[N8VEM-S100:7717] RE: SD Card Question

[John Monahan]

Actually I only running the 80386 reliably at 8MHz on the S100 bus Josh
(16Mhz on extended RAM/Protected Mode). I can get to 10MHz but from time
to time the IDE access fails. It's random am so frustrating to debug. I'm
only suspecting the 82C55. What worries me is that adding wait states (on
the V2 IDE board) does not seem to help however.



The problem with adding a DMA controller is that most people will be using
the 80386/80486 boards as an IEEE-696 style bus slave. Boot on Z80, then to
80386. Back and forth if need be, but the specs do not allow for a lower
sub-level DMA controller from the slave. In theory you have to go back up
to the master and back down to the slave. Now I actually addressed this on
our 80286 board in slave mode (slave 80286 to 68K and back) by using two of
the bus undefined lines. Alternatively you could just use the 80386 as a
bus master. However all this is not great and probably not worth the
complexity.



Related.

Complicating matters is the fact that my Saleae Logic 16 unit is not really
accurate for displaying signals around 8-10MHz. For example a 10MHz
oscillator is not evenly spaced apart. I'm thinking of getting a Zeroplus
unit (recently advertised in Nuts & Volts)
<http://microcontrollershop.com/product_info.php?products_id=5346>
http://microcontrollershop.com/product_info.php?products_id=5346



Has anybody tried this one, or have suggestions for a fast logic probe with
at least 8 inputs.



Josh I'm sending this to the whole group - hope you don't mind to get more
input

John





From: Crusty OMO [mailto:***@hotmail.com]
Sent: Friday, October 2, 2015 7:02 AM
To: John Monahan
Subject: RE: SD Card Question



Hi John,

Sorry I haven't been keeping up with all your new boards. I'm spending my
time on the old tech (8 bits or less).
With that said, what's the bus speed for the 386/486? I imagine you're
pushing 16Mhz?

I'm not a CF expert, but doesn't that interface in a parallel fashion?
Aren't you able to talk to it just like one big non-volatile RAM chip? I'm
clocking the SD card at the 8080 bus clock speed of 2Mhz. It takes 8 cycles
to transfer 1 SPI byte. That's 500KB/s (not counting the over head to
transfer it to memory). Since the transfer is all in software, I think it's
doing about 25KB/s by the time you count all the overhead.

In general, yes, SD are supposed to be faster than CF. But to get to those
speeds, you need to talk to the SD card through it's own licensed SD format.
This format allows for 4 bit transfers at speeds in the 50 to 100Mhz range.
But the free format is the SPI way, which is limited to 1 bit and up to 20
or 25Mhz. I am going to guess that the CF card and SD card (SPI way) are
about the same speed.

I'm not familiar with low level MSDOS access to the disk. Does it use INT
10h BIOS calls? Are those BIOS calls using DMA? Perhaps you need more
hardware on your disk card to perform DMA? Sorry I can't help in the DMA
chips either, but I'm sure a small microcoded processor could handle the
task. By using an EPROM, some TTL counter chips, flip flops and logic, a
circuit can be made to fetch X bytes from CF card and write them to RAM
while acting as the BUS master.

First, take a look at the DMA controller... is that chip 8237? (going by
memory). When I get a chance I'll take a look also.
But the first - first is to tell me more about your disk calls, INT 10h?

Cheers,
Josh



_____

From: ***@vitasoft.org
To: ***@hotmail.com
Subject: SD Card Question
Date: Thu, 1 Oct 2015 05:20:16 -0700

Hi Josh,

I am toying with the idea of building a dual Micro SD card S100 board. I'm
finding that the bottleneck in my 80386/80486 prototypes for MSDOS etc. is
the 8255A driven dual IDE/CF board. Did you get the SD card working with
CP/M on your 8080 board. Any idea how "fast" it is relative to a CF card
for data access. I'm worried the interface is in fact slower that an
8255A. Perhaps a large Flash RAM board would be a better approach. What do
you think.



John