On 2013-01-17, at 11:00 PM, Dave McGuire <mcguire at neurotica.com> wrote:
What sort of hardware interface would you envision?
Two ports - Sync and Ethernet. Maybe a couple of status LEDs.
What do we need for config? Local IP address, Remote IP address, Baud rate (because we'd need to generate clock if I'm not mistaken).
How do we get parameters in? Web interface or telnet CLI ? Maybe DHCP extra parameters (no need to write a GUI, but some people will find configuring their DHCP server to serve additional parameters very difficult).
UDP means not having to worry about master and slave.
Your thoughts?
Ian
On 01/18/2013 01:37 AM, Ian McLaughlin wrote:
I'd do it with an HDLC/SDLC-capable chip like a Z8530...offload as
much from the firmware as possible, leave those cycles for
protocol handling. Unless you meant wrapping bare ?DLC frames in
packets?
I was thinking that we don't even need to decapsulate the serial
frame at all - just detect the start and end. All we're doing is
emulating a piece of wire, after all. Unless we're putting in a
fancy LCD status display, but I think that's overkill.
Ohhhh ok, that's an interesting idea; make it a "dumb" transparent
bridge. I was thinking we could do interesting translational stuff at
the protocol level, but one can't beat the utility of a dumb
get-bits-from-here-to-there bridge. I like this idea!
I have that problem as well, but I got really excited when you
mentioned it, because I am doing a lot of designs with Ethernet
now. I have a "base" design, a "hardware macro" if you will, that
I've recycled several times. It's a Philips..erm, NXP
LPC2300-series ARM7 with an on-chip Ethernet MAC, and all
supporting hardware. I typically run FreeRTOS and uIP on them.
The only things I have dev platforms are on my end are PIC and Atmel.
I do a lot of AVR stuff for work, but mostly that's maintenance of
previous controllers. All of our current stuff (and my personal stuff)
is ARM7 now. If (and I'm not pushing, or even suggesting!) you ever
want to get into it, I can give you a big head start.
I have some Zilog stuff, but I gave up on them several years ago.
Same here...after a significant investment in time, code, and blank
PCBs. Their management is incredibly, unbelievably bad. I used the
eZ80F91 in several commercial products. Great chip. Crappy tools.
Crappy company management.
However, if you're able to put together the hardware layer, and
there's a C compiler available for your platform, I can help on the
software side. I've also done some software work for a local company
that uses a TI Davinci processor.
I am, and there is, that would be great, and...wow that's cool! :-)
What sort of hardware interface would you envision?
Interfacing an ?DLC engine to that would be a cakewalk starting
from that base design as a head-start. I have a few tubes of
Z8530s in PLCC-44 packages here.
Oh oh. I think there's another project starting... :)
Oh there are just too many. ;)
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
On 2013-01-17, at 10:48 PM, John Wilson <wilson at dbit.com> wrote:
Careful -- UDP doesn't guarantee in-order delivery.
How does HDLC feel about little surprises?
HDLC information and supervisory frames have sequence numbers. There's also unnumbered frames used for control purposes.
The commercial units I found used UDP as their ethernet transport.
Ian
On 01/18/2013 01:48 AM, John Wilson wrote:
I'd do it with an HDLC/SDLC-capable chip like a Z8530...offload as
much from the firmware as possible, leave those cycles for protocol
handling.
FWIW, I'm in the middle of writing a Z8530 driver for E11 (in heavily
commented x86 code), which I'd be happy to contribute (after modifying
as needed) if you aren't already neck-deep in SCC code.
I built a Z8530-based AX.25 PAD for ham packet radio...but that was 25
years ago! I remember being thoroughly impressed with the Z8530, and
I've used it in several projects since then...but only in async mode.
So yes...if you're doing sync stuff with it, that might be very handy if
we do this and it goes this route.
We could certainly do the framer in firmware; there's a certain appeal
to that...but I'm a big fan of distributed processing, and as the iOS
device saying goes, "there's a chip for that".
Wrapping the data in UDP is simple. The HDLC layer could deal with
error detection/correction.
Yes!
Careful -- UDP doesn't guarantee in-order delivery.
How does HDLC feel about little surprises?
I would imagine it'd be offended. ;) However if memory serves, HDLC
frames contain a few bits' worth of sequence numbers. These could be
checked and buffered/re-issued to recover from most bad situations, I
would think.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
From: Dave McGuire <mcguire at neurotica.com>
I'd do it with an HDLC/SDLC-capable chip like a Z8530...offload as
much from the firmware as possible, leave those cycles for protocol
handling.
FWIW, I'm in the middle of writing a Z8530 driver for E11 (in heavily
commented x86 code), which I'd be happy to contribute (after modifying
as needed) if you aren't already neck-deep in SCC code.
Wrapping the data in UDP is simple. The HDLC layer could deal with
error detection/correction.
Yes!
Careful -- UDP doesn't guarantee in-order delivery.
How does HDLC feel about little surprises?
John Wilson
D Bit
Think of this as a notification system and text message extension to VMS PHONE :)
PHONE HELP MAIL ?
That doesn't give you an overview of everybody logged on anywhere, an IM system does.
This is more of a system to see who's online rather than to necessarily communicate with them, PHONE etc are much better, which is why I should where the "buddy" is logged in.
Sampsa
This is way off-topic so I will be brief.
Does anyone here know if it's possible to connect an IBM bisync device
(like the host interface on a 3174-61R) to a serial interface on a Cisco
router, and talk it into bridging to Ethernet in whatever protocol(s)
the Ethernet-attached 3174s speak?
Thanks,
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
Half a year or maybe a year ago. Sampsa was running a program which used PHONE to check who was logged in on various machines. When he tried all machines in area 1, MIM got loads of entries in the log for Sampsa's machine trying to talk with machines that were down.
(I might be remembering things wrong, and I might be mixing things up, but it was not as if MIM crashed or anything because of Sampsa's code, but I asked him to stop, as he was filling up plenty of logs as far as I can remember.)
Yeah, I didn't mean crashed MIM or anything, but caused problems - basically the phone directory just went through the nodelist and did a PHONE DIR to them..If a node was down, the call would fail.
It wasn't a good idea from BQT's side :)
On 2013-01-17, at 10:20 PM, Dave McGuire <mcguire at neurotica.com> wrote:
I'd do it with an HDLC/SDLC-capable chip like a Z8530...offload as
much from the firmware as possible, leave those cycles for protocol
handling. Unless you meant wrapping bare ?DLC frames in packets?
I was thinking that we don't even need to decapsulate the serial frame at all - just detect the start and end. All we're doing is emulating a piece of wire, after all. Unless we're putting in a fancy LCD status display, but I think that's overkill.
Wrapping the data in UDP is simple. The HDLC layer could deal with
error detection/correction.
Yes!
The problem is finding surplus time to embark on this. Maybe a few
of us could collaborate on it?
I have that problem as well, but I got really excited when you
mentioned it, because I am doing a lot of designs with Ethernet now. I
have a "base" design, a "hardware macro" if you will, that I've recycled
several times. It's a Philips..erm, NXP LPC2300-series ARM7 with an
on-chip Ethernet MAC, and all supporting hardware. I typically run
FreeRTOS and uIP on them.
The only things I have dev platforms are on my end are PIC and Atmel. I have some Zilog stuff, but I gave up on them several years ago. However, if you're able to put together the hardware layer, and there's a C compiler available for your platform, I can help on the software side. I've also done some software work for a local company that uses a TI Davinci processor.
Interfacing an ?DLC engine to that would be a cakewalk starting from
that base design as a head-start. I have a few tubes of Z8530s in
PLCC-44 packages here.
Oh oh. I think there's another project starting... :)
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
---
Filter service subscribers can train this email as spam or not-spam here: http://my.email-as.net/spamham/cgi-bin/learn.pl?messageid=410C8376613711E2A…
On 01/18/2013 01:14 AM, Ian McLaughlin wrote:
I'm not working on anything now - I just came up with the idea.
I meant "do you design stuff with microcontrollers".
I did some Googling, and there seems to be commercial units, all around
$1K per end (async is much cheaper).
Good heavens!
Microcontrollers and digital electronics is another one of my
hobbies.
Excellent! Same here, though I also do it professionally.
Async serial is simple for a microcontroller to handle.
I've never actually looked to see if there's any synchronous
libraries out there.
I'd do it with an HDLC/SDLC-capable chip like a Z8530...offload as
much from the firmware as possible, leave those cycles for protocol
handling. Unless you meant wrapping bare ?DLC frames in packets?
Wrapping the data in UDP is simple. The HDLC layer could deal with
error detection/correction.
Yes!
The problem is finding surplus time to embark on this. Maybe a few
of us could collaborate on it?
I have that problem as well, but I got really excited when you
mentioned it, because I am doing a lot of designs with Ethernet now. I
have a "base" design, a "hardware macro" if you will, that I've recycled
several times. It's a Philips..erm, NXP LPC2300-series ARM7 with an
on-chip Ethernet MAC, and all supporting hardware. I typically run
FreeRTOS and uIP on them.
Interfacing an ?DLC engine to that would be a cakewalk starting from
that base design as a head-start. I have a few tubes of Z8530s in
PLCC-44 packages here.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA