Of polarization, beam combining, summation and power gains....

[norty303]

Ok, so this one has had me thinking for a couple of nights, and I can't work it out. Can anyone explain for me?

If we want to combine 2 beams, we have to use H and V polarization, otherwise the resultant output power is equivalent to just 1, correct?

So, with these 445 diodes (or indeed any wide stripe emitter diode), one of the popular ways of combining these is through knife edging.

As we know, the divergence is greater on the fast axis (fast means it diverges faster, yes? Or have I got that wrong?)

So, once past the knife edge mirror, both beams will continue to diverge (usually most in the horizontal axis, assuming a vertical 'side by side' knife edge setup) causing them to overlap.

Does this mean that the portions of the beam that overlap don't actually give us any gains as they are the same polarity? I would think it does, however the power outputs we see from dual setups don't seem to show these losses (or rather 'lack of gains')

[buffo]

If we want to combine 2 beams, we have to use H and V polarization, otherwise the resultant output power is equivalent to just 1, correct?

This is correct, assuming you're using a polarizing beam-splitting cube (in reverse) to combine the two beams. In this case, you need a horizontally polarized beam and a vertically polarized beam for the cube to work correctly. The advantage is that you're using the polarization difference to perfectly overlay one beam on the other. Thus you get a beam with roughly the same diameter as a single diode, but with nearly twice the power.

So, with these 445 diodes (or indeed any wide stripe emitter diode), one of the popular ways of combining these is through knife edging.

Yes, this is another way to combine the beams. You end up with a final beam that is twice as large, however. That's not a big deal when the beam is already fairly fat though, as is the case with these 445 nm diodes. People realize that you can stack them next to one another, and end up with a roughly square beam. Also, knife edging is cheaper than using a PBS.

Does this mean that the portions of the beam that overlap don't actually give us any gains as they are the same polarity?

No. The polarity has no bearing on the power. Polarity is only a factor when you are using a PBS, because one face will pass horizontally-polarized light and reflect vertically polarized light, and the other face is exactly the opposite. This is a function of the optical coating on the cube, not some fundamental property of polarized laser light.

Bottom line: if you knife-edge two horizontally-polarized beams, you will get (roughly) double the power, even as they diverge and overlap one another.

Adam

[norty303]

Ah, ok, so the relevant part as to whether you get doubling of power when using 2 sources is not the polarization per-se, its a function of the fact you're using a cube to combine them (which requires H and V to work)

So if you could combine 2 H polarized sources without a cube (fibre launcher?) then you would get doubling of power still?

[buffo]

Ah, ok, so the relevant part as to whether you get doubling of power when using 2 sources is not the polarization per-se, its a function of the fact you're using a cube to combine them (which requires H and V to work)

Exactly!

So if you could combine 2 H polarized sources without a cube (fibre launcher?) then you would get doubling of power still?

Well, here is the problem: If you want to perfectly overlay two beams, you need some way to differentiate them so you can reflect one and pass the other. One way is with a dichro, since it's reflection properties are frequency-specific. Thus you can mix red with green... But if the two beams are of the exact same frequency, then you need some other way to differentiate them. The only thing left is their polarization angle.

I'm not aware of any means of combining two beams of the same frequency and polarization that results in a perfect overlap like you get with a PBS or a dichro. Can't even see how fiber would help, though admittedly I'm no fiber optics expert... Still, I think you're bumping up against the limits of physics here.

Adam

[norty303]

Ok, poor example with the fibre, but theoretically we're talking the same thing.
I feel a little more enlightened today!

[badger1666]

Exactly!
Well, here is the problem: If you want to perfectly overlay two beams, you need some way to differentiate them so you can reflect one and pass the other. One way is with a dichro, since it's reflection properties are frequency-specific. Thus you can mix red with green... But if the two beams are of the exact same frequency, then you need some other way to differentiate them. The only thing left is their polarization angle.

I'm not aware of any means of combining two beams of the same frequency and polarization that results in a perfect overlap like you get with a PBS or a dichro. Can't even see how fiber would help, though admittedly I'm no fiber optics expert... Still, I think you're bumping up against the limits of physics here.

Adam

i know you can overlay two beams for the reasons you said, but if you could as far as i am aware you still would not get double power surely two beams at the same frequency
one would cancel the other out ?? the same as it does with audio, thats what i have been led to belive, or am i just reading your post wrong
or was i told wrong ?

[buffo]

surely two beams at the same frequency
one would cancel the other out ?? the same as it does with audio,

You are thinking about destructive interference. This is when one wave is exactly out-of-phase with another, so that as they combine they cancel each other out perfectly.

This is possible (with great difficulty) to achieve with audio. In theory, it would also be possible to do the same with light, but you have to understand the tolerances we're dealing with are *much* tighter. Audio wavelengths are measured in inches, or feet even. But with light we're talking about a wavelength of a few hundred nanometers. And remember, the two waves need to be *perfectly* aligned in all three axis in order for the destructive interference to work. We're talking about nanometer tolerances here, over all three axis. Not an easy thing to do!

Then too, remember that they'd have to be spacially coherent as well. (Well, actually anti-coherent, because they need to be exactly 180 degrees out of phase.) And finally, due to the relatively short coherence length of the diode's output, even if you managed to get the destructive interference to work, it would only be effective over a very short distance - probably no more than a few inches. And this is under the very best case in a carefully controlled laboratory.

Bottom line: There is far too much variation for destructive interference to ever have any measurable effect on combined beams at the wavelengths we're working with. It's not a problem with the theory, just the fact that in the real world you'd never be able to align the beams to that level of precision, nor be able to keep them that way long enough to get it to work.

Adam

[badger1666]

thanks ,i now understand this better