Update README.md with findings on GAL stability

With the current JED files it should be fine to use all Atmel files
in an A4091. YMMV, but even Chris' picky Amiga 3000D was happy, and
that is quite something.

README.md

@@ -39,10 +39,23 @@ The original chips use the following timings
 During the bring-up of the first ReA4091 prototype we used all ATF22V10C-10
 chips. This lead to a non functional ReA4091. Also, using a mix of ATF22V10C-10
 and ATF22V10C-15 as described above did not yield the desired result.
-Empirically we have found out that there are subtle differences in pin-to-pin
+Atmel chips have pin-keeper circuits that can influence the signal quality,
+particularly if there are a lot of inputs hanging off an output signal, like
+in the case of DOE and MTCR. There are also subtle differences in pin-to-pin
 propagation delay (and possibly other differences).
 
-In our tests we have found that the card works most stable if you use a
-Lattice 22V10-6 or 22V10-7 for U306 and a Lattice 22V10-15 for U205.
-All the remaining chips may be Atmel ATF22V10-10.
+### Using the original logic files from the Dave Haynie Archive
+
+If you want to use the original files, you will have to use two Lattice chips
+in combination with 6 Atmel chips. In our tests we have found that the card works
+most stable if you use a Lattice 22V10-6 or 22V10-7 for U306 and a Lattice 22V10-15
+for U205. All the remaining chips should be Atmel ATF22V10-10.
+
+### Using the latest version of logic files
+
+When using the latest versions of the logic files in this repository, you
+should be fine when using all Atmel ATF22V10-10 parts.
+
+There will also be a speedup of approximately 14% over the original GAL set.
+