2 Nov
2013
2 Nov
'13
1:17 p.m.
-----Original Message-----
From: cctalk-bounces at classiccmp.org [mailto:cctalk-
bounces at classiccmp.org] On Behalf Of John Wilson
Sent: 02 November 2013 17:59
To: General Discussion: On-Topic and Off-Topic Posts
Subject: Re: Looking for a MicroController Recommedation
On Sat, Nov 02, 2013 at 05:25:05PM -0000, Robert Jarratt wrote:
:-)
I have been mulling a project to emulate MFM disks
(DEC RD53 and RD54
in
particular) at the disk-to-controller interface.
You're part of a long tradition of mulling! Looking for recommendations for a suitable
microcontroller that does
not cost the earth ...
I never miss a chance to gush about the XMOS XS-1 chips. They're great this kind of thing -- once you're down into things
that happen a few dozen
ns
apart you'll definitely want to use assembly
language, but it's not a bad
architecture once you get over how C-centric it is. The main down side is
that
XMOS is proud of their tiny packages so you'll
need paste solder. But
oshpark.com is cheap and a toaster oven works nicely for soldering. The
chips themselves are mostly under US$20 and have a very nice architecture
for handling events. And they're hyperthreaded in hardware so you can
have one thread watch the RQDX3 side and another handle the SD card and
they both execute instructions no worse than once every two clocks (if
you're fully loaded at 8 threads -- faster if not), so it's easy to keep
up
w/o
driving yourself crazy.
I had a look and it seems close to what I would need, but possibly not
quite. It says it will do 400mips, which for a sampling frequency of 40MhZ
would give me 10 instructions per sample (crude assumption I know), but it
is not clear to me if each core does 400mips, or if that is 400mips
aggregated across 8 cores (which seems more likely), which would give my
just 1 instructions per sample. Do you know which it is?
There is a nice little affordable starter kit board, the XK-1A, which is
even available in the UK, but it doesn't have enough on board memory to
store a cylinder (which I now realise would need to be up to 500Kbytes per
cylinder if I am storing samples at 40Mhz, it could store a track though),
and I can't see any SD card addons on the xmos site, so I would have to
build all that. That may be a bit much for me.
Someone (was it Emanuel maybe?) suggested doing the emulation at the
level of the HD controller chip (I forget, is it an SMC part?) on the
RQDX3 board. Assuming you can find the data sheet, doing a register-level
emulation might *really* loosen up the timing quite a lot (hundreds of
nanoseconds to respond). Plus then you'd be dealing with cooked bytes
instead of raw flux transitions so you might be able to put something more
useful on your SD card (like plain old .DSK files, so it'd be easy to
insert/extract files with the SD card plugged into another computer w/o
having to hack the low-level format).
Yes, I have seen this suggestion. I don't really like it because I want my
RQDX3 to work "intact" and it is not very general for other controllers.
My thought was that if I can get this to work it wouldn't then be a big step
to modify it to work in the opposite direction and use it to image old disks
before they finally die.
Regards
Rob