exxos's Atari, Amiga & retro forum

Well, other than waiting on the QS8888-20P chip from China I think the 486 is as good as it's going to get.

Last night I attached a newer (2000) IDE drive as master, copied off the data and reformatted the "D:" drive to get rid of the bad blocks and then copied the data back. This also allowed be to set the best "translated" CHS parameters in the BIOS and hence get the largest size (201MB).

The machine still doesn't always start first time when switched on but a quick reset afterwards always works, which is good enough. The PSU is outputting a constant 5.04V on the +5V line and 12.07V (+-0.01V) on the +12V rail, except when the drives are spinning up, when it drops to 11.95V.

stephen_usher wrote: 13 Jan 2022 09:45
The machine still doesn't always start first time when switched on but a quick reset afterwards always works, which is good enough. The PSU is outputting a constant 5.04V on the +5V line and 12.07V (+-0.01V) on the +12V rail, except when the drives are spinning up, when it drops to 11.95V.

Sounds very like the problem I had with the Mac Classic.

Well, except that I can see that the power rails are coming up fine, though I've not tested the -12V. This feels like a motherboard issue now.

I thought that it might have been a PSU issue as every now and again the fan seems to slow and then after a few seconds speed up again, but the +12V is as steady as a rock (+- 0.01V).

stephen_usher wrote: 13 Jan 2022 10:39 Well, except that I can see that the power rails are coming up fine, though I've not tested the -12V. This feels like a motherboard issue now.

I thought that it might have been a PSU issue as every now and again the fan seems to slow and then after a few seconds speed up again, but the +12V is as steady as a rock (+- 0.01V).

Is there any current draw when it starts but doesn't boot?

From what I can see the power dips caused by load (which is all I can measure) are the same. I'm guessing that it's something in the reset circuit which is going out of spec., or the original design was close to the edge anyway and component age drift has just pushed it beyond the limit.

I found another QS8888 chip for sale on eBay yesterday which was from within the UK and delivery by the end of the week, so I ordered one.

Today I get an e-mail cancelling the order and saying, "Sorry, I can't locate the item." :-/

Why the heck would you create the sale if you don't actually have the stuff to hand? It just wastes everyone's time.

The chip from China still isn't due for about another month. (This assumes that (a) it's actually the correct chip and just not a random one with the correct number of pins which has been relabelled, and (b) actually works.)

stephen_usher wrote: 17 Jan 2022 13:58 I found another QS8888 chip for sale on eBay yesterday which was from within the UK and delivery by the end of the week, so I ordered one.

Today I get an e-mail cancelling the order and saying, "Sorry, I can't locate the item." :-/

Why the heck would you create the sale if you don't actually have the stuff to hand? It just wastes everyone's time.

The chip from China still isn't due for about another month. (This assumes that (a) it's actually the correct chip and just not a random one with the correct number of pins which has been relabelled, and (b) actually works.)

This sort of thing drives me mad!! I'm sure they do it just to get you in front of their 'other items' in the hope you'll want something they do actually have!!

Hmm... I think this might be why there's a smell of burning electronics when I plug the second PSU into our core switch unit...

stephen_usher wrote: 17 Jan 2022 14:58 Hmm... I think this might be why there's a smell of burning electronics when I plug the second PSU into our core switch unit...
IMG_2544.jpg

Ooooh, scorchio!!

This evening I've been thinking again about how to implement the Apricot hard disk emulation, using a Raspberry Pi to emulate the WD1010 controller an much of the logic.

The original interface mapped the WD1010 registers to the I/O port addresses 0x0e2 -> 0x0ee (even addresses, i.e. low byte only), then from 0x1e0 -> 0x1ee write-only 1 bit registers plus the address 0x1f0 as the location to read/write to the 8K data buffer (auto incrementing the internal address register per operation).

Thinking about this further, I'm going to have to use some SRAM to emulate the WD1010 registers and it complicates things a bit that two of the locations read and write different things, e.g. 0x0ee reads from the status register but writes to the command register. I can't just use the Pi for this as it would be too slow to respond.

Similarly, the 8K data buffer and counter wouldn't be feasible for the Pi to drive directly, so I'll definitely need a counter (12 bit?) and an 8Kx8 SRAM.

Now, fortuitously I've bought 4 32Kx8 SRAMs for the 486 repair so I could use one of those for both the registers and the buffer and possibly the latch values, but this will need quite a few GPIO pins (and bi-directional level convertors) on the Pi end. The Sparkfun one can do 4 bits for £3, but space may be an issue.

Hmmm....