The November Rougol meeting in a week later than usual. It will be looking at RiscPC and Archimedes repairs. It kicks off at 6.30pm in the Duke of Sussex and online at 7.30pm on Zoom (usual link or contact Rougol for one).
As always meeting full details are on the site.
| |
Rougol November 2023 talk is on RiscPC/Archimedes repairs |
| |
helpful (03:46 26/11/2023)
Elesar (07:36 28/11/2023)
CJE (11:56 29/11/2023)
Elesar (17:02 29/11/2023)
CJE (13:54 30/11/2023)
Elesar (22:49 30/11/2023)
CJE (11:35 5/12/2023)
CJE (19:02 15/2/2024)
timber (09:04 2/12/2023)
|
| |
|
Bryan Hogan |
Message #125531, posted by helpful at 03:46, 26/11/2023 |
Member
Posts: 254
|
Reminder that the meeting is this Monday (tomorrow!). Still happy to have more demos, either online or (even better) in the pub. Bring your old machines for a night out!
Hopefully see some of you on Monday 
Bryan. |
| |
 [ Log in to reply ] |
| |
|
Robert Sprowson |
Message #125532, posted by Elesar at 07:36, 28/11/2023, in reply to message #125531 |
Member
Posts: 45
|
There were questions about which compact flash interface I was showing in the A7000+ during the meeting, but tricky to post links out loud from memory, so here are the details.
The adapter
For the adapter you want one like this. At time of writing this is £5.25 including postage. Since that link wont last forever the main points to look for are:
* No chips - the conversion from IDE to CF is entirely passive, no controller chip is visible
* 40 pin IDE connector - matches the number of pins on the IDE, don't get 44 or 50 pin ones which also exist
* Straight pin male IDE connector - while female and right angled exist, intended so the adapter plugs straight into the motherboard IDE socket, these aren't suitable for the Risc PC or A7000(+) because the drive bay clashes mechanically
* A master/slave jumper - preferable but not essential, allows the CF drive to coexist with another harddisc
The cable
A short 40 pin IDE cable will be needed too, 6" is enough, to plug it into the motherboard and tuck the whole assembly in the bottom of the case. The Risc PC has conductive paint on the inside and the A7000(+) is metal, so you might need some insulation to stop it touching the case, this can be as low tech as a square of cardboard.
The card
I use a 16GB Sandisk Ultra Compact Flash card (model SDCFH-016G-A11). Pre RISC OS 5 there are diminishing returns using a huge card as the minimum allocation size gets bigger and bigger, 4GB or 8GB is probably enough.
Beware clone cards if buying online - if it seems unreasonably cheap (< £20) despite being "new" it's probably best to avoid it. Camera shops often sell Compact Flash cards so don't just think about computer shops.
The software
Then for extra modding points I took ADFS 3.33 which is the last version of ADFS before the work on 32 bitting and HAL abstraction, and applied the patch from ADFS 3.54 to it to avoid the DRQ timing problem that many CF cards will cause when ADFS is expecting a handshake during writes.
Then put that module into the spare space in the EPROM (or flash) on the NIC card. On power up a double seek of the floppy head can be heard - the first is the ADFS in the main ROM starting, the second is the patched ADFS starting. In theory the patched ADFS could be loaded off disc but I dislike the paradox that causes!
This is only needed for 26 bit versions of RISC OS, RISC OS 5 has the patch integrated already (since ADFS 3.54).
The transfer
Getting data on could be the trickiest part since the CF card will be blank when supplied. You'll need HForm to format the blank card, then a copy of the boot sequence. The options are:
* A floppy disc. Many floppy discs. So many.
* ShareFS to another RISC OS computer on the network, this is by far the easiest.
* A CD which includes a copy of HForm and boot sequence.
* Using a dual ribbon cable to copy in a computer which is already set up, with jumpers set to master on the CF and slave on the HDD |
| |
[ Log in to reply ] |
| |
|
Chris Evans |
Message #125533, posted by CJE at 11:56, 29/11/2023, in reply to message #125532 |
CJE Micros chap
Posts: 228
|
I think it's also worth mentioning that like IDE hard drives not all Compact Flash cards will work on the Acorn IDE interface. We have a box of Compact Flash cards that will work on most IDEFS podules but not Acorn's IDE interface. |
| |
[ Log in to reply ] |
| |
|
Robert Sprowson |
Message #125534, posted by Elesar at 17:02, 29/11/2023, in reply to message #125533 |
Member
Posts: 45
|
I think it's also worth mentioning that like IDE hard drives not all Compact Flash cards will work on the Acorn IDE interface. We have a box of Compact Flash cards that will work on most IDEFS podules but not Acorn's IDE interface. What do you define as not working? Disc errors on writing? Conflict if a NIC using the ready line is fitted? Something else? I've given the Sandisk Ultra model number as a known good example, do you have a list of others? |
| |
[ Log in to reply ] |
| |
|
Chris Evans |
Message #125535, posted by CJE at 13:54, 30/11/2023, in reply to message #125534 |
CJE Micros chap
Posts: 228
|
Nearly all have either failed to be recognised, gave gibberish heads and cylinder count to HForm or HForm fell over at the end. IIRC only one seemed to format but then gave errors when writing to. Yes we've seen the conflict with some NICs.
[Edited by CJE at 13:55, 30/11/2023] |
| |
[ Log in to reply ] |
| |
|
Robert Sprowson |
Message #125536, posted by Elesar at 22:49, 30/11/2023, in reply to message #125535 |
Member
Posts: 45
|
Nearly all have either failed to be recognised, gave gibberish heads and cylinder count to HForm Wild speculation time - they might be trying to signal at 3.3V, whereas the Acorn IDE interface is 5V, or not able to drive the bus to a valid logic level for those lines which are shared (unbuffered).
The write error problem can be solved by software. The NIC clash requires a track cut on the CF-IDE interface. |
| |
[ Log in to reply ] |
| |
|
André Timmermans |
Message #125537, posted by timber at 09:04, 2/12/2023, in reply to message #125535 |
Member
Posts: 5
|
I remember that not all HDs were usable on the RISC PC. If you tried to reproduce what RISC OS did to identify the HD, at some point it does request a special block of data from the drive providing a description of the drive (cylinders count, heads count, etc). For the Hitachi drive that I tried at that time, every odd (or was it even?) byte in that block was 0.
From what I remember is that these drives dropped the older specs and always transmitted data as 16-bit instead of 8-bit which the RISC PC IDE was asking. |
| |
[ Log in to reply ] |
| |
|
Chris Evans |
Message #125538, posted by CJE at 11:35, 5/12/2023, in reply to message #125536 |
CJE Micros chap
Posts: 228
|
Nearly all have either failed to be recognised, gave gibberish heads and cylinder count to HForm Wild speculation time - they might be trying to signal at 3.3V, whereas the Acorn IDE interface is 5V, or not able to drive the bus to a valid logic level for those lines which are shared (unbuffered).
The write error problem can be solved by software. The NIC clash requires a track cut on the CF-IDE interface.
Cut which line please? |
| |
[ Log in to reply ] |
| |
|
Chris Evans |
Message #125595, posted by CJE at 19:02, 15/2/2024, in reply to message #125538 |
CJE Micros chap
Posts: 228
|
The answer (as often is the case) is on Stardot:
https://stardot.org.uk/forums/viewtopic.php?f=16&t=14016&start=30#p189188%5B/url
& https://stardot.org.uk/forums/viewtopic.php?p=233287&hilit=compactflash+diode#p233287 |
| |
[ Log in to reply ] |
| |
|
