OT: Small / light NOT expensive PC with RS-232 port

Its near the bottom of the linked page.

hamilton

Put an ad in your local classifieds/ craigslist/kijiji to see if anyone has a win95/98 disk lying around...

I'd be surprised if the OS doesn't had that from applications. Did you look at the other options I posted. I didn't check the specs, but they may have built in RS-232.

It would help a lot, if the actual requirements would be known.

Unless some real time kernel is used that runs Windows/Linux as the lowest priority NULL task, bit banging is out of question for any real data rates above 300 bit/s.

However, if you need accurate data direction control in half duplex RS-485, at least some Ethernet/serial converters work quite well, but of course, even those do not exactly handle the Modbus 1.5/3.5 character time issues, but in most cases, it is not even required in practice.

Many Windows applications can be run in Linux. Look up wine and mono.

On Fri, 31 Jan 2014 22:05:08 +0100, Robert Roland Gave us:

WITH direct hardware access?

Copeth thyselfeth a clueeth.

On Fri, 31 Jan 2014 04:59:39 -0700, hamilton Gave us:

Doesn't look nice. Is 2 years old obsolete.

Everything now has USB 3. Lack thereof is an indication of years old HW design, even if the PC 'product' is only a year out the door.

On Fri, 31 Jan 2014 21:45:28 +0200, snipped-for-privacy@downunder.com Gave us:

You probably went way over his head.

Strange that he will not say (or has not said) what he is doing with it that makes a hard connection to it a requisite. He talks about a Windblows only software that he has to hook to it with.

Mutable... Like Putty.

I suspect that you will have to run your guest OS and application in a virtual machine.

?-/

version

luck.

OP (Robert Baer) has repeated stated that he needs to run legacy software that talks directly to the hardware port.

?-)

He may be out of luck - the classig 8250 / 16450 / 16550 -style serial port begins to be unobtanuim.

Some virtual machine implementations can do it in realtime. I recall an excellent emulation of the 16550 UART for OS/2 that would run on bog standard RS232 hardware and work at the then maximum speed 230kbaud ISTR as a bonus you could set the FIFO to more than 16 bytes as well.

At the time it was a demo of why OS/2 was better than Doze. Unfortunately the best technology seldom wins out over marketing clout.

On Sun, 02 Feb 2014 21:11:28 +0000, Martin Brown Gave us:

I had a multi-thread use example applet for my OS/2 I ran that showed how it ran multiple process threads more smoothly.

It was a definitely closer to the bone, more compact base level kernel. Not hard to beat Windows with that said.

You may well be right. For a single channel. Would one be able to synch up five machines via such ports, reliably?

Doesn't GIVEIO work ? It essentially maps the hardware I/O pages to the user mode virtual address space memory map ?

I think these programs simply remove the OS block so you can access the registers from user space. But, if you're at the mercy of the usb/rs232 interface, that doesn't help restore registers that aren't there. But, there are differences among the converter dongles that may allow some sort of control over the control lines...maybe...depends.

The task is still undefined. Running legacy software that talks directly to the hardware port leaves a lot of variability in exactly what is needed.

If it were me, I'd plug in a dongle, install the os unblocker software and see if i could peek/poke the registers needed. They may be simulated...maybe.

Unfortunately IBM screwed the (delayed) launch of OS/2 by conflating it with the PS/2 with their proprietory MCA lock-in architecture which galvanised the rest of the PC makers to produce the EISA standard.

Main reason for doing it was to keep up with the dialup modems of the day which due to ever increasing realtime compression ratios could handle the empty file (and some Usenet text files) at rates which would go haywire with overruns without a decent length FIFO.

Moschip MCS9904 is current, inexpensive, and apparently programattically fairly close.

It seems that the OP's code needs an interface that looks to the programmer just like the IBM PC Serial Adapter, with all the quirks.