Re: A new S-100 Prototype board to interface with IDE hard disk drives and CF cards
- From: Allison@xxxxxxxxxxxxx
- Date: Sat, 07 Nov 2009 09:29:03 -0500
On Fri, 6 Nov 2009 13:18:37 -0800 (PST), monahanz
On Nov 6, 12:59 pm, Alli...@xxxxxxxxxxxxx wrote:
On Fri, 6 Nov 2009 11:16:38 -0800 (PST), monahanz
It's almost easier to interface IDE/CF to S100 without 8255. It saves
a 40 pin chip and also requires less bit twiddling as all either IDE
or CF is is a series of registers. Having been through the exercise
the more direct interface has a much faster IO rate.
Also CF can do 8bit (rather than 16bit) IO and that also makes the
interface dirt simple for CF only and the result is very fast and easy
build using a few TTL to decode block address, io read and write
plus buffering the split bus to the common IO bus of the CF.
Allison- Hide quoted text -
- Show quoted text -
Allison I considered a straight 7400's IC chip layout and while most
of the IDE commands are 8 bits the actual sector data is sent/read as
16 bits and so will have to be latched on each read/write. Others
have done it without a 8255 type chip such as here:-
I am familiar, I tried it and it was about 3x slower for PIO transfers
due to the extra but twiddling.
There are a number of ways to skin that cat!
Alternatively you can really keep it real simple and just waste the
upper 8 bits of sector data and read in 2x512 byte sectors. Since the
IDE allows multiple sector reads the software overhead would probably
not be that bad. I just have not had the time to count clock cycles
I have tried that. The overhead is zero as the command allows for
number of sectors to read. The cost is tossing half the storage
but even small drives are bigger than needed.
I just opted for an 8255 approach because I knew the chip well. Is
there faster, probably yes, but I can tell you my running CPM3 here
right now is night and day different from an old ST-506 not to mention
Big time better. IDE compared to my Teletek board and quantum D540
31mb drives is far faster, quieter and cooler. Going to CF was the
next and easiest step as it eliminated the 8-16bit transfer problem.
The CF parts are easier as transfers can be done as 8bit (it's a mode
bit) and being both smaller and lower power is convenient.
Performance for both CP/M2 and 3 are much better and the old floppy
space issues are no longer an issue.
The last s100 board I did was to have a CF, 2 serial ports and boot
rom on one board as that reduces the whole system to CPU, RAM and the
disk/IO card. Since I can get 32 to 128mb CF cards both easily and
cheap the side effect is all those 500mb and smaller IDE drives I've
saved up are unused and I have a boatload of them.
Thanks for the suggestion however.
I figured it would save the cost of the LSI, though they are still
easy to get.
- Prev by Date: Re: A new S-100 Prototype board to interface with IDE hard disk drives and CF cards
- Next by Date: Re: A new S-100 Prototype board to interface with IDE hard disk drives and CF cards
- Previous by thread: Re: A new S-100 Prototype board to interface with IDE hard disk drives and CF cards
- Next by thread: Re: A new S-100 Prototype board to interface with IDE hard disk drives and CF cards