[vcf-midatlantic] portable terminal preferences

Ethan Dicks ethan.dicks at gmail.com
Mon Jan 4 05:48:56 UTC 2021

On Sun, Jan 3, 2021 at 10:43 PM Neil Cherry via vcf-midatlantic
<vcf-midatlantic at lists.vcfed.org> wrote:
> On 1/3/21 10:01 PM, Jameel Akari via vcf-midatlantic wrote:
> > Bringing us to USB-to-RS232 dongles, on which I’ll have to disagree, or at least say that your mileage may vary considerably.  If
> > you find one that works fine for you, it really opens up your options.
> I have to agree with Herb on this one. I have friends who work on Railroad equipment
> that needs RS232C (supports +/- 25V). We've seen a lot of equipment that doesn't work
> well with USB adapters because they can't source the current. I do think the good
> PL-xxxx RS232C dongle work but the lesser ones fail miserable. If you drop a line
> driver in between it then works.

I agree as well... about 10 years ago, I had the occasion to try to
bring a Bridgeport Series II CNC to life.  Inside the Bridgeport was a
real DEC LSI11 CPU  board mixed in with all the Bridgeport boards.  I
was handed a sketchy hand-build Tyco-round-to-DE-9 cable and told "we
can't figure out why the CNC won't talk".  After I built my own cable,
it still wouldn't talk to my laptop.  I did some digging and I forget
exactly what symptom pointed to the USB serial dongle we were all
using with our laptops, but something did.  I pulled out a desktop
machine, plugged in the DE-9 cable I had made, fired up minicom and
was able to push G-code to the CNC and move this massive CNC around.

In the end, it was voltages... the mid-70s EIA circuit in the
Bridgeport would not talk to any serial dongle, but worked fine with
an ordinary "real" serial port (75188 drivers, IIRC) on a modern

Additional to this, I find myself sometimes needing to talk to very
slow devices (below 300 bps, which is often the slowest divisor in the
USB serial dongles).  I do have gear at 45.45bps and 75bps and 110
bps.   *Some* USB serial dongles have tools that let you overwrite
entries in the factory divisor tables (600bps is frequently chosen
because that speed has very little practical use with real devices).
I have not done any such hacking, but I do know it's possible.

What does work is real UARTs.  Even the UARTs in AVR microcontrollers
support all the old standards including 5-bit words and 2 stop bits.
Sometimes there's divisor underflow for really slow devices, but you
can clock the MCU at 4MHz and reach those slower speeds.

If all you want to do is talk to stuff made since the 80s with 9600N1,
most of this doesn't apply.  If you want to talk to devices made in
the 60s and 70s, you really want a "real" serial port.

I've kind of wanted to take an old laptop (with a real DE-9 serial
port and a UART) and replace the BIOS code with a dedicated firmware
implementation of a VT100, but I've never wanted it badly enough to
make one.  I did gather the elements for one of Vince Briel's
Propellor-based dumb terminals, but that only gets you the CPU.  You
still have to haul around (or borrow) an LCD at the destination site.


More information about the vcf-midatlantic mailing list