<div dir="ltr">Back from a camping trip, lemme go through the queue here:<div><br></div><div>Romain: yes on the cas-before-ras and the cycle time, checked out with the scope at 78khz. I like your analysis, I'll make a "design" folder and start adding some of this stuff there, so I can itemize what changes we could try out. I did get a bunch of kit for the next step - 72 pin sockets, some blank 30-pin simm boards, and so on. I'll wire up an adapter with the 30 pin blanks first, and if that fails miserably then I'll partially fill in another board and patch in the larger socket. The chip clearance on the top means if I need to reroute signals on that side, I'd have to desolder the sockets anyway in the populated version.</div><div><br></div><div>It would be good to give names to the U114 state machine too, specifically C.S3</div><div>What's the formula for the refresh cycle, is it size dependent? with internal counter refresh, I would expect that as long as you get to all locations within the RAM spec, it wouldn't matter. I'll try putting in 4MB sticks and see if they still act like reliable 1MB ones.</div><div><br></div><div>I'll take a look at your CPLD pull request. If all the pins are there, I can put it on a TQFP to through-hole adapter and wire it up to some sockets to mimic the original chips, and we can test it that way. I should be able to solder it, thank you Dave for the offer. If we do start doing surface-mount redesign wholesale, then yeah we'll want some more professional service.</div><div><br></div><div>I saw the ethernet variants, I'll add it to the queue. I got a bunch of AM79C98PC and I might try a fully-on-board TPE solution first, on the "stay as original as possible" branch.</div><div><br></div><div>Dan.</div><div><br></div><div>PS: the negative logic *is* incredibly annoying</div></div><br><div class="gmail_quote"><div dir="ltr" class="gmail_attr">On Sun, Jun 1, 2025 at 9:25 AM Dave McGuire via rescue <<a href="mailto:rescue@sunhelp.org" target="_blank">rescue@sunhelp.org</a>> wrote:<br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">On 6/1/25 12:05, Romain Dolbeau via rescue wrote:<br>
> All three of the MMU (U23[012]) and FCDECODE (U102) (they all need<br>
> upper memory address bits, so it factorizes some inputs) fit easily in<br>
> an ATF1504 in a 100-pins package. Probably need to be soldered by a<br>
> professional house with tha massive TQFP package, JLCPCB worked for me<br>
> (mostly) with ATF1502 for my variant of the SEthernet [1], the<br>
> IisiSEthernet.<br>
<br>
I can take care of that soldering for anyone who needs help with it.<br>
<br>
-Dave<br>
<br>
-- <br>
Dave McGuire, AK4HZ<br>
New Kensington, PA<br>
<br>
_______________________________________________<br>
rescue list - <a href="http://sunhelp.org/mailman/listinfo/rescue_sunhelp.org" rel="noreferrer" target="_blank">http://sunhelp.org/mailman/listinfo/rescue_sunhelp.org</a><br>
</blockquote></div>