Should the Elk have been a 'full' Beeb?

on-topic acorn-related discussions not covered by the other forums
RobC
Posts: 2922
Joined: Sat Sep 01, 2007 10:41 pm
Contact:

Re: Should the Elk have been a 'full' Beeb?

Post by RobC » Sun Feb 24, 2019 1:13 pm

Commie_User wrote:
Sat Feb 23, 2019 10:52 pm
Schools bought the BBC for so long because it was the machine they received grants to buy and because of the time it took to gain ubiquity. Plus teachers took so long to persuade to buy a BBC in the first place, they would have been doubly reluctant to upgrade to the next superwonder before getting the maximum time of familiarity and service from the Beeb.
It's easy to forget that due to its position as the leader in education, there were a wealth of peripherals, high-quality educational software and documentation available for the Beeb that just weren't there for other machines.

Also, the build quality of many other home computers (which was often the result of penny pinching to keep costs down) just wasn't up to use in school environments. With tight school budgets, it made sense to invest sensibly for the long-term.

I can remember my secondary school still usefully using Beebs in the early 90s to teach about the electromagnetic spectrum in physics lessons, to monitor titrations and control lab equipment in chemistry and output satellite images in geography lessons.

Commie_User
Posts: 1309
Joined: Wed Jan 27, 2016 12:50 am
Contact:

Re: Should the Elk have been a 'full' Beeb?

Post by Commie_User » Sun Feb 24, 2019 2:24 pm

chinnyhill10 wrote:
Sun Feb 24, 2019 10:24 am
Having spoken to some older pupils our school went from Pets, to BBC's, to CPC's and then in 1991 to 286 PC's with Windows 3. But for our school the main purpose of the computers was to look good, not actually to teach. Teaching consisted of 'there's a computer, go dick around with it for 40 minutes because you know more than I do'.
But there must have been some IT syllabus? Into the 90s, there certainly was at our secondary school and we Microsoft Office'd the hell out of it. We had the RM Nimbus PCs dominate our area and I bet I know what the pitch was to have the local authority get those in, even in the 80s: 'Look, these machines are already becoming the dominant standard that people will go to work on, so look further forward than Podd and turtles'. That's probably why teachers in our schools had at least a more concrete idea for kids to get around on computers, though the 186 and 286 machines typically ran BBC-a-like learning games and graphics on them.

Commie_User
Posts: 1309
Joined: Wed Jan 27, 2016 12:50 am
Contact:

Re: Should the Elk have been a 'full' Beeb?

Post by Commie_User » Sun Feb 24, 2019 2:27 pm

RobC wrote:
Sun Feb 24, 2019 1:13 pm
It's easy to forget that due to its position as the leader in education, there were a wealth of peripherals, high-quality educational software and documentation available for the Beeb that just weren't there for other machines.

Etc. ..
Those are good points too. But there still comes a point of market obsolescence where a high price can look like an LEA is being taken advantage of. Though by the 90s, I think prices fell anyway.

User avatar
BigEd
Posts: 3184
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: Should the Elk have been a 'full' Beeb?

Post by BigEd » Sun Feb 24, 2019 6:15 pm

B3_B3_B3 wrote:
Fri Feb 22, 2019 10:47 pm
Would using 2 or more smaller ULAs have helped with the deadline / avoided the timing problems?
Interesting thought. I think only in one case: if the super-large ULA was having yield problems. The statistics of chip yield are exponential, so it would be advantageous from this perspective to have two smaller chips. (But the parts cost would surely have increased, and possibly also the design effort.)

I think I've read that the timing problems were disputed: Acorn designed according to how they thought the chip would behave, but were they designing within the specified behaviour of the chip? Had they perhaps first seen some prototypes which were 'fast' rather than 'typical' or 'slow'? These days fabs are well-controlled and for a first run you can dial in a 'split lot' which delivers some fast and some slow. This is good because both fast and slow can show up design problems which will ruin yield. (Just possibly I've mixed up stories about the VidProc and the Elk.)
If Spectrum like contended video was used, could the ULA have done any caching to hide the delay from the CPU?
The ULA was full, so no chance of any kind of cache or buffer, unless of course you throw something out.

Acorn were being very ambitious with the Elk project, and maybe if they'd been any less ambitious the proposition wouldn't have been worth executing. It's very very difficult to lock in a spec 6 or 12 months ahead of time and keep your engineering development on track. There's enough scope for delay when things go wrong without moving the goalposts. And if you do have a problem, there's no way to know if it's going to be an easy one or one which takes forever to fix. And if your product has a 3-4 year life, very difficult to predict component costs and market demand so far ahead. Basically, we tend to see the winners, who took the right risks and were relatively lucky. The others we tend not to be discussing.

As for the Elk being a Beeb - to be an actual Beeb it would need to meet BBC specs, and that was very unlikely to happen. Even if it did, it would take much time to get agreement. To have a mode 7 would mean having a teletext chip, and that means a bytewide RAM access, and that probably means giving up the cost advantage of just four DRAM chips. Remember, every dollar on parts cost gets quadrupled by the time you reach the shops. Staying under £200 would be very difficult, and even that turns out to be undercut by Sinclair. A price war is a pretty ugly thing, especially if your competition's product has been designed from the start with low price in mind.

User avatar
BigEd
Posts: 3184
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: Should the Elk have been a 'full' Beeb?

Post by BigEd » Sun Feb 24, 2019 6:15 pm

BigEd wrote:
Sun Feb 24, 2019 6:15 pm
B3_B3_B3 wrote:
Fri Feb 22, 2019 10:47 pm
Would using 2 or more smaller ULAs have helped with the deadline / avoided the timing problems?
Interesting thought. I think only in one case: if the super-large ULA was having yield problems. The statistics of chip yield are exponential, so it would be advantageous from this perspective to have two smaller chips. (But the parts cost would surely have increased, and possibly also the design effort.)

I think I've read that the timing problems were disputed: Acorn designed according to how they thought the chip would behave, but were they designing within the specified behaviour of the chip? Had they perhaps first seen some prototypes which were 'fast' rather than 'typical' or 'slow'? These days fabs are well-controlled and for a first run you can dial in a 'split lot' which delivers some fast and some slow. This is good because both fast and slow can show up design problems which will ruin yield. (Just possibly I've mixed up stories about the VidProc and the Elk.)

Edit: so, if there's a design problem in a ULA, if the process delivers chips which don't perform as expected, it's not clear that having two smaller ULAs helps much.
If Spectrum like contended video was used, could the ULA have done any caching to hide the delay from the CPU?
The ULA was full, so no chance of any kind of cache or buffer, unless of course you throw something out.

Acorn were being very ambitious with the Elk project, and maybe if they'd been any less ambitious the proposition wouldn't have been worth executing. It's very very difficult to lock in a spec 6 or 12 months ahead of time and keep your engineering development on track. There's enough scope for delay when things go wrong without moving the goalposts. And if you do have a problem, there's no way to know if it's going to be an easy one or one which takes forever to fix. And if your product has a 3-4 year life, very difficult to predict component costs and market demand so far ahead. Basically, we tend to see the winners, who took the right risks and were relatively lucky. The others we tend not to be discussing.

As for the Elk being a Beeb - to be an actual Beeb it would need to meet BBC specs, and that was very unlikely to happen. Even if it did, it would take much time to get agreement. To have a mode 7 would mean having a teletext chip, and that means a bytewide RAM access, and that probably means giving up the cost advantage of just four DRAM chips. Remember, every dollar on parts cost gets quadrupled by the time you reach the shops. Staying under £200 would be very difficult, and even that turns out to be undercut by Sinclair. A price war is a pretty ugly thing, especially if your competition's product has been designed from the start with low price in mind.

chinnyhill10
Posts: 135
Joined: Mon Apr 27, 2015 8:37 pm
Contact:

Re: Should the Elk have been a 'full' Beeb?

Post by chinnyhill10 » Sun Feb 24, 2019 8:59 pm

Commie_User wrote:
Sun Feb 24, 2019 2:24 pm

But there must have been some IT syllabus? Into the 90s, there certainly was at our secondary school and we Microsoft Office'd the hell out of it.
Well technically there was as they included it in "CDT". But they didn't actually teach anything. You got a term on the computers, the rest of the time you were firing nail guns behind the teachers back, melting bits of plastic and wondering if you could cut the annoying kid in half on the big saw like in a Bond film. They really didn't care less or understand the machines. All that mattered was they had computers, NEW computers, and this looked good to outsiders. It was for show, not for us.

When I transferred schools in '93 they had a suite full of Archies and an alcoholic IT teacher. We didn't have formal IT as 5th years at that school but we had a free period in the IT suite and Eddie the IT teacher did help those of us who were keen to do stuff. It was my first exposure to the Archie. I'd devoured details of them in New Computer Express. Used a DTP package and a rather lacklustre art package.

Poor Eddie, he was a chronic alcoholic. Reeked of mints/booze and had bottles of scotch 'hidden' under the desk in his office. On the way home from school most nights, Eddies car would be outside the boozer in the next village. At 15/16 this seemed hilarious. At 40 it seems bloody tragic.

So yeah, the IT at secondary school was somewhat lacking. :-)

chinnyhill10
Posts: 135
Joined: Mon Apr 27, 2015 8:37 pm
Contact:

Re: Should the Elk have been a 'full' Beeb?

Post by chinnyhill10 » Sun Feb 24, 2019 9:01 pm

Commie_User wrote:
Sun Feb 24, 2019 2:27 pm
RobC wrote:
Sun Feb 24, 2019 1:13 pm
It's easy to forget that due to its position as the leader in education, there were a wealth of peripherals, high-quality educational software and documentation available for the Beeb that just weren't there for other machines.

Etc. ..
Those are good points too. But there still comes a point of market obsolescence where a high price can look like an LEA is being taken advantage of. Though by the 90s, I think prices fell anyway.
A lot of people did VERY well exploiting that market. I'd hope things are tighter these days

B3_B3_B3
Posts: 97
Joined: Sat Apr 08, 2017 10:42 pm
Contact:

Re: Should the Elk have been a 'full' Beeb?

Post by B3_B3_B3 » Sun Feb 24, 2019 9:02 pm

BigEd wrote:
Sun Feb 24, 2019 6:15 pm
B3_B3_B3 wrote:
Fri Feb 22, 2019 10:47 pm
Would using 2 or more smaller ULAs have helped with the deadline / avoided the timing problems?
Interesting thought. I think only in one case: if the super-large ULA was having yield problems. ....
I was thinking it might have allowed use of older more established ULA technology whose timing properties were reliable/ matched the documentation etc.
Last edited by B3_B3_B3 on Sun Feb 24, 2019 9:28 pm, edited 1 time in total.

User avatar
BigEd
Posts: 3184
Joined: Sun Jan 24, 2010 10:24 am
Location: West
Contact:

Re: Should the Elk have been a 'full' Beeb?

Post by BigEd » Sun Feb 24, 2019 9:36 pm

Ah, yes indeed, that's a possibility!

B3_B3_B3
Posts: 97
Joined: Sat Apr 08, 2017 10:42 pm
Contact:

Re: Should the Elk have been a 'full' Beeb?

Post by B3_B3_B3 » Sun Feb 24, 2019 10:31 pm

BigEd wrote:
Sun Feb 24, 2019 6:15 pm
If Spectrum like contended video was used, could the ULA have done any caching to hide the delay from the CPU?
The ULA was full, so no chance of any kind of cache or buffer, unless of course you throw something out.......
Others have suggested providing Elk sound via the bbc sound chip, which could have left some space and saved some development time, as well as retaining compatibility?

EDIT: Also, the contended video suggestion was as suggested as a way of making 8 bit wide RAM affordable by reducing(halving) the required access speed, which would then remove the need for a ULA nibble to byte section, so more free ULA space:
but perhaps someone else's suggestion of slower 8 bit RAM with 1MHz CPU with interleaved 1Mhz video access (presumably only 10K modes then?) would be simpler and speed more predictable)?
Last edited by B3_B3_B3 on Mon Feb 25, 2019 7:43 pm, edited 2 times in total.

Commie_User
Posts: 1309
Joined: Wed Jan 27, 2016 12:50 am
Contact:

Re: Should the Elk have been a 'full' Beeb?

Post by Commie_User » Sun Feb 24, 2019 10:43 pm

chinnyhill10 wrote:
Sun Feb 24, 2019 8:59 pm

Poor Eddie, he was a chronic alcoholic. Reeked of mints/booze and had bottles of scotch 'hidden' under the desk in his office. On the way home from school most nights, Eddies car would be outside the boozer in the next village. At 15/16 this seemed hilarious. At 40 it seems bloody tragic.
Well, talk about the importance pinned to computers being variable at best.

Our computer teacher, Hupe, was a gross lardbucket who was easily set into a near-bellowing temper. Yet we got stuff done. Though looking back, given resetting a machine would crash the server virtually every time and other niggly crap, I'll at least give him some leeway.

User avatar
algenon_iii
Posts: 85
Joined: Sat Nov 25, 2006 6:49 pm
Location: Cardiff
Contact:

Re: Should the Elk have been a 'full' Beeb?

Post by algenon_iii » Wed Apr 24, 2019 8:21 pm

I'm sure I've posted some of these thoughts before but here goes anyway.

I've got copies of old Acorn User magazines (83ish) advertising the BBC A for £299. Far bigger PCB and case than the Elk, plus stuff like the teletext chip should have provided possibilities for cost cutting. Design out all of the PCB and sockets / pads that aren't required (have the same standard ports as the model A), get a smaller case, perhaps a chiclet keyboard like the PCjr or spectrum+, with perhaps an enhanced ULA to replace some board logic. Admittedly upgrading the RAM to 32k would have had a cost.

As well as the hardware over the years I've heard rumours about the financial side of things. Did Acorn pay the BBC to use the brand name which I'm sure read/heard that they did? Did Acorn actually offer schools discounts and make a profit, I've heard figures between 20-50% mentioned over the years. If so it might well have been possible to cost engineer the price down of a stripped down Acorn electron badged BBC A (with 32k) to the magical £200 mark.

Another interesting question is did Acorn pass on the design and development costs of the ULA directly onto the Elk buyer or did they just spread the cost as overheads?

We all know the delay hurt the Elk badly and made a real issue out of price and spec (esp. the amount of RAM) because fundamentally the Elk was a cut down version of a 1981 computer (albeit a very good computer). The Spectrum 48 by 1983 was retailing at just below £130 and with the Elk listed at £200 in '83 if you could get one - the Elk had one chance and missed it. By mid-1984 the C64 was retailing at around £200. In mid-1984 the CPC 464* came out, £250 for computer with built-in tape deck and a mono monitor. The Boots advert from Christmas 1984 (courtesy of @russty_russ on twitter) is quite sobering, the competition it was facing was tough.

https://pbs.twimg.com/media/Cy8Vkd5W8AAk06r.jpg

At the time of that advert in late-Nov early-Dec the PCN sales charts of the time were C64, Spectrum (not including the new Spectrum+), CPC (colour?) and the Elk. Then you'd have the beeb, spectrum+, C16, MSX etc. Just six months later in the last issue of PCN it was Spectrum, C64, CPC, Beeb, Atari 800XL, Elk.

* Amstrad had ULA problems with Ferranti as well, legend has it that Alan Sugar picked up the phone and gave one of bosses a right earbashing and eventually LSI Logic made the chip for them.

chinnyhill10
Posts: 135
Joined: Mon Apr 27, 2015 8:37 pm
Contact:

Re: Should the Elk have been a 'full' Beeb?

Post by chinnyhill10 » Sat Apr 27, 2019 12:54 am

algenon_iii wrote:
Wed Apr 24, 2019 8:21 pm


* Amstrad had ULA problems with Ferranti as well, legend has it that Alan Sugar picked up the phone and gave one of bosses a right earbashing and eventually LSI Logic made the chip for them.
They commissioned LSI after Sugar used some 'industrial language" down the phone to Ferranti when the engineers revealed they had fallen behind with the chips. But he didn't sever the contract. He kept it as an insurance policy and both LSI and Ferranti ended up making the Gate Array. As a result the 464 came out on schedule.

Post Reply

Return to “general”