rasz_pl wrote on 2024-03-04, 02:29:
mkarcher wrote on 2024-03-03, 11:22:
Someone might have a schematic of a mainboard using the SiS 85C407 at hand. If you are lucky, that schematic includes pin names and numbers.
yes 😀 File doesnt compress because diagram is embedded as a picture in the pdf 🙁 so link instead of uploading https://cdn.hackaday.io/files/183453784885315 … ual-issue-D.pdf
Thank you very much for finding and linking the schematics. They prove stuff wrong I said before.
rasz_pl wrote on 2024-03-04, 02:29:"SCPC486 STEbus 486DX PC AT Compatible Board Technical Manual" lists it as :
"CHIPset SIS 85C471 Green PC ISA-VESA Single Chip R […]
Show full quote
mkarcher wrote on 2024-03-03, 11:22:
The chipset encyclopedia mentions that the 85c407 might be an "I/O chip".
I expect the 85c407 to be an improved replacement for the ubiquitious 82c206
"SCPC486 STEbus 486DX PC AT Compatible Board Technical Manual" lists it as :
"CHIPset SIS 85C471 Green PC ISA-VESA Single Chip Reference Manual"
"SIS 85C407 Buffer Chip"
page 52 on a sheet called "ISA BUS BUFFERS and RTC". Buffers ISA address/data bus, multiplexes DMA/IRQ requests, RTC. Replaces at least 5 chips.
As I knew from the chipset encyclopedia that the 85C471 datasheet was available, I could have checked at least this quote. To me it clearly explains that this is not a 206 replacement, because the 206 functionality is built into the 471 chip, so my expectation was wrong. I got right that this chip contains the RTC, though. The RTC is an ultra-low-power device, whereas the 85c471 is a high-speed (runs at local bus clock) device. Most chip vendors have difficulties integrating high-speed and ultra-low-power technology on the same silicon, so it makes sense to leave out the RTC from the so-called "single-chkip" 486 system controller. The 407 only operates at ISA speed, not at local bus speed, and thus is a better fit to host the RTC. You see the 407 in action on page 52 in the PDF file linked by rasz_pl. In that PDF, it is split into three logical parts: The data buffers ("-A"), the multiplexer(s) ("-B") and the RTC ("-C").
I already discussed why you would have the RTC not in the 471 chip, but why would SiS put the data buffers and multiplexer(s) in a dedicated chip while advertising the 471 as "single chip" solution? The idea is the same as all other "single chip" manufacturers had at that time: The 471 is a 208-pin PLCC chip, and that's the case with the most pins that was cheaply available in manufacturing and soldering processes, and you run out of pins on that chip if you want a fully functional 486 system. So most vendors that tried to provide a "single-chip" solution integrated as much control logic into the single chip, but left "simple tasks that require a lot of pins" to external logic. As you see: "Replaces at least 5 chips" is not a whole lot of chips, but I recon this estimate to be quite low. Looking at the schematic, the 407 replaces (on a full-fledged AT-compatible mainboard): A bidirectional 8-bit host-to-X-Bus data buffer (74LS245), a bidirectional 8-bit ISA-to-X-Bus data buffer (74LS245), an 4-bit address latch (1/2 74LS373), a multitude of multiplexer chips, mapping 24 inputs into two outputs with a surprisingly low number of control pins (i.e. there needs to be some flip-flop / shift register logic in it, too), and finally the RTC. If I would have written the marketing blurb, I likely would have tried a claim like "replaces around 8 chips in typical AT-compatible computers".
So the 471 does all the local-bus stuff (including RAM and cache) and it fully controls what the 407 does, but the 471 doesn't get in contact with all the ISA bus signals, reducing the number of pins required on the 471 a lot. The 407 is produced using a proven cheap technology for operating frequencies up to 20MHz (the highest speed signal processed by that chip is a 14MHz clock controlling the multiplexer) and handles the "low-speed only" stuff.
rasz_pl wrote on 2024-03-04, 02:29:
mkarcher wrote on 2024-03-03, 17:38:
[*]pin 82 of the 85c407 is the RTC supply pin. The RTC is integrated in the 85c407.
82 looks like #reset? PSTRB hmm power strobe? 86 is power
On that schematic: Pin 82 is part of the RTC part of the 85c407. It is powered by RTC power, but it is not the RTC power pin (which is pin 86, indeed). So R17/C11 on the board of the OP again looks like a "power-on reset generator", as initially suspected, but it doesn't generate a power-on reset for the PC system, but a power-on reset for the RTC. If pin 82 goes from low to high, the RTC is reset (possibly clearing the CMOS, possibly just kicking the oscillator, possibly just setting the "power failed" bit in the RTC status), and this reset happens when you connect an external battery.
