Combining beams

A discussion was going on on LP about combining beams. It was the usual let’s serial several pbs cubes to make ridiculous energies. However it got me to think and here is what I thought.

we use one cube to combine two beams. There is no reason you could not send multiple beams through the cube and stack them like a knife edge. So say put 4 S polarized and 4 Polarized beams into one cube. This eliminates the mirrors needed for the knife edges and with a quarter wave plate leaves you with a stack half as wide.

ok great no great revolution. Now use a broadband cube and withe a single 1 inch cube you could combine all the rgb lasers into a very compact set of 3 coplanar beams. Those go to the usual dichro for combining.

a good application would be quad red, dual green, dual blue with mixed wavelength of each to enrich the color space all in one cube. Think 3d use of the cube surfaces. Reds at top green in middle blue on the floor or turn it sideways.

it would also shrink the rgb source a lot and best of all intrinsically fix a lot of the geometry so things won’t drift. In fact even heat is canceled since all beam will drift the same amount. I don’t think index of refraction would change enough to matter. If it did back to three cubes with a single red quad cube.

ok that’s the idea. Have fun and if you build it let me know.

I think I need to see a diagram regarding how you'd align three colors with a standard broad-spectrum polarized beam splitter/combiner cube. I also can't visualize how'd you'd stack without mirrors.

Buffo combined RGB with a RGB combining prism from a video projector several years back but the losses were hideous as I recall. (I have a couple of those prisms if anyone has a use for them.) I'm pretty sure the broadband PBS cubes are all very 'lossy' as well which may explain why this isn't commonly used in practice.

I look forward to chatting more about this in person in a couple weeks!

-David

Quote:

---

Originally Posted by kecked

LPF says: let’s serial several pbs cubes to make ridiculous energies

---

Common suggestion, with an equally common caveat that makes it a non-starter.

Here's the point that most everyone forgets: The output beam from a PBS cube is randomly-polarized (assuming H and V polarized beams at the inputs).

So now if you feed that randomly-polarized output beam into one of the input faces of a SECOND PBS cube, half of the energy will be reflected and the other half will pass straight through.

Thus even if you capture that second beam (normally that face would be glued to the mount so that pass-through beam would be absorbed), all you've done is to un-do the combining step of the first cube - you're left with the same H and V polarized beams you started with, minus the power lost through the optics.

Now, admittedly there are a few hugely expensive metamaterial optics that can re-polarize a beam with low losses (how? Creepy non-linear quantum magic... Just push the "I believe" button), but the cost is not practical for our application.

Quote:

---

we use one cube to combine two beams. There is no reason you could not send multiple beams through the cube and stack them like a knife edge.

---

There are designs that use knife-edge mirrors to stack several beams together in both H and V polarization and then use a single PBS cube to combine them. There are even designs that use a 1/4 wave plate to rotate the polarization angle while maintaining the beam profile, so instead of your V beam being like this: || and your H beam like this: =, you can have both beams be like: ||. (Hope the ascii makes sense!)

Note that there are broadband PBS cubes that have low losses - they're just really expensive!

Adam

Its pretty simple but easier to see if drawn. essentially you stack the beams into the cube, combine them and then dichro combine the result. Only advantage is less moving parts but much higher tolerance needed in machining to align beams to start. Once done though nothing should move. If you only do duals you don't even knife edge anything. three beams in each side of cube dichro combine the resulting beams. The only thing new is using a single PBS cube instead of three.

Logsquared made me an rgb block that has not drifted in ten years so it can be done. His is the reverse of this. He dichro combined three didoes. Two of those are pbs cubed. In this case the didoes are pbs combined then dichroic. Same thing just backwards.

Something important to take into account is removing that adjustability removes degrees of freedom. If it stays aligned you get away with it. If not your hosed.

I'm describing two things here. One is stacking four reds with four reds for a combined 8 beams. The other is using one cube for rgb.
Two different thoughts same principle.

As bastard <Bra|Ket> states:
<RRRR|P90>:<RRRR|S>=2|<RRRR|P+S>| leaving waveplate out too many brackets. just notate 90 degree turn. abs value required for intensity.
<RGB|P90>:<RGB|S>=2|<RGB:P+S>|

Well Really {[<R|P90>:<R|S>],[<G|P90>:<G|S>],[<B|P90>:<B|S>]}={2|<R|(P+S)>|,2|<G|(P+S)>|,|2<B|(P+S)|>} ------->Dichro------>2|<R|(P+S)>| + 2|<G|(P+S)>| + 2|<B|(P+S)>|

Generalizing <Wav(sub i)|P>:Wav(Sub j)|S>=Sum[|<Wav(Sub ij)|(P+S)>|] Now you can combine all you like When I=J =2|<Wav|(P+S)>|

Why did I just type that....Sign of a sick mind.

No matter how you "stack" the input beams, you are still using the knife-edge mirror technique. You need the degrees of freedom to ensure the beams are parallel and as close as possible to one another.

Adam

The cube acts as the knife edge mirrors. Yes you set up the beams so they are knife edged prior to the cube but they are just aligned in a metal housing without the mirror. The mirror is the center of the PBS cube. and yes each stacked side is P or S and yes rotated with a waveplate. The result is the same its just a cute way to do it.

Forget that.

Simplicity. Just combine two of each color with rotated polarity You say H V, I say S P same idea.

The idea is you use a single cube instead of three cubes. One cube combines each pair dislocated in space on the cube. The result then goes to the dichro line to make them collinear. like normal.

The question is, is one broadband PBS cheaper than three wavelength specific PBS cubes. Second is the amount of space required more important than the extra cost.

thor 1" broadband is 200.00 thor sells the individual line cubes for the same price. So really it's a matter of ordering a broadband cubes from a more affordable seller. No reason they would cost more. In fact should be cheaper to coat.
Swisslabs or whatever the name was could do it.

I emailed the drawing. Can't seems to attach images these days.

Quote:

---

Originally Posted by kecked

The cube acts as the knife edge mirrors.

---

This isn't going to work for knife-edge stacking of multiple diodes of the same color on each input face because you can't mount the diodes close enough together in a block. You will still have to use a knife-edge mirror on the output. True, it reduces the number of knife edge mirrors by half, but it doesn't eliminate the need for them.

Although your example below with just 2 diodes of each color will work without any knife edge mirrors, I don't see how it's going to be easier or cheaper than using 3 cubes. It also forces you to be very precise with your diode mount, since you have no other adjustment anywhere. I foresee lots of alignment problems...

Quote:

---

Forget that. Just combine two of each color with rotated polarity

---

This will work, but alignment will be a huge challenge. You're talking about mounting 3 diodes (red, green, blue) close together in a single block so that all 3 colors can hit the same broadband cube with all beams at the same height and parallel to each other.

With two diode blocks (one for each input face), that gives you a pair of diodes for each color, so your output from the cube would be three separate beams, each one having essentially double the power of a single diode, and all of them being randomly polarized and positioned very close together. Then you'd need a set of closely spaced dichros at the output to combine this into a white light beam. (I attached your picture of this arrangement below.)

The problem is that the diodes can't be mounted very close together in the block due to thermal considerations. And even if you did mount them as tight as possible (so the cans were literally touching), you would still be hard-pressed to get all 3 beams to fit on a 1 inch cube. By the time you add a tiny bit of spacing between the diodes, you'll definitely need to up-size the PBS to 2 inches, and at that point the cost skyrockets. ($800 and up)

Quote:

---

is one broadband PBS cheaper than three wavelength specific PBS cubes.

---

For a 1 inch broadband PBS cube the cost is probably the same, or maybe only a little more expensive, compared to using 3 separate cubes. For a 2 inch broadband PBS cube the cost will be significantly higher vs using 3 separate cubes.

Quote:

---

is the amount of space required more important than the extra cost

---

I own several modules that contain 6 red diodes, 4 green diodes, and 2 blue diodes, all with secondary correction optics, and including all the pbs cubes, dichros, and mounts for everything. The entire package is about 8 inches long, 5.5 inches wide, and 3 inches tall. I don't see any arrangement with a large broadband PBS cube and the same number of diodes being significantly smaller.?.

Quote:

---

thor 1" broadband is 200.00 thor sells the individual line cubes for the same price.

---

10 mm single-line cubes can be purchased from Chinese suppliers for between $50 and $75 each. That's only because these cubes are ubiquitous in the laser industry so they're basically a commodity item. In contrast, a broadband cube with high efficiency is far less common, so your only choices are Thor or Edmunds (or similar), which drives the price up.

Quote:

---

it's a matter of ordering a broadband cubes from a more affordable seller. No reason they would cost more.

---

Except that there isn't much demand for them, which is why they aren't being made in bulk and why they cost more. It would be nice if we could find someone to make them at the same cost as the single line cubes, but I doubt anyone will be willing to do so.

Quote:

---

I emailed the drawing. Can't seems to attach images these days.

---

Here's the image:

https://i.imgur.com/2EPRND2.jpg

And the descriptive link to imgur.com: https://imgur.com/a/ORMnbIB

Adam

... I have a good sample for "knife-edging" multiple diodes into one beam with a distributed diode array.

The 7 diodes with 5 emitters each are first collimated into a horizontal "line" ... then this "line-beams" are slightly "tilted" the same amount by the 45°-mirrors ... and "stacked" onto the last mirror, which beams the combined beam up vertically ...

Viktor

Whoa, where does one find multi-emitters like that?
Wavelength?

Never saw those before.

Thanks for posting!

...Mike

... this are "pumping diodes" for multi-kilowatt fiber-lasers or DDL (Direct-Diode-Lasers) -- they emits @976nm and the combined beam is inserted into a 0,2mm fiber and outputs up to 200 Watts CW! (up to 600 Watts pulsed, if averaged to 100Watts max and below 3 microseconds pulse-time)

I've bought 8 of this diodes (around 400 Euros each) as "salvaged" at ebay - two for an actual R&D-project, six for my "private use" :cool:

Viktor


PS: -- see here for some other diode types, I've collected and used in the last 10+ years - https://reprap.org/wiki/Laser_Cutter_Notes

Nice! I wish I could machine metal like that.

Well it was a thought. Good or bad I share them all.

... two (or up to six ;)) of this 200Watt-beams on a spot of 0.4mm is enough power to melt/fuse metal wires on a metal surface ... and you doesn't need such a stiff/solid frame, as used with metal machining.

So my plan is to hook a head with 2 to 6 fibers on my CNC-mill and perform some tests with "additive fabbing" with all sorts of materials, I can melt with the diodes :)

Attached is an older idea from this area, where I've combined 6 blue 1.5Watt-"beamys" on a spot of 0.3mm ... today there are blue 6Watt-diodes available ... or the 200Watt IR ones :cool:

Viktor

Ok so I think I solved the cost issue.

There is another way to do the same damn thing I just noticed. Just use three regular cubes aligned on their tips in a diagonal.



___
|__|__
.....|__|__
..........|__|
________Waveplate


Now you can use cheap cubes and they can be 5mm Same as if it was a 15mm cube.

One issue is packing in the dichros so tight. Can't think of an adjustable mount that would fit in less than 5mm of space so what does this really do...... No one wants to adjust the diodes to align.

I was thinking you could get a watt of single mode in a space under 2" unlikely unless mass produced with a jig to align it all and if it drifted your toast.

Well ideas are free.... i had to add the .... lines to make the cubes show right

ASCII art! Brilliant.

...Mike