Fault finding index

[Winston]

BBC Master 128k with ADF10 (econet) and RetroClinic internal datacentre + Retroclinic internal Tube 6502 copro.

After a few hours of use locked up with the message "Type mismatc" (the whole string not printed). Ctrl-Break left a cursor flashing in the top corner of the screen. Cycling power resulted in continuous sound tone. Machine would occasionally start but would lock up again usually within 30 seconds (but while not locked up, ADFS worked well enough to load something off the internal CF card).

Checked state of the internal CMOS battery (bad, low voltage) replaced cells, didn't expect it to fix the problem and it didn't.

Removed the ADF10, this did not fix the problem.

Disconnected the internal Datacentre, machine came up OK and was stable. Checked security of the 1MHz bus connectors, all the connectors on the Datacentre PCB and reconnected, back to the continuous tone and no video (LCD monitor reports no signal).

Removed the USB daughter board off the Datacentre, problem remained.

Removed the two 512k x 8 RAM chips off the Datacentre, machine came up OK and was stable, ADFS worked on the internal CF card.

Plugged the same 512k x 8 RAM chips back in, machine came up OK and was stable, RamFS worked.

Conclusion so far - probably a poor connection to one of the Datacentre's RAM chips since removing and plugging them back in seems to have solved the problem.

[Halifask]

BBC B+ standard machine with 1770 and Acorn DFS 2.26.

This had just recently been purchased and came with MMFS but this had been removed and a new GOTEK fitted.

Any attempt to access the drive, Shift Break or any * disk commands result in a flashing cursor and silence, no drive stepping, just nothing.

Resolution can be found here:

viewtopic.php?p=422269#p422269

Hope this helps

Arthur

[arnoldemu]

Master 128 issue 1. Powered on and showed signs of life. Good vsync and hsync and csync. Resetting CMOS with crtl+R did nothing. UHF didn't tune in, composite output appeared black but increasing brightness to max showed an image. RGB unaffected. Modulator/aerial case (UM1233) was showing +3.7V and so was the gnd connection on right side of it. Initially there was gunk around C85 which was cleaned but continuity checking found no connection between Q11 and C87. Patch wire fixed the contrast and restored 0v at the modulator case.

https://www.stardot.org.uk/forums/viewtopic.php?t=28822

[maniacminer]

BBC B Issue 7 factory B/32K with disc interface.

Problems: disc interface was removed at some point in the past and the piece of paper inside the machine suggested that the disc interface was faulty. I didn't need disc, so I installed Econet - the machine worked perfectly well before adding Econet.

After adding Econet the machine would not boot, a single beep from the speaker and the cursor (in MODE7) would sit in the top left hand corner flashing away, no keys on the keyboard had any effect.

Starting by pulling the DNFS ROM, the machine immediately booted up. This suggested to me that the NMI was not being cleared. Scoping the 6502 I could see a lot of activity around the NMI. I traced it back and found that the CTS and DCD on IC89 (68B54) were filled with high frequency waveforms and this was causing it to not respond properly. Following back from these pins, it became immediately apparent that IC87 (74LS123) was producing rubbish.

A quick look at it and the ah-ha moment, it had become corroded, a white and green powder was all over the legs of the IC and the socket. I pulled the IC and the contacts in the socket turned to powder. I removed the socket and replaced it, however, several of the tracks were corroded and had to be rewired, one of these went to the disc interface and that explains the original fault. Testing the 74LS123, it was just plain dead, I replaced that and now the machine boots with DNFS and can see the network, yay!

The 74LS123 fails on the tester

The socket is heavily corroded and breaking up

Bodge wires are underneath

[maniacminer]

Issue 4 Model B 32K (upgraded from a Model A at some point)

Issue: very quiet speaker

I checked that the potentiometer was set to zero resistance and the signal arriving at the amplifier was similar to another "loud" Beeb I had on the bench repairing a failed Econet installation. I replaced the speaker and that improved the intensity of the sound by a small degree. I then increased the default gain of IC19 (LM386) from 20 to approximately 40 by soldering a 10µF electrolytic capacitor in series with a 1k5Ω resistor from pins #1 to #8 (datasheet https://www.ti.com/lit/ds/symlink/lm386.pdf)

Capacitor and resistor between pins 1 and 8 control the gain

The speaker was plenty loud enough now, but the buzzing background noise was annoying, so taking a leaf from the omitted B+ 1nF capacitor, I soldered one between pins 2&3, then increased it to 2n2F almost eliminating the noise entirely, leaving an eerily silent, but also loud, speaker. A quick go on Starship Command and this Beeb is plenty loud enough now!

Back to the troublesome M5000's - I have quite a pile of them and all-bar-one have problems...

[maniacminer]

BBC model B factory 32K and DFS

It appeared to be working OK and I put Econet into it, only later on did I discover that there's snow on the screen in MODE7 and random clumps of garbage in other modes.

The noise only appears when the machine is cold, as it warms up the noise disappears. I tried freezer spray on each IC and it seemed to be a memory issue. I pulled out all the RAM and tested it and there was no RAM fault. I put the RAM into another Beeb and no fault, so I put it back into the original Beeb and... no fault. My guess is the capacitors in the DRAM had got "tired" and heating them up to extract them solved their problem. It was definitely there as I could see it on the scope and the behaviour of the machine.

The faulty RAM

X marks the spots

RAM out

Cleaned and straightened

Sockets yay!