Author Topic: EEVblog #382 - Cambridge Z88 Teardown  (Read 11315 times)

0 Members and 1 Guest are viewing this topic.

Offline EEVblogTopic starter

  • Administrator
  • *****
  • Posts: 37661
  • Country: au
    • EEVblog
EEVblog #382 - Cambridge Z88 Teardown
« on: November 13, 2012, 03:47:53 am »


Dave.
 

Offline smugtronix

  • Contributor
  • Posts: 28
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #1 on: November 13, 2012, 03:51:33 am »
I have to ask: What was the dominant home computer in Australia in the 1980s? The UK had the ZX Spectrum and Amiga, and we Yanks had the Apple II and the Commodore 64. What was it in Daveland?

Excellent teardown of something I didn't know existed. Interesting to see what Clive Sinclair did outside of Timex Sinclair.
 

Offline Zad

  • Super Contributor
  • ***
  • Posts: 1013
  • Country: gb
    • Digital Wizardry, Analogue Alchemy, Software Sorcery
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #2 on: November 13, 2012, 05:31:27 am »
I know I said this on the Youtube comments, but it seems we have people enjoying a handbag fight on there at the moment, so I suspect comments may get lost.

For anyone interested in industrial design, the bloke who designed the Z88's mouldings has uploaded a load of his drawings and photos to Flickr.

http://www.flickr.com/photos/9574086@N02/collections/72157608812198325/

It was originally going to look like this, with a flat-screen CRT in the lid:



This is what mine looks like (inside its vinyl dust cover). Yes, a rare colour Apple Z88 notebook!:



I was on work placement at Uni at the time, and we had just taken delivery of some Mac FX machines for CAD work. So I took the opportunity to borrow some stickers. You would not believe the number of people who Oooh'ed and Aaaar'ed over what looked like a super compact Apple portable computer.

At the same time, in another part of Cambridge, a company called Acorn were working on their Acorn RISC Machine, which nowadays really does power Apple portable computers!

On mine, there is no expansion connector in the side, but you can see where it should be in the moulding. I wonder what they had planned for that. Presumably floppy drives etc. The serial number on mine is 47,000 higher, so presumably a design revision happened somewhere along the line.

The "hard reset" switch is intended to protect the system when the memory modules are removed/inserted. Normally there would be a transparent plastic cover which flips down. Doing so would flip the switch and so power the system down (or at least put it in a sleep state) while the user was messing around.


Offline johnboxall

  • Supporter
  • ****
  • Posts: 652
  • Country: au
  • You do nothing, you get nothing.
    • Books, services and more:
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #3 on: November 13, 2012, 10:08:01 am »
Sir Clive Sinclair is a fascinating man, and one heck of a poker player to boot.
There's a good dramatised show about his battle with Acorn in the early years - http://www.bbc.co.uk/programmes/b00n5b92
It's on YouTube if you look for it.

Offline peter.mitchell

  • Super Contributor
  • ***
  • Posts: 1567
  • Country: au
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #4 on: November 13, 2012, 10:29:20 am »
Bah, every time i see one of these i expect you to overclock it :(
 

Offline EEVblogTopic starter

  • Administrator
  • *****
  • Posts: 37661
  • Country: au
    • EEVblog
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #5 on: November 13, 2012, 11:32:43 am »
The C64 was popular here too but we also had our own homegrown Z80 based machine called the MicroBee.
http://en.wikipedia.org/wiki/MicroBee
It was popular in schools too.

The Microbee was mostly popular in schools, and that's what made the company successful.
The other usual suspects were more popular because they had big advertising dollars, and were sold in Dicks Smith and Tandy stores et.al
So the Tandy Trash-80 machines were very popular, as were the Commodore's and Apple's sold in the department stores.
Hard to know what the best seller in numbers was? My guess would be either the C64 or one of the Tandy's.
Dick Smith had his own line of machines like the CAT (Apple clone), and VZ-200/300.

Dave.
 

Offline grumpydoc

  • Super Contributor
  • ***
  • Posts: 2905
  • Country: gb
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #6 on: November 13, 2012, 11:45:15 pm »
Silent keys for road warriors my arse! Whatever was cheapest more like.

I suspect that it was a Z80 (= 2MHz) not a Z80A (=4Mhz) but "Uncle Clive" bought loads of the lower frequency chip and just tested them to see which ones ran at 3 and a bit MHz and used those in the Z88. Probably threw the others in a ZX81 or something.

The inductor is presumably part of an inverter for the +/- 12V required for the RS232.

The bursty clock is neat though to get the battery life up. Not sure if that was Clives personal input - his products were nearly always designed down to a price but there were some genuinely ingenious features in them. Unfortunately not always reliable features :-/
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 16272
  • Country: za
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #7 on: November 14, 2012, 04:32:38 am »
The Z80A was a unit that was tested to run at 5MHz, so any that did not were branded Z80 as long as they ran at 2.5MHz. Sinclair just retested them to see those that could do 3.5, so most likely he did get a good yield, and could sell or use the failures elsewhere.
 

Offline Shas-O

  • Contributor
  • Posts: 10
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #8 on: November 14, 2012, 07:42:48 am »
Shame about that tear down, its worth £100k un opened, dead or alive :)
 

Offline nitro2k01

  • Frequent Contributor
  • **
  • Posts: 843
  • Country: 00
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #9 on: November 14, 2012, 08:14:43 am »
Shame about that tear down, its worth £100k un opened, dead or alive :)
Were you joking or do you mean £100? That's what this reseller is apparently selling them for anyway.

http://www.rakewell.com/z88/z88hardware.shtml#this_compact
Whoa! How the hell did Dave know that Bob is my uncle? Amazing!
 

Offline grumpydoc

  • Super Contributor
  • ***
  • Posts: 2905
  • Country: gb
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #10 on: November 14, 2012, 09:29:12 am »
Quote
The Z80A was a unit that was tested to run at 5MHz, so any that did not were branded Z80 as long as they ran at 2.5MHz. Sinclair just retested them to see those that could do 3.5, so most likely he did get a good yield, and could sell or use the failures elsewhere.
By 1987 (the year the Z88 was introduced) Z80's were available up to 8MHz - just pulled my SGS "Z80 Microprocessor Family Databook" 6th edition, March 1987 off the shelf to check (yikes, that dates me!)

Plain Z80's were, indeed 2.5MHz though I never saw a design that ran them at that speed which is why I always thought of them as a 2MHz chip.

Z80A's were specified as 4MHz parts (though I could well believe that they were tested at 5), Z80B's at 6MHz and Z80H's at a massive 8MHz

But Clive was well known for that type of stunt. There was a rumour at the time that the DRAM for the Spectrum 48k were 64k parts with errors is the last block that were bought cheaply. Certainly if a machine was populated with "48k" chips and didn't work it could be "downgraded" and sold as a 16k machine - no change to the RAM chips, of course, just a configuration change so only 16k was used.
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 16272
  • Country: za
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #11 on: November 14, 2012, 09:56:11 am »
Intel does the same with processors. If the L1 cache fails self test it is chopped in half and marked Celeron. Speed is tested and binned after assembly, each wafer is similar in max speed so they get tested, marked and multipliers set and sold accordingly. AMD units had the jumpers outside the package so they were easy to reset to overclock.
 

Offline Unixon

  • Frequent Contributor
  • **
  • Posts: 396
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #12 on: November 14, 2012, 10:40:46 am »
Look at the part in a glass tube at 15:00 to the left from QFP chip! There is another one around DB9 connector. Seems that it's an SMD ceramic capacitor inside a glass diode-like tube with axial leads. Not a very common part nowadays... Well, at least I haven't seen them before. Nice little through-hole thing  ;)
 

Offline grumpydoc

  • Super Contributor
  • ***
  • Posts: 2905
  • Country: gb
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #13 on: November 14, 2012, 11:09:56 am »
Quote
Seems that it's an SMD ceramic capacitor inside a glass diode-like tube with axial leads. Not a very common part nowadays... Well, at least I haven't seen them before.
I'm just repairing some Williams System 7 pinball boards for a friend and they're stuffed with them - I hadn't seen them before either and it really threw me at first!
 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 16272
  • Country: za
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #14 on: November 14, 2012, 11:42:20 am »
Chip ceramics in a glass tube, been around since the 1980's. IBM used them a lot as decoupling caps. They do have a marked tendency to catch fire though.
 

Offline Shas-O

  • Contributor
  • Posts: 10
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #15 on: November 14, 2012, 10:50:19 pm »
Yes,  ;)
Shame about that tear down, its worth £100k un opened, dead or alive :)
Were you joking or do you mean £100? That's what this reseller is apparently selling them for anyway.

http://www.rakewell.com/z88/z88hardware.shtml#this_compact
 

Offline Hypernova

  • Supporter
  • ****
  • Posts: 655
  • Country: tw
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #16 on: November 16, 2012, 03:22:14 am »
Why is through-hole resistors used when the thing obviously went through a pick-n-place for the QFP glue logic?
 

Offline Zad

  • Super Contributor
  • ***
  • Posts: 1013
  • Country: gb
    • Digital Wizardry, Analogue Alchemy, Software Sorcery
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #17 on: November 18, 2012, 06:07:26 am »
16K Spectrums definitely only had 16KB worth of chips in. I saw enough of the insides of them when I was installing upgrades. They were mostly in the early part of the machine's market life, and they really weren't going to solder more chips in than they had to. It wasn't like they were making a few hundred of them in a back room, they were made by a third party company and messing around with marginal parts would not be worth the cost of replacing them. Issue 1 Spectrums didn't even have the sockets, but space for a daughterboard. See just below the edge connector. Iss 1 really was a horrible looking hand-drawn PCB.



By 1987, the Z80 had been a cheap commodity 4MHz part for years.

The concept of under-clocking hardware wasn't Clive Sinclair's, but that of Jim Westwood. When developing the first pocket calculator for Sinclair, he had the idea that running the processor in bursts could lower the current consumption to a level where battery life became acceptable. Rather than get the story third hand from me, see here:

http://www.reghardware.com/2011/11/15/heroes_of_tech_jim_westwood/

Offline grumpydoc

  • Super Contributor
  • ***
  • Posts: 2905
  • Country: gb
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #18 on: November 18, 2012, 08:36:04 am »
Quote
16K Spectrums definitely only had 16KB worth of chips in. I saw enough of the insides of them when I was installing upgrades
I fixed a fair few myself in the 80's - they (and the ZX80/81) were not the most reliable of beasts. The expansion connector on all three had the Z80 address, data and control lines brought out unbuffered along with the power supply rails which meant fried Z80s were very common. Either that or a dead DRAM.

IIRC the 48k machine had a bank 4116 DRAMs for the base 16k and a bank of 4532 DRAMs for the expansion memory - 32k isn't a "natural" size for DRAMs given the multiplexed address bus so the folklore of the day was definitely that these were "failed" 4164s

As to Sinclair shipping machines with faulty RAM reconfigured, agree I never saw that myself but I suppose it wouldn't have taken too many like that to start rumours.

Quote
By 1987, the Z80 had been a cheap commodity 4MHz part for years.
It was available as a 4MHz part certainly - my Nascom 2 had the 4MHz processor but a "plain" Z80 rather than a Z80A was 2.5MHz. As I said I still have the 1987 SGS datasheet, it's possible that it was difficult to find a Z80 that wouldn't run at the higher speed but they were binning them down from 6 or 8MHz by then.

Quote
Rather than get the story third hand from me, see here:

I think "The Reg" probably already counts as 4th or 5th hand...  :)

I presume that the Z88 "woke up" just long enough to refresh the DRAM when idling.

 

Offline SeanB

  • Super Contributor
  • ***
  • Posts: 16272
  • Country: za
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #19 on: November 18, 2012, 11:34:57 am »
The Sinclair flat screens are still made though, for use in intercom systems, though they are being replaced by LCD displays rather fast.
 

Offline Don Hills

  • Regular Contributor
  • *
  • Posts: 159
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #20 on: November 21, 2012, 01:00:37 am »
Early Spectrums had a jumper to select upper or lower 32K bank. Apparently early 32K chips had a a "H" or "L" marking.
 

Online wasedadoc

  • Super Contributor
  • ***
  • Posts: 1308
  • Country: gb
Re: EEVblog #382 - Cambridge Z88 Teardown
« Reply #21 on: December 21, 2013, 11:31:44 pm »
The Z88 was switched on or off by simultaneously pressing both Shift keys.

The Z88 was 'soft reset' by pressing the small switch on the left hand side (with the memory card flap closed).

The Z88 was 'hard reset' by pressing that switch on the left hand side with the memory card flap open.

As already mentioned in another post above the 'internal' switch put the machine into hibernate mode when the card flap was opened to swap cards.  (They were not intended to be 'hot-swapped' with the machine operating.)  That switch also determined hard or soft reset.

The serial connector had a non-standard pinout

1     NC
2    TxD
3    RxD
4    RTS
5    CTS
6    NC
7    GND
8    DCD
9    +5v

Although designed with an expansion connector, it was never used and the case was changed on later production runs to eliminate the removable cover.

The Spectrum was sold in 16K and 48K versions.  Both had 8 off 16K x 1 DRAMs.  The 48K versions had an additional 8 off 64K x 1 DRAMs but only 32K was used.  They were indeed chips that had fault(s) in the high or low half.  The highest address line of all 8 chips went to a link that was soldered to 0 or 5 volts to select the good half.  Initially those chips were on a plug-in daughter board which plugged into the two empty DIL sockets seen in the photo in Zad's post above.  A later revised PCB layout provided enough space for them to be fitted on the main board.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf