Thursday 23 May 2013

Testing 2 - Fixing the IQ with machine polishing

The image quality is poor due to the cumulative errors caused by inaccurate surface polishing. The easiest way of fixing this is to remove the human factor from within the polishing process.

Current Process

Right now I use a variable speed small lathe to spin the elements, which are either secured directly or though a collet into the machines chuck. As they spin I use a range of papers to gradually smooth the surface down. At the start there are usually a few larger scratches. The most difficult to remove are at the extremity of the elements profile. Once the surface quality is even I use the next grade paper. The grades I use are P1000, P6000, and P12000. It is essential that I use lots of spray lubricant while polishing to ensure the debris is taken away. I attempt to keep pressure even across the surface of the lens, and move my hand back and forth slowly as it spins. Once finished with the papers, I often check with a super bright LED light to see if I have missed any scratches. If I have I go back to the appropriate grade paper and continue from there. After this I take a damn cloth and use UNIPOL polishing paste and rub it on, and then off again with a dry cloth. By the point the lens is usually suitably polished. Some convex lenses and occasionally some shallow concave lenses can be taken and further polished on the buffing wheel with acrylic buffing soap. This is usually the end of the process.

There are lots of areas for errors to occur here. Firstly, to remove specific scratches, it is tempting to put pressure only on that area instead of moving the pressure across the entire face of the lens. If you hold pressure in one spot as the element spins it causes the surface of the lens to become uneven, resulting in distortion with in the final image. This is what can be seen in some of the close up images of the lens elements during manufacturing. A second problem is the centre spot. Material travels a shorter distance closer to the centre, as if spinning faster than the outside. Therefore it is natural for any polishing papers or cloths to pull away from the centre and not polish it properly due to the speed up. Also at the centre, you encounter material spinning both down on one side and up on the other, adding to the resistance against the polisher. Thirdly, it is absolutely the most boring thing to do ever. Standing over a lathe polishing lens after lens is boring and you very quickly lower your standards of what you consider a finished lens to be. I know for a fact I could have polished some of the lens elements to a higher standard than I did, however life is to short and I am sure I would have been driven insane had I have carried on.

New Process

The easiest way to fix all of these errors is to pass the process over to a machine. If the CNC lathe can run a tool in exactly the right path to cut the optics profile, then why can it not do the same but with a polishing tool substituted for the cutting tool?

The idea is to make a custom tool for the Boxford CNC lathe onto which polishing pads to be placed on the end. This tool can then run on a repeated path identical to that of the cutting path, providing even abrasing of the surface with whichever polishing pad is installed. Given that I program the cutting paths for the concave faces myself, I would not be difficult to program in a repeat loop to allow the polishing tool to run across the entire face of the element over and over and over again until the desired finish is achieved.

The biggest advantage of this approach is that the polishing is guaranteed to be even. There will be no risk of the surface distorting, and the polishing can be run constantly until the surface has the desired finish. In this scenario, an important thing to note is that the polishing process must immediately follow the cutting process. If the optic is removed and replaced back into the machine there is no guarantee that the surface will be in exactly the same place, so it could result in uneven polishing. Therefore it needs to be an all in one process with the acrylic billet going in and the finished surface coming out. The only time a human will interact with it is to take the optic out and reverse it for the rear to be cut and polished, and then to take it out and add it to the assembly. This is how I would expect it to be done professionally, as it minimises the chance of any human error.