Econet: Line jammed! (Solved)

[thecellartroll]

Today I have fitted a BeebMaster Econet kit to my Issue 4 BBC B.

I have KenLowe's econet bridge sitting on his hub, with a Pi3B. I have installed the bridge software as per the instructions and the software seems to be running correctly. There is a nice blue light on the bridge.

The Beeb recognises the clock, and displays the station number. But the good news ends there.

*I AM SYST returns (after a few seconds) a Line Jammed error. I don't really know where to start diagnosing this.

With my scope I can see that when the Beeb is connected to the bridge the D+ and D- lines are permanently high. Even if the Beeb is turned off. If the Beeb is not connected to the bridge then the D+ and D- lines are permanently low.

I have also tried a different Pi and different Pi PSU.

So probably I made an erse of fitting the upgrade kit, but is there a recognised procedure for hunting down the issue? There are a lot of solder connections to poke at.

The Beeb also has one of Marks 177x boards in, which is functioning fine.

[KenLowe]

Oh, that's interesting. Is that both a beeb and your Piconet that fail to communicate with your bridge?

Do you have anything that is able to communicate with your bridge?

Can you check that the hub termination is switched off. Also, if you split the bridge from the hub, and just connect your devices directly to the two bridge ports, does that make any difference?

The blue LED just indicates that the clock is running.

[thecellartroll]

Oh, that's interesting. Is that both a beeb and your Piconet that fail to communicate with your bridge?

Do you have anything that is able to communicate with your bridge?

Can you check that the hub termination is switched off. Also, if you split the bridge from the hub, and just connect your devices directly to the two bridge ports, does that make any difference?

The blue LED just indicates that the clock is running.

I’m not sure how to check whether the Pi is communicating with the bridge.

The termination is off on the hub and on on the bridge. I have just split the two and plugged into the bridge and the behaviour is the same.

I have the other econet module, so later on I can try it out in my A5000. I could plug it into the Master I guess, but I’ll need to find and burn the right ROM.

[arg]

You should look more closely at the voltages on the data lines rather than just 'high' or 'low' - my tool of choice here is a multimeter, but the scope should suffice.

With the system idle, you should see the voltage applied by the terminator. The critical thing is that D+ is about 0.4V greater than D- with the exact value being less important but around 2V. So 2.2V and 1.8V or something close to that. If both are very close to zero or both above 3V then something is wrong.

Start by measuring just the network with no stations (in your case, you've got a station combined into your source of clock/termination so you can't unplug that, but at least unplug the BBC.

Now start plugging in the station(s). If plugging in an idle station affects the network, something is wrong with that station.

In a situation where it had been working previously and now not, the 99% certain cause would be the 75159 line driver chip (IC93)blown up and needing replacement; in your case there's scope for more complex faults, but unplugging IC93 is still the place to start if plugging in the station affects the rest of the network. Conceivably there could be something wrong around IC91 that enables IC93 - IC93p9 should be low.

If the station isn't jamming the network but is just internally giving a false report of Line Jammed, then there's other things to look at. To transmit, the station requires that RxD on the 6854(pin3) is high for 15 consecutive bit times. On an idle network, it should be permanently high due to the termination voltage that we've just checked out. So if it is stuck low or bouncing up and down, then there's something wrong around the line receiver circuit (IC94 - but we know IC94 isn't totally failed as the other half of it does the clock and that's working). R47/R48/C18 are supposed to generate a constant bias voltage of about 2V. All the inputs of IC94 are then biassed to about this voltage, with each input of the comparators in IC94 getting one of the clock or data lines via a 10:1 attenuation (R38/39/40/41)

[thecellartroll]

Thanks @arg! That’s a lot of useful information and I’ll start testing when I’m next off work

[maniacminer]

I just had a freshly made Beeb with "Line jammed" and found I had inadvertently bent pin#3 on the 68B54 underneath the IC. I checked all over the analogue circuitry and couldn't find the problem, then pulled the IC for the ah-ha! moment.

[thecellartroll]

I just had a freshly made Beeb with "Line jammed" and found I had inadvertently bent pin#3 on the 68B54 underneath the IC. I checked all over the analogue circuitry and couldn't find the problem, then pulled the IC for the ah-ha! moment.

It was actually a PITA to fit that into its socket but I'm pretty sure I didn't bend any pins. I may extract it and try again though

[maniacminer]

I just had a freshly made Beeb with "Line jammed" and found I had inadvertently bent pin#3 on the 68B54 underneath the IC. I checked all over the analogue circuitry and couldn't find the problem, then pulled the IC for the ah-ha! moment.

It was actually a PITA to fit that into its socket but I'm pretty sure I didn't bend any pins. I may extract it and try again though

Get an IC leg bending tool, it's so handy for making the legs "just drop in" (provided you're not trying to do it in almost darkness - like I did!)

[daveejhitchins]

I found one (there were two, actually - read-on) of the 'bearing' type, catering for both 0.3" and 0.6" parts, on eBay. So snapped it up for £15. Darn if it didn't get lost in the post. Had it refunded and bought the second one (last) which did arrive. They work so well

Dave H.

[maniacminer]

I found one (there were two, actually - read-on) of the 'bearing' type, catering for both 0.3" and 0.6" parts, on eBay. So snapped it up for £15. Darn if it didn't get lost in the post. Had it refunded and bought the second one (last) which did arrive. They work so well

Dave H.

The only problem with bearing type straighteners is they don't work with board pulls. I use the bearing type for ICs coming out of a tube and going into my stock box and the palm of the hand crusher for board pulls, like the 30+ ICs pulled from each of the six Music 500s - I've got a good grip

[daveejhitchins]

I found one (there were two, actually - read-on) of the 'bearing' type, catering for both 0.3" and 0.6" parts, on eBay. So snapped it up for £15. Darn if it didn't get lost in the post. Had it refunded and bought the second one (last) which did arrive. They work so well

Dave H.

The only problem with bearing type straighteners is they don't work with board pulls. I use the bearing type for ICs coming out of a tube and going into my stock box and the palm of the hand crusher for board pulls, like the 30+ ICs pulled from each of the six Music 500s - I've got a good grip

But they're brilliant for production work! I've broken 3 of the plastic ones, so this one should last a while longer. I'll have to get Mark Hillier to publish the .stl files for the one he uses - it looked good.

Dave H.

[maniacminer]

Someone made one for me, the STL files are already out there. I had a steel Texas Instruments branded leg former, but it vanished in the museum years ago. It was just a series of fluted channels and you just dragged the IC through it, small and simple.

[1024MAK]

Can we have less quoting of entire posts please.

It makes it less easy for members that browse the using their 'phone.

Either edit what you are quoting to only the relevant part, or just use the reply button instead of the quote button.

Also, if you wish to continue discussion of IC pin tools, it would be better in its own topic.

Mark

[maniacminer]

Can we have less quoting of entire posts please.

It makes it less easy for members that browse the using their 'phone.

Either edit what you are quoting to only the relevant part, or just use the reply button instead of the quote button.

Also, if you wish to continue discussion of IC pin tools, it would be better in its own topic.

Mark

Ok I've edited the post. Amusingly, I was using a phone browser and find it quite tedious to remove the nesting of comments, it's much easier on the desktop/laptop. Also I find I don't get any notification to my comment unless I've made quoted post.

P.S. as a mod can you move the bottom chunk of this onto an OT thread?

[thecellartroll]

You should look more closely at the voltages on the data lines rather than just 'high' or 'low' - my tool of choice here is a multimeter, but the scope should suffice.

OK so: (tested with multimeter)

Voltage from the Pi Bridge, turned on and running but not connected to the Beeb:
Data+ 2.56v
Data- 2.07v

Voltage from the Beeb with no econet cable to the bridge:
Data+ 4.91
Data- 4.95

Voltage from the Beeb when connected to the bridge:
Data+ 2.6
Data- 2.13

Should the voltage for the Beeb be so high with no cable attached?

[arg]

Should the voltage for the Beeb be so high with no cable attached?

No.

That suggests either IC93 blown up (common on machines pulled from the field, unexpected on a newly-upgraded machine), or problems around R47/R48 etc.

IC93 probably unlikely in this case as it isn't forcing the voltages when connected to the network, but might as well unplug it and test again just in case.

Then measure the voltage at the junction of R47/R48/C18 relative to GND - this should be exactly 2V.

If that is OK, remove IC94 and re-measure the inputs.

If still no joy, must be some wrong value resistors around there.

[thecellartroll]

OK, I will start to test shortly.

I have fitted Econet to my Master and it talks to the Pi Bridge just fine, so it is definitely I who stuffed up the upgrade somehow.

[BeebMaster]

Also check that S2 is completely broken and that there is no hint of continuity between the two pads.

I just had some very unusual behaviour testing with S2 closed on my Issue 2 BBC B. Normally this results in constant activity on the two data lines, a Line jammed error on any other station trying to transmit, and the BBC B never responding to a network transmission.

However, I found that if another station is transmitting, and then I attempt to transmit on the BBC B with S2 closed, I get a line jammed error on the Beeb. The other station carries on. When the other station has stopped transmitting, the BBC B still gives a line jammed error, although there is no activity on the data lines at all. This persists until the next reset when the original behaviour returns. So it's as if the ADLC goes into a state where it thinks the network is permanently jammed without actually having tried to transmit, and is cured by resetting the ADLC (ie. by pressing BREAK).

Possibly a similar thing is happening here, with S2 being a bit intermittent and wavering between open and closed, and confusing the ADLC in the way I just described?

[thecellartroll]

If still no joy, must be some wrong value resistors around there.

I have checked all of the resistor values and they're all correct according to Beebmaster's instructions.

Also check that S2 is completely broken and that there is no hint of continuity between the two pads.

To be sure I removed the board from the case and desoldered the remains of the wire link so there is now nothing left for continuity.

However... Doing that I seem to have broken something else as the board has now returned to a continuous beep and no life. That's how it was when I first met it. I'm not sure that I have the patience to diagnose and repair it again

[arg]

Then measure the voltage at the junction of R47/R48/C18 relative to GND - this should be exactly 2V.

Did that really come out at 2V with the pins on the network connector showing 4.xxV ?

[thecellartroll]

Did that really come out at 2V with the pins on the network connector showing 4.xxV ?

I don't know. The board became a corpse before I checked.

[thecellartroll]

Did that really come out at 2V with the pins on the network connector showing 4.xxV ?

*ignore this

[thecellartroll]

Did that really come out at 2V with the pins on the network connector showing 4.xxV ?

Fixed the beep of death (user error!) and some more testing:

With the beep conected to the network:

IC93 in place: D+ 2.44 D- 2.00 Junction 1.85
IC93 removed: D+ 2.44 D- 2.00 4.95 Junction 1.84
IC93 &IC94 removed: D+ 2.44 D- 1.99 Junction 1.84

Beeb not connected to the network

IC93 in place: D+ 0.00 D- 1.68 Junction 1.85
IC93 removed: D+ 0.00 D- 1.68 Junction 1.85
IC93 &IC94 removed: D+ 0.00 D- 1.68 Junction 1.85

Therefore something has changed as I'm not seeing the same variance between connected and disconnected that I was before.


*Edited because I should have had my glasses on!

[arg]

IC93 in place: D+ 4.95 D- 4.95 Junction 1.85
IC93 removed: D+ 4.95 D- 4.95 Junction 1.85
IC93 &IC94 removed: D+ 4.95 D- 4.95 Junction 1.85

This looks like multiple faults. The junction of R47/R48 is just a division of the 5V supply in the ratio of those resistors - so should be 2V : 3V. The only way it can be low is if some of R34/R35/R45/R51/C18 are pulling it down; if the other end of all of those resistors were shorted to GND then it would come out at 1.61V. But in fact your measurement of D+/D- shows that those are if anything shorted high rather than low. Maybe C18 is bad (or in backwards?).

Then we have the D+/D- being high. Maybe useful to look at C+/C- as that's a simpler circuit.
But with IC93 and IC94 removed, the only route for current to pull those pins to anything other than that 1.85V mid-point is either Via R46/R36 (but R36 is meant to be 1M5, so shouldn't be able to have much effect at all), or via R64 (again 1M5), or via fault in IC95 (but it's again got 56K in series so shouldn't possibly be able to pull all the way up to 4.95V in competition with R34/R35 10K.

So: short somewhere around those resistors? Accidentally used 1K5 instead of 1M5?

This is really a very simple circuit with those chips unplugged, so should be easy to debug.

[arg]

IC93 in place: D+ 0.00 D- 1.68 Junction 1.85
IC93 removed: D+ 0.00 D- 1.68 Junction 1.85
IC93 &IC94 removed: D+ 0.00 D- 1.68 Junction 1.85

That's looking a bit more plausible: there's a short to GND somewhere on the inside of R40/R41, both pulling the D+/D- to zero and causing the mid-point generator to come out low.

[thecellartroll]

That's looking a bit more plausible: there's a short to GND somewhere on the inside of R40/R41, both pulling the D+/D- to zero and causing the mid-point generator to come out low.

More testing based on this:
viewtopic.php?p=72570#p72570

IC93 Pin 12 > GND 100K (80K if leads are inverted)
IC93 Pin 13 > GND 109K (93K if leads are inverted)
IC93 Pin 12 > Pin 13 200K
IC93 Pin 12 > +5V 87K (100K if leads are inverted)
IC93 Pin 13 > +5v 93K (109K if leads are inverted)

With IC 93 removed:

IC93 Pin 12 > GND 100K (100K if leads are inverted)
IC93 Pin 13 > GND 109K (109K if leads are inverted)
IC93 Pin 12 > Pin 13 210K
IC93 Pin 12 > +5V 100K (100K if leads are inverted)
IC93 Pin 13 > +5v 110K (110K if leads are inverted)

R34 580
R35 9.1K
R36 1.1K
R40 100K
R41 100K
R59 56K
R60 56K
R62 30K
R63 30K
R64 30K

R34, R35 and R36 seem wildly different to the figures given in that thread.

[1024MAK]

In circuit testing of values like resistance may be affected by parallel paths that the test current from the multimeter can take (including via power rails). For example, two 1000 ohm resistors wired in parallel will each test 500 ohms.

Which issue board?
Have you compared the resistor colour codes with the required values and positions?
Have you inspected all the solder joints/pads?

Mark

[thecellartroll]

Maybe C18 is bad (or in backwards?).

Could C18 be backwards? Conveniently the + sign is right in the middle so I went by leg length.

[thecellartroll]

Which issue board?

Issue 4

Have you compared the resistor colour codes with the required values and positions?
Have you inspected all the solder joints/pads?

Yes, I have checked the colour code of each resistor. I find it impossible to tell which end is which when the tolerance band is brown and there's no gap, but assuming that Beebmaster read them correctly when he put the kit together the resistors are in the right place.

I have also been going over the soldering points with my microscope and I can't find any problematic looking ones. I've done a lot of soldering on this board already, and apart from a few vias that were filled by some crusty crud that needed cleaning out, it hasn't been too bad to work on.

[1024MAK]

I find it impossible to tell which end is which when the tolerance band is brown and there's no gap <snip>

That's when you have to compare to the "standard" values. Normally the E12 range is used. If it's not a value from the E12 range, it's possible that it's from one of the larger ranges or it's being read backwards...

E12 Standard Resistor Series
1.0
1.2
1.5
1.8
2.2
2.7
3.3
3.9
4.7
5.6
6.8
8.2

See this web page for more details

assuming that Beebmaster read them correctly when he put the kit together

So far, for the kits that I have installed from him, everything has been correctly marked.

Mark