Fault finding index

[maniacminer]

Installed Econet into Issue 3 model A->B with no disc interface fitted. Econet showed the station number and returned to the BASIC prompt on boot. However, any attempt to access the network would result in the machine waiting indefinitely for a response. The fault lay in nINTON not being asserted at IC97 pin#2, despite being created on IC26 pin#6. Tracing the track from IC26 the signal was lost on the other side of the cassette motor relay. A visual inspection of the area showed someone had performed a very poor repair and destroyed the track. A wire has now been run to join the pins together and the network access problem has now gone.

Damage to PCB

Wire run to link two ICs

[maniacminer]

Put Econet into an Issue 4 board, everything on the board functions perfectly, except the newly installed Econet...

Initially the Econet wasn't being detected, nothing on the banner.

I checked the select and nINTON, STATID etc and they were performing normally. Oddly, the 68B54 wasn't creating any useful signals and it is a known good IC. I then checked each pin in turn against my reference and discovered that the +5V pin although connected to the IC and to the solder side of the board, was not connected to power.

A bit more poking around and I discovered that the flood area on the top of the board has no through hole plating, putting a multimeter from the top to the bottom is megohms. I then soldered the socket from the top as well as the bottom for that pin and yay, I have the Econet banner.

But wait, it's station 000 !?! Turns out that the same problem I had with the 68B54 I'm having with the station ID buffer IC, as pin #20 is floating, supposedly the +5V rail. With my tongue at the right angle, a blob of solder on that pin and finally, station 076 - yes!!!

This board is trash, the solder mask on the top is thin and patchy, the sockets are poor quality (although not the hideous white sockets), the tracks and pads are very thin and lift very easily.

Worse than ICL, I give you BEPI... anyone know who that is? It feels like Acorn squeezed suppliers hard to pinch every penny possible?

[atsampson]

Machine: BBC Master 128 with Astec PSU

Symptom: Previously-working machine didn't power up. There was about 1V on the 5V rail, so the PSU hadn't started up.

Solution: Replaced C101 (220µF, 10V) on the vertical daughterboard in the PSU.

[ferrisio]

Machine - BBC Model B iss 4

Symptom - machine running slow at approx 1/4 speed

Fault - faulty VIA 6522 IC3 causing spurious IRQ activity, thread here - https://stardot.org.uk/forums/viewtopi ... =3&t=26388

[hatman72]

Machine - BBC Model B issue 7

Symptom - When machine turned on, one low beep and the screen filled up with white blocks.

Fault - Initially this seemed to be caused by a faulty DNFS 1.20 ROM - after removing this ROM the system started as normal. However the same fault occurred with a new MMFS ROM too. It was suggested that the Tube host code (in ADFS and MMFS) could be wrongly detecting that a second processor is present due to something connected to the data bus pulling bit 0 high. The white blocks are character &FF which is likely being caused by the Tube host code printing what it thinks is the reset message.

It subsequently transpired that the Serial ULA was not working on this machine either, and after this was removed the Beeb started as normal with both the ADFS and MMFS Roms installed and has been working fine since.

[maniacminer]

An Issue 4 - no DFS - upgraded to Econet - any attempt to access the network results in a hang and eventually "Net Error" reported on the screen.

Checked clock and data from the socket to the ADLC and everything checked out ok, all ICs swapped for known good ICs - no difference. Checked all the resistors for shorts, none found and all measure more-or-less exactly what a working Issue 4 board reads. Checking the logic and found a similar fault to an Issue 3 board I fixed earlier where the INT OFF/STATID (pin #4 on IC97) was not receiving any signal. Tracing the track back to pin #9 on IC26 I found that the underside of the PCB on IC97 is not connected to the top side (where the track originates)

Re-soldering the socket of IC97 pin #4, from the top side, solved this problem.

[HHH]

BBC PSU -5v rail only pushing out -1.78v ( Low / out of specification voltage from -5v rail )

RIFA caps and C9 replaced tested -5v rail only seeing -1.78v

Replaced Caps C18 and C21

Afterwards -5v rail showing -4.38v (unloaded)

[maniacminer]

BBC Issue 4 continuous tone. Worked through the board, CPU coming out of reset but nothing happening. Checked with scope and the CPU isn't receiving valid data. Checked IC14 - 74LS245 and it's bad.
Desoldered and replaced it. More testing and found IC76 - 74LS163 to be enabling more than one ROM at once and IC24 - 74LS138 enabling more than one bus connected device at once. Desoldered and replaced. Now boots to BASIC prompt as normal.

[maniacminer]

Randomly crashing BBC B Issue 7 with half the RAM socketed. PSU completely recapped and load tested. Econet installed (I should really do a more thorough test before installing Econet!)

After around five minutes of running with the lid off any program would freeze and garbage appears on the screen. CTRL+BREAK would then show a garbage startup banner.

Switching off the machine and on again would randomly result in the death tone. Leaving the machine off for a few minutes, it would start up as normal, but gradually get more and more unstable. The soldered in RAM seemed to be to blame, as it wouldn't boot in 16K mode. Something caught my eye, an unusual 6502A processor. A UMC branded chip, no idea if they are any good, obviously no good in this case

I grabbed it out of the board and put it in the BackBit Tester - it said "OK" but I wasn't 100% convinced, so I put it into another Beeb and lo-and-behold it started behaving like a box of frogs.

Turns out the CPU is thermally sensitive. I replaced it with a period Rockwell 6502A and the machine is rock solid. I did the drop test and zero problems

EDIT: it died overnight and now is a continuous tone. The CPU and buffer are very hot. I am suspecting a bus clash due to timing/logic problems.

[Gelpack]

Well Whaddya Know IC20

Fix Link

[tomato]

Installed Econet into an issue 7 motherboard. Econet hardware was recognised by DNFS and clock was found but trying to communicate with anything over the Econet resulted in "Station not present" / "Not listening". Machine also did not respond to *NOTIFY requests from another working Econet machine. Running the econet-monitor utility on a Pi Econet bridge showed the Beeb was however successfully sending out packets.

Tracing through the circuit revealed that pin 3 of IC94 (one of the LM319s) was not connected to ground as it should have been (presumably due to a broken PCB trace). Fixed by installing a mod wire on the underside of the motherboard connecting pin 3 to pin 8 which had the correct connection to ground.

[Gelpack]

EDIT: it died overnight and now is a continuous tone. The CPU and buffer are very hot. I am suspecting a bus clash due to timing/logic problems.

You get all the good faults. I think the boards know its you and decide to give you good workout on the repairs .

Beeb's definitely can and do give you a workout when they go wrong sometimes, but I never had the same level of problems on other vintage machines, but maybe I just got lucky.

Keep on truckin, you'll get it.

[maniacminer]

Tracing through the circuit revealed that pin 3 of IC94 (one of the LM319s) was not connected to ground as it should have been (presumably due to a broken PCB trace). Fixed by installing a mod wire on the underside of the motherboard connecting pin 3 to pin 8 which had the correct connection to ground.

There's a huge ground plane around that IC, my experience is the through hole plating was missing or the solder didn't get enough heat to break through the oxide layer on the board due to the heatsink nature of the large ground plane area.

Troublesome Beebs and long stares it's all going to come out on the psychiatrists couch

[maniacminer]

EDIT: it died overnight and now is a continuous tone. The CPU and buffer are very hot. I am suspecting a bus clash due to timing/logic problems.

You get all the good faults. I think the boards know its you and decide to give you good workout on the repairs .

Beeb's definitely can and do give you a workout when they go wrong sometimes, but I never had the same level of problems on other vintage machines, but maybe I just got lucky.

Keep on truckin, you'll get it.

I've worked on a lot in a short space of time. Many have been straight forward and others have taken the time of five The video ULA clock divider was the underlying fault in the end. The 2MHz clock output disappears over time.

[andyruwilliams]

Hi Guys
Has anyone ever tried to collate all these solutions into a book, database, spreadsheet.

I was thinking there were target areas like boot problems or video problems, communication problems, I'm sure you can see my drift here. Rather than trolling through 6 long pages of solutions looking for a near enough problem and then a solution we could use something which groups typical problems.

I remember when i worked for GTM having this enormous circuit board with huge 40 pin crocodile connectors and connectors up the wazzoo. I cant remember it ever finding a solution other than telling me that the beeb was broken in a an area i already knew. I didn't have an encyclopedic insight into the bios or circuit diagram but slowly worked along a chain with a Logic Probe and a Logic Pulser. I could really have used an engineers trouble shooting solution guide written by engineers. The problem then was all the engineers jealously guarded solutions as part of their worth. Micro Power was also in leeds and the idea of George Turnbull allowing me to chat with Micro Power was laughable.

So keep sharing guys I applaud you.
Andy

[marcusjambler]

Hi All

I've just fixed an Issue 7 loud audio buzz.
Replaced C5,C2,C9 C15 and C16 and snipped the top of R11 leg.
( Actually, I reinstated the resistor connection and noise increased slightly. )
Solved the buzz and slightly increased the loudness

Marcus

[maniacminer]

BBC issue 7 no sync on RGB output. Traced the horizontal and vertical sync from IC2 and found the track broken going east from IC2 and composite sync now appeared on the RGB connector, but was very weak, traced signal back to IC48 and replaced it. Sync back to 5V instead of 1V, picture good now.

[maniacminer]

Fully recapped and tested PSU and BBC Issue 7 that worked for around 2 hours before dying. Switching off and on got tone of death. Went off and did something else, came back, switched it on again and amusingly it worked for around 10 mins, then died again. I tried it as 16k with S25 South, it worked for a few minutes, then it didn't recover. Started to go through the board, desoldering IC45 as some of the outputs didn't look right, replaced IC45 and the outputs were different, but still incorrect. Followed back into the CAS0 and CAS1, they were interfering with each other, removed all the DRAM and found five bad. Removed the address bus buffers and tested them, all OK. With no RAM and using Tricky's TestROM, I started to look for other issues, I found that IC34 would start to oscillate randomly without any noise or superfluous signals on its inputs - replaced that, now the outputs on IC45 looked reasonable, still not correct. Traced back to the video ULA losing 2MHz and outputting 2MHz where the 1MHz was supposed to be. Replaced the video ULA with another known good part and the 2MHz clock went back to being square and the 1MHz clock was now present. Still no joy, there was still weird artefacts on the 2MHz signal that came and went. I traced out the PCB tracks of the 2MHz signal and checked all the places it went under or near anything else - then found the culprit - should have known... Someone had removed IC4 and there was residue left on the board, I removed IC4 and cleaned both sides of the board, the problem went away and the Beeb started up with Tricky's boings I inspected the track between IC4 pin 4 and the 2MHz clock line going to IC34 and found that the edge of the track wasn't covered with solder resist, thus allowing the data clock pin to short out the 2MHz clock - thus destroying a whole lot of stuff. I have painted a couple of layers of lacquer around the IC, refitted it and still no problems. Now to find and fit the remaining 16 DRAMs

TLDR: if you see flux/cleaning residue on the board, clean BOTH sides thoroughly, especially between IC legs.

It was this bad on the underside ...

[andyruwilliams]

Hi Guys
I Spent the last few weeks collating this index.

I derived 2 books

One was A Collection of all the service manuals i could find.

One was a broken down and collated copy of all the entries on this sub.

I then collected these documents together and imposed them intotwo seperate files (pdf) which can be printed on a double sided printer using short edge collating. Effectively an a5 Booklet with 4 pages per A4 waiting to be printed.
They are quite big files
However
When i tried to upload them the board says the combined Service mnauals is to big to upload. (17mb)
Any clues as to how i can share this please?

[gdavidson]

BBC model B Issue 7

Repair thread: viewtopic.php?f=3&t=27145&p=396021#p396021

BBC B hangs at boot -

BBC Computer 32K
DFS 1.A
BASIC
_

No second beep, no “>” prompt, no keyboard, keyboard LED lights appear correct but do not respond to key presses.

/IRQ line gets stuck low a few hundred mS after power on reset and for a much shorter period if break is pressed.

This was traced to a marginal logic “low” signal on databus line D0 during 2nd /Tube strobe assertion pulse. It’s only about a microsecond long so may need scoped whilst triggered by the /Tube selection line. There are only two /Tube low pulses in the boot up sequence so not hard to miss if the scope is triggered on the pulses. If using a logic analyser the marginal low may appear low to the analyser but still as a high to the CPU.

It’s then a case of removing devices on the CPU data bus to find device(s) dragging D0 high. In my case the 6850 was to blame.

—
Graeme

[maniacminer]

BBC Micro Issue 7 - with 8271 and speech

Beeb dies after being on for an hour or so, either solid lock up or a screen filled with garbage and then a lockup. It will restart and stay working for around 5 mins. Looking with the thermal camera, IC78 was glowing hot and running at around 90C, using the Tube snooping device, I suddenly see random bits being set on the data bus... My guess was the 8271 so I put a large heatsink (big stepper motor) on top of it and the Beeb stayed working for over 8 hours. Replaced IC78 and it runs at around 55C and the Beeb no longer freezes. Another duff 8271.

[techfury]

Symptom:

BBC Master hangs while formatting discs (both DFS and ADFS) about 40% of the time, reading and writing works fine. Symptoms are same with Gotek or known-good NEC FD1055 80 track drives. Point of hanging is random, can be as soon as track 0 or the very last track.

Solution:

Replaced WD1770/WD1772/VL1772 with new WD1770. Upon removing the original VL1772, corrosion was observed where the INDEX signal pin meets the chip package, and the pin fell off during removal.

[browndog]

Symptom: No video on RGB, flickering white every few seconds on composite and RF.

Issue: This was traced to no VSYNC output on pin 40 of IC2 6845 VIDEO CRTC.

Solution: Replaced the 6845 and all three video outputs work perfectly.

Original post with images and oscilloscope data below.

viewtopic.php?f=3&t=27220&p=398250#p398250

[atsampson]

The January 1993 issue of Television included "BBC Model B Computer Fault Notes" by Arthur Rumbelow, with 59 common faults and several suggestions for PSU servicing. A letter in the March issue added the PSU C9 startup fault.

I've attached a copy, extracted from World Radio History's excellent scans.

[Maddog Bailey]

Issue 4 random Language? on hard reset.

Mostly boots to BASIC prompt OK. Randomly halts at language? but only on hard reset.

Changed D1, C10, R20 as suggested. No change.

Scope on 6502 reset pin shows about 107ms regardless.
No difference if mainboard is hot or cold.

Another post in Acorn Hardware titled "*FINALLY FIXED* - ODD BBC B problem" suggested IC20 74LS139. Had one spare, swapped it out and now boots to basic everytime so far.

Not sure how to add a link to that particular post.
If anyone can tell me how, I will be happy to edit.

Hope this is helpful.

Best Regards,
Stewart.

[!FOZ!]

Issue: Rolling screen in Mode 7 - all other modes OK

Solution: Use the Hitachi CRT 6845S in IC2 not the Motorola MC6845P.

Thread: viewtopic.php?t=27498

[maniacminer]

BBC B issue 7 otherwise fine except the keyboard randomly producing characters and other times not accepting any key presses. I swapped the keyboard from another known BBC and the fault followed the keyboard. I removed all the ICs from the keyboard and tested them. I found the 74LS251 to be faulty. A Fujitsu part made in Indonesia, no standard date code on the package. Replaced with a modern equivalent and all is good. My, oh, my was this keyboard bad, filthy and lots of dead keys

[maniacminer]

BBC Model B Issue 4 tape only - before installing Econet I'm testing everything and came across an odd problem with the tape interface. The machine would become deaf after three minutes or so. Traced it back to IC35 (LM324) the signal disappeared when it left pin #1. Removed and replaced with a new IC and the problem disappeared, I can load Super Invaders now. Yay!

[maniacminer]

BBC Model B Issue 7 - continuous tone - checked the usual signals on the scope and read the databus using the Tube Snooper - got lots of garbage on the databus as soon as I powered on the machine. Worked out the CAS lines were sluggish and crossing over each other also the memory address buffer control lines were doing the same. This pointed to IC45 being faulty, so I pulled that and tested it, yup, dead. Replaced IC45 (74LS139) and continued. Now getting much less garbage so checked what was causing the ICs on the databus to trigger. The usual suspects, IC24 (74LS138) and IC26 (74LS139) were giving out marginal signals, removed them and tested them, both duff, replaced. IC23 (74LS30) was creating an incorrect output for its input states - removed, checked, also duff, replaced. Got closer, it almost boots, but freezes after a few characters. Checked IC20 (74LS139) and it was giving a weak output on a couple of pins - checked, duff, replaced. Now almost perfect except the serial port doesn't work. Removed and checked IC4 (6850) and it was duff, replaced. Now all is good. Yay! Tricky's Test ROM is very useful for a minimal boot

[maniacminer]

BBC Model B Issue 7 - the Texas Tone - got quite far with Tricky's Test ROM, so I wasn't expecting a lot of trouble here. There was definitely something awry with the RAM. The Frogger screen was corrupted in a way that had bit 0 missing. With the scope I traced the missing bit all the way to the RAM - zero all the way. I replaced the 8118 DRAMs (IC53 and IC61) and started to get random working bit 0 - repeatedly whacking BREAK, I'd get a more-or-less OK Frogger screen. Checked back again through the chain from IC5 (SAA5050) through IC15 (74LS273) still nothing and then to IC14 (74LS245) and here was another problem, D0 was marginal the other bits were strong. Thankfully IC14 is in a socket (a wear part, I guess) I removed it and tested it, yup, duff. Replaced it and the Beeb was perfect. Another Beeb saved using Tricky's Test ROM