[rescue] sun 3/60 update

[Dan Moisa]

Great analysis!

The SRAM consolidation is definitely on the menu. The bipolar PROM can be
replaced with a few things including a GAL:
http://www.retroclinic.com/leopardcats/galprom/galprom.htm
We can of course go overkill and put in a flash part, and tbh even a modern
MCU bit banging the output - see below for some examples. Here's another
one for the boot prom:
https://github.com/wickerwaka/PicoROM
I'm trying to convince this guy to make a 28 pin version, though likely
we'll just bump up the implementation on our side before that happens.

TRANSF is a PE_64102 3-pair 1:1 isolation transformer and might be
skippable in a pinch. I'm still debugging gremlins in the Ethernet section.

The Zilog SCC can likely be swapped with an Atmel. It's*extremely*
underutilized as far as I can tell. Not so for the network, but there it's
part of QEMU right now, and might fit again in a MCU that already has an
Ethernet impl, or FPGA/etc.

I actually had weird experience with the tl7705 voltage supervisor from
Digikey - it simply didn't work per spec -neither of the 10 parts in the
lot did. I actually made an ATTiny drop-in replacement that just waits for
800ms for the inrush to stabilize. Eventually got a smd part on adapter
that actually works as expected. Still don't know what happened but another
reason I was glad I kept fidelity with and could experiment on an original.

On Wed, May 28, 2025, 06:28 Romain Dolbeau <romain at dolbeau.org> wrote:

> Le jeu. 22 mai 2025 à 08:28, Dan Moisa via rescue <rescue at sunhelp.org> a
> écrit :
> > Putting the BOM together is a PITA because about 40% of the components
> are not on Digikey/Mouser
>
> Just checked the 74xx chips for stock or alternative at Mouser, and
> there's hope:
> * The ALS are in stock in SOIC, all but the ALS652 in DIP (and for
> that one, the FCT652 is a cheaper option that might work as a
> substitute)
> * The F00, F04, F175 are all there, the F374 in SOIC only, the F534 is
> MIA but there's an ABT that might work as a substitute; also it's just
> a register for BERR, and the 3/160 uses a different part (slower,
> non-inverting), so there's options to substiture/reengineer
> * The AS are mostly not in stock/no made or expensive; however it's
> possible they can be substituted by F, which are cheaper and usually
> available. The exception is the AS286, but the FCT11286 has similar
> functionality though different pinouts. Also, it's for the parity
> generation/checking, required for real workstation/server usage but
> could be disabled in a replica if some components are no longer
> available.
>
> On the SRAM side, updating the vintage stuff will be wasteful but
> doable; pair of 7188 (u200/u201) could be replaced by IS62C1024AL-35
> (8x the needed size...) while 2168 could be replaced by IS62C256AL-25
> (idem), though while u205/6 and u207/8 can be merged pair-wise,
> u202/u203 have different /WE so need different chips (8-bits SRAM dont
> have 4-bits granularity).
>
> The ICM7170, Am7990 (lance), AM7992, Z8530 I don't see a way to
> replace (unless by re-implementing in programmable logic, or having a
> go at recreating them with open-source EDA & PDK perhaps as FOSH, but
> that's not happening anytime soon).
>
> The Am25LS2518 is just a dual-output register, it might be possible to
> replace it by two registers in parallel with the same input and
> control, one for the always-on output and one for the three-state
> output.
>
> I'm not sure what "TRANSF" is ?
>
> (the PALs are a problem of their own, discussed in the GitHub already).
>
> Cordially,
>
> --
> Romain Dolbeau
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sunhelp.org/pipermail/rescue_sunhelp.org/attachments/20250528/a8ccd99b/attachment.html>