Sinclair QL Design issues

[1024MAK]

Continued from here

Let's be honest, the notoriously game-o-phobic Sir Clive Sinclair wasn't much of a businessman and as much as toes curl at what Alan Sugar is these days, it's Amstrad that breathed new life into the Speccy. People want games, that's that. Sir Clive got the pricing right but the marketing very wrong. The QL was a nice idea on paper - an affordable serious computer with significantly better tech specs than a Spectrum. It's just that the QL was fudged.

Incidentally, does anyone have any evidence for the widely-held view that Clive Sinclair disapproved of the Spectrum being used for games? The fictional Sir Clive in Micro Men is portrayed that way, but I am unaware of the real Sir Clive ever saying anything like that.

I think the best evidence is all of Sinclair's publicity and catalogues up to about early 1985 or so. All of them pitch the Spectrum as a general purpose computer with a lot of scientific and educational uses. I'd also say the QL is a big indication -.it's widely considered true that Sinclair was angry the Spectrum wasn't taken more seriously. The only post-takeover machine he ever released was the Z88 as well - try gaming on that!

In October 1984, Sinclair paid £100,000 for Bandersnatch to be made for the QL with an option on Psyclape. "Ideal for the type of QL user we envisage" said Alison Maguire, Sinclair Software Manager. Doesn't sound very "game-o-phobic".

https://spectrumcomputing.co.uk/page.ph ... 650&page=5

Having said that, I do wonder why Amstrad did not have a go with an updated QL. Similar approach to the +3. Add a decent keyboard, tidy up the hardware and firmware, slap on a 3” disk, provide one of their colour TV monitors, expand the RAM to 512K and boom you have a QL+. They could have even packaged the QL as per the PCW especially as it came with the Psion business packages. That could have been an interesting machine…

Kind regards

Paddy

It would have been interesting, but they already had the PCW range servicing the budget home office market, and the decision to pull the +3 early shows they were more loyal to their own in-house products than stuff they'd bought elsewhere. The +3 was seen as a threat to CPC sales and was pulled. There's not much point in competing with yourself unless you can absolutely guarantee growing the market. Look at all of British Leyland's sub-brands selling multiple cars in the same market segment - it was a nightmare, and they only recovered from near-oblivion when they dumped a lot of the old brands and sold only one model in each size/segment.

Ahh, market research. Yes, a company should carefully consider the market they are developing a product for. But, sometimes, in a market that does not exist (maybe in its current form) is hard to judge. And paying for market research may be a bit of a money pit if you are not careful.

However, sometimes there may be a way around this.

Remember, Sinclair launched the humble ZX Spectrum in two versions, a 16K model that Sinclair thought would be the bigger seller and the more expensive 48K version. And which turned out to be the most popular?

And even before that, the ZX81 was available either in kit form, or fully built form.

Hence, when it came to his next project, the QL, I would have thought that having more than one model may have been an idea to spread the risk. The model as launched would have been the base model. And a more expensive version, say, with D connectors for the RS232 and Joystick ports, and a built in 3.5” floppy drive, would have been the higher spec model.

With things like the C5, yes, it definitely should have been launched on a nice warm sunny day. Possibly at a location where people could try it out in a public area that had a nice wide path or similar. Like a promenade (seaside) or park. Offer people free rides and then afterwards an interview with them on film/video.

Then, use a campaign for suitable facilities (cycle lanes) to boost both Sinclair’s profile and in turn use this to help market the C5.

But, as always, hindsight is a wonderful thing (pssss pass me that time machine!).

Mark

As far as amstrads activities with the spectrum.

If they had announced the release of the QL and advertised it's serious uses alongside the games, I probably would have bought it.
Instead of the aforementioned megadrive.

I didn't know the QL existed until two years ago. So Sinclairs advertising for such an interesting computer was woeful.
I even remember the advert where sir Clive jumped over the other computers. But that didn't tell anyone what the computer did , and didn't even show the computer on the screen for barely more than a second.

I can see the logic of using microdrives to seriously undercut the competition, it's just a shame they had the faults.
I wonder how much a ql would have cost with the 3.5" drive?
The QL could actually display sprites that had more than one colour per 8x8 square which would have been really welcome then.

If they advertised the QL as being able to make spectrum games and showed someone using it as a dev kit that would have sold loads. If you put that alongside the word processing for homework. It would've been quite an enticing package.

The C5 did have use in some edge cases like disabled people living close to heavily paved areas. People who live closeish to the shops but not close enough.
But they got the &-+( taken out of them so much.

IMHO, there was no place for QL in Amstrad product line, they had home computer CPC and small business computer PCW already.
And QL really fails to tell what it is.
It could be beefed home computer, with 68008, 128K RAM it would be nice machine, if only it had better keyboard, floppy and less ambitious video. With 8-bit data bus, they could have horizontal resolution 640 pixels monochrome, 320 with 4 colors or 160 with 16 colors. They choosed video modes that needed almost double memory bandwitdth and they needed to slow down 68008 that was already choked by narrow bus. They could even read ram during blanking to provide sampled sound and they could read video line start to make scrolling easier. With a 64 color pallete it would be really nice machine. They could do what amstrad did and add a tv disguised as monitor. And if they switch to 256K model in 1985 or 1986, QL could survive till PC ragnarok by the end of 90s,
Or they could aim for bussines machine and it that case 128K was bare minimum, 256K or 512K would be much better. And it that case they probably should switch to 16bit data bus and offer better video and 68000. By 1984, bussiness already proved that it is willing to pay $1000+ for Apple or IBM PC.

https://en.m.wikipedia.org/wiki/CST_Thor


The QL lived on after the Amstrad takeover.

One of the guys who reverse engineered the Z88 OZ operating system told me a little about it once. Sounds far out.

It's often been said that Sinclair's genius was making things cheap, and all the clever stuff in his products was directed towards that end - and it's what often makes them quite annoying. Much as I love the Spectrum, there's a lot you can't do with it, and there's a lot you can do more easily on machines with fewer compromises. The BBC Micro is a classic example. It cost an arm and a leg but it was incredibly well-specced, if a bit short on RAM.

Of course, my perennially skint parents could never afford a Beeb.

The home computer market could tolerate the compromises that made Sinclair products cheap, though. Owning a computer with limitations and shortcomings was better than not owning a computer at all, and the Spectrum software library was huge, so there was always plenty of choice on what you could do with it.

The QL was also built down to a price, though, and that was a fatal error for a high-end machine aimed at businesses and serious home users. Quite apart from the botched and premature launch, the machine had so many really serious flaws and was completely non-standard. I often wonder how it would have turned out if more sensible decisions had been made. Probably infinitely more boring and a lot more expensive, but it might have been a success.

If it was a cheaper ST , and advertised as such it might have found a gap In the market.

Businesses seemed to go with IBM or something very reliable .

In a way the Amiga was a dream computer as it had gaming and serious applications but it just fell short in both areas to be a massive success.
I owned two Amigas but only saw them as a way to play RPGs that didn't appear on consoles

I didn't do any coding,or college work on my Amigas cos I didn't know they were any good for that.

I can't help thinking that if the QL had included a Spectrum compatibility mode it would have flown off the shelves, and possibly created more demand for Spectrum software on Microdrive cartridges.

Yes, being “affordable” was most definitely a high priority aim as far as Sinclair was concerned.

The ZX Spectrum is a general purpose microcomputer. Just like many of its competitors. The limitations are in keeping with its intended market - home users. Hence, cheap keyboard, tape storage, a reasonable display and simple sound.

It’s possible to connect a far better keyboard, a different BASIC/OS ROM, floppy disk drive interface, or a better display system (all due to the full Z80 bus plus the /ROMCS line being available on the edge-connector/expansion port).

In practice, everything I mentioned above became available apart from the better display system.
I agree with most of the above. However, apart from CP/M, Shugart floppy disk drive interface, Centronics parallel printer ports and similar, there were little in the way of industry standards at the time that the QL was being developed. I don’t think the concept of a business computer from Sinclair was flawed.

The bundled software with the QL was certainly useable. The biggest problems were the microdrives. Not just the perception that they were unreliable (not all were), but more that they were an absolute pain for commercial software distribution.

Mark

If, that’s a very big IF.

The hardware was VERY different, so only software emulation would have been possible (which would be rather slow). Unless more hardware was added or the QL substantially redesigned (see the Commodore 128 as a comparison). Keep in mind that most ‘new’ computers at the time were not, or ever intended to be software compatible with their predecessors.

The QL was also not intended (going on the marketing) as a games machine (although for some reason it did include a ROM cartridge slot and two joystick ports… answers on a post card please…!).

It is rather bizarre that a company that originally was supposed to be aiming at selling business computers, became mostly known for games computers (Commodore Business Machines), so much so, that even the very capable Amiga was also mainly used for games (it did get used for the Babylon 5 TV programme CGI).

So, Sinclair with the QL was really going well against the grain, it had to be attractive to the one man band / small business owner. It wasn’t…

I don’t think, even if it had been marketed as a games machine, it would have done any better. Again, because, as I’ve said, of the problems of duplication of microdrive cartridges. And with no tape interface, and no Sinclair standard for floppy disks, how else are you going to duplicate and sell your games in large quantities?

Mark

@1024MAK very interesting, thank you.

But why bother to add microdrives at all? They could have allowed for an expansion and given people a choice of disk Vs microdrive.

I think the point you make about lack of standards is very valid. It's easy to look back with hindsight.

I often think Japan was mad to attack USA when they did. But actually they were very closely matched and probably only lost due to poor discipline with their encryption. However hindsight makes there actions seem ridiculous. Therefore hindsight is not our friend in this analysis. (Minor deviation from the topic )

As someone on this thread already pointed out, the PC standard didn't come to dominate until late 90s, at the launch of the QL the cement was still wet.

I know very little about the QL, and suspect the technology is under rated, and lacked a clear market. For example, I often hear criticism of its keyboard, but I tried one a few years ago and I was surprised how good it was. It looks like a 48k+ keyboard, but the single layer membrane makes a TONNE of difference.

I believe the QL had a multi tasking OS, I'd love to know more about that. Is it similar to the Z88, in that you can suspend one program and launch another, switching seamlessly between the two? But then that is way off topic for this thread!!

Ps spectrum compatibility would not have sold it to me. People were being strongly driven by graphic enhancements, and games that went beyond the 8bit era.

I bought an Atari ST - it wasn't so I could play manic miner with better graphics.

PPS was the QL ever in the Argos catalogue??

Well, microdrives were Clives baby…. There was no way he was going to abandon them and not use them. Of course, a business needs something reliable, hence even though floppy disk drives were expensive, they were generally considered to be the most reliable mass media at the time for personal computers.

Not too many people do know much detail about the QL. The actual concept was fine. And some of the ideas were in fact good. The hardware is however, a mess. They went through various issue boards before ending up with issue 5 being reliable enough. But issue 6 boards are also fairly common.

The size of the combined OS and BASIC exceeded the allocated 32K bytes of ROM, hence the dongle that went in the ROM cartridge slot. The extra ROM being 16K bytes in size. So the total size of the OS and BASIC is 48K bytes.

It was in fact possible to have not needed the dongle if they had fitted a modification on the main board. Which is what eventually happened in later issue boards. Before incorporating the amended design properly in later still issue boards. This is also the reason that the ROM cartridge slot is more or less useless. Half of it’s capacity has been used for the OS and BASIC ROM, leaving just 16K bytes for any ROM cartridge to use (not that a 32K byte ROM would have been that useful for applications or games, but definitely better than only 16K, and keep in mind this is for a 16/32 bit processor).

The SuperBASIC is much better than the ZX Spectrum BASIC. The OS is a proper OS. It’s called QDOS, and yes, it’s multitasking. But this is not apparent when using SuperBASIC as normal. To see the multitasking, you have to know what you are doing. And not many programmers made use of it.

Yes, the OS development and bug fixing continued long after the first machines were sent out to customers, which did not help with the reputation.

So, plus points:

• eventually a good OS that could multitask had had basic windows on screen
• a good and improved BASIC
• the best Sinclair keyboard (although I still don’t like it much, and it’s still not good enough for many) with space bar
• the expansion port used a proper connector (no edge-connector wobble) and ‘standard’ size cards could be wholly contained within the computer
• twin joystick and twin serial ports as standard
• had a microcontroller to deal with the keyboard, reducing main processor workload
• the video system was pixel based, so no attribute clash. 80 column text was both possible and nice and clear
• RGB port so good quality monitors could be used
• uses a 68008 processor, hence has a much better instruction set than the Z80
• using modern languages which are compiled is much easier with a 68XXX processor
• has built-in mass storage system (microdrives)
• has built-in network ports

Unfortunately there are too many problems:

• touch typists don’t like the keyboard, it really needed a better keyboard
• the joystick and serial ports use non-standard BT type connectors which were and still are, hard to get
• its not practical to try to use both serial ports at the same time due to design limitations
• most printers at the time use a Centronics parallel printer port, which the QL did not have
• to use 80 column mode, a special monitor was needed, otherwise due to overscan, the left hand side and right hand side of the displayed image were off screen (TV mode just does not print to these areas of the screen)
• having and using microdrives, this was never going to sell it to businesses and professionals
• software companies hatred microdrives because only Sinclair could duplicate them, so software distribution was difficult and expensive
• the style was not really going to look at home in an office
• the machine being released long before it was ready, and hence it got a bad reputation. Same for the microdrives.
• no standard or provision for a Sinclair floppy disk drive
• because they used a 68008 with a 8 bit data-bus, the memory throughput to the processor was not as quick as a 68000. Don’t get me wrong, its a lot quicker than a ZX Spectrum, but it could have been better…
• only 128K of RAM built-in, of which 32K bytes is used for the screen display
• the microcontroller produces the sound, as the QL does not use a purpose designed sound chip, hence only one sound ‘channel’
• totally incompatible with the other Sinclair computers, including the microdrive format/files

If some of these issues had been addressed and resolved, it would have been a really nice machine.

I don’t know for certain, but I don’t think the QL was in the Argos catalogue.

Mark

[Jbizzel]

Wow, this is great analysis and detail.

I think the QL was also a great looking computer too

[Matt_B]

The main reason it's so slow isn't really due to the 68008. That'd be a decent performer in a well-designed machine. Rather, the problem is that it's losing about half its cycles to memory contention with the ULA that displays the screen.

This is actually quite similar to the original Apple Macintosh, which is also cripplingly slow but features a full 68000. The Atari ST and Amiga 1000, in contrast, had very little memory contention, but flew along in comparison.

[catmeows]

Yeah, exactly. The annoying thing about QL that it could be good computer. 68008 was a good choice for a cheap but powerful computer.

[Jbizzel]

Personal computer magazine had it in for Sinclair too I think.

Whenever they featured a Sinclair product they used a chimp with it.

I saw a benchmarking video they did where the cpc464 beat the Sinclair QL.

The Sinclair might be slow for what it could have been, but was that benchmark unfair?

I think a lot of people would have been influenced by that sort of 'independent' review. But I think PCM were a load of Amstrad fanboys.

Isn't there an elite demo that shows a full shaded elite running on a QL super fast?

Sinclair should have made a technical demo that showed what the QL could do graphically.

Something designed specifically around what the ql could do.

[1024MAK]

Yes, the video system steals memory access cycles from the 68008. Fixing that in a system that does not have separated RAM or uses much faster RAM is far from simple.

In fact, nearly all eight bit bus systems (and some 16 bit bus systems) where the display RAM is memory mapped into the processor address space suffer from the video system delaying the processor.

Now, keep in mind that for the 128K of RAM, they used sixteen 4164 1 bit x 65536 (1 bit wide x 64K) DRAM chips and in the end, the OS and BASIC ended up on two ROM chips (some issue boards have provision for various combinations of one or two ROM/EPROM chips as ‘standard’).

Hence if the PCB had been designed a little better (maybe a four layer board), it would have been possible to use a 68000 with its 16 bit data-bus, with all the memory (ROM and RAM) accessed 16 bits at a time.

The two ROMs would have been connected with one on the lower eight data bus lines and the other connected to the upper eight data bus lines.

Each of the sixteen 4164 DRAM chips would provide one data bit.

All the I/O could have stayed 8 bit, including the display system. To stay 8 bit wide, ahead of the ULA you would use a couple of glue 74 series logic chips to convert from the 16 bit data bus.

Sinclair had already had to move away from using a single ULA, and ended up using two ULAs plus the microcontroller. So, it’s not like they had not already had to adapt and change the design during the development of the machine.

Hence the extra cost of using a sixteen bit bus may not have been substantial. Of course, what we don’t know is the price that Sinclair got the 68008 chips for…

Mark

[Jbizzel]

A demo like this at the opening would have helped.

To summarize, the arguments neatly, if possible...

They compromised on quality to get cost down, failing to find a market. But there was probably a market for a QL that was expensive and without compromise.

Did they learn any lessons for their next computer do you think?

[AndyC]

Yeah, exactly. The annoying thing about QL that it could be good computer. 68008 was a good choice for a cheap but powerful computer.

It really wasn't. The cost savings from the 68008 by the time of the QL were incredibly marginal. And not being a "proper" 16-bit system was a big negative when everyone else was moving forward.

But honestly, I don't think many of the decisions in the QL made sense. Microdrives were a poor choice over some sort of disk technology, the keyboard was awful, graphics that only supported 8 colours (so even worse that the spectrum!) - it all amounted to a machine that didn't really make a lot of sense. It's not a surprise that Amstrad didn't want the machine when they bought out Sinclair.

[1024MAK]

Some further reading:

The glossy launch press kit which was handed out at the launch press conference on January 12th 1984.
From THE REPOSITORY of the Sinclair QL Preservation Project (SQPP)

More than 350 pages with unique original Press Releases and Bulletins from Bill Nichols, former Sinclair Research PR manager.
From THE REPOSITORY of the Sinclair QL Preservation Project (SQPP)

SinclairQL_microcomputer_case_study_OpenUniversity_1986_T362_Block_2_Marketing-SQPP (note this is a 14MB file) - This is a scanned copy of an Open University unit from their Design and Innovation course from 1986. It is a study on why the marketing of the Sinclair QL failed at the time
From the Dilwyn Jones Sinclair QL Pages

Mark

[Timmy]

My guess about the reason why they didn't use a Z80 was that simply couldn't get any, so they had to adapt it to use a different processor.

By the time they were making the QL, the Z80 was becoming so popular that everyone wanted one for their own computer. All the Spectrum clones, the MSX, the CPC, even the C128 had a Z80 inside. Sinclair probably couldn't buy it for cheap anymore so they had to use something else.

Which was a bit unfortunate, but on the other hand, none of the other Z80 based machines were as popular as the Spectrum, so if they could have wait another year he probably could have gotten them for a lot less.

[catmeows]

It really wasn't. The cost savings from the 68008 by the time of the QL were incredibly marginal. And not being a "proper" 16-bit system was a big negative when everyone else was moving forward.

Depends. I read somewhere that 68008 was about 1/3 cheaper than regular 68000. When you make computers in quantities like higher tens of thounsands or lower hundreds of thounsands per year, that is a big difference in your cash flow.
For sure, 68008 performance is about half of 68000 but there are still things that it can offer.
First, the computing power is way ahead any 8bit cpu, ByteSieve benchmark shows 68000@8MHz to be about 13 times faster than Z80@4MHz. 68008 has about half power but it is still 6 times faster than Z80.
Second thing is flat memory model. It makes step beyond 64K boundary much more natural than anything banking or segmentation can offer.
And finally 68008 is much better target for compilers.

"My QL" would be still with 68008, using single set of 256K memories that emerged about that time. I would use stock 68xx chips for interfacing and just single ASIC for video and sound. With single set of memories and 8bit data path board could be much simpler than any 16 bit design. Also, while 256K is half of what be offered soon by Atari ST, Amiga and MacIntosh, it is still 2 or 4 times more than what was offered by 8 bits. And memory was big constraint for most of productivity software, much bigger than cpu power or graphic capabilities.
Unless you want to slow down CPU, shared 8bit data bus doesn't allow to have better bitmaped graphics than what was seen in BBC Micro/CPC. But there is no penalty for bigger color pallete (64 or 512 colors). Also, ASIC could read pointer to start of pixel line during horizontal blanking to allow vertical and limited horizontal scrolling. And finally, there would be time to read few bytes for sample playback.
Regarding price, it could be above level of more expensive 8bits and there had to be well articulated gap between QL and true cheap 16 bit - Atari ST.

[1024MAK]

Maybe this article helps…

The Quantum Leap - to where?

Mark

[catmeows]

Another article about QL hardware and its ULAs:
http://qlforum.co.uk/viewtopic.php?t=17 ... 104#p20102

[Wall_Axe]

How much did the QL cost at launch?
and how much did it cost after a year or so on the market?

How did that compare to the Atari ST?

I suppose one idea would have been to make it have a similar case as the spectrum+ but with the QL innards.
Then advertise the idea of the expandability. Pictures of 3.5" drives, printers, and maybe hard disks attached.

So the base unit would be as cheap as possible. Maybe even using casettes through the EAR port for people who didnt want to expand it at all.
I think someone else may have mentioned that idea in this thread already

[Jbizzel]

Maybe this article helps…

The Quantum Leap - to where?

Mark

Yes, this answers a lot.

And blows away everything I thought was the case about how the the ql was made.

The flawed cost saving is crazy....

Firstly...

A decent keyboard, a proper CPU and a proper disk drive would have hardly altered the costs! AND Sinclair new that at the time!!

Secondly....

I always assumed the lack of board structure was responsible for Clive doing mad things as a solo actor. BUT in fact there was a board and clive was hardly involved in the QL.

His main involvement was to insist on the crap keyboard and microdrives, and built in 2inch flat screen ctr. All bonkers things.

And it was his lack of involvement also hampered the project. They needed him to drive the r&d, which he was actually good at. But he had thrown his toys out of the pram because he didn't want the board structure.

The whole story is quite mad!

[Jbizzel]

Why have 2 micro drives? What was the point?

And is the final QL - after all the s/w and h/w revisions a decent micro?

Feeling like the answer is no. But I want it to be yes.

[Waldroid]

Why have 2 micro drives? What was the point?

The usual way of working for business machines of the time was to have one drive with your application software in it, and one drive to store the data you were working on.

[1024MAK]

Why have 2 micro drives? What was the point?

And is the final QL - after all the s/w and h/w revisions a decent micro?

Feeling like the answer is no. But I want it to be yes.

Depends on what you want from it.

As a hobby machine, yes. It’s actually quite useful. But out of the QL machines that I have, the one with a replacement keyboard and which has a combined disk drive interface, RAM and 68000 processor card (see Miracle-Systems-Gold-Card), obviously with connected floppy disk drive is the one I use the most.

But even without that wonderful card, the machines with issue 5 (up to the latest hardware and firmware version) and later issue boards were useable.

I don’t use the microdrives.

There are some games available for it. But this is a very, very small number compared to the Speccy.

I suggest you try an emulator, at the very least to experience SuperBASIC.

Mark

[Matt_B]

The idea is that you'd use one drive for running the application while another for storing your data, as was starting to become something of a standard on business PCs at the time. Given the high latency and unreliability of Microdrives, keeping a strict separation between the two seems additionally valuable on the QL.

The drives were dirt cheap to make too. Sinclair could put a couple in a QL for a tenner, where a dual disk drive would cost them several hundred.

[Wall_Axe]

I saw in the Argos magazine that microdrives cost £50 each for the public. They were that cheap to manufacture?

[Matt_B]

I saw in the Argos magazine that microdrives cost £50 each for the public. They were that cheap to manufacture?

If you look inside one, the hardware is very simple, mostly comprised of off-the-shelf parts similar to what you'd see in a cassette recorder.

The expensive part was getting them to work. Sinclair Research sunk a fortune into that.

In all honesty, they weren't that bad when I had them - the early bugs haven't been worked out - but the failure rate was still higher than floppy discs and the media hasn't aged very gracefully either.

[Wall_Axe]

Ah trying to recoup the costs.

If cassette tape is made from the same stuff it would be very unreliable.


I kept tapes on top of the tv and they developed loading errors .

[Turtle_Quality]

I think the limited colour and sound backs up the stories about Clive being uninterested in the game market. I've never been a keyboard connoisseur, but I don't think it was much different to the wireless keyboard I'm using now.

Superbasic was great. QDOS was neat, Linus Torvalds said it inspired him to create Linux. It wasn't a GUI though - no mouse and you start at a command prompt, which is daunting for non-nerds. The Mac was launched only days after the QL, and then there was the Amiga and ST, all letting you point and click. Of course if it hadn't been delayed so much it could have carved out some market share before the others arrived.

Though I didn't have much to compare with at the time, I like the bundled Psion Xchange software (a pre cursor to MS Office) and learnt my way around word processing and database programming with it (came in handy for coding SQL 20 years later).

I remember IBM started selling PCs in the early 80's, and anyone going up against them would be at a serious disadvantage, their machines weren't great, but you knew they would be supported for years. Why invest in Sinclair business machines that might disappear from the market next year ? Save a few 100 pounds on machines, but spend a month on developing your application, then maybe start all over again a year or 2 later - too risky.

The decision to stick with microdrives was just stubbornness over reason - there were reliability, capacity and sourcing concerns compared to other media available - 5 or 3 inch floppies.

Someone asked about prices, as I remember the QL started at £400 and fell to £200

[1024MAK]

If cassette tape is made from the same stuff it would be very unreliable.

I kept tapes on top of the tv and they developed loading errors .

The microdrives IIRC, used the same tape as video cassette recorders used.

CRT TVs and CRT monitors are most definitely not good for magnetic media like magnetic tapes or disks. Firstly, they always have a powerful line output transformer and the horizontal deflection coil on the tube (that’s what makes the 15625Hz (TVs) ‘squealing’ (well, if you have young ears) and it’s magnetic field.
Then you have the horizontal deflection coil and it’s magnetic field.
But worst of all, for colour CRTs, around the front, around the outside of the tube, you have the degaussing coil. Every time the TV is switched on, this coil is energised to remove any magnetic field in the steel shadow mask just behind the screen.

Mark

[Jbizzel]

The QL team made mini changes to the microdrive.

They seemed minor, but caused lots of problems.

This was seen as a major mistake by Sinclair. If he had been more involved in the project this might have been avoided.

Changing the microdrive:

1) cost development time
2) introduced reliability issues
3) delayed launch
4) made them incompatible with spectrum drives

And by doing so they gained....

1) "Very little" - Clive Sinclair.