Forums

I am documenting the SF-7000. The interesting thing is that it's made only of off-the-shelf (for their time) parts, no custom ICs. I know that for the sake of emulation what is needed is already understood, but it'd be nice to completely document it.

I attach my progress so far. I based my schematic on the following sources

Original SF 7000 schematic, 1/3 pages available
SF 7000 pictures
SF-7000 user's manual
Enri's cart schematic

The official schematic is illegible in several parts, however by tracing the pcb pictures I identified most of the gates.
Unluckily many of the vias end under an IC, so it is not possible to reconstruct other parts of the schematic.

Is there anyone with more information / willing to probe the sf-7000 with a tester / willing to do it together (I live in Osaka)?

Findings so far:
1)PA2 _is_ index. This is missing the the SF-7000 user manual, but hinted in meka's source code (it has a question mark) and verified from the pcb photos.
2)The chip select for the FDC,8255,8251 is derived only from addresses A2,A3,A4.
The IO/R IO/W reported in the original schematic must be a function of A5,A6,A7 as well, otherwise there would be confilict between the SC-3000 internal PPI for addresses $C0 to $DF.
3)There are physically many gates (IC14,IC15,IC16,IC17,IC21,IC23,IC25,IC26) whose purpose is not clear. Some are clearly aimed at multiplexing the address lines (ram takes 8 bit column and row addresses, while the address bus is 16 lines) and a gate must be used to invert the rom enable output of the 8255, but there are way more ICS then one would expect for those 2 functions.

sf7000.zip (18.25 KB)

sf7000.pdf (632.43 KB)

If you didn’t already see it, I made a disassembly some years ago of the IPL: https://github.com/maxim-zhao/smsbioses/tree/master/SF-7000%20IPL which heavily depended on the FDC documentation from a clone controller.

Thank you, yes, I saw that awesome disassembly :) It confirmed the PA2 and it made the datasheet much easier to understand

Two more findings:
4) There was a post some time ago which asked whether the SF-7000 is compatible with the SG-1000+SK1100. The answer seems to be no. The whole address space $C0-$FF is taken by the keyboard/the joysticks depending on the joy/sel line. The keyboard does not receive A5, therefore it cannot leave the $E0-$FF addresses to the SF-7000. Reads over that range would lead in both devices (keyboard and sf-7000) trying to write on the bus.
5) meka forces a NMI when writing to port E6 (line 300 of ioports.cpp). When this is removed, the IPL fails. However, that appears to be physically impossible,as NMI is not present on the cartridge connector

I obtained a prototype SF-7000 some type ago. I forgot if the IC/board were identical but I'm pretty sure the BIOS was on an EPROM with a <1.0 version number. Will let you know when I pull more info or a dump from it.

New finding
6) PC3 on the SC-3000's 8255 at some point was probably intended to be a "tape motor" signal. It is disconnected on the real hardware, but both the standard basic and disk basic set it when the tape motor should spin (load/save and cload/csave, respectively) and clear it when the motor should stop.

Are the disks used by this system the same Mitsumi Quick Disk format used in other 3 inch floppy disk implementations? Are they sectorless "stringy floppy" format with 64 K per side or so? And (other than the housing shape) compatible with Famicom Disk System?

I'm wondering whether the FDS Stick might be adaptable somehow (with different firmware I guess?) to replace the drive using modern flash memory

Based on the schematic it's NEC's clone of the standard PC floppy drive controller (µPD765A)

bsittler wrote

Are the disks used by this system the same Mitsumi Quick Disk format used in other 3 inch floppy disk implementations? Are they sectorless "stringy floppy" format with 64 K per side or so? And (other than the housing shape) compatible with Famicom Disk System?

I'm wondering whether the FDS Stick might be adaptable somehow (with different firmware I guess?) to replace the drive using modern flash memory

AFAIK, totally unrelated. This are the same 3" disks used in the Amstrad CPC or Spectrum +3 machines.

bsittler wrote

I'm wondering whether the FDS Stick might be adaptable somehow (with different firmware I guess?) to replace the drive using modern flash memory

I am not sure if that may work, but I am pretty confident the gotek would work, since it works on amstrad CPCs (http://www.cpcwiki.eu/index.php/Gotek) which have practically the same drive of the sf-7000

Author	Message
kamillebidan Joined: 25 Feb 2013 Posts: 384 Location: Osaka	Request for help - documenting the SF-7000 Posted: Sun Mar 21, 2021 5:26 am
	I am documenting the SF-7000. The interesting thing is that it's made only of off-the-shelf (for their time) parts, no custom ICs. I know that for the sake of emulation what is needed is already understood, but it'd be nice to completely document it. I attach my progress so far. I based my schematic on the following sources Original SF 7000 schematic, 1/3 pages available SF 7000 pictures SF-7000 user's manual Enri's cart schematic The official schematic is illegible in several parts, however by tracing the pcb pictures I identified most of the gates. Unluckily many of the vias end under an IC, so it is not possible to reconstruct other parts of the schematic. Is there anyone with more information / willing to probe the sf-7000 with a tester / willing to do it together (I live in Osaka)? Findings so far: 1)PA2 _is_ index. This is missing the the SF-7000 user manual, but hinted in meka's source code (it has a question mark) and verified from the pcb photos. 2)The chip select for the FDC,8255,8251 is derived only from addresses A2,A3,A4. The IO/R IO/W reported in the original schematic must be a function of A5,A6,A7 as well, otherwise there would be confilict between the SC-3000 internal PPI for addresses $C0 to $DF. 3)There are physically many gates (IC14,IC15,IC16,IC17,IC21,IC23,IC25,IC26) whose purpose is not clear. Some are clearly aimed at multiplexing the address lines (ram takes 8 bit column and row addresses, while the address bus is 16 lines) and a gate must be used to invert the rom enable output of the 8255, but there are way more ICS then one would expect for those 2 functions. sf7000.zip (18.25 KB) sf7000.pdf (632.43 KB)

Maxim Site Admin Joined: 19 Oct 1999 Posts: 14740 Location: London	Posted: Sun Mar 21, 2021 8:46 am
	If you didn’t already see it, I made a disassembly some years ago of the IPL: https://github.com/maxim-zhao/smsbioses/tree/master/SF-7000%20IPL which heavily depended on the FDC documentation from a clone controller.

kamillebidan Joined: 25 Feb 2013 Posts: 384 Location: Osaka	Posted: Sun Mar 21, 2021 9:27 am
	Thank you, yes, I saw that awesome disassembly :) It confirmed the PA2 and it made the datasheet much easier to understand

kamillebidan Joined: 25 Feb 2013 Posts: 384 Location: Osaka	Posted: Wed Mar 24, 2021 7:19 am
	Two more findings: 4) There was a post some time ago which asked whether the SF-7000 is compatible with the SG-1000+SK1100. The answer seems to be no. The whole address space $C0-$FF is taken by the keyboard/the joysticks depending on the joy/sel line. The keyboard does not receive A5, therefore it cannot leave the $E0-$FF addresses to the SF-7000. Reads over that range would lead in both devices (keyboard and sf-7000) trying to write on the bus. 5) meka forces a NMI when writing to port E6 (line 300 of ioports.cpp). When this is removed, the IPL fails. However, that appears to be physically impossible,as NMI is not present on the cartridge connector

Bock Site Admin Joined: 08 Jul 2001 Posts: 8649 Location: Paris, France	Posted: Wed Mar 24, 2021 1:05 pm
	I obtained a prototype SF-7000 some type ago. I forgot if the IC/board were identical but I'm pretty sure the BIOS was on an EPROM with a <1.0 version number. Will let you know when I pull more info or a dump from it.

kamillebidan Joined: 25 Feb 2013 Posts: 384 Location: Osaka	Posted: Thu Apr 29, 2021 3:01 pm
	New finding 6) PC3 on the SC-3000's 8255 at some point was probably intended to be a "tape motor" signal. It is disconnected on the real hardware, but both the standard basic and disk basic set it when the tape motor should spin (load/save and cload/csave, respectively) and clear it when the motor should stop.

bsittler Joined: 22 Apr 2018 Posts: 530	Posted: Thu May 13, 2021 5:57 am
bsittler Joined: 22 Apr 2018 Posts: 530	Are the disks used by this system the same Mitsumi Quick Disk format used in other 3 inch floppy disk implementations? Are they sectorless "stringy floppy" format with 64 K per side or so? And (other than the housing shape) compatible with Famicom Disk System? I'm wondering whether the FDS Stick might be adaptable somehow (with different firmware I guess?) to replace the drive using modern flash memory

lidnariq Joined: 24 Mar 2021 Posts: 120	Posted: Thu May 13, 2021 6:22 am
lidnariq Joined: 24 Mar 2021 Posts: 120	Based on the schematic it's NEC's clone of the standard PC floppy drive controller (µPD765A)

kusfo Joined: 29 Mar 2012 Posts: 886 Location: Spain	Posted: Thu May 13, 2021 7:43 am
kusfo Joined: 29 Mar 2012 Posts: 886 Location: Spain	bsittler wrote Are the disks used by this system the same Mitsumi Quick Disk format used in other 3 inch floppy disk implementations? Are they sectorless "stringy floppy" format with 64 K per side or so? And (other than the housing shape) compatible with Famicom Disk System? I'm wondering whether the FDS Stick might be adaptable somehow (with different firmware I guess?) to replace the drive using modern flash memory AFAIK, totally unrelated. This are the same 3" disks used in the Amstrad CPC or Spectrum +3 machines.

kamillebidan Joined: 25 Feb 2013 Posts: 384 Location: Osaka	Posted: Thu May 13, 2021 11:07 am
	bsittler wrote I'm wondering whether the FDS Stick might be adaptable somehow (with different firmware I guess?) to replace the drive using modern flash memory I am not sure if that may work, but I am pretty confident the gotek would work, since it works on amstrad CPCs (http://www.cpcwiki.eu/index.php/Gotek) which have practically the same drive of the sf-7000

Forums

View topic - Request for help - documenting the SF-7000