Most other computers of that era tend to use something similar to a 27c512, 27c256, or smaller.

Since you mention 8K, a 27C64. disable chip id

Which chips have you already tried to select in tl866 software?

See if you can find the schematic for the Amstrad and check the pinot looks compatible with one of the 27xxx series ROM datasheet. I just dumped the Acorn Electron ROM last night using my TL866 and I just picked the AM27256 after matching the pinot to a datasheet. Worked great. Also interestingly the Acorn Electron ROM is not fully accessible from the Electron itself so there is a nice 'hidden' thank you message stored on it at the inaccessible locations, which is nice!

Thank you for your input so far!

I feel a bit stupid now... AT28C64B gave me a pin detect error on pin 27. I was reluctant to "force" reading so I never tried that before posting. Based on what you guys wrote I went ahead with it and it read the ROM successfully. I compared both dumps with what I got from splitting up what I got from DEBUG and the hashes match. So that solves this mystery.

Still do not understand the pin detect error, though! Attached the schematics just fyi. I know they also are on the net.

Thanks again!

Jens

Nice one glad you got them dumped. Its something that also stumped me when dumping ROMs from EGA cards and your instinct is right, you don't want to just override and fry something. Cheers!

Hey

Today I tried to dump the Paradise EGA ROM of a 1640 (dubbed “40100” by Amstrad). To my surprise it was a 27(!)256. So on the PC1640 there are actually 2864s and 27256s on the same board. Like with the 2864s the ROM reader reported loads of “bad” pins but read the ROM fine when forced.

Just thought I’d add that here for future reference.

jewesta wrote on 2022-07-07, 20:08:

Hey

Today I tried to dump the Paradise EGA ROM of a 1640 (dubbed “40100” by Amstrad). To my surprise it was a 27(!)256. So on the PC1640 there are actually 2864s and 27256s on the same board. Like with the 2864s the ROM reader reported loads of “bad” pins but read the ROM fine when forced.

Just thought I’d add that here for future reference.

Sorry to dig up an old thread, but have a PC1640 & both bios roms have failed, along with some of the RAM...
So I need to burn a new pair of bios roms.
The RAM is easy enough to find, but I just need to know what type of eeproms or eproms I should be using for the job ?

Vadership wrote on 2025-04-23, 14:33:

The RAM is easy enough to find, but I just need to know what type of eeproms or eproms I should be using for the job ?

Hey! 🙂

I hate to be that guy but it basically says so in my post you quoted: The EGA ROM is a 27256 and the two BIOS ROMs are 2864s. Any ROM (eprom / eeprom) fitting the general specs should work.

I can see how the numbers can be a bit confusing at first though and tbh what I know also mainly comes from experimenting and interweb research. So take what I write with a grain of salt. You will find these ROMs usually labelled in the form 27x256x or 28x64x. The first two numbers specify the pinout type. So 27s and 28s are not pin compatible. The second set of numbers specifies the capacity. The X mainly represent vendor codes and can vary, though from experience all 27s work where 27s are required and vice versa with 28s. Make sure to get the right package, meaning DIP (dual inline package = classic thruhole pins) and not e.g. SMD variants.

ROMs are graded by speed which is a number appended at the end. Get the fastest you can get, though I usually use -150s (150ns) and that mostly works fine. With the Chinese knockoffs you never know anyway. Best to buy from a trusted electronics seller even though these will be vastly more expensive.

Now the capacity should match what your PC1640 requires, though usually you can go bigger. If you only have access to bigger ROMs, e.g. a 27512 instead of a 27256, then you “simply” need to put the image on the ROM (in this case) twice in succession. Either the PC picks up the first or the second (or third or fourth for even larger ROMs) copy, depending on the wiring. It can(!) work if you only put it on address zero, but it also can not.

With the popular TL866 uou first load the rom at zero, checking “fill rest with zeroes”. Then you load it at the second (third, fourth…) address but this time do not check “fill with zeroes” but select leave rest as-is.

For convenience I use Atmel AT28C64 and Winbond W27C512. These are EEPROMs, so electronically erasable, without having to erase them with UV light. They are relatively easy to get and cover most of my scenarios but they can be expensive. Like I said, for the W27C512, when used as an Amstrad EGA BIOS, you’d have to load the 256 image twice with your EEPROM writer, first at address zero and then at address 8000 wich marks the first of the second half (if memory serves!) and then burn it to the ROM.

Hth!
Jens