Missing toshiba HX-F101 cartridge, reverse engineering schematics for it?

Page 1/4
| 2 | 3 | 4

Par popolon_

Master (132)

Portrait de popolon_

30-12-2020, 22:18

I have toshiba HX-F101 (japan) external floppy drive unit, that is missing the cartridge part. Does anybody have this unit and could take good photos from both sides of the PCB? But even with pictures, there would be some signals hidden under the ICs still.
Based on the other topic in here, https://www.msx.org/forum/msx-talk/hardware/toshiba-hx-f101-... , the HX34 diskrom was used successfully in HX-F101 cartridge. And it was mentioned that same addressing is used in both HX34 and HX-F101 cartridge.
So I tried to check the implementation between toshiba HX34 mainboard floppy circuit and the external unit circuit that I have, and then the one picture from that cartridge component side:
http://www.retrocomputacion.com/e107_files/public/1404066639...

I tried to draw schematics, but there are missing pieces.
hx_f101_cart_1st_draft.jpg

the connector cable between cartridge and unit has 34 pins (marked in schema as PB1-PB34, odd pins are ground)
I traced backwards the signals from these PPB2-PB34 signals, and almost same was found from mainboard, but the 74LS133 input signal logic (for ENABLE signal) was not exactly same. Would this be working same way if the PB16 input would be generated via different logic in HX34 and HX-F101 cases? I assumed same logic to generate PB16, but it cannot be true as all the same signals in mainboard case are not available in slot interface. So what would be logic for this PB16 signal, i.e. 74LS133 input signals?

And in that cartridge component side picture, it can be seen that OR output (U4C) output is connected to NAND input (U8A), but I did not find that kind of connection in HX34 mainboard.

Does anybody have clues how to complete this schematics?

!login ou Inscrivez-vous pour poster

Par popolon_

Master (132)

Portrait de popolon_

01-01-2021, 20:17

the logic difference for those 74LS133:
hx_PB16_differences.jpg

Par popolon_

Master (132)

Portrait de popolon_

01-01-2021, 22:06

There seems to be many different ways to set for example that rom enable signal (_CE).
In some philips fdd cartridge case it is set by using A15+A14 + _SLTSL + A13-A3 + A2 (these signals are used but complex logic is used).
https://hansotten.file-hunter.com/uploads/files/philipsvy001...

In talent TDC-600 case it is set straight to _SLTSL signal.
https://www.msx.org/wiki/Talent_TDC-600

In toshiba hx34 mainboard case it is connected to _RD.

So why so different mechanism in module designs? Is there any recommended way to do it?

In this toshiba case the I/O address would be that 7FFF0, in philips case it was 7FFF8. So would NAND with A3-A14 with inverted _SLTSL signal and inverted A2 be enough to decode the address between 7FF0...7?

Where are these all decodings and activation of rom for module cart design really explained, i have not found any documents excect some software point of view things but not hardware.

Par Manuel

Ascended (17513)

Portrait de Manuel

01-01-2021, 22:16

I'm not sure if this is helpful, but there are several 'connection schemes' of the 2793 used. The F101 uses a very specific one, which matches the HX-33/34. Zeilemaker figured it out and it's documented in code here: https://github.com/openMSX/openMSX/blob/master/src/fdc/Toshi...

Par popolon_

Master (132)

Portrait de popolon_

01-01-2021, 22:32

have to check that github page through.

Current scheme assumption is this kind of:
hx_f101_cart_v0.2.gif

Par gdx

Prophet (4045)

Portrait de gdx

02-01-2021, 01:48

Thanks Popolon_ I enjoy your investigations on MSX hardware.

Par popolon_

Master (132)

Portrait de popolon_

02-01-2021, 21:32

I looked at that github code, but that did not help now.
I draw also the hx f101 unit side schematics (without some resistors/transistors in fdd side), to hope to see the missing logic. But there is one very different logic inside, if i compare it to philips fdc cartridge.
In toshiba IC12E + IC8B inputs (P6= A2 <-> P16= A3-A13 + STSTL NAND output) are reversed
hx_f101_unit_schema_part.gif

when comparing to philips schematics (IC11 + IC12):
philips_fdc_logic_part.gif

Philips logic ( makes sense as 7FF8-7FFB are for FDC, and 7FFC-7FFF are for that flip flop part ):
A13=1,A12=1,A11=1,A12=1,A11=1,A10=1,A9=1,A8=1,A7=1,A6=1,A5=1,A4=1,A3=1,A2=1), WD2793 _CE = 1 (so chip disabled), write and read to flip flop parts possible if also _RD=0 or _RW=0 signals.
A13=1,A12=1,A11=1,A12=1,A11=1,A10=1,A9=1,A8=1,A7=1,A6=1,A5=1,A4=1,A3=1,A2=0), WD2793 _CE = 0 (so chip enabled),no write and no read to flip flop parts

Toshiba logic (to get 7FF0-7FF7 addresses):
A13=1,A12=1,A11=1,A12=1,A11=1,A10=1,A9=1,A8=1,A7=1,A6=1,A5=1,A4=1,A3=0,A2=1 -> WD2793 _CE = 1 (so chip disabled), no write and no read to flip flop parts.
A13=1,A12=1,A11=1,A12=1,A11=1,A10=1,A9=1,A8=1,A7=1,A6=1,A5=1,A4=1,A3=0,A2=0 -> WD2793 _CE = 0 (so chip enabled), no write and no read to flip flop parts.
one of A4-A13 is 0 or A3 is 1, and A2=0 -> WD2793 _CE = 1 (so chip disabled), write and read to flip flop parts possible if also _WE=0 or _RE=0 signals.

So it is strange why would this even work that when address is not even the correct, even what ever, as long as A2=0, then it would read and write to flip flops. This address decoding was also totally different in the hx34 mainboard case, where A9-A14+_A15 were only used. So maybe I have then invalid logic for these two signals that I need from cartridge ( P6 + P16). I assumed that P6 is A2 and P16 is that address decoding without A14,_A15,A2. But maybe that is not correct. Based on philips schematics, the assumption could have been correct if that one place would not have been different.
this is current cartridge schematics:
hx_f101_cart_v0.31.jpg

Par msxholder

Champion (370)

Portrait de msxholder

02-01-2021, 21:48

I could provide foto's from hx-f101.

Par popolon_

Master (132)

Portrait de popolon_

02-01-2021, 22:06

Of course pictures would help. But better would be if the signals could be measured with multimeter (what IC pins are connected to what IC pins/cartridge slot pins) inside the cartridge.

Anyway I have been looking at too long at this schematics, as the P16 and P6 signals would have to in real life reversed (so logic comes then same as in philips, at least for the flip flow read and write).

Anyway the rom enable (_ce) uses now same logic as in philips case (except the A3=0, as address is between FFF0-FFF7). Also I am not sure what to do with those RD and WR signals, shuold those be OR:ed with _SLTSL or not. And that 74LS245 direction signal is done with only that _RD signal. Then there is this enable signal for this 74LS245, it is using now _A15= + A14=1 + _SLTSL=1 logic, this logic was copied from HX24 mainboard. Both these signals goes from cartridge to the external unit, to the 74LS245 ic.

Par popolon_

Master (132)

Portrait de popolon_

03-01-2021, 14:25

I put 74LS245 direction signal behind OR ( _SLTSL , _RD ), as in MSX Technical Handbook (there is only one hw example in that document, and even that is for some slot expander, so not good doc at all for this case).
It seems msx.org wiki page https://www.msx.org/wiki/Hardware_Design and that MSX Technical Handbook are not really giving good examples.
Anyway here is latest kicad schematics:
https://github.com/popolonx/MSX_STUFF/tree/main/hx_f101
inside there is separate directory for cartridge and for drive unit side (note! only contains logic chips parts).

Par popolon_

Master (132)

Portrait de popolon_

03-01-2021, 16:46

Maybe RD and WR signals could be used as such(no OR against _SLTSL), as in philips case. And maybe 74LS245 _CE signal could be just that _SLTSL.

Page 1/4
| 2 | 3 | 4