Talk:TransWarp GS

Please post your support questions here and ReActiveMicro or someone from the Community will be glad to assist you!

User: Step 3 summary contains "replace 10uF electrolytic capacitors (four units)" but there is no further mention of this (why/where). I've never replaced mine. I suggest that this part of Step 3 is split out into a later step and given further elaboration. Henry: Agreed. A past user was going to help with this section and never did. Basically the caps are C35, C36, C37, C38. One is on the end of the board closest to the IIgs front. One is one the lower edge of the board by the Cache Card connector. And two are on the end of the board closest to the IIgs rear, one near the edge connector, and one near the oscillator.

User: Can I mention competitors products? The ByteBoosters 32k cache is not recommended from my testing with stock PSU. As I understand it, they did their testing with 5.25V. It doesn't work for me at all in a ROM3 and the top cpu speed I can achieve with it in a ROM 1 is less than the other 32k options (including a home made 8->32 upgrade). Henry: Sure, feel free to mention other vendors. And of course feel free to edit the Wiki page anyway you see fit. The idea is users like yourself can easily add notes, share experiences, correct things, and even create pages to keep notes or share projects and ideas.

Transwarp GS and GALs

I came across this in an ebay.com listing for a transwarp gs:

"Here is some interesting trivia: Once upon a time (before 1992), the Western Design Center could not make 65C816 chips that reliably ran above a speed of 7MHz. In an effort to make faster chips, the company produced models designated as "engineering chips" (identified by part number W65C816PL-ENG). These chips varied in their top speed and tended to go no more than 9MHz to 10MHz reliably. They were really not much better than the production parts. It was discovered that chips based on this older design had difficulties executing 16-bit instructions above a certain speed. The TransWarp GS's GALs were designed to intercept these instructions and add extra cycles to slow down execution and overcome the processor's limitations. Apparently, when Western Design Center came out with their engineering chips, hitting 10MHz produced even more 16-bit instructions that had trouble executing. Thus, for a time Applied Engineering offered a replacement set of GALs to allow the TransWarp GS to reach high speeds with those chips.

Applied Engineering is now gone, so what do users do about the GALs? Nothing.  Since the time of the engineering chips, the Western Design Center had Sanyo completely redesign the 65C816 (chips of the new design all have an "S" in their names, incidentally).  The new chips properly reach speeds of 16MHz (according to the technical specifications on WDC's world-wide web page).  16-bit and 8-bit instructions fail simultaneously if the chip is overclocked.  This means that the GALs are no longer necessary.  A stock TransWarp GS with ROM version 1.5 or greater can now be upgraded to 10MHz-15MHz speeds without replacement GALs.

Does this make designing a V.2 version easier?

Cheers, Andy

[HENRY-RM] It's good to know this info, and thanks for posting it. Pretty sure however it won't impact v2 much or at all. I'm using the latest GALs. Even testing an 832 core for the replacement of the CPU. The 'hard' part is trying to move away from the old Xilinx array. The software to help decode the bitstreem is nebulous at best. And even when I do decode a path it's hard to test that finding as the array usually is doing combinational logic from all inputs. So I can only test most or all, or none at a time. Which is slowing things down a lot. And if I guess, then it means that needs to be fully retested to ensure it's a correct guess before I can move on. Lots of lab and bench time, but getting closer each month to the goal.