Topic: BBC Domesday Computer repair log  (Read 2364 times)

Author Message

0 Members and 1 Guest are viewing this topic.

Offline Phu

  • RCM Workshop
  • Committee
  • Amiga 4000
  • *
  • Posts: 2084
  • Kudos 41
  • Gender: Female
  • Pay no attention to that PCB....
    • View Profile
    • ZX Spectrum Laptop Project
BBC Domesday Computer repair log
« on: September 09, 2012, 05:48:27 PM »
Those in attendance to Snibston in August 2012 will have witnessed the BBC Master Turbo driving the Domesday system being somewhat uncooperative. After making crackling noises from the speaker, it flat refused to start.

On getting it back to d'Workshop, the machine was more co-operative. Starting, but being confused. It's CMOS RAM was suspiciously corrupt. It also wouldn't beep. Or sometimes it would beep, sometimes it would try to beep and make crackling noises, and other it would just sit silently.

Now all of this is part of one subsystem in a BBC Micro, called the "Slow Peripheral Bus". Essentially, one of the 6522 chips acts a slow-access device controller. It has two latchable I/O ports which are used as address and data buses. The "B" port of the 6522 is connected to a 74259 chip called an "addressable latch". To explain simply, this has 8 output lines to which you can "write" a 1 or 0 to, and the value sticks. One of these is connected to the 76489 sound chip.

To summarise: CPU -> 6522 VIA -> 74259 latch -> 76489 sound -> amp -> speaker. Got that?

So I figured that perhaps the addressable latch was faulty. I initially ruled this out because the other things it controls (screen start address, CMOS RAM access) were working fine (while CMOS was corrupt, it did still work - more on that later). But putting an oscilloscope on it showed no drop on the sound write enable line.

A swap out from a spare BBC B board later and..... no difference. ???

Ok, lets concentrate on the crackles. There's no familiar low-tone on startup, so perhaps the sound chip is faulty? Swapped out and... no difference. !!!!?!

OK, if it's not the sound chip screwing up, it must be one of the amplifier chips. The LM324 is the pre-amp and is connected directly to the sound chip. Since I'm seeing crackles there on the scope, it would be logical to believe that's the faulty chip. I think you know what happened when I swapped that.

NOTHING!

At this point I'm somewhat confused. Everything logical seems actually to have been intact. Looking at the circuit diagram shows the LM386 amplifier only connected to the sound circuitry via its input, not its output.... unless...

This 8-pin amp chip was backfeeding noise into the sound circuit and making it appear at the sound chip. On top of that, it wasn't doing its job and was creating silence/crackles. Once I replaced the LM386...

BOOOOP BEEP!

Turns out that WE line *was* going low, it was just doing so so quickly that the scope could barely see it. If I looked closely I could see a dot for a blink of time at the 0 line during sound writes.

So what caused this mess? Well, the battery pack in this machine was less than professionally put together. Three energizer AA batteries, taped together with insulation tape, and then the wires to the motherboard taped on too. This was used to connect batteries 2 + 3 together:



I should point out that blob of solder was just "there" - it wasn't being used to hold it to the batteries - that piece of metal was just taped on.

Unfortunately, during the movements of the machine to various events, the positive contact slipped a little from battery 3, and cut through the plastic covering of both itself and battery 1. This is the impact mark on battery 1, where the casing (negative) of battery 2 contacted with the negative from battery 3, causing a loop through batteries 2 and 3:



At some point it scraped itself, as can be seen on the negative edge of battery 3:



I shall be building a somewhat better battery pack with fresh batteries to restore the system :)

-- Richard
8 End of File, RCM:1

Offline AndyRCM

  • >=))))º> GO FEED THE FISH! <º((((=<
  • Administrator
  • Amiga 4000
  • ******
  • Posts: 9660
  • Kudos 50
  • Gender: Male
  • Manic Jet Set Willy
    • View Profile
    • Retro Computer Museum
Re: BBC Domesday Computer repair log
« Reply #1 on: September 09, 2012, 06:22:12 PM »
Flipping 'eck mate - nice work . . . :)

Does this mean it is a simple fix or is it still not working? ;)

A
"I could see the faces of those who led pissing themselves laughing" - Funeral Pyre by The Jam

Offline Phu

  • RCM Workshop
  • Committee
  • Amiga 4000
  • *
  • Posts: 2084
  • Kudos 41
  • Gender: Female
  • Pay no attention to that PCB....
    • View Profile
    • ZX Spectrum Laptop Project
Re: BBC Domesday Computer repair log
« Reply #2 on: September 20, 2012, 11:04:58 PM »
Part II - Modem's revenge.

So the previous fix didn't last long... 3 minutes into running and Beeeeee....e...k..eee......kk...kkrrkckck. Silence.

Clearly, while the battery was uh... "unprofessional", it wasn't to blame for the sound chip dying. Something on the board was killing one or more components.

A consultation with knowledgeable facebook friends suggests replacing the filter capacitor. This didn't work. So that wasn't it. All other answers dried up, so some probing was in order.

A little bit of electrical knowledge regarding capacitors. They will conduct until they are charged, or more specifically until they are charged in the same polarity as you are trying to make them conduct. Then they stop conducting. This is useful, as it allows you to test capacitors as to whether they need replacing.

A capacitor should make your continuity test beep for a short period. The larger the value, the longer the beep. Then it should abruptly cut off. If you switch the polarity then it should do the same thing as the original charge discharges, then the capacitor recharges in the opposite polarity.

Thus a capacitor that beeps continuously is either in excess of 3300uF and the size of texas, or faulty. Guess what? None of the capacitors were faulty. They all pulsed a beep as expected. GRR.

Circuit diagram consultation time. Part of the audio circuit is tapped off to a "modem" connector. The Master 128 is equipped with internal modem support, in the form of a digital bus (which in the Domesday system is where the SCSI board connects) and a 4-prong analog connector for mixing the phone line sounds with the internal speaker.

I did some tests with the pin's connectivity to other parts of the circuit, and by chance happened to touch one of the speaker pins (speaker was disconnected at this point, as from a continuity point of view it forms a deceptive low-value resistor) and got a beep.

OK I thought, that's clearly where the audio signal mixes in in some way, so I test the other pin which I assume to be ground, then I'll get nothing....  silence. Good.

Not good. The pin I touched accidentally was ground, not audio. The "audio" pin of the modem connector was connected to ground by a 10ohms resistance, which it shouldn't be.

Much examination, rechecking of capacitors, etc. later I found myself staring at the modem connector and thinking... I wonder if it's the connector itself? I don't see how, but let's check.

After desoldering it, the resistance went up from 10 ohms to 40 ohms. Curious. Some liberal application of RCM's favourite cleaning agent (nail varnish remover) and the resistance went up to ..... too high for my multimeter to read in that mode ;)

Could this be the fault? I fitted two sockets to the motherboard (one for the sound chip, and one for the remaining amp chip) and installed two sockets. A test program to send every tone to the sound chip has been running for 20 minutes without fault now.

So it seems a conductive "something" got between "audio" and "ground" on the modem connector, and crosswired the entire audio circuit. The "gunk" had a greenish hue to it, so I do wonder if some corrosion transferred atmospherically to the pins from the manky battery (cf. the delights of Amiga batteries for how nasty that can get), though there are no other signs of battery gunk getting in.

At least now the audio works I can put the lid on a test the whole Domesday system.

-- Richard

8 End of File, RCM:1

Offline porchy

  • Moderator
  • Amiga 4000
  • ******
  • Posts: 825
  • Kudos 33
  • Gender: Male
    • View Profile
    • JAMMArcade
Re: BBC Domesday Computer repair log
« Reply #3 on: September 21, 2012, 06:22:44 AM »
Curious stuff.
Jimbobs megacd and 32x also have a greenish tint issue but that's been cause by some kind of liquid being present.

Good fix fella

Offline WizRider

  • ZX80
  • *
  • Posts: 28
  • Kudos 3
    • View Profile
Re: BBC Domesday Computer repair log
« Reply #4 on: September 21, 2012, 10:29:48 AM »
A lot of it goes over my head but I do love reading these repair logs (and also ones for old arcade machines).
It really highlights the need for RCM to keep these machines alive as long as possible (and for good, accurate, non-HLE emulators).

Keep 'em coming!  :)

Cheers...