Originally Posted by Riempie
So what's different with the aca 1232 compared to the others?
The CPUs are used, and they're mostly scratched. There's hardly a chip that doesn't have a dent or scratch. The 030 processors are tested @40MHz before they are mounted on the 33MHz board. All 020 processors are tested @25MHz before they are mounted on the accelerators.
If you take a closer look at the card, you'll find *lots* or larger ceramic capacitors. This is accounting for high ripple on old power supplies. These 1206-size caps are 10µF (yep, the expensive stuff). Also, the PLL that generates the fast logic/memory clock has a separate filter for it's power.
Apart from that, the logic is pretty much identical to the 1231 logic. Memory timing has minor tweaks, but these are accounting for the differencies between Winbond and Etron. No difference in access speed. If Winbond hadn't increased their prices so much, I'd have used them on the ACA1220/1232 as well. This way, I found a way to push the Etron chips to their limits without affecting stability.
I'm still "officially a fan" of the 200MHz Winbond SD-Rams, but their performance is not required on these accelerators. I only buy them for the Chameleon, where the Minimig core makes use of the low first-access-penalty.