So... What I want to ask is whether there's an easy way of knocking up
some custom hardware that takes a RJ45 as input and spits out Econet
through a DIN at the other. The RPCs in question would only ever be
connected to the Econet, so there's no need to check for anything
other than AUN...
Hardware to talk to Econet is now hard to come by (ie, producing your
own PCB to do it.)
If you actually want to do this, I suspect the "nicest" way might be to
use a RISC iX box as a router.
B.
I recall discussing with someone the economics of producing 10-100 units and
the cost we thought was going to be to high at �40+
Chris Evans
--
CJE Micro's / 4D 'RISC OS Specialists'
Telephone: 01903 523222 Fax: 01903 523679
ch...@cjemicros.co.uk http://www.cjemicros.co.uk/
78 Brighton Road, Worthing, West Sussex, BN11 2EN
The most beautiful thing anyone can wear, is a smile!
> In article
> <8899167c-5b95-4976...@v36g2000yqv.googlegroups.com>,
> Gazza <URL:mailto:use...@garethlock.com> wrote:
> > I've just about given up trying to find Econet NICs for the two RPC
> > class machines we have at the club. Apparently these NICs are like
> > gold-dust. Anyway, from what I can make out, AUN is just Econet over
> > IP.
> >
> > So... What I want to ask is whether there's an easy way of knocking
> > up some custom hardware that takes a RJ45 as input and spits out
> > Econet through a DIN at the other. The RPCs in question would only
> > ever be connected to the Econet, so there's no need to check for
> > anything other than AUN...
>
> I recall discussing with someone the economics of producing 10-100
> units and the cost we thought was going to be to high at £40+
More than that, I'd have thought, these days. Perhaps 60 quid a
pop at a real squeeze.
You can most likely find a microcontroller with both 10Mbps ethernet
and the right serial standard on, but there's still the software and
PCB to design and test.
B.
I have not as yet spotted a microcontroller with a serial port capable
of the HDLC/SDLC type framing that Econet uses. (That is at least
amongst the Atmel AVR micros). I am on the look out for a candidate
for USB<->Econet.
I contemplated that an external GAL chip or the like might be able to
work in conjunction with a SPI port where the GAL does the zero
insertion and removal & start/end frame detection transparently to SPI
could work, but I have no pressing need and even less time.
Last time I thought about this the AVR chips were too slow to manage a
software emulation of framing that a 68B54 uses for Econet (at
256KBPS), and lacked the RAM necessary to buffer a econet packet,
which can be up to 8K I think. Newer chips may now have enough ram.
My Econet module in my RISC PC has not worked since RISC OS 3.7 or
so. Never with any version of 4 I don't think.
The unipod promised that it would be possible to fit an ADF10 (Master
128 Econet interface) but to my knowledge that promise was never kept,
certainly I have not had any success probing the podule bus looking
for the one plugged into mine. I suspect that the signals required
are simply not programmed into the fpga chips on the unipod.
Alan
> Last time I thought about this the AVR chips were too slow to manage a
> software emulation of framing that a 68B54 uses for Econet (at
> 256KBPS), and lacked the RAM necessary to buffer a econet packet,
> which can be up to 8K I think. Newer chips may now have enough ram.
Try looking at STM32s; they are available up to 72MHz and 20kB of RAM.
Still may need a CPLD or similar to handle the framing, though.
> My Econet module in my RISC PC has not worked since RISC OS 3.7 or
> so. Never with any version of 4 I don't think.
>
> The unipod promised that it would be possible to fit an ADF10 (Master
> 128 Econet interface) but to my knowledge that promise was never kept,
> certainly I have not had any success probing the podule bus looking
> for the one plugged into mine. I suspect that the signals required
> are simply not programmed into the fpga chips on the unipod.
There are no FPGAs on the Unipod :) The socket is certainly there, and
it does appear somewhere in the memory map. Somebody posted on Drobe
in the past year or so with some information that might be useful;
http://www.drobe.co.uk/article.php?id=2476
(Last comment.)
B.
Does Econet work on RO4 in general? ISTR the RPC Econet card had to
softload its own Econet module, which was modified to look at the netslot
rather than the Master-style Econet module. I think the Econet module
itself disappeared from RO ROM some time ago (3.x?).
I think ROL did some Econet-related changes in RO4 or Select, but I don't
recall them dropping Econet completely. If it's merely bitrot, perhaps a
bit of tweaking of the ROOL Econet module might be in order?
> > The unipod promised that it would be possible to fit an ADF10 (Master
> > 128 Econet interface) but to my knowledge that promise was never kept,
> > certainly I have not had any success probing the podule bus looking
> > for the one plugged into mine. I suspect that the signals required
> > are simply not programmed into the fpga chips on the unipod.
>
> There are no FPGAs on the Unipod :) The socket is certainly there, and
> it does appear somewhere in the memory map. Somebody posted on Drobe
> in the past year or so with some information that might be useful;
>
> http://www.drobe.co.uk/article.php?id=2476
Oooh, nice. That looks quite similar to the address map I deduced from
the Simtec USB card the other day:
http://www.riscos.info/index.php/Simtec_USB_technical#Hardware
Theo
> > There are no FPGAs on the Unipod :) The socket is certainly there,
> > and it does appear somewhere in the memory map. Somebody posted on
> > Drobe in the past year or so with some information that might be
> > useful;
> >
> > http://www.drobe.co.uk/article.php?id=2476
>
> Oooh, nice. That looks quite similar to the address map I deduced
> from the Simtec USB card the other day:
> http://www.riscos.info/index.php/Simtec_USB_technical#Hardware
Something tells me Simtec are creatures of habit >:)
B.
So... If this is correct, I could stick a VPod in the RPC and use the
regular LAN card in the A7000+. Hook the two together using a standard
hub/router and hook the Econet end to the RPC??
> So... If this is correct, I could stick a VPod in the RPC and use the
> regular LAN card in the A7000+. Hook the two together using a standard
> hub/router and hook the Econet end to the RPC??
No, nothing to do with V-Pods. It's the Unipod that has the 8 bit
expansion socket. There's certainly the possibility of putting an
Econet module in the Unipod, and writing some software to bridge the
Econet to an Ethernet, using the Unipod's NIC.
But there would be a lot of software work, and a lot of trial and
error, I suspect.
B.
Would that be a lot of software work? A5000s were used as Econet-Ethernet
bridges years ago, and the software exists. The difficulties about doing
that now might be:
Does the bridge software work on RO4?
Does the Econet stack work on RO4?
Does the bridge software talk DCI4 to the Unipod's NIC driver (and use the
Inet5 API)?
Can the Econet module be easily hacked to the new address (yes it can, IME)
That doesn't need more than a few instructions to be written, just needs
some luck that all the components will work together.
Theo
ok true, I simply read XILINX and made an assumption, XC9572XL is
really a CPLD.
Thank you for the link to the address map information.
Theo Markettos <theom+n...@chiark.greenend.org.uk> wrote:
> Does Econet work on RO4 in general?
From memory, and its quite some time ago that I encountered the issue,
on RISC OS 4 the Econet nic conflicts with the floppy drive when the
Econet clock box is turned on. I have just inspected my RPC and it
still has the Econet nic in it with RISC OS 4.39, but I don't have any
Econet or clock box set up these days.
*rommodules lists Econet 5.70 as from podule 8 and 'Dormant'.
There was discussion here at the time and it was been suggested that a
module could be removed from RISC OS 3.7 and soft loaded into 4, which
would resolve the IRQ problem, but the details escape me.
The Econet lead supplied with the Econet NIC is also correct for the
unipod with an ADF10 plugged in.
The dil header at the rear of the unipod seems to break out the extra
parallel port correctly with a old PC ISA multi IO card's parallel
port 25pin D to IDC lead. The parallel port is part of the NIC chip,
but I haven't actually managed to twiddle its bits. With the PC lead,
pin one with the stripe goes towards the USB connectors.
Alan.
> Would that be a lot of software work? A5000s were used as
> Econet-Ethernet bridges years ago, and the software exists. The
> difficulties about doing that now might be:
>
> Does the bridge software work on RO4?
> Does the Econet stack work on RO4?
> Does the bridge software talk DCI4 to the Unipod's NIC driver (and
> use the Inet5 API)?
> Can the Econet module be easily hacked to the new address (yes it
> can, IME)
>
> That doesn't need more than a few instructions to be written, just
> needs some luck that all the components will work together.
But of course, if you're unlucky, you've got a lot of software to
write. :)
(And I suspect strongly that the answers to some of those questions
will be "no".)
B.
> ajw wrote:
> > 256KBPS), and lacked the RAM necessary to buffer a econet packet,
> > which can be up to 8K I think. Newer chips may now have enough ram.
>
> An Econet packet can be infinitely long - that's why it has
> specific start and end markers. It's the transmitting and receiving
> programs that decide on what packet length to use for a particular
> communication.
AUN uses UDP, if I recall, so one assumes it'd be relatively simply to
produce a device that sent a UDP packet after every 16kB or so of
Econet packet that arrived.
B.
The FIQ setup was rejigged slightly for StrongARM, so that could be a reason
(details in the StrongARM appnote).
> *rommodules lists Econet 5.70 as from podule 8 and 'Dormant'.
>
> There was discussion here at the time and it was been suggested that a
> module could be removed from RISC OS 3.7 and soft loaded into 4, which
> would resolve the IRQ problem, but the details escape me.
http://www.riscosopen.org/viewer/view/castle/RiscOS/Sources/Networking/Econet/
is there if anyone feels keen ;-)
Hasn't been changed since 3.71.
> The dil header at the rear of the unipod seems to break out the extra
> parallel port correctly with a old PC ISA multi IO card's parallel
> port 25pin D to IDC lead. The parallel port is part of the NIC chip,
> but I haven't actually managed to twiddle its bits. With the PC lead,
> pin one with the stripe goes towards the USB connectors.
I'm slightly surprised that no-one was interested in adding extra parallel
ports. I added one on my prototype Super IO podule, but of course ran into
the problem that there was no software support for a second port (so I'd
have to bodge up a parallel2: device and tell people to manually configure
Printers, and things that hung off Parallel_Op wouldn't work)
I imagine from the Unipod and USB podule layouts that have been deduced so
far, it wouldn't take someone too long with *Memory to poke around and work
out the rest of the Unipod layout. I don't have one, I'm afraid.
Theo
> I imagine from the Unipod and USB podule layouts that have been
> deduced so far, it wouldn't take someone too long with *Memory to
> poke around and work out the rest of the Unipod layout. I don't have
> one, I'm afraid.
One could always try asking Simtec and/or Advantage Six for them.
B.
Is that the royal One? Since you work for them, perhaps you could ask
yourself? ;-)
Seriously, I'm quite happy to ask if the answers are likely to be
forthcoming. I've emailed STD about various USB-related things over the years
and never had a single reply from their advertised contact addresses (Matt
Edgar is helpful, when you know how to get hold of him).
Theo
> Rob Kendrick <nn...@rjek.com> wrote:
> > One could always try asking Simtec and/or Advantage Six for them.
>
> Is that the royal One? Since you work for them, perhaps you could ask
> yourself? ;-)
I'll ask Gareth in the morning :)
B.
It seems that he has two RPCs, which since he wants an Econet to
Ethernet bridge, I presume already have Ethernet interfaces with RJ45
ports.
If we knew more about what was on the Econet we may be able to find a
way to connect the RPC's. If there was any other ARM kit with Econet
and a spare podule slot then that machine needs an ethernet podule and
connection to the RPC via a switch etc. Candidates for this would be
A310, A410/20/40, A3000, A5000, A540, A3020 and A4000, though the last
two would need rarer parts.
It would also be handy to actually understand what was to be achieved.
The talk of bridging Econet and AUN suggests that arbitrary RPC needs
to talk Econet protocols to arbitrary machines on physical Econet. I
am struggling to remember if that was ever possible.
Rob raised the spectre of using RISC iX with Ether and Econet
interfaces, but that only gives you IP routing between devices on
ethernet with ip stacks, and devices on Econet with DCI2 IP over
Econet stacks, which means RISC OS or RISC iX machines. It does not
give you a Econet bridge between the two media types.
If in reality its only fileserver access that is required then a
server with both Econet and Ethernet interfaces will talk to AUN
Ethernet clients and physical Econet clients. Level 4 should do this,
awServer might too, I can't remember ever testing it though. Any of
the machines listed earlier could be used. An A540 or A5000 would
probably be my pick.
It would be tricky for the RPC to get access to native Econet file
server in this model though. (Filestore, Level2, Level3, MDFS etc)
Alan
> Rob raised the spectre of using RISC iX with Ether and Econet
> interfaces, but that only gives you IP routing between devices on
> ethernet with ip stacks, and devices on Econet with DCI2 IP over
> Econet stacks, which means RISC OS or RISC iX machines. It does not
> give you a Econet bridge between the two media types.
Assuming RISC iX exposes the Econet via the standard BSD sockets
interface (which I assume it does; it's been a while!), the program to
do the bridging is perhaps a couple of hundred lines of C at most.
B.
Well... We have an A310 that doesn't power up ATM. Don't know why this
is. The power light on the front comes on, but nothing else happens,
of the working machines, we have an A4000 and an A3020, the machine
I'm planning to use as the new FS is an A5000 with the seperate SCSI
media box stacked on top with a couple of LVD drives. One fitted in
the media box and another daisied via an external SCSI housing that
used to contain a dead CD-ROM drive that was donated to the club a
while back. Not sure what the capacities of these drives are, but I
suspect that they might be 18GB or so and will need partitioning to
work with RISC OS. The FS will be running LVL4 software.
> Assuming RISC iX exposes the Econet via the standard BSD sockets
> interface (which I assume it does; it's been a while!), the program to
> do the bridging is perhaps a couple of hundred lines of C at most.
True indeed.
> (which I assume it does; it's been a while!)
You are not kidding me! I think the first C program I ever wrote for
unix of any kind took stdin and fed it to an Econet print server. I
don't think it was sockets, I didn't learn about them until much
later. I suspect it was an ioctl interface. I also suspect that
'promiscuous mode' doesn't exist in the Econet stack, so to exhibit
generic 'bridge' behaviour might be quite tricky in an os that's not
open source. This is ringing very dim bells as something I might have
had to think about before, probably 20 years ago. I suspect that at
the time I figured it was too hard, and that application layer proxies
(as one would now call them) might be a more practical solution.
Apart from file and print there really isn't a lot else that's common
on Econet. Unfortunately a proxy for NetFS/NFS has to know nearly as
much about the protocol as an entire server does.
I remember now there are other issues with the way an Econet to Econet
bridge works, which are quite difficult to replicate in an Econet to
Ethernet bridge. The Econet bridge has to send flag fill out to the
source net while it forwards the packet and waits for a reply on the
destination net. This prevents any other transactions starting on the
source net while the bridged transaction is in progress. This would
be tricky if one side was Ethernet.
Alan
Anyone got a suitable ethernet card for either the A4000 or the A5000
that I'm planning on using as an FS, then the A5000 could take on both
jobs and the 4000 wouldn't have to be ON all the time. Also, I hear
that the A5000 hardware would be easier to come by. Please correct me
if I'm wrong.
Here's the ioctl interface described:
http://www.chiark.greenend.org.uk/~theom/riscos/docs/Acorn-Unix-Econet-Device-Driver-and-Network-Device.pdf
(I think it was JGH who originally scanned that, and I think he might have
more details somewhere on his site)
Theo
That would be great. However, that would mean that the A4000 would
have to be going to provide the bridge, unless these cards can also be
used on the A5000. The other question is whether I could use both a
Master 128 Econet (IIRC) and the Ethernet card in the same machine.
Particularly in the same A4000. I seem to remember that space in these
was a little tight.
Yes it is, you won't get an ADF10 in underneath the Ethernet
interface, you would need one modern Econet interfaces designed for
the A3020 and A4000 era machines.
OK... That's another thing I'm gonna have to put on the wish list
then... lol
> The A4000's essentially a single-slice A5000. All my A4000s have
> ethernet and Econet cards in them.
No, no no. Only in appearance. The A4000 is essentailly a two-box A3020. It
has the A3xxx type expansion socket whereas the A5000 has a 4slot 'podule'
backplane.
--
David Holden - APDL - <http://www.apdl.co.uk>
It's quite possible to fit ethernet and Econet to an A3020/A4000: Acorn
Econet card in the special network slot, then a standard ethernet minipodule
on top in the single expansion slot. But that's all your expansion options
used up.
Theo
That was another reason I was gunning for the A5000 part. Doing it
that way, the FS could also provide the bridge between the Ethernet
and Econet machines.