Econet

[marcelaj1]

Tomorrow I will put my scope on the tx line and see if there is data going out, I assume it is sending out a "hello" packet.

[ChrisO]

I'm having some similar issues with my 2 BBC set-up, one is reporting 'not listening' the other 'net error' when I send a notify. Looking at the data lines the 'message' is very short (just a few bits) from the one reporting 'net error' so I guess its aborting the transmission for some reason ?

Both are running DNFS 1.20/3.60.

Had enough for today, back on the case tomorrow.

[BeebMaster]

"Net error" usually indicates some sort of fault or other on the network. It's difficult to pin down as it can be caused by many things, including faulty interfaces, dodgy wiring or terminators or clock, duplicate station numbers. Unfortunately everything will have to be checked! Even if you add a perfectly good station but the clock speed is no longer sufficient to run the whole network can cause it, so maybe try slowing down the clock speed. Also make sure that the earth lines are connected as well as the 2 clock and 2 data lines.

If the network is working correctly but there is no communication possible between stations then you will get "Not listening". This means either the receiving station doesn't exist, or it is not running as a file server or its expected receive port isn't open. If you get "No reply" then that means that the receiving station acknowledged the packet from the sending station, but didn't send any other reply. Usually that happens with a file server where the remote station's request causes some sort of file server fault, like a disc error.

If it's a two station network you may be able to narrow things down a bit by just having one station connected to the clock only, and see if you get anything other than Not listening errors when you try to transmit, then try the same with the other station. If the Net error returns when both stations are back on the network then it suggests, although it isn't conclusive, that the fault is in the wiring between the stations, including the clock and any terminators.

[marcelaj1]

OK need to wrap my head around this...

I have a cobbled together 555 based clock running at about 130Khz with a 20/80% mark space ratio. The "clock" is supplying d.c. biasing, this is directly connected to my beeb via the econet port - would I need a terminator on this setup, its a bit like having a beeb plugged directly into a filestore, if the terminator is required where would it go? would I need to cobble together a terminator/biasing circuit?

I read an article that suggests I can use midi cables anyone tried this?

[BeebMaster]

I commend you for trying to cobble together an Econet, that's how I started with an issue 4 BBC B which eventually became, and still is, Station 128, although I did have an actual Clock. My first Econet lead was the wire from a mouse with bits of paper clip soldered onto the wires and then poked into the Econet socket.

5-pin DIN 180 degree leads are often described as "Midi" leads, they will work as long as they are wired straight through pin for pin, and not cross wired.

If you are after instant Econet wiring infrastructure and terminators then this may help.

[marcelaj1]

I think what I will do is commit the clock to veroboard as its a bit unstable on the breadboard and then if I can prove to myself that the on board stuff works build a proper one maybe using the ATtiny45 that was used in other posts - my thinking is to use readily available modern day (cheap) components rather than using recovered originals.

[ChrisO]

I've made some progress with my setup. No joy with the 2nd BBC, the window comparator that drives CTS is producing erroneous pulses during a transmission. Thats whats causing the Net Error - can't find anything wrong with the board though so a bit of a mystery ?

I'm using an original Acorn Clock (2-tone grey Vero box) with the active terminator also enabled, data and clock look good on a scope.

I then turned to one of my System computers, installed NFS3.40 switched on and it seems to work ! I get an OK from the notify program running on the BBC. I thought notify would actually display the message sent to the System computer but that doesn't seem to happen ? If I send a *notify from the System computer to the BBC I get a Busy message ?

Anyway some small progress, work will now get in the way for a couple of day so maybe back to it later in the week.

[BeebMaster]

The original Acorn Clock and terminators share the same board, but I don't think both sets of components are supposed to be fitted to the same one. Could that be causing a problem?

[IanS]

The original Acorn Clock and terminators share the same board, but I don't think both sets of components are supposed to be fitted to the same one. Could that be causing a problem?

No, the different parts only share a power supply. It would meean the combo unity would need to be at one end of a (large) network, as it now includes a terminator/line biasing.

[marcelaj1]

Just when I thought I had it sorted in my head I go and read the "Acorn Econet Advancd User Guide". Page 78 has a cct diagram of a clock/terminator box. It shows that the clock is generated by a "MC14411 Bit Rate Frequency Generator" according to the spec' sheet it boasts of a 50% duty cycle. I thought that we did not want a 50% duty cycle as that restricts the data transmission window or is that irrelevant and I was being sidetracked.

[IanS]

Just when I thought I had it sorted in my head I go and read the "Acorn Econet Advancd User Guide". Page 78 has a cct diagram of a clock/terminator box. It shows that the clock is generated by a "MC14411 Bit Rate Frequency Generator" according to the spec' sheet it boasts of a 50% duty cycle. I thought that we did not want a 50% duty cycle as that restricts the data transmission window or is that irrelevant and I was being sidetracked.

The early clocks were 50/50 mark/space ratio. It was the later (SJ research design?) clocks that had configurable mark/space options. On a small (test) network either will be fine.

[arg]

I've made some progress with my setup. No joy with the 2nd BBC, the window comparator that drives CTS is producing erroneous pulses during a transmission. Thats whats causing the Net Error - can't find anything wrong with the board though so a bit of a mystery ?

If you are using system machines, check that they have the right component values (for example, the feedback resistor in the circuit you mention should be 1M5 not 2M2). There were discoveries made in early 1982 which fed into the BBC Issue 4 board design, and, component choice at least, applied to all BBC micros as none had been built with econet by that stage (issue 2/3 BBCs with Econet will have been upgraded subseqently). However, there were lots of System/Atom interfaces already built by that stage with the wrong values, and several years later I encountered more (being built by a 3rd party rather than Acorn) that were still using the original design.

Even with the right components, there's an issue with the layout on BBC iss 2/3 - with the resistors mounted vertically as they are, you can get crosstalk from one resistor to another, particularly the high impedance node around those 1M5 resistors. Performance (getting rid of erroneous pulses on CTS) can be improved by carefully bending the resistors apart. Iss 4 lays them flat and without 5V swing signals adjacent to high impedance analogue nodes.

[ChrisO]

I've just had a quick look at both of my BBC's, they are both Issue 7 and are fitted with 1M5 flat on the board so sadly thats not it ? I also had a quick look at the System board schematics and you're right that some version schematics show 2M2.

[BeebMaster]

I've made some progress with my setup. No joy with the 2nd BBC, the window comparator that drives CTS is producing erroneous pulses during a transmission. Thats whats causing the Net Error - can't find anything wrong with the board though so a bit of a mystery ?

That sounds like one of the LM319s has gone faulty, have you tried replacing them?