Dead BBC Masters - ROM screwup leading to main system damage

discuss both original and modern hardware for the bbc micro/electron
Post Reply
mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Thu Dec 22, 2016 5:38 pm

So I've had a mega disaster with my two BBC Masters. I have all the details of what I've done and the faultfinding done so far to try and rectify things.

I pulled them out of the loft to restore them, it's been a long time since they've seen the light of day. Dismantled, cleaned everything thoroughly and restored both. Replaced PSU caps, cleaned the mainboards and sockets, repaired the keyboards and replaced the memory backup battery packs. Configured them all and they were working great. Got the 5 1/4in drive units out and some disks, cleaned them, discovered the disks surface was knackered on most of the disks, cleaned out the oxide that had come off and tested on some good disks - all working great. I could even read the original Master utility disks and make a few backups onto the few disks that were ok.

I ordered a TurboMMC kit for reliable storage and fitted it to one in IC27 location. It booted fine, I could see the new DFS rom, but no BASIC V2. Connected up the card PCB and it worked great. So very pleased so far.

Here is where I made the mistake.

Knowing that the rom was a 27C256 with two roms in, I read that IC27 could only contain a single ROM. So I put it into IC41 location, leaving IC27 free and moved the jumper LK18 west. I chose IC41 location, under the plastic shroud, so I could easily fit any other ROMS. I booted the Master and it came up, but after maybe 10-15s, it crashed, then the video was interrupted. I turned it off and on, the long beeeeeeee of death :(
I removed the ROM, set LK18 east and booted, but still long beeeeee of death.

I checked the ROM install notes again, and it says IC27 must be fitted first. So it looks like the ROM has conflicted with something else on the data bus and killed something.

I checked the service manual and did some of the initial tests. I discovered that the CAS output from IC36 pin 3 74F00 was high with the multimeter. I checked back through the NAND chain, 6 seemed not quite right and 11 was definitely oscillating. So maybe bad IC36. But the question remained, how would that IC be damaged with a screwup at the ROMs? Perhaps static from changing LK18? Unlikely though. I checked on the other BBC master, and the readings seemed the same and that was definitely working. So probably a red herring.

So I swapped the system ROM into the working system and that was OK.

I looked on the circuit diagram to see what could possibly be damaged by mess ups at the ROM socket. IC20 memory controller perhaps, poss the 6512, IC14. The address lines are always in to the other chips, so damage not likely from clashing devices. So I looked at the data bus CD0-CD7 to see what they were connected to, and it was the main RAM ICs IC17,18,23,26 and the CPU. So screwups on the data bus at the ROMs could easily kill the RAM ICs (or the CPU). As the unit isn't booting at all, it may be damage to memory IC17 or IC23. I have checked the outputs of the RAM ICs pins 2,3,15 and 17 and they are oscillating, nothing held high or low.

Finally, against the voice in the back of my head yelling 'The ROM is probably damaged too!' I put the ROM into the working Master to see if it would boot. It did, so I reassembled it. But when used, it crashed after a short time and also became damaged :( :( Very likely with the same problem. I removed the rom. It does boot, but crashes after 20s or so at which point I turn it off to prevent final death.

Welcome to my Christmas pain.

I would be very grateful if a beeb master guru could work through my description with the diagram and tell me any thoughts to confirm that I'm on the right path to repair. I've ordered some new 41464s, they are NEC 41464-10s. Also a 74F00 too, just in case. Luckily, the mem controller IC CF30058 is still available from China. I will replace the four RAMs first. It has just occurred to me that it could be the CPU data bus, if the RAMs aren't held high or low.

I have an old scope with rudimentary storage facility and a spectrum analyser to help.

User avatar
hoglet
Posts: 9299
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by hoglet » Thu Dec 22, 2016 6:00 pm

mw0uzo wrote: Here is where I made the mistake.

Knowing that the rom was a 27C256 with two roms in, I read that IC27 could only contain a single ROM. So I put it into IC41 location, leaving IC27 free and moved the jumper LK18 west. I chose IC41 location, under the plastic shroud, so I could easily fit any other ROMS. I booted the Master and it came up, but after maybe 10-15s, it crashed, then the video was interrupted. I turned it off and on, the long beeeeeeee of death :(
I removed the ROM, set LK18 east and booted, but still long beeeeee of death. F*ck.

I checked the ROM install notes again, and it says IC27 must be fitted first. So it looks like the ROM has conflicted with something else on the data bus and killed something.
It doesn't seem to me that what you have done with this ROM should have caused any damage.

In fact, setting LK18 to the west position and putting a 27C256 ROM in IC41 seems completely valid.

My guess is that some unrelated failure has occurred.

Or possibly static damage (but that would be very unlucky to have that happen twice).

Maybe start by verifying the power supply voltages are correct, and the battery is good?
mw0uzo wrote: I checked the ROM install notes again, and it says IC27 must be fitted first. So it looks like the ROM has conflicted with something else on the data bus and killed something.
Can you post that section of the ROM install notes?

I can't see how a conflict can have occurred, and even then they are rarely fatal.

Dave

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Thu Dec 22, 2016 6:17 pm

hoglet wrote:
mw0uzo wrote: Here is where I made the mistake.

Knowing that the rom was a 27C256 with two roms in, I read that IC27 could only contain a single ROM. So I put it into IC41 location, leaving IC27 free and moved the jumper LK18 west. I chose IC41 location, under the plastic shroud, so I could easily fit any other ROMS. I booted the Master and it came up, but after maybe 10-15s, it crashed, then the video was interrupted. I turned it off and on, the long beeeeeeee of death :(
I removed the ROM, set LK18 east and booted, but still long beeeeee of death. F*ck.

I checked the ROM install notes again, and it says IC27 must be fitted first. So it looks like the ROM has conflicted with something else on the data bus and killed something.
It doesn't seem to me that what you have done with this ROM should have caused any damage.

In fact, setting LK18 to the west position and putting a 27C256 ROM in IC41 seems completely valid.

My guess is that some unrelated failure has occurred.

Or possibly static damage (but that would be very unlucky to have that happen twice).

Maybe start by verifying the power supply voltages are correct, and the battery is good?
mw0uzo wrote: I checked the ROM install notes again, and it says IC27 must be fitted first. So it looks like the ROM has conflicted with something else on the data bus and killed something.
Can you post that section of the ROM install notes?

I can't see how a conflict can have occurred, and even then they are rarely fatal.

Dave
Thanks for the reply. Both systems were working great prior to this, I've checked all the simple stuff and reset CMOS on the partially working unit. The ROM has been damaged somehow, which has damaged the computers. I remember a long time ago, my old computer buddy said to me 'Don't put a ROM in that socket under the plastic shroud, it killed my BBC master.' Unfortunately I didn't remember until after it all happened, it was so long ago. There must be a good reason that doing so breaks it... maybe rom CE not controlled correctly with no IC in IC27?

Here's a link to the rom install notes:
http://www.8bs.com/submit/acornappspdf/042.pdf

User avatar
danielj
Posts: 8245
Joined: Thu Oct 02, 2008 5:51 pm
Location: Manchester
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by danielj » Thu Dec 22, 2016 6:58 pm

/OE is the same on all the sockets. Any of those sockets can be used, it really doesn't matter which order you use them in. The only real issue with using 37 and 41 is you lose some of your sideways RAM. The installation notes just say to use one of the other sockets if IC27 is already used and you need to put more ROMs in. The reason for that is simply because of the sideways RAM thing - you don't want to lose sideways RAM to a ROM if you don't need to.

I agree with Dave - something else has gone wrong :(

Before you send money to China, I believe the mem-controller (IC20) is a custom IC - all of the CF ICs in the M128 are. The only source for a another one of these will be another BBC Master. I'd guess it's a standard "customisable" gate array though, hence other ICs exist with the same number.

User avatar
hoglet
Posts: 9299
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by hoglet » Thu Dec 22, 2016 7:02 pm

I'm currently using IC41 in my Master to hold a 27C256 EPROM containing two ROM images: MMFS 1.27 and ADFS 1.57.

LK18 is in the east position, disabling sideways RAM in slots 4 and 5 and enabling the 32K ROM in IC41.

LK19 is in the west position, enabling sideways RAM in slots 6 and 7 and disabling the 32K ROM in IC37.

There really is very little difference between IC37 and IC41, apart from the slot number they map to.

Even if you install an ROM and set the link incorrectly, pull-up resistors ensure the ROM is disabled so no bus conflict can result.

It is conceivable in your case the EPROM itself is faulty and is ignoring the CE input, or is slow to response to the CE input, and this has caused a bus conflict. But my gut feeling is still that something else is going on.

Lets see what other people think....

Dave

User avatar
hoglet
Posts: 9299
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by hoglet » Thu Dec 22, 2016 7:08 pm

mw0uzo wrote: Here's a link to the rom install notes:
http://www.8bs.com/submit/acornappspdf/042.pdf
The information in that app note is incorrect:
6. If you put the chip in as IC 41, you will need to change link 18 to its west position. If you put the chip in
as IC 37, you will need to change link 19 to its west position. Both of these links are three pins with a
movable connector. Link 18 is just above link 12 so make sure that you do not confuse the two.
Specifically, to enable the ROM (and disable sideways RAM) the link needs to be in the east position.

The Master 128 Service Manual has it the correct way around:
http://chrisacorns.computinghistory.org ... sterSM.pdf
LK18 plug, made West: Paged ROM/RAM Select. Two position link.

When fitted in the West position, this link causes 16Kbyte of RAM to appear in each of the "sideways" memory "slots" 6 and 7.

When fitted in the East position, a 32Kbyte ROM occupying slots 6 and 7 may be plugged into socket labelled 1C41.
But as I said earlier, incorrect setting of the link does not result in a bus conflict, as pullups disable the ROM.

Dave

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Thu Dec 22, 2016 7:20 pm

Ok, looks like my thinking on that doesn't match up with either of yours. Scratch that then - very unlucky damage to ROM for unknown reason/static. Hopefully its not the mem controller CF30058... Luckily the 65sc12 is available from sprow :D

Thanks for the extra information Dave, wow the note is wrong!! And wrong selection by itself would not have damaged.

I'll have a good poke around with the scope to check further, but probably the next major step is to change RAM ICs or the CPU. Desoldering those will be fun... :lol:

User avatar
1024MAK
Posts: 10101
Joined: Mon Apr 18, 2011 5:46 pm
Location: Looking forward to summer in Somerset, UK...
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by 1024MAK » Thu Dec 22, 2016 7:20 pm

As said above, let's start with the basic checks, tests and information.

With most 1980 to 1990 computers, the most common faults are related to DRAM chips failing, problems with electrolytic capacitors (in the power supply), titanium capacitors going short circuit, oxidation or dirt on connectors, pins, sockets and chips. In addition, where X2 suppression capacitors are used on the mains side of the PSU, these fail. Some of these are more likely to fail a short while after power up if the machine has not been used for very long recently. So it is possible that your Masters may have failed regardless of what you did.

So the first question is, you say both your Masters were working okay beforehand, but how long have you actually had them turned on and used them for? Minutes, or hours?

Second question, how did you clean the motherboards? And how long ago?

Third question, you have been asked about the +5V supply voltage, but I don't see a reply where you say what the actual measured voltage is? Also use the mV AC range to see if there is any ripple (or better still, use a 'scope).

The ROM sockets on the Master, as with all ROM/EPROM sockets, have both /CS and /OE control lines. The links / jumper shunts just select between enabling the ROM socket or the address space to switch the DRAM chips so as to provide sideways RAM. It is unlikely that a simple bus clash would cause your problems. Much more likely is that at some stage the 27C256 got put in the wrong way round. This will result in the incorrect polarity, damaging the chip. If this was then put into another machine, it's possible that it will then short out address, control and/or data lines.

Are you able to attend any of the meet ups as advertised in the upcoming events section?

Mark

User avatar
1024MAK
Posts: 10101
Joined: Mon Apr 18, 2011 5:46 pm
Location: Looking forward to summer in Somerset, UK...
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by 1024MAK » Thu Dec 22, 2016 7:29 pm

I know that apart from the ROM chips, all the other chips are normally soldered. But we are not yet at the stage of desoldering or cutting out chips.

Please use the service manual to check that all the links / jumper shunts are in the correct positions. And check that they are clean and making good electrical contact.

Also make sure that the OS ROM (the mega ROM) is in the correct socket and no pins are damaged.

A test ROM can be made for a BBC model B (there is a thread about this on the forum). I'm not sure however if one for a Master 128 exists. But it should not be that hard to make a very basic test ROM. This can then be used with a 'scope to prove the CPU and a lot of the circuity.

Mark

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Thu Dec 22, 2016 8:24 pm

Ok thx all for replies and help :) I'll rewind and start from the beginning and reply a little later on.

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Fri Dec 23, 2016 2:01 pm

1024MAK wrote:In addition, where X2 suppression capacitors are used on the mains side of the PSU, these fail. Some of these are more likely to fail a short while after power up if the machine has not been used for very long recently. So it is possible that your Masters may have failed regardless of what you did.
Yes these were replaced at the start, one had the type that fail and were badly cracked and the other had a different type that had no signs of failure so they were left as is (they were the red, epoxy dipped type and different design of psu).
1024MAK wrote:So the first question is, you say both your Masters were working okay beforehand, but how long have you actually had them turned on and used them for? Minutes, or hours?
They had both been working for days beforehand, one used to copy the adfs disks and the other to play a few games with the MMC interface.
1024MAK wrote:Second question, how did you clean the motherboards? And how long ago?
Good question. Cleaned last week. The rom sockets were sprayed sparingly with switch cleaner on both. One was sprayed with IPA and blown off with an air compressor. The other was sprayed with IPA, the board started to show signs of white residue on the underside, so it was immediately sprayed with butane/methanol, scrubbed with a toothbrush in the affected areas and blown off with air. Both boards were left to dry before being refitted to the case. I will remove them and double check for any residue on the rear, there is none on top but there could be some non-visible on the bottom of the pcb. - I have removed and inspected the boards, there was a small amount of white residue at the econet header on one and at one of the rom cartridge sockets on the other. I removed it with a toothbrush with a little butane/methanol with no change in fault on either.
1024MAK wrote:Third question, you have been asked about the +5V supply voltage, but I don't see a reply where you say what the actual measured voltage is? Also use the mV AC range to see if there is any ripple (or better still, use a 'scope).
Ok one master at a time, this is the unit with the beeeeeee. The 5V line is 5.05V on all three supply lines. The -5V line is -5.03V. Ripple undetectable on the AC multimeter mV range, any will be too high frequency. I'll check with the scope later.
1024MAK wrote:The ROM sockets on the Master, as with all ROM/EPROM sockets, have both /CS and /OE control lines. The links / jumper shunts just select between enabling the ROM socket or the address space to switch the DRAM chips so as to provide sideways RAM. It is unlikely that a simple bus clash would cause your problems. Much more likely is that at some stage the 27C256 got put in the wrong way round. This will result in the incorrect polarity, damaging the chip. If this was then put into another machine, it's possible that it will then short out address, control and/or data lines.
Ok on the bus clash. I am 99% sure that the rom was never inserted the wrong way around or offset by a pin either way, always double if not triple checked. The pins on the OS rom are not damaged and making good contact.
1024MAK wrote:Are you able to attend any of the meet ups as advertised in the upcoming events section?
This should be possible if needed :) Thank you again for help, it really is appreciated.

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Fri Dec 23, 2016 2:05 pm

The only jumpers changed on either are LK18 and LK19. I will double check with the default link settings in the service manual.

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Fri Dec 23, 2016 4:54 pm

LK60 made between A and B
LK12 B, east
LK18 and 19, west, no ROMs in sockets

That's it, only four selection link headers. The rest are PCB tracks.

User avatar
hoglet
Posts: 9299
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by hoglet » Fri Dec 23, 2016 5:13 pm

What scope do you have?

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Fri Dec 23, 2016 5:15 pm

It's a Gould Digital Storage Type 1421 20Mhz scope.

User avatar
1024MAK
Posts: 10101
Joined: Mon Apr 18, 2011 5:46 pm
Location: Looking forward to summer in Somerset, UK...
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by 1024MAK » Fri Dec 23, 2016 5:41 pm

Is this a dual channel type?

Do you have switchable X1/X10 probes for it?
If yes, set the 'scope up and check that the probes are adjusted correctly (use the calibration test output on the 'scope).

Find a nice solid 0V / ground connection (one of the PSU leads is best if you can get a good connection). Then test the ripple on the +5V line. The -5V is not important.

It's recommended to connect one input of the 'scope to the CPU/system clock, then you can trigger the 'scope on this input.

Display both traces, and set the vertical gain/setting so that the trace from the probe that is testing a signal is at least half the height of the screen.

Then download a pin-out of the CPU, test the clock pin, the R/W pin, each address pin, and each data pin. The actual level does not matter, at this stage we are looking to confirm that the signal is toggling between logic zero and logic one and that a logic zero is 0.8V or less, and that a logic one is 2V or more.

Then do the same on the address pins on one DRAM chip in each group (in each group, they are connected in parallel, no no need to test each address pin on each chip...). But do test the data pins on each chip. As with the tests on the CPU, at thus stage we are only looking to see if the signal is toggling between logic zero and logic one.

Mark

User avatar
hoglet
Posts: 9299
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by hoglet » Fri Dec 23, 2016 5:43 pm

mw0uzo wrote:It's a Gould Digital Storage Type 1421 20Mhz scope.
Can you trigger off the rising edge of RST (pin 40) and display what happens to R/W (pin 34) when the break key is released?

Set the time base to 100us per division so about 1ms of activity is displayed.

Dave

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Fri Dec 23, 2016 5:56 pm

@Mark - Yes I have dual switchable probes. I'll get this set up after Christmas and check it all in detail and take some pics if necessary.

@Dave - I might be able to do that, there is an Arm switch which must enable storage on a trigger input. I'll take a look at the manual.

Have good Christmas everyone :D

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Wed Dec 28, 2016 2:21 pm

Back with the Masters now. Will be able to get some measurements over the day, i'll log them here progressively until I finish detailing all the data and address bus signals.

- Checked 16Mhz clock at IC43 pin 2 - present, it's lower in amplitude than expected and more of a sin wave. I think this is the limitation of the probes/scope, rather than a fault.
- Checked 1,2,4,8 Mhz signals from IC42 4,5,6,7. 1Mhz, 2Mhz, 4Mhz square and amplitude of just over 4V pk-pk falling to under 4V pk-pk at 8Mhz. OK.
- 2Mhz clock to CPU - Good
- RAS to main DRAMs IC 17 and 23 appears good
- There may be a problem with CAS. More to follow.

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Thu Dec 29, 2016 11:36 am

Ok I have investigated further and found some interesting behaviour. :D

Here is a digital capture of the 2Mhz CPU clock at pin 37 and the RAS control signal to the DRAMs. The 2Mhz clock seems OK and so does RAS. I zoomed into the capture (not shown) and the timings seem as expected from the timing diagram in the service manual page 33.
IMG_20161229_101110.jpg
Next I compared the 2MHz clock and CAS. There was some 'ghosting' on the signal, shown:
IMG_20161229_101131.jpg
I took a capture and found some high frequency switching on CAS - not right!
IMG_20161229_101151.jpg
Zoomed in:
IMG_20161229_101238.jpg
So I traced the signal back via the circuit diagram, through the nand gates in IC34 and IC38. Output at IC34 pin 6, CAS was bad, so I checked the 4Mhz clock at pin 5, OK, and the signal at pin 4, bad. The CPU clock at IC38 pin 12 was OK, and source signal DRAMEN at pin 13 was bad, shown:
IMG_20161229_101741.jpg
The bad signal DRAMEN is generated by the memory controller IC20. So ah, maybe bad memory controller. I looked at the IC block in the diagram and figured that DRAMEN is going to be constructed from some signals going into the memory controller, specifically pin 11, 8Mhz clock - OK, pin 13, 2Mhz clock - OK and maybe pin 9, SYNC signal from CPU - very very bad:
IMG_20161229_102841.jpg
That signal is meant to be high when the CPU is fetching an opcode, so high frequency content is not expected, it should be high for at least several clock cycles at a time? * So bad CPU?? *

Finally, I checked the ripple on the 5V supply line at the spade connector nearest the CPU. There was more noise from CPU operation than I expected, spikes at 300mV. No obvious ripple from bad cap in the PSU. The spikes might be from the bad CPU drawing excessive current while it operates incorrectly?
IMG_20161229_104141.jpg
I'll stop here for comments before checking further the CPU. Its possible there could be a bad control signal into the CPU.

User avatar
BigEd
Posts: 3199
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by BigEd » Thu Dec 29, 2016 1:00 pm

Sync should be high for single cycles at a time. It looks like it's somehow ANDed with a 16MHz signal, and that's a surprise to me.

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Thu Dec 29, 2016 1:23 pm

BigEd wrote:Sync should be high for single cycles at a time. It looks like it's somehow ANDed with a 16MHz signal, and that's a surprise to me.
I double checked my work and swapped probes around, because in standard scope mode, the high frequency content was not so easily visible. I got the scope to trigger on a good SYNC pulse - you can see a single pulse followed by the rest which have the faulty signal on.
Attachments
IMG_20161229_131758.jpg

User avatar
BigEd
Posts: 3199
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by BigEd » Thu Dec 29, 2016 2:26 pm

If Sync is sometimes good and sometimes chopped, that's even more surprising. Are you able to change the CPU? Sync is, I think, not even used in a Beeb, so if Sync is bad, it's either shorting against something else (like an adjacent signal) or it's a bad CPU.

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Thu Dec 29, 2016 2:53 pm

A0-15 contain similar signals, a few stable level changes but all mainly containing the high frequency content.

Here is what's on D7, something tried to go high, but with all the noise on the address lines it's junk. Similar pattern on D6, D5. I didn't check further.
IMG_20161229_142424.jpg
And zoomed in
IMG_20161229_142440.jpg
If Sync is sometimes good and sometimes chopped, that's even more surprising. Are you able to change the CPU? Sync is, I think, not even used in a Beeb, so if Sync is bad, it's either shorting against something else (like an adjacent signal) or it's a bad CPU.
Yes I can change it, i'll need to order a few from sprow and desolder [-o< Haven't decided yet on whether to chop the pins and remove or try to desolder the whole IC. I haven't seen any high frequency content on IRQ, RST, NMI or RDY. I am wondering though, if there are some valid outputs without the noise, whether its IC20, the mem controller IC which is knackered and leaking the high frequency rubbish onto the address lines and SYNC.

User avatar
jms2
Posts: 2622
Joined: Mon Jan 08, 2007 6:38 am
Location: Derby, UK
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by jms2 » Thu Dec 29, 2016 3:19 pm

Desoldering without chopping legs is quite easy actually. The way I've taught myself to do it is:

- Support the board vertically. This is quite tricky - I actually made a stand out of plywood to help.
- Put the soldering iron on an IC leg. Wait for the solder to melt. If you have very good hearing, and it's quiet, you can sometime hear this happening.
- On the non component side, put a solder sucker over the leg and suck. About 75% of the time this will work.
- Repeat for all legs.
- Ease the IC out, heating any positions which are sticking.

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Thu Dec 29, 2016 4:16 pm

Thanks for the advice :) My usual method is to use desolder wick and higher heat, but this often fails when there is a heavy track or gnd on the other side of the board. It can sometime tear off a pad with no track if impatient. I've never had success with the solder sucker, but there's always a first time as you make it sound easier!!

User avatar
danielj
Posts: 8245
Joined: Thu Oct 02, 2008 5:51 pm
Location: Manchester
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by danielj » Thu Dec 29, 2016 4:24 pm

Wick is good for cleaning up after sucking, but a good sucker with a flexible nozzle and keeping it emptied is better as, as you said, it's quite easy to damage things with braid. I confess though, for a 40 pin IC which I have a replacement for I'd be tempted just to hack it out and do the pins. The alternative is hot air rework station, or a dedicated through hole desoldering station (and I've still managed to wreck through hole plating with the latter!).

d.

User avatar
1024MAK
Posts: 10101
Joined: Mon Apr 18, 2011 5:46 pm
Location: Looking forward to summer in Somerset, UK...
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by 1024MAK » Thu Dec 29, 2016 7:11 pm

Or an electric paint stripper...

Hold the board vertically (90o to the desk) by a rig or stand.

That's what I use. But don't keep it pointing in the same spot all the time. Move it backwards, forwards and around the chip aiming for the IC pins. At some point, all the solder on all the pins will melt, then a set of dental type pliers can be used to gently ease it out of the PCB. Gentle to prevent hot solder spraying around...

You do need to be careful around nearby components, and practice on a scrap PCB first. Oh, and don't be in a rush, but at the same time, don't take too long or you will cook the board and the chip :lol:

Mark

mw0uzo
Posts: 202
Joined: Wed Dec 21, 2016 9:00 pm
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by mw0uzo » Fri Dec 30, 2016 12:20 pm

Having a tidy up here then I'm going to desolder the CPU. Mark, from what you've seen happening from the signals seen at the CPU would you agree that it's likely CPU death? Is it unwise to power on the computer without the CPU inserted, in case of no RAS/CAS to the DRAMs, damaging them? (That feels like a silly question :lol: , really I just want to be sure that the high frequency carp is definitely not coming from the mem controller. Probably its a case of no further checks can be done, just change CPU and see if it's improved/working. And the mem controller translates the address bus into enables and pipes around the data bus, so no realistic way it could be interfering with such a wide selection of the CPU signals.)

User avatar
hoglet
Posts: 9299
Joined: Sat Oct 13, 2012 7:21 pm
Location: Bristol
Contact:

Re: Dead BBC Masters - ROM screwup leading to main system damage

Post by hoglet » Fri Dec 30, 2016 2:04 pm

mw0uzo wrote:Having a tidy up here then I'm going to desolder the CPU. Mark, from what you've seen happening from the signals seen at the CPU would you agree that it's likely CPU death?
I would explore a bit further before desoldering the 65SC12 CPU.

I just don't see how the 16MHz "junk" could be caused by a faulty 65SC12 CPU, as there is normally no 16MHz signal into the CPU.

Also, make really sure your scope probes are properly grounded.

If you expand the time base on your scope, is the 16MHz "junk" on SYNC precisely 16MHz?

If you put the second scope channel on the real 16MHz clock ( IC42 pin 8 ), are they in phase?

What is pin 39 (O2 out) of the 65SC12 CPU doing?

It seems to be unlikely the 65SC12 would somehow be generating a 16MHz clock that happens to look just like the 16MHz clock that exists elsewhere on the Master. I think something else must be going on.

Also, you said you started with two Masters. Are they both now behaving in this way?

Dave

Post Reply

Return to “8-bit acorn hardware”