General Category > LisaList2

Really super weird things about the Lisa

(1/3) > >>

rayarachelian:
So there's a lot of weird stuff about the Lisa's design, and not just the GUI, but also the hardware.
For example, the latches for various things like selecting the MMU segment can be triggered by READING from certain addresses. Sure, this was done because it's easy enough to decode a bus address and then trigger a flop without relying on the R/!W line, so that's clearly a cost saving hack.
The MMU design itself is pure genius, created by Rich Page of course. This adds an MMU to a CPU that wasn't designed for one, and thus allowing virtual memory and memory protection to work, and it was done with just some adders and some extra RAM.
But there's something really weird about the VSROM. The Video State ROM is a wonderful hack, and this isn't all that weird on the surface. It implements a state machine where rather than use up a ton of logic chips to build a state machine, a ROM was used a short cut. And you could design lots of things in this way. For example, any gate you look at, such as OR, AND, NAND, XOR, etc. can be implemented simply in ROM, or even half adders, but this isn't usually done because those gates are much cheaper than a masked ROM or PROM.
You'd take the inputs and tie them to the ROM's address lines, and your output(s) would be on the data lines, and you could do with one chip what could take dozens normally.

So, in this case the video state ROM does save on a ton of circuitry by putting all that on a single ROM, so it's quite cost effective for the 1980s. So far so good, it all makes great sense.

Given a bit of timing input on the address lines, the VSROM outputs whether or not to signal the horizontal or vertical retrace signals which define the size of the screen. and the number of vertical lines, Indeed this chip is what allows the Screen Modification kit to work, because its timing is different, and thus produces a different resolution and square pixels instead of rectangular ones.

But one thing does not make sense.
The Lisa's serial number is actually hidden in the video state ROM. And this machine is pretty much the only one that had such a feature. It would make more sense on a Mac or an Apple II where piracy would be far more rampant than on a Lisa. After all, most of the Lisa's LOS software except one or two titles was published by Apple itself. So anyone who had a Lisa, had all the Lisa software bundled with the Lisa, so really there wouldn't be much need for anti-piracy measures. At least not in that context.
But then, there's this: https://en.wikipedia.org/wiki/Van_Eck_phreaking. This isn't really common knowledge, but it was known amongst some circles in the mid-to-late 1980s. With old CRT TVs and computer monitors it was possible to use a bit of circuitry, the right antenna and another TV. This allows the eavesdropper to view the victim's screen from a distance, and so it could have been, but wasn't, the stuff of James Bond movies. This was surely known of in the three-letter-agencies, such as the CIA and NSA, and funnily enough in the UK it was known by the TV detector van people - in the UK, you'd have to pay a tax to the BBC to watch TV, and there were vans going around looking for unlicensed TVs. ( https://en.wikipedia.org/wiki/TV_detector_van ) - when the first early computers arrived in the UK there was an anecdote about someone's computer monitor being detected as an unlicensed TV, but that they couldn't figure out which channel was being watched.  ;D (The page mentions that optical emissions are used but it's more likely that radio signals were being used.)

But the real implication here of the VSROM is that the serial number was being pumped out one bit at a time on every frame is that someone with the proper TEMPEST equipment could identify the serial number of the Lisa from a distance (as well as get a fuzzy image of what was on the display.)

Now the real fun thing is that Van Eck didn't publish his paper until 1985, and by then the Lisa was already out, as work on it began in 1980 and it was on the market by 1983. So whomever designed the video circuitry and decided to put the serial number in the VSROM had information about TEMPEST, so they probably previously worked at one of these agencies (or had spoke with Van Eck and known of his work before deciding to add this to the Lisa)

In fact, there were such things as TEMPEST hardened computers, and you can even find an example of a Mac 512K listed on eBay incorrectly as a Lisa right now. ( https://www.ebay.com/itm/Apple-512K-Lisa-Computer-ValueTec-2900T-Tempest-Shielded-RARE/143191209815? ) Various agencies used such machines to guard against eavesdroppers.

Interestingly, the Lisa 1 and 2, had an AMD9510 floating point math coprocessor, but as John of Vintage Micros pointed out, (see: https://www.ebay.com/itm/Apple-Lisa-2-10-XL-WD2001-Data-Encryption-Chip-with-Socket-MINT/141829462056 ) the 2/10 has the ability to add an and WD2001 encryption chip, which could do DES, but on the 2/10 there's no soldered in socket. So perhaps Apple had some relations with Uncle Sam for these. Of course, there's no currently available Lisa software that uses the coprocessor, nor the encryption chip, which is sad...

So, anyway, either the serial number on the VSROM was designed with TEMPEST in mind, or it was an insanely huge coincidence, you decide.  ;D
Has anyone played with TEMPEST detection on here? Perhaps it would be a nice electronics project to make a Lisa serial number detector and potentially see if you could read a Lisa's screen from across the street?
Edit: added some source code from the H-ROM that reads the serial number. Interesting thing is that LOS doesn't use this serial number which is written to low memory, instead, it re-reads it again after it boots up! (I suppose the logic is that you could power on your Lisa, go into service mode, and overwrite the serial number, so they re-read it again to ensure you can't do that.)
Edit2: Since the SN is presented in the vertical retrace, it might be at the top of the screen on video tape recorded off the Lisa in the same way as captions were sent. see: https://en.wikipedia.org/wiki/Closed_captioning
--- Quote ---For all types of NTSC programming, captions are "encoded" into line 21 of the vertical blanking interval - a part of the TV picture that sits just above the visible portion and is usually unseen. For ATSC (digital television) programming, three streams are encoded in the video: two are backward compatible "line 21" captions, and the third is a set of up to 63 additional caption streams encoded in EIA-708 format.[24]
--- End quote ---

stepleton:
Could it be that the Apple management just wanted a cheap way to make piracy harder for a fantasised software ecosystem that never arrived?

If the Lisa designers had put the serial number on a chip sitting on the CPU or I/O boards, it would have been "too easy" for people to swap their original serial number out for a duplicate of someone else's. If this "personality module" were socketed (as it might be to make the service tech's life easier) you wouldn't need any tools at all. If not, you could still imagine an office of full Lisas (lol) where people trade around a single seat license for ExpensiveCad or whatever by passing a "blessed" card cage to whoever needs to use it next. Either scenario loses the software publishers money.

Putting the serial number on the video board locks it in an unpleasant, difficult-to-access chamber with greasy CRT dust and a "Beware of the Leopard" sign (well, a yellow triangle with a lightning bolt at least). Pirates have to be more determined (especially if they want to remove that *@!$ chunky connector at the bottom of the video board---you can only do a few of those before your fingers start to bleed). Best of all, Apple doesn't need to add another electronic assembly to the CRT bay---they can just kludge it into the existing video board, which was using a ROM anyway.

Too easy for me to explain this one with McDuck's Razor  ;D

--Tom

rayarachelian:
I'm afraid McDuck will have to keep swimming in his vault of gold coin with beard stubble.

No, it's not part of the video board, (the video board is all analog) it's part of the CPU board and the VSROM on there is socketed.

Here's the CPU schematic sheet 5, the VSROM is on the top left, the 371-0072 chip. Note the VSYNC, HSYNC signals, and also a bit to the left, the line labeled "SN" on pin D7 or 14. It's that one.

On the right you can see the latch that implements the START and SEG1/SEG2 MMU selectors I talked about earlier.


So, yes, it literally sits on the CPU board, and is exactly as easily replaceable as you say, and that was by design because if your CPU board died, you'd want to move the VSROM to your new service part CPU board - infact Apple's technical procedures describe that this must be done or the user won't be able to print, see: http://bitsavers.org/pdf/apple/lisa/service/072-0085_Level_1_Lisa_Technical_Procedures_Mar85.pdfsection 1.8 on PDF page #19:


--- Quote ---INSTALLING A NEW CPU BOARD: TRANSFERRING THE VIDEO STATE MACHINE ROM The Video State Machine ROM (630.9 chip) is found at position C6 on the CPU board (see Figure 1, below). This ROM contains the serial number of the Lisa, which is used as its address for any network applications. This chip is also used in the software protection system. Whenever you change the CPU board on a Lisa, you must transfer this chip from the old board to the new board; otherwise the user's software will not run on the Lisa.
--- End quote ---


The serial number could have as easily been hidden in the I/O ROM, or even more securely in the COP421 microcontroller, but the VSROM was picked specifically.

The whole video state machine is on the CPU board because it shares the RAM with the CPU and has a register that selects which page to display, this is also why the CPU is slowed down to 5MHz instead of the full 8MHz the 68000 is capable of. In the Mac Burrell Smith ( https://www.folklore.org/StoryView.py?project=Macintosh&story=Square_Dots.txt ) found a way to access video RAM while the CPU was idling by alternating accesses without slowing it down, and it appears this was shared with the Lisa team, but they ignored his upgrade.  ::)

Fun watch: https://www.youtube.com/watch?v=5VJh5Nea33s

rayarachelian:
Now, I didn't mean to make this thread all about the serial number, there's plenty of other weird things to explore.

For example, the Lisa seems to support a 3 button mouse! In the schematic below from 2nd the I/O board sheet, we see the COP421 controller at U9F, and to the right of it, we see an LS153 at U8E, which is a dual 4 input MUX, this is used to add more I/O lines to the COP421 microcontroller, which is used to manage the Lisa's clock, power, keyboard and mouse interface. To the right of this MUX we see something really interesting, labels! You see the expected keyboard line, sure, and up, down, left, right. All good and expected. But, look at those labeled SW0, SW1, SW2! SW in schematics stands for switch, and SW0 would be switch #0, and so on. And SW0 goes to the mouse's button.

But our Lisa's mouse has only one button. So this means the Lisa was designed to have up to 3 buttons, just like the Xerox Alto or Xerox Star. Of Course if you watch the youtube video in the CHM thread I posted a couple of days ago, you'd have caught the bit where Larry Tesler mentions the removal of the other buttons. But the hardware is still there. Or at least for the second button. The pin for the 3rd button isn't on the DB9 connector of the mouse.


A fun experiment, then, would be to obtain a DB9M-F cable, and splice it open, finding the pins that map to SW2, and SW3 and see what happens in LOS, Monitor, MacWorks, Xenix, and UniPlus when SW1 and SW2 are triggered.


You'd need to first see how SW0 works, whether it's a pull up or pull down resistor, or if it just floats when it's not pressed, and implement the same with those extra wires on the remaining two lines. (Hint, from the resistor network to the right of the mux, you can clearly see that these are pulled up to +5V, so to trigger these, likely the buttons would be shorted to GND.)

Most likely nothing will happen. But it would be an interesting experiment to see what codes the COP421 produces when these extra two mouse buttons are triggered, if any. (It may well be that while the hardware is still there that Apple removed the code from the COP421, in which case, you could never take advantage of the extra buttons.)



Now, maybe you've seen this before - a "Shining Super Mouse": https://www.ebay.com/itm/9-pin-Mouse-for-Macintosh-128k-512k-XL-Mac-Plus-Lisa-Apple-IIc-IIe-Laser-128-/264315459498?nordt=true&rt=nc&orig_cvip=true
If  not it's worth a look. This is a third party Apple II, pre-ADB Mac, and Lisa compatible two button mouse, but, the 2nd button is a click lock button, much like the caps lock on your keyboard, and not a true right click button.
I didn't buy the one from this listing, but I did buy one like it earlier this year, and yes, it does work very nicely on the Lisa. I thought it was a PC serial mouse and was skeptical, but it isn't, it's actually a Mac mouse.



Are you aware of other weird Lisa hardware things? if so, please reply to this thread so we can document them all.

rayarachelian:
It seems that the Lisa's expansion slots use 56 pin ZIF sockets, but Apple - at least on the 2/10 motherboard, not sure about the 2/5 which I don't have infront of me right now, could accommodate upto a 64 pin socket. Wonder what plans they might have had for those extra pins? I don't think those are wired at all though.

(Weirdly it's labeled 63, while the existing one ends at 56, not sure why the odd end, so possibly one pin was missing on 64 pin ZIF connectors?)
Also as an aside, I've had the yellow plastic that rotates with the pin on a couple of my motherboard break off a bit, it didn't affect the slot, but it seems these are starting to age.
Yet, another weird thing, since the Lisa uses the MC68000 MOVEP opcode to do I/O, even though the Lisa's expansion slots are capable of doing 16 bit I/O all the expansion slot cards I've seen are all 8 bit. This is likely to the fact that the chips used on these, such as the VIA6522 were all 8 bit devices. Further adding to the weirdness, in fact, the ROMs on these are all 8 bit, and the boot ROM copies them to RAM before executing the code on them! I wonder if they ever even had a 16 bit slot? Perhaps ROMs or chip sockets were more expensive back then?
(While the 68000 is internally 32 bit, it has a 16 bit data bus so it can't move 32 bits at a time, so a 32 bit slot wouldn't have been possible until things like 68020.)

Navigation

[0] Message Index

[#] Next page