Here we go again:

I'm researching a 486-able target for MASM development. This could be an embedded 486DX2 micro with a DMA controller. I still need to determine the DMA controller and suitable OS kernal methods to R/W memory and ports in the SUPER I/O chip or just plain 16555 UARTS.

I'm leaning away from the 368-PGA device at ZFMicro because I want to actually draw and etch my own board with 12-16 mil traces! I use the TONER TRANSFER method from Digikey to get 16 mil traces HIWIRE II and autorouter to find connections and recheck my work after manual routing. Creating the artwork is not the problem. Its whether it can be done with a reasonable number of component-side jumpers on a single-side board. My 2-sided masterpieces are shy of that "professional results" descriptor. Single-sided boards are getting pretty good though. A 168-pin PGA seems OK for my process but it will be pushing it.

The DMA hardware implementation, OS and KERNAL methods appear to be the most difficult. This project requires only the DMA and serial memory transfers with no keyboard, mouse, display, IRD, USB or printer. Just 2 115,200kbs serial ports.

Any suggestions on how I might actually implement such hardware with DMA controller, minimal system memory (some flashable via the serial port,) and serial ports?

-AV- : grin:

4th option in poll was to say yes, with the toner-transfer method.
Posted on 2002-03-20 10:33:39 by avrster
Arg, my head hurts.

What purpose would this embedded computer have?
Posted on 2002-03-20 13:58:23 by bdjames
Hi !

Aren't there enough of those high integrated micro-controllers on the market already ? At my work we build devices with AMD-microcontrollers ...

Greetings, CALEB
Posted on 2002-03-20 14:32:17 by Caleb
Smartcard emulation and WIN32 smartcard OS development.

:o

Its about the simplest hardware project you could have for a dedicated 486. Not much for peripherals and I/O.

To use a 486 in development, I'm under the impression that the BOOT EPROM gets loaded to the cache and executed.

EEPROMS are slow, and generally I don't believe there is a FAST EPROM that can be interfaced directly to the 486 address bus. I could be wrong though. Saw a Pentum II with the BIOS ROM all the way across the MB. A 486-80 had it right next to the processor though.

Also a serial port peripheral will be slow, and need some kind of controller to buffer signals between its latches and the processor.

If I could just put devices on the address lines and get valid response times from everything I would.

-
Posted on 2002-03-20 14:37:27 by avrster
I found PCI methods in WIN32 source for tinyBIOS an open source project.

There are a few open source developments out there. Its a matter of picking out a good AT 486 hardware implementation of the PCI DMA controller.

Even a good ISA 8732 implementation would be a good place to start, but if its available I'll implement the 33Mhz bus.

I'm still looking for the hardware schematic that implements the 32-bit PCI bus controller.

-AV

www.pcengines.com/tinybios.htm

Actually I learned that the tinyBIOS source is edited to be compiled by the A386 assembler by Eric Isaacson http://www.eji.com

So I'm naturally looking at how much work it will be to edit the source to compile with MASM32. I wonder if someone has already done this?
Posted on 2002-03-20 16:02:35 by avrster
All is not as it first seems.

Distribution of components supported by tinyBIOS is very tight. TinyBios is somewhat experimental, and people who want to use it would do well to first go out and find the correct hardware. Tiny BIOS supports ALi M1487, M1489, M6117 and the Winbond Winbond W83977EF super I/O chips that implement ISA or PCI bus interfaces to the local bus.

ALi is actully one of the most prolific of suppliers of these components and their products are found in many small 386 and 486 P_104 SBC cards selling today. But ALi does not post ANY information about the products on their website. This is troublesome. It could be that parts are in supply, but not encouraged for new designs.

Winbond is very elusive. Can't track it down.

So,, I'm keeping an eye open for ALi compatible PCI components and researching those which might be currently available in small quantities. (Hey its possible, someone might have a few too many.)

The A386 Assembler-Debugger at $190 seems like a pretty hot deal when you couple it with tinyBIOS. You think, "Hey this stuff might get easier..." Heh,, things change when you find out about supply problems.

As I find new PCI controllers, the immediate issue then becomes, "Can I convert the tinyBIOS methods to support these new controllers?" After all they use the same register locations on the local bus.

I could call up Insight or Hallmark or someone and get a sales person to pitch a product to me somehow... but I don't think that is my best option. There is always a backdoor to everything. You just have to FIND it.

The link below was made from an OS2 perspective it gives you an in dept perspective. Especially when you note that there is no information about ALi products, (almost a mystery.) Japan-based ALi (ACER LABS Inc.) has NO U.S. distributors! Their PCI and ISA product information is almost TOP SECRET.

http://www.os2forum.or.at/english/info/os2hardwareinfo/pci_chips.html

ACER LABS components do exist (there are photos of them.)

http://www.plasma-online.de/index.html?content=http%3A//www.plasma-online.com/english/identify/picture/acerlabs.html


OOPS!!! Just found Winbond. Web Ferret is wierd if you are searching for several words. I finally tried "Winbond" by itself and it popped up. If I was going to design a x86 ISA or PCI board, based on what I learned today I would recommend using a Winbond product for I/O and System control. It sortof looks like they are actually in business to sell a product.

http://www.winbond.com.cn/e-winbondhtm/index.asp

www.winbond.com.cn

Base specifications:

www.winbond.com.cn/e-winbondhtm/partner/PDFResult.asp?Pname=175

Search for 83977 or 83877 and there are two pages of data sheet links.

Looks like this is THE company that is trying to sell the I/O and system
control chips for x86 development in the U.S.

-AV- :grin:
Posted on 2002-03-20 20:59:00 by avrster