On 8/28/06 10:30 PM, "graywolf", <[EMAIL PROTECTED]> wrote:
> Are you talking about a Step and Repeat Machine? * > > That takes a photograph of the IC artwork, and reduces it in size, then > prints 4 copies of the artwork 1/2 size. Then it does it again and you > get 16 copies 1/4 size, again and you have 64 copies 1/8 size, etc. When > you get the IC's down to the desired size you use the image to photo > etch the IC's onto the wafer. I can see how there may be a minimum size > limit, but not how there can be a maximum size limit. After all the > original image is many times larger than the wafer. Of course there may > well be a size limit on the automatic cutters that cut the wafers into > individual IC's. > > *My knowledge of this is ancient, they undoubtedly use digital imaging > now, but the principle should be the same. I could do some research on > modern IC production methods. But why? No one is going to hire me to > design IC's for them <grin>. Hi Graywolf, Several people posted on this subject, and please let me take the liberty of lumping them together and respond with what poor knowledge I have. Like you said, no one is going to hire me on this matter :-). I saw the article a year or so ago, explaining how the current APS size became the de facto standard, and it was NOT performance related, and thought interesting. I do not even remember if it was in Japanese or English. So, I hastily googled, trying to find out the site which might be talking about specifically about this subject (not about the general IC making etc). But so far, I do not have much luck (strange). I googled "stepper" to see if there is any article talking about the origin of APS sensor size, but the sites explaining the stepper machine itself is relatively few. Anyway, here is what I know. Yes, the stepper is indeed the Step and Repeat Machine for the photolithographing the minute circuit detail on the wafer with photoresist, which by itself is a well known process. But today's stepper became so sophisiticated and cost tens of millions of dollars per unit. I think Nikon, that currently dominate the market sell a few thousand units a year of various sizes as I understand it. It projects the circuit detail through various reticles (circuit pattern) and then a huge "reduction glass" to "demagnify" (20cm dia and often 1m length etc). After each shot, the wafer is moved minutely and repeat this step, thus step and repeat. I am sure many people know about this process. The reduction glass obviously requires an extremely high resolution with no distortion and other imaging defects. Now getting to the point, and this is where I only guess as I could not find a "dead-on" explanation (my disclaimer :-). People in the know should chime in. There is a huge size difference between the IC chips and optical sensor, and the reduction glass must have certain limited range regarding the minimum size (for IC chips) and the maximum size (for optical sensor). I think the sensor mfg probably require the largest projection image from a given stepper. This is probably the limiting factor of the size of the sensor which can be produced in "one shot" projection process as in IC chip making. To manufacture a larger sensor, the process has to be repeated and it is said that the FF size sensor requires 3 shots to achieve the required photo etching. Then they have to precisely "stitch" (probably the wrong term) them together to make into a single FF sensor. This obviously requires a lot of redundant steps/process and smaller yield (sometimes zero yield?) and when translated into a commercial production, it is going to cost so much. But above part is only my guess but should not be too far from the truth. The only thing I thought I knew was that the APS-H size sensor was derived from the stepper driven limitations, but NOT by its optimum performance as a product etc, as some people pointed it out recently. I too thought that the APS sensor size was based on the performance-driven considerations but when I read the article (which I am still searching), I remember I was struck to find out that it was actually hardware-driven size. But I could very well be wrong. If and when I found the 'dead-on" article (this is not any sort of engineering secret etc, and there must be tons of article about it. Only if I have time...), I will of course post the link. Cheers, Ken -- PDML Pentax-Discuss Mail List [email protected] http://pdml.net/mailman/listinfo/pdml_pdml.net
