Fault finding index

[maniacminer]

BBC B Issue 4 - factory fitted disc and Econet

Unfortunately, neither of these interfaces work, but the machine, otherwise, is perfectly fine.

"Line jammed" on Econet and Disc fault 18 on disc use.

I traced the NMI - seemed to work OK.

Next I checked the dual monostables on IC87 (74LS123) and found they were dead.

Replaced with a new part (easy-peasy as it's in a socket) and both disc and Econet are now working.

[maniacminer]

BBC Model B Issue 3 - factory 32K - randomly won't start up, with Tricky's Test ROM the bongs are really slow and the Frogger screen is mangled, very garbled RAM test screens.

Firstly, I noticed the BREAK key didn't work.

Checked from the 555 timer for the RESET line from pin #2 of IC16, it was +5V there, followed it to the keyboard connector and it was there too.

Checked the keyswitch for BREAK and there was just noise on both sides of the pin, checked pin #2 of the keyboard connector expecting +5V, but there was nothing.

Checked keyboard ribbon and the wire to pin #1 and pin #2 was broken. Repaired the connector and tried Tricky's ROM again, this time normal speed bongs and no Frogger screen corruption.

The RAM test is still a bit wild, with flickering all over and noise.

Wriggled the video ULA (IC6) with its giant heatsink (I think only early Beebs without the ceramic package have this heatsink) and the display became normal.

Replaced the socket for IC6 and the problem disappeared.

Now to add Econet to it (after my tired eyes problem on the last Issue 3, I'm going to be extra careful!)

[maniacminer]

BBC Micro Issue 7 model A->B upgrade, tape only - continuous beep, no display in MODE7 (no flashing cursor from the 6845)

Someone's replaced the OS ROM socket with a fancy Augat turned pin (done at some point in the distant past) checked - all good here.

Tricky's Test ROM runs but there is nothing initially, then a cursor flying all over the screen.

Using the Tube Snoop board, I am able to see huge amounts of corruption on the databus.

Checking the address buffer drive signals, I find that they are all enabled at the same time! Check IC45 (74LS139) and lo-and-behold - the outputs on pins #9,#10,#11 and #12 are all low - impossible combination. The chip is duff and needs to be replaced.

Replaced IC45 and another Beeb starts working (ready for Econet)

[maniacminer]

Something I've noticed on a few boards is the solder resist not being well defined around solder pads. That means it is easy to create solder bridges between a pad and an adjacent track or creep allowing an alternative current path between a track and and a pad causing all kinds of problems. I have to repair this manufacturing defect before adding Econet to the board...

[maniacminer]

BBC Issue 7 Model B with factory DFS worked intermittently for a few minutes and now continuous tone.

Worked through the board with Tricky's test ROM, reset, clocks, RAS/CAS and noticed that both CAS were lagging causing the RAM access to cross over from the CPU to video clocks. I removed and replaced IC45 and gained a marginal improvement. Following the schematic I found IC41 (74LS02) to be the culprit I measured the lag with the scope. On removal, the chip tester said it was OK, but with a sig. gen. and scope I could see a lag that changed with frequency, but only on the one gate. I replaced IC41 and now the Beeb is working and reliable.

[maniacminer]

BBC Micro model B Issue 4 - stops working after being powered on for 20 mins with the lid on. It also has a horrible MODE7.

I checked the 6MHz clock and although wobbly, isn't bad enough to create distorted characters. When the machine locked up I used the thermal camera and noticed that IC43 was getting noticeably warmer than other logic gates. Looked at the output of IC43 (74S04) and it was collapsing with temperature. I replaced it, half expecting its replacement (74F04) to exhibit the same heating problem, but no, it was cold and worked perfectly. The MODE7 problem went away, as did the inability to assert RAS into the RAM which was making the machine lock up. It is also responsible for generating the system clock (16MHz) and the 6MHz Teletext clock, these were not affected as much compared to the RAS signal.

[maniacminer]

BBC Micro Model B Issue 4 with DFS - totally dead - I had rebuilt the PSU first and on connection to the board the rails collapsed to 1V. I put the motherboard onto the bench supply and it was pegged at 3A - so I dug out the thermal camera to see what was getting hot. When I stepped up to the 10A-100A range, the traces around Vcc2 were glowing and the capacitor marked A next to IC4 was glowing brightly and then a crack and the PSU dropped to 1.5A and the Beeb sprang to life. Yay! I removed it and replaced it with a 33n metallised polyester cap.

[maniacminer]

BBC Micro Issue 4 Model B DFS and Econet - note written inside the lid says "Econet not working" - so I cleaned the board and case, nothing wrong with the DFS or the Beeb in general, however the Econet was "Not listening" - so I replaced the LM319s - no change, I replaced the 68B54 - no change, I replaced the monostable 74LS123 (IC87) still no change. I put the scope onto output of IC87 and noticed there was no action despite changing the IC. I checked R50 - 39kR - ok, then C17 - no reading - replaced it with a 2n2F metallised polyester capacitor and we have Econet. Yay!

[Ben M]

BBC Micro: Model B issue 7 : Continious beep, weird acting LEDs, keyboard not functioning untill pressing break.

Sympthoms:

At the beginning, the BBC worked, but the LEDs for shift-and capslock, were behaving strangely. When i pressed random keys, the shiftlock or capslock LED flickered. And they where sometimes dim or did not work.

A day later, I powerd the BBC up, and i got a continious beeb. You need to hear two beebs, the second beeb was ok, but the first was continious.( booooooooooobiepooooooooo.....)
So when i powerd the BBC up, the boot message was normal, i got the continious first beep and the keyboard dit not work, until i pressed the break key. It changed the tone of the continious beep, but it kept beeping. But i could use the bbc like it was normal. When I pressed the copy key, I could not use the keyboard anymore.

What dit i check?:

At first i checked the LEDs and keyboard cable, they had no problems.
Then i swapped the 6522 VIA's, still the same problem.
Then i put my problem on stardot....

Solution:

I got the suggestion that maybe IC32 (74LS259) was defective. It is conected to the keyboard, LEDs for shift-and capslock, and the sound chip.
It seems a logic reason, because the leds were also direcly connected to IC32 and if the 6522 is not defective, but the sympthoms are almost the same as with an defective 6522. It was worth a try to change IC32. So I got the old one out, put a socket in and the new IC.

Result:

Everything is working perfectly fine, also the LEDs. And no continious beep! So, problem solved.

Link to Original Post: viewtopic.php?p=406600#p406600


Regards, Ben

[sbadger]

BBC Micro: Model B

Symptom

Picture slow rolling via RGB to SCART cable, vsync problem and some hsync stability jitters

Picture was stable via Composite

Checked
Picture roll present in all MODES
Link 31 was in the normal west position
Pulled and checked IC48 in logic tester, it failed

Solution

Replaced IC48 74LS86 (TTL Quad XOR gate)

Picture now rock solid and super clear

[maniacminer]

Very dodgy Watford 12 RAM/ROM board. Even typing on the keys caused the Beeb to randomly crash. The board has three connections to the Beeb PCB, into the A/D socket, the OS ROM socket and a link header. I checked each pin of these connectors whilst banging the Beeb and discovered a few pins of the two sockets were bad. I resoldered the pins and that helped a bit, I then replaced the sockets on the Beeb PCB and it made a small improvement. Then I noticed that one of the pins had broken on the Watford board and using a pair of tweezers gently moved each pin side to side and yeah... about a third of the pins snapped off. So I removed the pin headers from the Watford board and the A/D socket on there too. Now I had to build two lots of an extended 28 pin DIL header. I did that, soldered it all together and yay, Beeb is now rock solid. Just got to find my stash of 6264 SRAMs.

ADC pin headers and socket removed

OS ROM header pins removed

What's left of the old extension pin headers

Installed ADC socket and extension header

Header pins installed

Board installed into a Beeb for testing

[maniacminer]

BBC B Issue 4 won't read discs or format them and randomly doesn't see the Watford DDFS board. Turns out it has the same problem as the Watford 12 ROM/RAM board in that the pins were really brittle. I found it would work if I placed a heavy object on the daughterboard but would still be a bit dodgy. I removed all the extension pins and recreated a new one from strip pins and a 40 pin DIL socket. Once reinstalled it worked first time, read and formatted discs perfectly.

Extension socket removed (top view)

Extension board removed (bottom view)

Bodge coming in...

Soldered in (top view)

Soldered in (bottom view)

[maniacminer]

BBC Micro Issue 7 32K - upgraded to Econet - added DNFS 3.0.2 and the moment I did the machine booted with all kinds of oddities.

For a start, there is one beep and the cursor flashing in the top left corner. Hitting BREAK gets back to the BBC Computer 32K Acorn DFS BASIC prompt; but any attempt to do star commands e.g. *NET results in screens of garbage. I managed to do *HELP and got DFS and NFS repeated four times...

I rechecked the EPROM in the burner with the image and it matched. Then I checked the 'scope and the lightbulb moment, the chip is far too slow, turns out the unmarked ICs run at >400ns and that's too slow for the Beeb in its default setting. I burnt a new IC and the problem vanished.

[maniacminer]

BBC Micro Issue 7 model B, factory DFS - a couple a minor issues that I fixed, then I installed Econet.

After installing Econet I am getting a "Line jammed" error and no activity on the serial data pins of the ADLC. I then checked the resistances of the various points of the schematic for shorts and incorrect resistor values. I then found I had got distracted and fitted a 1kR resistor instead of a 100kR resistor for R40. Replaced with 100kR and Econet now works.

The photo has the offending part with an arrow of shame pointing at it and the correct part has a ring around it - obvious eh?

[maniacminer]

BBC Issue 3 model B

No sound, just the gentle buzz from the speaker - the usual idle BBC Micro noises.

Checked around IC18 and IC17 for signal - seemed OK - check for supply to IC17/LM324 and noticed that the negative supply rail was at 1V, the PSU was producing -5V so checked each side of R18 and found that the -5V side was not connected to the board. I cut the remains of R18 and bent it over to bridge the pins and the sound came back, yay! I'm not sure what the 10R resistors are doing in line with the power for the LM324? Maybe there's an inrush problem?

[wiggy]

I'm not sure what the 10R resistors are doing in line with the power for the LM324? Maybe there's an inrush problem?

They, together with the decoupling electrolytics, form some additional filtering to cut the noise down thus improving the PSRR. It's a very common trick, although with newer (faster) designs the resistor is often replaced by a ferrite bead for better rejection at higher frequencies.

[maniacminer]

They, together with the decoupling electrolytics, form some additional filtering to cut the noise down thus improving the PSRR. It's a very common trick, although with newer (faster) designs the resistor is often replaced by a ferrite bead for better rejection at higher frequencies.

Ah, yes, I asked a guy at work about it and he showed me some simulations of various resistor/inductor combos, but suggested that in this case it wouldn't make any difference and I could just leave it bridged or put in a small inductor in series to fill the holes. I showed him the schematic and the PCB layout, his comment on it suggested that it came out of the backside of a dog

[marrold]

BBC Issue 7 Model B - Disc Error 50 with 1770 + DFS 2.2X and Disc Fault 18 at 00/00 with a 8271 + DFS 1.20

Received with a 1770 Disc Controller and DFS 2.2X ROM, but when I attempted to read discs I got the error "Disc Error 50". I converted it to an 8271 to make my 3 machines identical hoping it would also fix it, but it didn't. With DFS 1.20 I got the error "Disc Fault 18 at 00/00"

The IC87 16 pin socket required replacement - when someone installed the 1770 and fitted the link/jumper between pins 9 and 12 the socket was damaged leading to an intermittent connection.

How I arrived at this solution is mostly documented in this thread, but in summary I attached a logic analyser to various points on the Gotek and BBC main board and ascertained that whilst Read Data was coming out of the Gotek, it wasn't arriving on IC87p9. I removed IC87 from the board to perform more testing and noticed the damaged socket.

[1024MAK]

<snip> ...his comment on it suggested that it came out of the backside of a dog

Maybe from Muttley... That would explain the hiss from the speaker

Yes, the design is not the best

Mark

[derek]

Hi,

Here is a nice video from Adrian's Digital Basement, where Adrian, before fault finding on a non-working Electron and gets it working.

https://www.youtube.com/watch?v=sukuk-m ... alBasement

Goes from unboxing to repair, the video is 1 hour long, but worth watching.

All very interesting and may help others struggling with Electron repairs.

[maniacminer]

BBC Micro Issue 7 - factory disc interface fitted

Continuous tone from the speaker, display flashing white/black randomly.

Checked all the usual suspects, reset, clock, NMI, IRQ, address bus activity, data bus activity etc. all looked good.

Looking at the trace from the Tube Snooper it was obvious that an address pin wasn't being recognised by the OS ROM - so I removed it and put it into the tester and yup, it's dead. I took another from my stash, checked it first, got a checksum and put it in the Beeb - success!

Good OS ROM

Bad OS ROM

[maniacminer]

BBC Micro Issue 4 - factory disc interface fitted.

Everything was OK until I added Econet, then the Beeb wouldn't start up, just a flashing cursor at the top left hand corner of the screen.

Looking at the scope, everything was reasonable except that the NMI line to the 6502 was very busy indeed. I removed the disc controller IC and made link S9, but no difference. I then removed IC91 and made S2 and the problem vanished and the Beeb booted up with no disc or Econet enabled.

I then removed S9 and reinstalled the disc controller and got DFS back.

The source of the NMI was not the 6854 ADLC, as its IRQ line was high during this time. By process of elimination with a thermal camera, it turned out that IC27 was bad - no-one would have known until Econet was installed. I removed IC27 (7438 - yup, old school TTL) and replaced it, then removed S2 and reinstalled IC91 and the Beeb booted up successfully with CTRL+N+BREAK bringing up the usual BBC boot screen with Econet station ID. Two out of the four NAND gates tested as bad, one for the strobe on the printer port and the other for Econet.

[roganjosh]

Acorn Atom

Fault: System unresponsive after ~5min from cold after which subsequent boots would mostly fail.

Diagnosis: Performed using freezer spray.

Solution: Replaced defective component X2 (4MHz crystal). Passed 2hr soak test.

[maniacminer]

BBC Micro model B issue 3 with Econet runs slowly, lots of noise from the speaker.

Checked over the PCB and found a few things, the OSROM socket has been punished with a ROM board in the past and now needs replacing, I temporarily bent the pins to get the OSROM to have a good, temporary contact. That didn't affect the speed problem.

Checking the 6502 NMI and IRQ - the NMI was continuously high, so that's not the source.

However, the IRQ line was very heavy with action. Going around the ICs, I removed the 6850 (IC4) and the user VIA (IC69) and still no change. Removing system VIA (IC3) and replacing it with IC69 and the amount of IRQ action dropped significantly.

This suggests there is a spurious fault in the system 6522 as putting it into the user VIA socket, I don't get the same storm of IRQs.

[JonnoW]

BBC issue 3, DFS, MMFS.
Initially working.
Suddenly started repeatedly printing carriage returns on pressing the return key after hard reset.
Keyboard OK in another machine, same symptoms with substitute 6522.
No sound issues ruled out IC18 leaving IC32 as the only other likely cause.
Socketed IC32 and issue fixed.

[maniacminer]

BBC B Issue 7 - just added Econet to a working board - now continuous tone.

Sometimes I get a narrow cursor in the top left, but most of the time I nothing but a blank screen and a continuous tone from the speaker.

I checked all the usual, supply, clocks, reset, the scope traces look particularly bad on the databus, with sloping edges, castellations and glitches. The address bus was nice and square. I removed the OS ROM and replaced it with Tricky's Test ROM and got nothing at all, just a blank screen and the Texas Tone. I tested continuity between the pins of the OS ROM socket (with the IC in place) back to the processor. I discovered a dodgy pin! Pushing up and down on the pin made it come and go at different resistances. So I removed the socket and replaced it.

Snapped pin, ever so common when the pin of the IC sticks to the pin in the socket

Almost like a Morse Key...

With Tricky's Test ROM in the new socket, I now get the bongs from the speaker! Now I put the Acorn OS ROM back in and I got the cursor in the top left - a small cursor, like MODE0. Ah-ha, I know what that is, whip out the 74LS251 on the keyboard and replace it. The tester reported it as a failure.

Tester couldn't auto identify the IC and direct selection showed it was bad.

Now I can get it to boot if I wait for around a minute and whack BREAK for a while. Oddly, the machine starts up immediately if it has been on for a a few minutes. Busted out the thermal camera and looked for something different and yup, IC45 (74LS139) is having a bad day. Replaced it and now the machine boots up every time.

74 Life Short at 139F?

Unfortunately, there's another problem, printing ?&FE18 to get the Econet stationID results in a blank. On checking the chip select for the station ID, IC26, another 74LS139 is also bad, with one output hovering at 1V and another at 0V... Removed and replaced it. More testing...

Checking further into the Beeb, the ADC doesn't work, I removed and replaced the socket for IC73 and got some wild results when updating the ADC registers. I then scoped the lines and noticed NWDS was always low, contradicting the inputs to IC77 (74LS00), so I removed IC77 and yup, it was bad, then replaced it. Now the ADC kind of works.

Connecting up a Unilab 540.006 "Computer Module" that lets me test the user port and printer port, I get blinkenlights whenever I press a key on the keyboard. On the scope it becomes apparent that the chip select lines from IC24 (74LS138) are misbehaving. Pressing a key has pin#13 and pin#12 firing at the same time (yeah, shouldn't be possible) also the ADC and TUBE pins are filled with random pulses. Remove and replace IC24 is next on the list - I'm almost there!

I replaced IC24 but I was still getting random chip selects on IC69, oddly these were now only at IC69 and not at IC24, IC69 is getting quite hot. So I replaced the 6522 with another known working 6522 and the problem went away. The chip tester passed the faulty IC69, but the new 6522 has no activity on the chip select when it isn't being selected and works as expected.

After 24H of endurance tests the Beeb failed again, this time I knew what it was immediately, the screen was full of corruption and the BASIC program running had the good old OSCLIOSCLIOSCLI at line 30 kind of error. IC14 (74LS245) had become heat sensitive, replaced it with a new 74F245 (yeah, I know, I shouldn't use F's here, but it works and doesn't heat up) and it has been running perfectly for a few hours so far.

So to summarise, eight problems, broken OS socket, two duff 74LS139's, a duff 74LS00, two duff 6522s, a duff 74LS245 and a duff 74LS251 on the keyboard.

[maniacminer]

BBC B+ Issue 1 continuous tone from the speaker, nothing on the screen.

Went through checking the address/data bus and found the RAM to be producing either nothing, or wild pulses.

Removed the 8x 6164 DRAMs (not all came out in one piece)

These break apart and are always bad

I then tested them...

Partial failure

Total failure

I then added sockets, then put in new 6164-100 DRAMs - boots normally now.

Oddly, this B+ has no floppy controller, there is provision for a 177x or a 8271, but neither are fitted. The is contrary to this entry in the BeebWiki for the B+ https://beebwiki.mdfs.net/Model_B_Plus

The Model B+ also included floppy disk support as standard

Where's my promised floppy controller?

The internal speaker is just as buzzy as a Model B so I don't see they've gained much experience in the analogue layout of this circuit

Now to install Econet into it...

Econet done! Now for some Starship Command

[marcusjambler]

BBC Micro Issue 4 motherboard with Econet and 8271 DFS
Fault : Keyboard unresponsive and continuous (polyphonic)tone.
Culprit : IC32 74LS259
Now working time to test Econet and DFS.

EDIT: Also swapped VIDPROC all colours displaying as RED in mode 2/5? ( bonecruncher 2nd splash screen )

Marcus

[maniacminer]

BBC B+ Issue 1 lots of noise on the internal speaker.

I replaced everything in the audio chain and it had no effect.

https://www.stardot.org.uk/forums/viewtopic.php?t=28844 my explorations and frustrations are in this forum post.

To cure it, solder a 1nF capacitor into S21 - it is right next to the speaker output on PL15.

[vanekp]

BBC model B issue:-
Tape and RS423 both gone deaf
i.e. save worked fine and sending from the RS423 port no problem, but you could not load tapes or receive data from the RS423 port.
After lots of testing as I 1st though it was only the tape system changed the LM324 IC35 (scoped data coming out of pin1 to the serial ULA) and also changed the serial ULA IC7 (both are in sockets), nether fixed the fault and could see data getting as far as the output of the Serial ULA up to the ACIA IC4 (which I think is soldered in on most BBC's), but not getting to the BBC irself.
Socketed the ACIA chip IC4 and installed another MC6850P and the BBC can hear once again on both the tape system and the RS423 port.