BBC B+ 64K

[Tony L]

I took a fairly risky punt on another BBC B on eBay (I already have a really nice Issue 4 with DFS). I did so based purely on spotting the serial was ANB51 xxxx so theoretically a 64K machine and figured I wouldn’t lose much if it wasn’t. It paid off as it was indeed what the serial suggested, and after a long day deep-cleaning and doing a basic PSU recap (just the three caps as that’s all I had to hand, the Rifas had already ecploded) and voltage test it looks to be working and is actually in superb condition:



Here’s the mainboard:



I thought I’d start a thread on it as I suspect it is a rather interesting machine as it is obviously the later 64K variant, which seems rare in itself, but is nobbled by not having DFS, which to my mind makes it pretty much useless for its time! Why would someone pay for 64K but want to put up with a cassette deck?!

I’m curious what I can learn from this machine. I’m very tempted to install a 1770 DFS, though I guess that might devalue it from a ‘collector perspective’. It isn’t socketed and there is no sign of rework at all, to be honest I don’t think the mainboard had ever been out until I stripped it down to thoroughly clean the case (I’m totally obsessive with computer hygiene, e.g. I’ve pulled and washed all the keys etc). I do have a 40 track 5 1/4” disk drive and MicroSPI SD card on the other machine.

So, what would folk recommend I do with it? I assume it can teach me a fair bit about sideways ROM/RAM etc, a topic I’m still a noob at.

On a related topic does anyone know if it is possible to burn ROMs using a Mac? All the software and hardware I’ve seen seems to be PC based, which is rather annoying. This machine has Prestel installed, not that that’s any use to me!

PS One thing I’m noticing is it is very noisy, quite a bit of white noise and other background sounds from the speaker. I will eventually recap it fully (e.g. mainboard as well as the rest of PSU), I did that to my Issue 4 and it seemed to make it quieter and better behaved all round.

[Tony L]

Damn it, I spoke too soon. Just had a quick play and all other modes aside from the default mode 7 are corrupted. Flashing bars, malformed characters etc. The machine is still responsive, e.g. will change modes even though the screen is garbled, but something is clearly very wrong with it. Annoyingly only one chip is socketed so diagnosis will be a pain. Any ideas?

[Pernod]

I’m curious what I can learn from this machine. I’m very tempted to install a 1770 DFS, though I guess that might devalue it from a ‘collector perspective’.

The ANB51 did indeed not have the 1770 and DFS fitted. I don't think these are very common at all, it's the first I've seen.

[Tony L]

I’m still very much on a learning curve here, but likewise it is the first one I’ve seen evidence of too. As I say I can’t really see the point of it as given the choice I’d have picked a standard B with DFS at the time which I guess would have been about the same price. Again I run out of knowledge pretty fast here, but I assume the 64 and the standard B ran concurrently for a while? As far as I can tell neither the 64K B+ or the B+ 128 were around for long at all as the Master arrived within months and was so much more powerful. My B+ is dated Jan 85 according to a sticker on the DIN sockets and the Master arrived early 86.

PS Here’s a quick YouTube video of the display corruption. Easier to film it than try and describe it! As you can see the machine remains responsive through mode changes, it hasn’t crashed. https://youtu.be/qljQIG933Vk All the noise on the video is coming from the B+ speaker!

[DutchAcorn]

What about shadow modes, do they behave the same? E.g. MODE 128?

[Tony L]

Yes, all modes are corrupt aside from 7 and 135. The nature of the corruption looks to have changed since I turned it back on and it now isn’t doing the whole screen blinking thing, but the text is blinking, corrupt or missing in all the non-teletext modes.

PS Even though I’m seeing different things from earlier the modes look the same, e.g. mode 0 and mode 128 are broken in the same way, as are mode 1 and 129 etc.

[cmorley]

Bad video ULA?

[Tony L]

Hope not, I assume that is pretty unobtanium being a bespoke chip!

To be honest I’ve no idea how different this thing is from a normal B. It certainly looks different, though I’ve not gone through identifying specific parts yet. The only thing in a socket is the 6512, whatever that is. I tried lifting, cleaning and reseating that with no difference. I’ve not gone through checking all the jumpers etc, but I assume they are ok given it boots into mode 7 fine?

PS I have got a Hakko FR301 desoldering gun so can probably get chips out cleanly enough (I use it for recapping audio kit), but I don’t have much diagnostic knowledge nor any understanding of this specific machine. My other standard B is far easier to deal with as it is nearly all socketed plus I have a lot of spares! Given the potential rarity of this B+ I’m tempted to get a grown-up to deal with it, e.g. RetroClinic, but if we can narrow it down to something simple I’ll give it a go!

[gfoot]

The 6512 is the CPU, very similar to the 6502.

It looks like that board supports both 1770 and 8271 floppy disk controllers, I didn't know they did that.

The flashing effect sounds to me like an issue with the video address transceivers or maybe the tracks leading from the RAM to the ULA, assuming it's mostly similar to a regular beeb in that respect. Or the ULA but let's hope not.

[Tony L]

I’d be amazed if it was a track issue, the board is absolutely immaculate. No sign of rework, corrosion etc, it wasn’t even dusty so I haven’t cleaned it - the picture is how it was and still is (aside from I’ve since pulled the Prestel ROM). Really looks like a new one. The whole computer is superb condition wise, e.g. all keys have their full texture etc, no wear. I suspect I was the first in there aside from the Prestel ROM being fitted by the original owner.

Is the V-ULA bespoke to a B+ or the same as a later B? My Issue 4 has the old Ferranti one with the big heatsink. I’ve stockpiled a fair few spares for my Issue 4, but I’ve little that would be relevant here I think.

PS Much better resolution of the mainboard on my Flickr account here: https://flic.kr/p/2mYrwjx

[danielj]

Is the V-ULA bespoke to a B+ or the same as a later B? My Issue 4 has the old Ferranti one with the big heatsink. I’ve stockpiled a fair few spares for my Issue 4, but I’ve little that would be relevant here I think.

The ULAs are the same up to the Master. Obviously the later ones that don't need the heatsink are better, and you don't want the very early one that needed a patch to invert it.

Which ones need a heatsink are detailed here: viewtopic.php?p=43650#p43650

[Tony L]

Excellent, thanks. That suggests finding one shouldn’t be impossible. If that looks like the most likely culprit here I’ll stick a wanted ad up for a nice clean one of a the same era and then socket it.

[arg]

As I say I can’t really see the point of it as given the choice I’d have picked a standard B with DFS at the time which I guess would have been about the same price. Again I run out of knowledge pretty fast here, but I assume the 64 and the standard B ran concurrently for a while? As far as I can tell neither the 64K B+ or the B+ 128 were around for long at all as the Master arrived within months and was so much more powerful. My B+ is dated Jan 85 according to a sticker on the DIN sockets and the Master arrived early 86.

The B+ was originally intended to be just the next revision of the B - the RAM change due to the relative cost/availability of the chips used. So it should have retailed at the same price as the older B which would have been withdrawn.

However, it came at the point when Acorn were in financial trouble and stuck with excess stock of existing machines. So the B+ was initially sold only with DFS fitted as a higher price-point machine alongside the old-stock standard B. And as you note, the Master was along shortly afterwards.

So your machine is a bit of an oddity and it's not clear what it would have cost. Did they eventually manage to shift all the stocks of standard B's and start selling base-model B+? Maybe after the Master was out the base-model B+ was still available as the cheaper option?

Is the V-ULA bespoke to a B+ or the same as a later B? My Issue 4 has the old Ferranti one with the big heatsink. I’ve stockpiled a fair few spares for my Issue 4, but I’ve little that would be relevant here I think.

Video ULA is nominally the same between B/B+, but you probably can't swap the one out of your Issue 4: the Ferranti parts needed a big resistor and some diodes to generate an internal supply voltage, which wasn't needed for the later VTI vidprocs. I believe later production omitted those parts, especially the B+ that was always built with the VTI vidproc.

[Tony L]

So your machine is a bit of an oddity and it's not clear what it would have cost. Did they eventually manage to shift all the stocks of standard B's and start selling base-model B+? Maybe after the Master was out the base-model B+ was still available as the cheaper option?

Yes, it has me baffled, and even more so now I see the date inside (8 Jan 85). I found Wikipedia and the mention on Chris’s Acorns site: https://chrisacorns.computinghistory.or ... CB+64.html

Both suggest that the B+ didn’t arrive until mid-85 and did so with the 1770 DFS fitted as standard, i.e. it was only around for a few months before the Master arrived. The date on my mainboard obviously predates the magazine announcements and reviews of June and July 85 where the assumption is the B+ has DFS fitted as a standard feature. I have googled-up other references that the ANB51 (no DFS model) existed at least theoretically, though I couldn’t find any hard evidence of another actual example. I wonder if it was a prototype? It came in a really manky set of polys but sadly no box or manual (by saying that can’t complain as that’s why it slipped through eBay as a bog-standard B!). The function key strip has no mention of 64K, but it seems the ones linked in the reviews and Acorn’s own publicity didn’t either, so I’ve no reason to believe this has been swapped. My best guess is it is early, not late in the admittedly short production run. Full serial is ANB51-3503282 if anyone knows how to pick that apart.

Video ULA is nominally the same between B/B+, but you probably can't swap the one out of your Issue 4: the Ferranti parts needed a big resistor and some diodes to generate an internal supply voltage, which wasn't needed for the later VTI vidprocs. I believe later production omitted those parts, especially the B+ that was always built with the VTI vidproc.

I’m definitely not going to botch it. If I can be sure it is the V-ULA I’ll go hunting for a perfect period correct example, I don’t mind paying a good price for the right one as it is an interesting machine in otherwise beautiful condition that deserves saving IMO. I want the only change to be that it is socketed!

Has anyone any other tests I could do in mode 7 to establish health/narrow things down, e.g. is there a simple Basic type-in I could do to check the RAM or anything?

[paulb]

As I say I can’t really see the point of it as given the choice I’d have picked a standard B with DFS at the time which I guess would have been about the same price. Again I run out of knowledge pretty fast here, but I assume the 64 and the standard B ran concurrently for a while? As far as I can tell neither the 64K B+ or the B+ 128 were around for long at all as the Master arrived within months and was so much more powerful. My B+ is dated Jan 85 according to a sticker on the DIN sockets and the Master arrived early 86.

The B+ was originally intended to be just the next revision of the B - the RAM change due to the relative cost/availability of the chips used. So it should have retailed at the same price as the older B which would have been withdrawn.

However, it came at the point when Acorn were in financial trouble and stuck with excess stock of existing machines. So the B+ was initially sold only with DFS fitted as a higher price-point machine alongside the old-stock standard B. And as you note, the Master was along shortly afterwards.

It is interesting to read the critical reception of the B+ in the Acorn-related media at the time ("B+ Grading", Acorn User, July 1985), not to mention the broader computing press ("D- for BBC B+", Your Computer, June 1985), as well as the response from various dealers. I think I read that Watford were doing a good trade in selling off Model B stocks at a discount and were generally unhappy with the prospect of having to sell the B+. Their advert in the same issue as the B+ review gives a price of £289 plus VAT for a Model B, so I suppose the B+ may have helped Beeb sales along somehow.

The following article is interesting in that it publicly states that the B+ was based on the ABC mainboard, but it also makes some remarks about the change from Intel 8271 to WD1770 disk controller:

"BBC B+ Premium Product"

Alongside the need to switch to 64-kilobit RAM chips (following the Electron), cost reduction is mentioned in the article in relation to the motherboard. I think I read an opinion from Guy Kewney in PCW to the effect that Acorn had largely failed to reduce costs in line with its competitors. Certainly, the chip count in the Beeb range would have been a hindrance when combined with the need to preserve compatibility and feature parity, even though one could have imagined deleting certain aspects of the Model B specification (the speech upgrade capability, for example) from the B+. And if the B+ did largely leverage the ABC mainboard design, I guess the specification would have been locked in some time before, and cost reduction would not have been the primary goal, anyway.

[Tony L]

Yes, pretty damning reviews! As stated I really can’t see the point of the ANB51 model I have which lacks DFS, that one must have looked hopelessly out-dated and pretty pointless by 1985. I’d have thought the DFS equipped B+ had something to offer, but 16 bit stuff was starting to appear at that time.

The B+ build is very interesting given the earlier Electron. It seems they actually took very little from that project as whilst the B+ looks a little more simple than the B it is still far from the comparative simplicity and LSI of the Elk. I assume they were a bit on the back foot due to the Elk’s well documented ULA issues etc. I don’t know enough/anything much about the Master to know how that’s laid out, I’ll have a google sometime. It is a machine I don’t know at all. I’ve got an working Elk and matching tape deck, but no add-ons which makes it a bit of a faff to use.

PS One thing one of the reviews got spot on was the relocation of the ROMs in the B+ is really not good. Very hard to get them out compared to the standard B due to the case back being in the way. They certainly didn’t think that one through!

[gfoot]

Has anyone any other tests I could do in mode 7 to establish health/narrow things down, e.g. is there a simple Basic type-in I could do to check the RAM or anything?

It might be interesting to see a close up of what happens in modes 1 and 2 with other colours.

[tricky]

I was also going to suggest exercising the palette to see what happens.
You would get the effects that you are seeing if you only partially program the palette.
Does the flashing stop if you hold break or change the flash periods.
There is a thread to a great new website all about the versions of manufacture and serial numbers, but I can't find the thread.

[eelco108]

The problem might well be one of the DRAM chips - it looks like (from the image) that one of the eight DRAM ICs has already been replaced, as the text on that iC is shifted somewhat (the bottom right one in the image) compared to the other seven, and the colour seems somewhat different too (although that could be the lighting when the picture was taken).

This also happens to be exactly the same one that died in my own BBC B+ last year ... and was a pain to replace. Did fix it, though.

Of course if it has already been replaced, the culprit might be one of the other ones.

[Tony L]

Thanks. The screen flashing does stop if I hold break. The background colour seems to work as expected, the foreground being the issue. I stuck another video up: https://youtu.be/hXGDCXEpBRQ

If there is a possibility it is a RAM chip is there a way of testing that in mode 7 in Basic with a type-in, e.g. can you just read/write every byte somehow? I have to admit my Basic is really rusty and was never great at all. I learned the real rudiments of programming from a friend back in the ‘80s who had a Beeb, but really just variables, condition statements, loops etc. Nothing fancy. I later ended up working with COBOL, which obviously does nothing fun at all.

[arg]

Thanks. The screen flashing does stop if I hold break. The background colour seems to work as expected, the foreground being the issue. I stuck another video up: https://youtu.be/hXGDCXEpBRQ

If there is a possibility it is a RAM chip is there a way of testing that in mode 7 in Basic with a type-in, e.g. can you just read/write every byte somehow?

This really doesn't look like a RAM problem, as the screen doesn't become permanently corrupt - you could confirm by filling the screen with more stuff, but it seems fairly clear that the RAM is 'remembering' completely correctly.

It looks very much like the palette register isn't being set correctly - and only one bit of it as shown by the behaviour in Mode 0, where instead of the whole character cell flashing it flashes in stripes (since the palette hardware still works on 4-bit chunks as for 16-colour modes; in 2-colour modes the 'colour' settings are spread out over 4 pixels instead).

You can almost certainly find a palette setting (via VDU 19) that gives you a steady display and eventually figure out which bit isn't working.

Points the finger towards the Vidproc, but not necessarily a dud vidproc as such - could be failed connection preventing the CPU writing to it correctly.

[1024MAK]

The 6512 is the CPU, very similar to the 6502.

Yes, it’s a 65x2 class CPU.

It looks like that board supports both 1770 and 8271 floppy disk controllers, I didn't know they did that.

As far as I am aware, all the B+ PCBs had this.

The flashing effect sounds to me like an issue with the video address transceivers or maybe the tracks leading from the RAM to the ULA, assuming it's mostly similar to a regular beeb in that respect. Or the ULA but let's hope not.

Yes, I tend to agree. The fault is in the digital circuitry. It looks like it’s in either the part that addresses the DRAM (so from the 6845 CRTC to the DRAM chips via the various multiplexers, or the control circuits for the multiplexers) or maybe in the data bus from the DRAM to the video processor (Video ULA). Or maybe the video processor itself.

Keep in mind that some people may have bought a B+ to replace a tape only BBC B. One of the reasons for the introduction of the B+ was that the 16k bit DRAM chips used in the model A and model B had become obsolete. The more popular other 16k bit DRAM chip used in other computers, the 4116, was also obsolete, but Sinclair for example had excessive stock of built ZX Spectrums, which is why the launch in the U.K. of the 128K model was delayed…

Mark

[arg]

So your machine is a bit of an oddity and it's not clear what it would have cost. Did they eventually manage to shift all the stocks of standard B's and start selling base-model B+? Maybe after the Master was out the base-model B+ was still available as the cheaper option?

Yes, it has me baffled, and even more so now I see the date inside (8 Jan 85). I found Wikipedia and the mention on Chris’s Acorns site: https://chrisacorns.computinghistory.or ... CB+64.html

Both suggest that the B+ didn’t arrive until mid-85 and did so with the 1770 DFS fitted as standard, i.e. it was only around for a few months before the Master arrived. The date on my mainboard obviously predates the magazine announcements and reviews of June and July 85 where the assumption is the B+ has DFS fitted as a standard feature. I have googled-up other references that the ANB51 (no DFS model) existed at least theoretically, though I couldn’t find any hard evidence of another actual example. I wonder if it was a prototype?

Yes, that date does emphasise what was going on here - the B+ was really just the "Issue 8" of the original BBC, and the Master was the redesign for greater integration etc. But everything got pushed back as Acorn went through the shock of switching from "can't build them fast enough" to having vast excess inventory. So your Jan '85 B+ entered the back of the queue, with the higher spec ones jumping the queue as they could be sold at higher price.

Also worth noting that by this time ARM was in full swing, and there was something of a 'brain drain' into ARM-related work depriving the rest of the business: the Master was largely done by different people from the original BBC.

I don't have any firm info about what delayed the Master. Most likely money wasn't available to authorise the start of work, but I'm just speculating about that.

[BeebMaster]

Very interesting, one I have also escaped as a tape-only exhibit:



Mine also has Prestel, so does this suggest a limited run of B+ Prestel terminals, not needing a disc interface, perhaps?

I don't think anyone has ever seen one with 8271.

[Tony L]

Thanks folks. I haven’t figured out how to do multi-quotes on this forum software short of manually editing tags etc (seems different to vBulletin or XenForo) so a bit general in response!

I’ll need to go and learn the VDU command parameters!

I will at some point pull the board out again and have a very close look underneath the V-ULA with a loupe and maybe reflow it. I won’t be able to get to that today but I’ll certainly have a better look and take a decent picture or two. I saw nothing yesterday that suggested anything had been reworked, certainly no flux, discolouration or any of the usual signs. I was surprised by how good it looked.

I do have a good multimeter so could continuity test any tracks anyone wanted me to. I’ve also got an oscilloscope, but I’m a total noob there and would prefer to practice on something I didn’t care about. I’d be nervous working on a live board I could kill with an imprecise probe position! To be honest if we can pin this down to say the V-ULA and I can find one I’ll replace it. If it is looking much more complex than that I’ll just pay a pro to do it. I’m very reluctant to get to a point where I risk making things a lot worse.

PS I’ll obviously type in anything anyone suggests, just don’t assume I know what I’m doing!

[1024MAK]

Thanks. The screen flashing does stop if I hold break. The background colour seems to work as expected, the foreground being the issue. I stuck another video up: https://youtu.be/hXGDCXEpBRQ

If there is a possibility it is a RAM chip is there a way of testing that in mode 7 in Basic with a type-in, e.g. can you just read/write every byte somehow? I have to admit my Basic is really rusty and was never great at all. I learned the real rudiments of programming from a friend back in the ‘80s who had a Beeb, but really just variables, condition statements, loops etc. Nothing fancy. I later ended up working with COBOL, which obviously does nothing fun at all.

There are various BASIC programs listed or linked to in the Fault Finding Index topic. None specifically for the B+, but from the CPU’s point of view, the memory map is effectively the same for the ‘main’ RAM.

And the vast majority of DRAM chip faults cause a failure across the full address range of that chip. This is why this fault is less likely to be a DRAM chip fault, as if it was a typical DRAM chip fault, the system would not be capable of running any BASIC commands.

Mark

[Tony L]

Mine also has Prestel, so does this suggest a limited run of B+ Prestel terminals, not needing a disc interface, perhaps?

That is very interesting! To my mind that makes more sense than selling a machine without DFS in 1985, especially (AIUI) the high initial price of DFS in the B was down to the price of the already obsolete and unavailable 8721 chip, the 1770 was apparently far cheaper, so no good financial reason not to include it.

Good to see another one anyway, and with a Prestel 1.1 eprom in the same socket!

[1024MAK]

Thanks folks. I haven’t figured out how to do multi-quotes on this forum software short of manually editing tags etc (seems different to vBulletin or XenForo) so a bit general in response!

To do multi-quotes, use the quote button for the first message you want to quote. This then opens the text editor box.

Now scroll down the screen / web page, and again use the quote button for the next message you want to quote. This will then be added to the text editor box:

Mine also has Prestel, so does this suggest a limited run of B+ Prestel terminals, not needing a disc interface, perhaps?

That is very interesting! To my mind that makes more sense than selling a machine without DFS in 1985, especially (AIUI) the high initial price of DFS in the B was down to the price of the already obsolete and unavailable 8721 chip, the 1770 was apparently far cheaper, so no good financial reason not to include it.

Good to see another one anyway, and with a Prestel 1.1 eprom in the same socket!

Continue as required…

With only two examples, it’s not really enough for a pattern yet, but Ian (BeebMaster) may have a good point. Acorn machines have been known to be sold as a package for specific markets.

Mark

[Tony L]

There are various BASIC programs listed or linked to in the Fault Finding Index topic. None specifically for the B+, but from the CPU’s point of view, the memory map is effectively the same for the ‘main’ RAM.

And the vast majority of DRAM chip faults cause a failure across the full address range of that chip. This is why this fault is less likely to be a DRAM chip fault, as if it was a typical DRAM chip fault, the system would not be capable of running any BASIC commands.

Thanks, I just tried the shorter one:

10 ?&4000=0
20 A=?&4000
30 IF A=0 THEN GOTO 20
40 PRINT A

It doesn’t ever seem to break out of the loop, which I assume means it hasn’t hit an error. The longer one might be problematic given I have no way to save/load at present and I’m not sure my typing is that accurate! I didn’t include the ‘mode 0’ statement to keep it in mode 7.

[gfoot]

Two tests would be good:

Code: Select all

MODE 2:VDU 19,0,7;0;19,7,0;0;

It should just invert the screen (black text on white background) but it'll be interesting to see whether the background flashes instead now, I'm expecting it to.

And something like this, as close up as you can while in focus and keeping the whole pattern in view:

Code: Select all

10 O%=0
20 FOR I%=0 TO 255
30 ?(&5800+(I% AND 63)+640*(I% DIV 64)) = I%+O%
40 NEXT

It will exercise all the combinations of byte values and let us see which work and which don't. I'm most interested in MODE 2 but MODE 5 and MODE 4 could be interesting as well.

O% is an optional offset. O%=8 might be interesting but I don't want to stack up too many tests at once!