Results: Collimating with cylindrical lens instead of Asphere

[Jubathoph]

I have been experimenting with different setups to reduce divergence on multimode diodes. One idea was to just use 1 cylindrical lens for each axis (rotated 90 degrees) and eliminate the aspheric collimation lens completely. Doing this allows you to use an extremely long focal length cylindrical lens for the slow axis (i.e. the problem axis for MM diodes). This is simpler than the "traditional" way of improving divergence; collimating with a short FL asphere, then expanding with a concave/convex cylindrical lens pair that operate on the SAME axis.

I machined up a setup that allows me to swap optical elements and diodes, and compare back to back. I plan to use this to test out some ideas before implementing in my builds. It consists of a track with T-nuts, to which I bolt L-brackets that I can swap different optical elements or diodes, and can move around as needed, and rotate around Z and Y axes. This is going to come in handy when I test different diodes to fine-tune optical setup for my eventual RGB build:



For this experiment I used a Ushio 1.2W red diode. This has very bad divergence, the worst I have ever used, similar to the high power 445. Using the G8 lens, this gives a spot length of 4.75" at 36 feet.

G8 setup & spot size at 36':





For the cylindrical setup I used a 9.7mm focal length cylindrical lens for the fast axis (adjacent to diode), and a long 25mm focal length lens for the slow axis. Doing this I was able to get a spot size of 0.930" at 36'. That is over 5x improvement vs the G8. I expect proportionately more improvement could be had with replacing the 25mm FL lens with a 30 or 40mm..



The width of the spot is driven by the FL of the "fast axis" lens, which is 9.7mm in this case. That is somewhat higher than the G8 which is why it is about twice as thick. The Height of the spot is driven by the FL of the Slow axis lens, which is 25mm.

For power I had measured 860mw for the G8 and 850mw with the cylindrical setup. The tradoffs here is a larger beam at aperture, and more "noise" around the spot (probably due to some spherical aberration). I think this setup would be pretty awesome on a high-power 445 where the improvement in divergence is much needed, and IMO a slightly fat beam at that power is not a problem. The noise can be cleaned up with an aperture window. The major advantage with this type of approach is you can choose any focal length lens for either axis, independently. It is also much simpler and less losses than you would get with the 3 lens approach (1 asphere, then 2 cylindrical to expand PCV/PCX).

Cool thing about this is I can machine this all into a host with Inside diameter of 0.75"; too many projects not enough time.

 

[RedCowboy]

I have thought about doing this, it's nice to see it in practice, so how wide is your beam leaving your 2nd lens ?

 

[Jubathoph]

Just measured it, it was .365" at aperture, .930 at 36', calcs to .67mrad if I did that right.

 

[kecked]

Put a pinhole inbetween the lens to form a spatial filter. That will clean up the beam a lot In terms of side splash.

 

[RA_pierce]

Really nice work.
You can get a less elliptical beam (i.e. more circular/square) by matching the focal length ratio of your lenses to the divergence angle(s) of the diode's fast and slow axes.
I built this system into a laser pen back in 2018 (with some aberrations due to space limitations) and have version 2 in the works right now... planning to post about it next month.

 

[Jubathoph]

You can get a less elliptical beam (i.e. more circular/square) by matching the focal length ratio of your lenses to the divergence angle(s) of the diode's fast and slow axes.

That's right, but only at a certain distance; if you match the size on fast & slow axes at aperture you would still have different divergences & the beam will eventually turn into a long rectangle.

 

[kecked]

The only lasers with good beams are dye, gas, opsl, and some dpss lasers. Yes I am sure I missed a few others. Diodes are just not great quality lasers. Single mode lasers are better but never as good as the others but they are cheap, in many wavelengths, available in power, and for most purposes just fine. Unless you want to pay a lot more money or have a small space heater what you are achieving is about as good as it gets.

 

[Singlemode Laser]

That's right, but only at a certain distance; if you match the size on fast & slow axes at aperture you would still have different divergences & the beam will eventually turn into a long rectangle.

This is why cyl lenses are used for astigmatism correction. Same divergence for every distance

Singlemode

 

[Jubathoph]

Update, I did a back-to-back comparing to a long FL asphere lens (Thorlabs A240, 12mm diameter & 8mm FL), and 6x cylinder lenses operating on the slow axis. This is similar to the tried and true approach for correction with cylindrical lenses, and ultimately is superior in that there is much less "noise" in the area surrounding the central spot. The amount of power in the "noise" part was significant (~20%) as I figured out when I moved the LPM back several yards to measure only the power in the central beam. I think this is due to the fact that spherical aberration is less of an issue when the beam is already collimated.

So bottom line, I do not recommend using just normal Cylindrical lenses for collimation. In the future I'm going to try out some diodes with Fast Axis correction & extra long focal length lenses. If people are interested in different diode/lens/correction scenarios, let me know. The modular "test setup" is very useful for back to back comparisons.

 

[RedCowboy]

When working with these MM blue laser diodes such as nubm44, I prefer using a short FL asphere ( G2 ) followed by a 6X c-lens pair to deal with the aggressive axis then a beam expander but I have also had good results with a FAC modified laser diode and a single longer FL asphere, the difference is the near field beam width when not using the beam expander, so for a laser show application the asphere and c-lens pair would be better, but for getting a tight beam waist in the far field the FAC diode sure is simple, I did a basic test here of what's essentially a FAC enhanced nubm06 with a window can.