Re: [cosmacelf] 1802 Lithography
- On 06/07/2013 02:24 AM, Lee Hart wrote:
> The transistors that drive the output pins have to be bigger so theyIt is even worse than that.
> actually have useful amounts of drive. So they would stay about the same
The drive capability of the transistors drops as they get smaller but
the parasitic capacitance in these output transistors stays the same.
The textbook cure for this is a series of transistors with a ratio of
sizes equal to e. (base of the natural logarithms)
The 1802 has a set of about 10 transistors in parallel in the output
drivers. This isn't too horribly far off from e that you need to worry
about it but if you dropped the internal transistor size down to 20nm
then you would have a big slow problem. So add some space for ~5
(ln(6u/20nm) - 1) stages of output pre-drivers.
> But even when you take all these into account, you would still wind upWhat seems more likely is that the pad frame would drive the size with
> with a chip that is quite a bit smaller, faster, and lower power! The
> chip may be physically too small to easily fabricate; so you would add
> more "stuff" on the chip, like memory or peripheral, just so the
> resulting chip isn't too small to handle.
the 1802 being a parasitic core. 40 I/O pins with ESD protection,
buffers, and drivers require a fair amount of space. A quick search
turns up a 0.5u 40 pin MOSIS Tinychip that is 1.5mm square with a 0.9mm
square area left for the device core.
David W. Schultz
Returned for Regrooving