Emulate a 9316C ROM for Currah Microspeech (µSpeech)

[greenonline]

Would it be possible to use a One ROM to replace a 9316C ROM IC, that is found in a Currah µSpeech for the ZXSpectrum?

A photo of the ROM on the PCB is here (taken from currah_uspeech_tests):

The 9316C appears to be a rather obscure ROM, and I've not found much out about it.

This is the pinout (apparently, not 100% sure):

 9316C
          +---v---+
    A7  1 |       | 24  Vcc (+5V)
    A6  2 |       | 23  A8
    A5  3 |       | 22  A9
    A4  4 |       | 21  CS3 (Chip Select 3)*
    A3  5 |       | 20  CS1 (Chip Select 1)*
    A2  6 |       | 19  A10
    A1  7 |       | 18  CS2 (Chip Select 2)*
    A0  8 |       | 17  D7 (Data 7)
    D0  9 |       | 16  D6 (Data 6)
    D1 10 |       | 15  D5 (Data 5)
    D2 11 |       | 14  D4 (Data 4)
   GND 12 |       | 13  D3 (Data 3)
          +-------+

There are suggestions on the web to use a 2716, which has a similar pin out:

 2716
          +---v---+
    A7  1 |       | 24  Vcc (+5V)
    A6  2 |       | 23  A8
    A5  3 |       | 22  A9
    A4  4 |       | 21  VPP
    A3  5 |       | 20  /OE
    A2  6 |       | 19  A10
    A1  7 |       | 18  /CE
    A0  8 |       | 17  D7 (Data 7)
    D0  9 |       | 16  D6 (Data 6)
    D1 10 |       | 15  D5 (Data 5)
    D2 11 |       | 14  D4 (Data 4)
   GND 12 |       | 13  D3 (Data 3)
          +-------+

but apparently a couple of the select lines can cause issues, and a custom harness using a DIP socket is required.

Note:

• Inverted logic on two of the enable pins
• VPP replaces CS3

Here are a couple of reports that I found:

From HOW TO READ 9316 ROMS

I made an adapter for 9316 to 2716 as follows:

1. Obtain a 24 Pin wire wrap socket.
2. Cut pin 21, leaving enough of a stub to solder to.
3. Solder a wire from Pin 24 to pin 18.
4. Solder a wire from Pin 12 to the stub pin 21.

Also, 9316B Rom replacement

For the base I am using a 2kb cartridge that contained a 9316B rom and I have replaced it with an at28c16 eeprom. I have successfully written to the eeprom using an arduino but for the life of me I can't get it to work. I have seen various sources quoting that the 9316B and 2716 are pin compatible (hence the 2816 should be too right?) and others saying pins 18 and 21 need to be swapped. I have tried both of these layouts and had no success.

So, after spending a couple of days wondering about this, whilst watching Turbo Mail Call #6: One mystery solved but new mysteries yet to be solved w/ tools, games and more., at 56:15, it dawned on me that the One ROM might be a perfect solution.

Is that possible? I assume that the IC is the standard width, etc.. Can the weird issue with the select lines be addressed? I haven't actually looked into the One ROM all that much, I've only seen what Adrian has shown in his videos. So, this is probably a naive question post, and the answer (I hope) is probably, yes?

ROM chip select

• Pin 18 - CS2 - Tied to GND, i.e. LOW
• Pin 20 - CS1 - ULA pin 7 (unknown behaviour, until ULA decoded)
• Pin 21 - CS3 - Tied to "Regulated 5V" line, i.e. HIGH

---

Background

I'm basically slowly getting around to reverse-engineering a µSpeech, see currah_microspeech, recreating the Schematic, PCB, and I will get around to dumping the ROMs. Not sure about the ULA though, but will tackle that last.

[piersfinlayson]

One ROM (24 pin) can support the 9316 natively. That's one of its big advantages. No need to be a 2716 and hack the CS lines.

So, configure as a 9316 (or 2316 - they are the same). You do need to figure out which direction each of the 3 chip select lines need to be for that care - each one could be programmed active high or active low at factory.

You might be able to figure out from the 2716 hack instructions you found, or if you've reverse engineered this bit of the schematic you'll now. Alternatively just try all 8 combinations one after the other - one should work. Then please come back here and document the answer!

I haven't ever actually seen a 9316C datasheet - but it's probably just faster than the 9316A and 9316B.

[greenonline]

I found a datasheet for the 9316, linked to from Unable to read older GI 9316b ROM in programmers - Suggestions?, that has the specs for the 9316C. It is indeed faster:

Type	Access time	CS response time
9316A	600-850	200-300
9316B	350-450	100-200
9316C	250-350	100-200

[greenonline]

FWIW, I have dumped the ROM, here: Currah µSpeech - ROM reading. The text format might not be quite right? Not sure.

[piersfinlayson]

Here's the ROM as a .bin file, more useful for programming One ROM.
ROM.bin.zip

Access and CS response times should be fine for One ROM.

[greenonline]

Ah thanks, nice one. I've actually just Pythoned up a script to make a .bin from that text file and also disassembled the code. It's all under the same link above. Busy night tonight :-)

[greenonline]

Please discard that ROM file/image. As the disassembly seemed a bit suspect, I decided to re-run multiple ROM dumps today, and while they are all 99% similar, there are a varying number of non-matching bytes, in different locations, each run, unfortunately. Plus it gets really hot. I think that the ROM is clearly defective, hence discard the image. I'll have to get another unit and crack it open and dump that ROM. Most disappointing.