How to create super bright laser beams

[planters]

This is the PM I recently sent out.



The approach I have toyed with is called wavelength division multiplexing. You have to get comfortable with temperature controlling the diodes and with hard cooling to allow as great a temperature range as possible. You test each diode at R.T., say 25C and measure the peak nm reading. I have a spectrometer for this measurement. You then set up a series of diodes and spread the temperature of each by lets say 5-10 degrees. You may even heat the warmest. You also take advantage of the native wavelength spread you measured earlier and then you knife edge the series of beams into a prism and as the divergence of the beams within the prism varies with wavelength, the combined beam has the dimensions of a single diode beam. This will probably work better for single mode diodes because they tend to have a narrower wavelength emission. The only limit in the number of "stacked" beams is the band-width of each diode vs the interval between each diode. They obviously have to be distinct.

Eric

In addition to this advantage the resulting beam will be quite broad band say 620 to 650 nm. This will give more color saturation to, in this case, the red beam. I do not believe there will be much advantage to individually modulating each diode to say get 642nm vs say 632 nm, but who knows. Even just three diodes co-aligned, combined with another doubling with a PBS as well would mean over 1W with the same beam specs as a single 170 nm (not over driven) diode.

[planters]

As I have now been running this over in my mind as well as the "envelope" it occurs to me that if you use a very short FL collimator the exit beams from the diode mounts will be very tiny and so when they are knife edged their convergence angle into the combining prism can be made tiny as well and this reduces the requirement for wavelength deviation and so, the requirement for temperature spread.

[flecom]

This is the PM I recently sent out.



The approach I have toyed with is called wavelength division multiplexing. You have to get comfortable with temperature controlling the diodes and with hard cooling to allow as great a temperature range as possible. You test each diode at R.T., say 25C and measure the peak nm reading. I have a spectrometer for this measurement. You then set up a series of diodes and spread the temperature of each by lets say 5-10 degrees. You may even heat the warmest. You also take advantage of the native wavelength spread you measured earlier and then you knife edge the series of beams into a prism and as the divergence of the beams within the prism varies with wavelength, the combined beam has the dimensions of a single diode beam. This will probably work better for single mode diodes because they tend to have a narrower wavelength emission. The only limit in the number of "stacked" beams is the band-width of each diode vs the interval between each diode. They obviously have to be distinct.

Eric

In addition to this advantage the resulting beam will be quite broad band say 620 to 650 nm. This will give more color saturation to, in this case, the red beam. I do not believe there will be much advantage to individually modulating each diode to say get 642nm vs say 632 nm, but who knows. Even just three diodes co-aligned, combined with another doubling with a PBS as well would mean over 1W with the same beam specs as a single 170 nm (not over driven) diode.

DWM would require extremely accurate temperature control, normally dwm and it's variants (DWDM) etc use different wavelength lasers but your theory makes sense to me, if it's practical to implement is a different story

[jors]

Sorry, I'm lost here
because prism refraction varies vs wavelenght...are you proposing to enter beams on prism at slightly different angles/diode?
id, for example: 620nm diode at 'x' angle, 625nm diode, at "y" angle etc...
If so, there are only few nm between diodes, so angle to enter prism could be very very narrow and critical, maybe impossible to achieve mechanically, isn't it?

[andy_con]

Eric is bang on the money for this one.



Put a single white beam into a prism and out the other side you get the full colour spectrum. So turn that around and put different frequencies of light into a prism and it acts as a dichro and you get a single white beam out, simple. Telecommunications companies use this method i believe with RGB light and some company has patented and copy righted etc...

Were not dealing with a laser company here were dealing with a big multi billion pound company, just bare that in mind.

So using the above principle you could add different frequencies of red light into a prism or other unnamed optic and get a single red beam out. The prism would act as a dichro so each beam would be perfectly over lapped even at aperture, its the holy grail of diode combining and would produce a more powerful and better beamed red than an opsl 5w red.

Think of it like this, if you could get 10 red diodes @ 170mw each with a 1nm difference then you could combine them using the above method into a 1mm by 1.5mm beam @ 0.5mR ( or whatever spec a 2mm fl lens gives on a single mode red – cant remember off the top of my head).

So thats 1.7w in a 1mm beam
Now stack 6 sets together so you have 3 wide and 2 high that would give a beam of 3mm by 3mm @ 0.5mR at a power of 10W in a tiny beam. Then cube another 6 in and you have 20w of red in a 3mm by 3mm beam at 0.5mR, beat that coherent!

It all sounds so easy and perfect ... yeah sadly its a right cunt to actually get to work.

Each diode would need to be temperature controlled to probably 0.1 degree

You can only set it up using a spectrometer, from testing every diode gives a slightly different frequency. At the same current and same temp diodes can be as far as 3nm different. you have to adjust the temp of each diode until the required frequency is reached, the spectrometer would need to be accurate to around 0.2nm to make it work properly.

The most nm difference i managed to get out of a diode with a temp range of 0-40 degrees was 8nm, so that means you could only combine 8 diodes into the prism.

The further the distance between the diodes and the prism the easier it is to get it to work. Im sure someone could do some sums and work out the distance and angle each diode would need to be. But the closer to the prism the smaller the angle, the further away the bigger the angle.

If you have 8 diodes on an aluminium base plate, at one end a diode is cooled to 0 degrees then 150mm away the furthest diode is heated to 40 degrees, with some more diodes in the middle at different temps. Managing the temp of the base plate is tricky. Not looked into this in the slightest.

When you first turn the unit on each diode is at the same temp, room temp, so out the prism would be lots of little dots. You would need a way to turn the diodes on once each has reached is pre set temp. So you couldn’t turn the a projector on and start using it straight away.


Ok now for the fun part some real world testing

3 diodes, 2x 110mw 683nm diodes tec’ed and a 660nm diode un tec’ed – all single mode. One 638 tec’ed to 0 degrees to other to about 20 degrees, no idea on what wavelength they were giving out as i don’t have a spectrometer.
All 3 diodes had 2mm fl lens on them



As you can see beams all tight together



The combining optic isn’t important at the moment, but i was able to get all 3 beams perfectly over lapped into a 1mm beam




The really fun part is dicking around with the temp of one diode and watching it slowly drift across the wall, kept me amused for hours.


Interesting shit, i just need a really really accurate spectrometer to complete the project now 

[planters]

Flecom,

That's what is neat. It wouldn't. The precision necessary is only proportional to the density of the stacking. For example, if a 1mm beam converges on a prism, 200mm away and it's neighbor does so as well then that means that the correction in the prism is 5mrad/ delta lambda and that is proportional to 5mrad/ delta temperature. If the step temperature is 10C then 1C temperature control equals 1/2mrad deviation. 0.1C is easy especially if all the diodes are modulated as a single unit.

[planters]

My spectrometer is that precise, if not that accurate, but it's the precision that matters as we are only trying to get a relative wavelength. All you have to do once they are ordered according to wavelength is to place the TECs for each into a voltage/current, computer mounted multi-output card and display program with sliders for each diode. My son eats and breaths this stuff. I also agree that 8 stacked beams is a reasonable guess, but this all depends on the typical diode bandwidth.

[andy_con]

ill take your word for it as ive no idea, I don't even own a spectrometer.

but providing each wavelength is different then it should work. I was working at a distance between diode and prism of about 400mm

[planters]

I'd have to get a single mode diode and test the bandwidth. It would be interesting to compare this with a multi-mode while I'm at it. I would have to confirm the wavelength shift with temperature as well. Once those numbers were known then the practical limits could be estimated.

Andy,
I think your proposed mirror mounts as knife edge prism mounts would be perfect as they could be equipped with extended screws to allow adjustment through the top of the enclosure that houses the diodes.

This is exciting enough that I will pursue it even if it interrupts my other project.

[jors]

So are you entering beams knife edged & paralleled to prism? and just tweaking temp on diodes to slightly change refraction id output angle?
I thought that each beam (@ different wavelengths) should enter prism at different angles...