collimated.
seems to me that if the cone was exactly the right shape, you could combine near infinite beams.

 

[viroy]

kinda like a jet engine... source comes from 360 degrees, meets in middle and is redirected

 

[Jimmymcjimthejim]

Something like this, but less expensive?

 

[Cyparagon]

You're not thinking in three dimensions. The laser is not hitting a planar surface and would thus not be reflected properly.

 

[viroy]

yes it would bend out a little but if you collimated right away you shouldnt lose much.
why else would they sell those cone mirrors?
Thats the first time ive seen those

 

[rog8811]

kinda like a jet engine... source comes from 360 degrees, meets in middle and is redirected

We had a go at the cone idea some time ago, I gave it up after a while as I felt it was a dead end.

Regards rog8811

 

[rog8811]

This was the concept drawing for it...

Regards rog8811

 

[Switch]

Like cyparagon said. If you look at a 2D drawing it makes perfect sense.But in reality the beam is hitting a curved surface.Furthermore, the cone tip is tiny and the curvature of the surface is huge.The beam is gonna spread out a lot, pretty fast. You'd need a fancy lens to recollimate that :