This is really heart crushing you know. A chip company can put these out in a matter of weeks considering their simplicity. Intel et al can already put billions of transistors on a few cm^2. Let alone diodes that are simpler and in this case pretty much straightforward. A CPU or GPU is a complete package to run a PC but this is just putting diode after diode which requires no brain work so to speak.

Instead we have to resort to making our own chip foundry at home because they are too stubborn to handle the request.

I agree!  It would just be nice to at least see a quote, I mean, even if the quote is highly inflated just to get rid of me, lol.  They could at least say, "Oh, it would cost you 2 million dollars to make that chip."

The good news is that a home brew garage project semiconductor fabrication setup is not too difficult or expensive. I was talking to a guy who made diodes and LED's while in college with home brew equipment his professor made. They used 1e-4 torr vacuum, which is very easy and inexpensive to achieve. To make a chip that could power a small light doesn't need to be 30nm technology. Even 10um ancient technology could achieve that much power.

Don't get me wrong, I'm not trying to suggest that if we slap together a diode array chip that it's going to produce usable amounts of power. This is uncharted territory. I can only offer the conventional semiconductor mathematics, in addition to my present diode arrays that use over the counter diodes.

What's nice about making our own equipment to make diode array chips is that we can make a *small* chip, test it, make another small chip with a slight adjustment, test, and compare. We could learn what diode parameters work best before making a large chip.

Also, it would be great to be able to make your own chips, any kind of chip. To say the least, you could easily make your own solar cells, which would be easier to make than a single small diode. 


PL

just looked on Ebay, you can get 700   1N4007 diodes for $15

that comes to about 2 cents each.   im sure there are other places to get them in bulk.
cut the legs short and solder them into rods of 20-30 diodes.
then connect them all in series to get the voltage you desire.
then duplicate the modules and connect them in parallel to attempt to draw usable power.

700-1k  shouldn't be much bigger than a large textbook,     and if you could even get it to light a tiny LED that would be enough to draw interest in the chip making industry

Thanks for info. Here's the mathematical evaluation:

There are only a few over the counter diodes that I could recommend because most diodes have too much capacitance and to high resistance. The 1N4007 has over 60 times more total capacitance than the SMS7630, and 2000 times more low signal resistance. We want diodes with the lowest capacitance and resistance. Also the 1N4007 has extremely high ideality coefficient, which is not good.

In short, this means each 1N4007 diode would produce 21 million times less DC power than the SMS7630. Each SMS7630 diode is 0.47/2 = 23.5 cents each.

I used the 1N4007 model from:
http://www.electronicspoint.com/ibis-models-wanted-1n4007-and-icm7556-t24976.html

As far as I'm aware, the SMS7630 is the most cost effective over the counter diode for this.


A few things I could promise with nearly 100% guarantee. A diode array built with 1N4007 is guaranteed to fail. It would require far too many diodes to produce sufficient DC voltage, *but* a 1N4007 diode array total resistance would be too high for the voltage meter.


PL

I have been told that the Japanese did some work like this about ten years ago using cable wire. I am waiting for more information on that and I will post it as soon as I get it.

I am also wondering what the outcome would be if triodes were used in the place of diodes. Any thoughts on this?