EEVblog #747 - PC Based Logic Analyser Project

[EEVblog]

Dave goes back 20 years and find an old PC Based Logic Analyser project of his that was published in Electronics Australia magazine back in 1996. He uncovers the original timing diagrams, schematics, and prototype. And tries to resurrect the old Borland Pascal 7 source code and Lattice ispLSI PLD chip code.
And the old Protel Autotrax for DOS PCB and schematic files.
Will it all work 20 years later?
A bonus side detour into the venerable Tektronix TDS210/220 oscilloscope.

Original articles: http://www.eevblog.com/files/PCLA-Article-ElectronicsAustralia.pdf
Original design notes: http://www.eevblog.com/files/PCLA-OriginalDesignNotes.pdf

[miguelvp]

There was no web in 1995? sure it was, maybe not for AOLers or later arrivals.

Btw nice video 

[ktulu]

Here are the two EXE files patched for Error 200. I hope Dave don't mind it.
(PCLA10 needs to be run under DOSBOX)

[EEVblog]

There was no web in 1995? sure it was

Yeah but it had bugger-all.
No such thing as online e-commerce, manufacturers with websites and PDF data sheets like we take for granted these days. Altavista was the search engine of choice, not that you ever found much. No such things as blogs or online video, or even modern forums, usenet still ruled that world.
There was basically nothing on there of real worth for a designer at the time.

[EEVblog]

Here are the two EXE files patched for Error 200. I hope Dave don't mind it.
(PCLA10 needs to be run under DOSBOX)

Wow, just ran it, totally forgot what the main menu or disassembly screens looked like!

[firewalker]

Is the source code available?

Alexander.

[gemby]

I was always wondering, how big and complex projects, like this one are made in those days. Today it is very much different, you can simulate almost everything, including hardware in hardware. In those days, it requires much more knowledge, time and devotion to bring out something like this. I remember, long time ago ( in galaxy, far, far away, off course ) i was trying to make programable digital timer for photography, made entirely in TTL logic, ok, this one came to documentation phase only, no prototype has been made, but it was hundreds, and hundreds of hours, shitload of printed documentation, and all done on paper and all simulations are done in head. Today, i would not even think of TTL logic, and not even try to assemble anything before simulation. Probably not even put anything on paper at all, it will all be in digital form.

[EEVblog]

Is the source code available?

I'd mail it to you on a 3.5" or 5.25" floppy back in the day 

[Fungus]

Your software looks nice...

Drill 717 holes by hand? Rather you than me.


PS: Tab indent size 8 on a DOS screen in Turbo Pascal? Fail! 3 or 4 indents and you're half way across the screen.

[vlad777]

How many Davids are there in your world?
Old Protel runs fine in DosBox emulator under W7.
Windows 7 and up doesn't run 16 bit programs.

[Grapsus]

Thanks a lot for the video! No nostalgia for me since I was five years old at the time. But I do know some of this stuff. Some time ago I helped my grandfather to run code that he wrote in the nineties using Borland Pascal 7. It was a pretty heavy graphical application for simulating thermo-dynamic systems, like power plants, somewhere around 30k lines of code. First we ran into all sorts of troubles to just launch the damn exe. Same error 200, lots of screwed up timings etc. And then we recompiled all the code to make sure we were able to reproduce the thing.

We tried a lot of different methods to run this stuff on modern machines: DOS on bare metal, Windows XP, Dosbox, Bochs simulator, QEMU, VMware and finally Virtual Box. All of them had some quircks, but in the end the absolute best environment was a VirtualBox VM. VirtualBox enables you to simulate hardware that's recognized by DOS without any workarounds, graphics are good, you can attach iso images on the fly to get files into the VM, slow down the CPU etc. I can export my VM with DOS installed and ready to go if anyone is interested.

[vlad777]

What 30k lines of code?
Tell us about this project, how many people ,did it sell.... ?

[EEVblog]

Drill 717 holes by hand? Rather you than me.

No drill press either, all with a small hand drill.

[vlad777]

This is one of my first projects ,26 March 1998.
I was in high school ,but did this in my own time.
Looks like I used a wrong layer on some components, ups, no color monitor at the time.
Spoiler alert , it doesn't work.

[hikariuk]

How many Davids are there in your world?
Old Protel runs fine in DosBox emulator under W7.
Windows 7 and up doesn't run 16 bit programs.

Windows 7 64-bit and up don't run 16-bit programs; Windows 7 32-bit still has WoW (as opposed to WoW64) for running 16-bit Windows programs.

The best way of running old DOS applications is probably just to use DOSBox though, as Dave appeared to be doing in some of the video.

[Grapsus]

What 30k lines of code?
Tell us about this project, how many people ,did it sell.... ?

Nope, it was purely an academic/research project, it was designed by professors, and many features were implemented by PhD students. I'm not sure how much I can tell in detail since the thing contains a lot of data about actual power plants. For the same reason the program had a copy protection to prevent people from taking it outside the university and it gave us a hard time to run the thing before we could recompile it. But my grandfather was very happy to see that this project could still work after all the time he spent on it back in the day.

[Tek_TDS220]

"TDS 220....  Who can remember that?"  Just used one yesterday.  It gave me the data I needed.

[vlad777]

What 30k lines of code?
Tell us about this project, how many people ,did it sell.... ?

Nope, it was purely an academic/research project, it was designed by professors, and many features were implemented by PhD students. I'm not sure how much I can tell in detail since the thing contains a lot of data about actual power plants. For the same reason the program had a copy protection to prevent people from taking it outside the university and it gave us a hard time to run the thing before we could recompile it. But my grandfather was very happy to see that this project could still work after all the time he spent on it back in the day.

Wait, I just realized...
PhD students writing in Pascal?
That doesn't sound right, how come?


(Pascal was always a noob lang at universities.)

[mushroom]

Pascal was very popular in french universities...

Why... ? Probably because of Philippe Kahn. (Google that name)

On my side, I got Turbo Pascal (don't remember what version, maybe 3 or 4...), and immediately switched to TurboC 1.5. I *NEVER* used again this "language" (as crappy as Basic)

[zapta]

Pascal was very popular in french universities...

Why... ? Probably because of Philippe Kahn. (Google that name)

And then another Frenchmen invented Ada which is Pascal on steroids.

BTW, Philippe Kahn was also one of the pioneers of merging cameras and cell phones which at the time looked as a kitchen sink approach.

[coppice]

There was no web in 1995? sure it was

Yeah but it had bugger-all.
No such thing as online e-commerce, manufacturers with websites and PDF data sheets like we take for granted these days. Altavista was the search engine of choice, not that you ever found much. No such things as blogs or online video, or even modern forums, usenet still ruled that world.
There was basically nothing on there of real worth for a designer at the time.

Didn't you ever come across sites like cdrom.com? That had lots of electronic design material in the mid 90s.

[Grapsus]

I wasn't working in the IT field at that time, but according to this article

http://en.wikipedia.org/wiki/Turbo_Pascal

TP had a lot of advantages: first integrated IDE, small code footprint, super fast compilation, TPU modules with reusable code were available for a lot of tasks, and 50$ retail price, even by today standards is unbeatable. Performance-wise it's closer to C than to Basic. I think Pascal programmers would take offense for putting the two in the same basket.

[Fungus]

TP had a lot of advantages: first integrated IDE, small code footprint, super fast compilation, TPU modules with reusable code were available for a lot of tasks, and 50$ retail price, even by today standards is unbeatable.

I dunno. My copy of Visual Studio was free to download and includes quite a lot of stuff.

Performance-wise it's closer to C than to Basic. I think Pascal programmers would take offense for putting the two in the same basket.

Pascal was a terrible language. Fundamentally broken: http://www.lysator.liu.se/c/bwk-on-pascal.html

[zapta]

I think Pascal programmers would take offense for putting the two in the same basket.

Don't pay attention to language war trolling.

TP set a new standard for developer productivity at the time.

[Carl_Smith]

I did a lot of programming in Borland's Turbo Pascal, and I liked it a lot.  I still have all the books taking up space in one of my bookshelves even though I hadn't looked at one of them in years until today.  One of the things I liked was all the extra books that came with example material, like the Gameworks, Graphix, Editor, and Database toolbox books.  I spent countless hours back in the 80's reading through these learning programming and algorithms.  It was just amazing to me to have the full source code for something like a fully functional DOS text editor that I could just go through and examine as I pleased.

[SeanB]

A shed load of business core is still running that was written in Pascal. Although most of the code is now almost unmaintainable, because the original creators are either long gone, dead or did not write any more comments other than "don't touch this, i don't know how it works but it does" on large swathes of the source.

[Grapsus]

I come here for electronics talk, and I find that there's not a lot of trolling, people seem happy with different tools and the main question is "what are you going to use this for ?". But man, every time it comes to computers, there are these stupid endless discussions. Last time it was "how can you do any serious work with this OS ?" and now it's "everybody knows that the language X is for noobs", damn.

[nctnico]

There was no web in 1995? sure it was

Yeah but it had bugger-all.
No such thing as online e-commerce, manufacturers with websites and PDF data sheets like we take for granted these days. Altavista was the search engine of choice, not that you ever found much. No such things as blogs or online video, or even modern forums, usenet still ruled that world.
There was basically nothing on there of real worth for a designer at the time.

IIRC companies started to put PDF files with datasheets online around that time (1995/1996). I recall having a list with links to datasheets on my 'homepage'.

Regarding Pascal: been there done that too but the biggest problem is the rigid structure of Pascal so it is hard to adapt the structure of a program to unavoidable requirement changes and Borland adding many C-isms later on. IIRC Borland's Delphi has about 7 different string types and one operation you can do on type A cannot be done on type B so you have to convert from B to A, do the operation and convert back from A to B.  After all Pascal was invented to teach people structured programming and not actually being used for any real software. However for many people everything looks like a nail if they only have a hammer...

[jancumps]

Pascal was very popular in french universities...

...

I had Pascal as the program language in my first year too (late 80's). To learn the principles.
Later we had c, cobol and assembler for the 68xxx.

[ivan747]

Episode #747 was about something I always wanted insight on!  Thanks Dave! 

And it even was in the year I was born 
Sigh... I should publish something 

[MartinX]

I remember the internet in 1995 that was the year I got internet, dial up modem of course. Data sheets were not so easy to find and took a long time to download I remember getting CD ROMs from distributors at trade shows , that was a huge relief, I still got some Texas Instruments and National CD ROMs and the Farnell catalogue.

[vlad777]

I come here for electronics talk, and I find that there's not a lot of trolling, people seem happy with different tools and the main question is "what are you going to use this for ?". But man, every time it comes to computers, there are these stupid endless discussions. Last time it was "how can you do any serious work with this OS ?" and now it's "everybody knows that the language X is for noobs", damn.

I didn't mean to troll, I was explicitly told (in school and faculty) that Pascal is a learning tool. Sorry.

EDIT: I did my final exam for high school in Pascal. It was a file manager with my graphical window system.

[Carl_Smith]

Episode #747 was about something I always wanted insight on!  Thanks Dave! 

I agree.   What else from Dave's past has he talked about but not made a video on yet?

[EEVblog]

Pascal was a terrible language. Fundamentally broken: http://www.lysator.liu.se/c/bwk-on-pascal.html

Tell that to Altium who built a half billion dollar business out of it.

[donmr]

We used to say "You need to archive the old software, the old OS, the old hardware and the old engineer who knows how it all works."

[EEVblog]

Didn't you ever come across sites like cdrom.com? That had lots of electronic design material in the mid 90s.

IIRC it had bugger-all in 1995.
https://web.archive.org/web/19961221082532/http://www.cdrom.com/
It was mostly shareware collections etc.
It took quite a few years for the main manufacturers to start releasing PDF's to the point where you went to the internet for info instead of your paper library.

[zapta]

I didn't mean to troll, I was explicitly told (in school and faculty) that Pascal is a learning tool.

Yes, it's also a great learning tool. 

[hikariuk]

I didn't mean to troll, I was explicitly told (in school and faculty) that Pascal is a learning tool. Sorry.

I was told the same thing when I was at college (UK style, meaning 16-18 education).  Although I was programming in C at that point anyway, mostly because - in common with a lot of people back then - I'd started teaching myself to program from computer magazines when I was young(er).

[zapta]

Kernighans comments are about how Pascal is unsuitable as a systems programming language are perfectly well considered. And in doing so he entirely avoids highly opinionated value judgements like "terrible" and "fundamentally broken".

K&R C was terrible,  that's why it required so many extensions over the years.

[boffin]

Borland Pascal 'fixed' most of the things you couldn't do with J&W Pascal like variable length strings.

However, for systems programming you could (alleast in the Borland implementation) do very cool things with sets.  Imagine you had an IO port that controlled 4 lights and 4 fans; you could do stuff like this

type IOPins = (light0, light1, light2, light3, fan0, fan1, fan2, fan3);

var myIO  : set of IOPins;

myIO := [light0, fan0];
Port[1234] := myIO;
myIO := myIO + fan1 - fan0;    (* turn on fan1, turn off fan0 *)
Port[1234] := myIO;

or you could do things like
if (light1 in myIO) THEN


laying out IO ports as sets was completely awesome, and sure beats
Port := Port || _mybit4;
Port := Port && (! mybit2);

[free_electron]

YAY. PCB 1.61  that's what i started with. i had schematic 2.7 though , not 3.something.
I programmed in Turbo Pascal 3.5 which did not have a text-gui with mouse. just command line. I bought Turbo Assembler as well. Still have the books of that one i believe. That was the bees knees back then. Combine that with Tom Hogans 'the PC Sourcebook' and you could flick any register, and bit and interrupt and interface with the BIOS.  And there was also 'the undocumented pc' that came with 'Sourcer' , and incredble commenting disassembler.

Later on i bought PowerBasic ( used to be Turbo Basic but borland gave that back to its designer ) Powerbasic is still around. It creates very fast code. Faster than microsofts compilers.

I remember writing a TSR ( who remembers those. Terminate and stay resident programs ) that hooked into the mouse interrupt so i had a mouse caret in the dos command line. i coudl select a line of text and cope it from the screen buffer by directly reading video ram.

the first VGA card came and had RAMDACs and palettedacs so you could spin the color palette. you could also upload new fonts . i made an upside down font. kinda fun. stick the call in the autoexec.bat and every character was upside down on the screen.

ahhh . memories...

[hli]

The "error 200" is caused by using the CRT unit. On start up it calculates how fast the CPU is (to time the 'delay' function), which ends up in a division overflow when the CPU approaches 200MHz (and is reported as division by zero for some reason).
See J.R. Stockton's page for more details and patches that might help.

[SL4P]

... long time ago ( in galaxy, far, far away, off course ) ...

I'm sure George Lucas didn't envisage that...!

[Stonent]

How many Davids are there in your world?
Old Protel runs fine in DosBox emulator under W7.
Windows 7 and up doesn't run 16 bit programs.

64-Bit Windows Vista and higher do not have a 16 bit subsystem.  I believe the 32bit versions still do.

[CatalinaWOW]

Borland Pascal 1.0 was a game changer.  I was writing commercial software for CPM machines at the time as a hobby, and had copies of Microsoft's language studio of the day.  Included Basic, Fortran, COBOL and one other language I can't remember now.  I was proficient in FORTRAN, OK in Basic and the unremembered language.  No contact with Pascal.  The developers window and the really fast compile time completely overcame the limitations of the language and my lack of familiarity.  I was able to port my program to Pascal, correct errors due to translation and have it running in less time than one compile cycle took with the Microsoft languages.  This was on a 2MHz homebrew Z80 machine with 8" floppy drives.  The compile cycle for the MS languages took about half an hour while it was a few seconds for the Borland Pascal.  I think it took about fifteen iterations to get the stuff working, but it still beat whatever MS was doing.  I immediately switched, and stuck with them until OOP took over, and Phillipe Kahn apparently got wealthy and disinterested.  Still have and occasionally use the language.

[miguelvp]

Before Turbo (Borland) Pascal, Pascal was used to teach proper procedural and well structured programming.

I went the Turbo C and later the Turbo C++ route since the later wasn't even on the curriculum but Fortran and Cobol where mandatory (so was the report writer part of Cobol which I really hated, at least it put me off for going to work for a bank or the like), I did take VAX assembly as well, not sure that was mandatory but heck I was a sponge back then (and so I'm now), but I took all the electives I was able to get into, including C++ which was introduced into the curriculum later on, and C was pretty new even if it was around for a while, Fortran was still king.

[boffin]

Before Turbo (Borland) Pascal, Pascal was used to teach proper procedural and well structured programming.

I went the Turbo C and later the Turbo C++ route since the later wasn't even on the curriculum but Fortran and Cobol where mandatory (so was the report writer part of Cobol which I really hated, at least it put me off for going to work for a bank or the like), I did take VAX assembly as well, not sure that was mandatory but heck I was a sponge back then (and so I'm now), but I took all the electives I was able to get into, including C++ which was introduced into the curriculum later on, and C was pretty new even if it was around for a while, Fortran was still king.

I thought I was the only person left on the planet that did VAX assembly. 
The ultimate complex instruction set with instructions like MOVTUC, POLYD and ACBL

[miguelvp]

Before Turbo (Borland) Pascal, Pascal was used to teach proper procedural and well structured programming.

I went the Turbo C and later the Turbo C++ route since the later wasn't even on the curriculum but Fortran and Cobol where mandatory (so was the report writer part of Cobol which I really hated, at least it put me off for going to work for a bank or the like), I did take VAX assembly as well, not sure that was mandatory but heck I was a sponge back then (and so I'm now), but I took all the electives I was able to get into, including C++ which was introduced into the curriculum later on, and C was pretty new even if it was around for a while, Fortran was still king.

I thought I was the only person left on the planet that did VAX assembly. 
The ultimate complex instruction set with instructions like MOVTUC, POLYD and ACBL

What impressed me the most was the Queue instructions manipulating a circular double linked list, very powerful.
I did never used Vax assembly professionally, and funny, my teacher was Canadian, forgot his name long ago.

[SL4P]

Aargh... does FORTRAN on VAX count?

[GonzoTheGreat]

I thought I was the only person left on the planet that did VAX assembly. 
The ultimate complex instruction set with instructions like MOVTUC, POLYD and ACBL

What did these instructions do?
How many were in the instruction set?

[miguelvp]

I thought I was the only person left on the planet that did VAX assembly. 
The ultimate complex instruction set with instructions like MOVTUC, POLYD and ACBL

What did these instructions do?
How many were in the instruction set?

MOVTUC: Move Translated Until Character
moves a string from source to destination using a 256 entry translation table and will stop when the given character is encountered or either the source or destination string are done.
Both source and destination strings have lengths supplied.

POLYD: Polynomial Evaluation using D_Floating point format.
VAX supported 4 floating point formats of different sizes and exponents
F_floating 1 bit sign, 8 bits exponent (bias 127) 23 bits fraction (IEEE 754 single-precision float)

D_floating 1 bit sign, 8 bits exponent (bias 127) 55 bits fraction
(could be converted to F_float by discarding the last 32 bits, great for high precision calculations and then you could convert it to float for the final answer.

G_floating 1 bit sign, 11 bits exponent (bias 1023) 52 bits fraction (IEEE 754 double-precision float)

H_floating 1 bit sign, 15 bits exponent (bias 16383) 112 bits fraction

for the full description on what POLY did, look at page 322 of the linked doc.

ACBL: Add Compare and Branch (Long)
opcode limit.rx, add.rx, index.mx, displ.bw

limit is what to compare to, add is what to add to the index operand, displ is a sign-extended value for the jump when the comparison is greater than or equal to 0. Used for For and Do loops.

Here is the full VAX Instruction Set:
http://h20565.www2.hp.com/hpsc/doc/public/display?docId=emr_na-c04623178

But really I don't think it's of any use anymore, but it was a really powerful instruction set, RISC superseded it in the early 90s and the Alpha chip just put the last nails in the coffin.

Other than in college I've never used it since since all the coding was done in Fortran or C, in the meanwhile DEC was switching to MIPS processors for their DECStations, final end of life for VAX was on the year 2000.

I worked using VAX until 1995, by that time we replaced our 8600 for a Dec Alpha server, original cost of the 8600 probably was around half a million and we got about $10,000 trade in value (after about 10 years) for the Alpha server that sat in the same huge room with halon system and power conditioning on a raised floor.

But that 8600 could have over one hundred users connected and no one felt a hitch. Awesome machine!

[Len]

I thought I was the only person left on the planet that did VAX assembly. 
The ultimate complex instruction set with instructions like MOVTUC, POLYD and ACBL

What did these instructions do?
How many were in the instruction set?

Too much.
Too many.

It turned out that RISC CPUs with simpler instruction sets got higher performance because the hardware required to implement the VAX instruction set was so complex. And a compiler could produce faster code by ignoring some of the fancy instructions and using simpler ones. The CALLS procedure call instruction was a prime example of over-engineering.

I have many fond memories of using VAXes, but they had the Ford Expedition of instruction sets.

[Carl_Smith]

The ultimate complex instruction set with instructions like MOVTUC, POLYD and ACBL

The best assembly language mnemonic is from the PowerPC, called Enforce In Order Execution of I/O, written as EIEIO.

[SL4P]

And the best assembly language (instruction set) for a long time was the PDP-11 family!!
Consistent addressing modes, all sorts of easy to remember functionality.

[kizzap]

Ignoring all the programming arguements, I have to note one thing:

It looks like the power input you have there has no protection what-so-ever 

No fuse, no EMI filters? Not even a Power switch? (unless the Power input is a combination device of some sort.)

-kizzap

[SL4P]

Well, it was a long time ago... declared as a rough and ready prototype... (With lashings of elec insulation tape!)
Apart from the transformer primary and technician's fingers, there isn't much more to protect I'm that plastic case !'

[st]

Does this look familiar at all?

Built this way back when I was still in school and Dave was _not famous_.
Didn't remember at all he had anything to do with it, but it's been sitting on my shelf all of those years.

Thought it would be fun to show this on the forum rather than just the youtube comment.

Should I keep this until it is worth as much as an Apple 1, or should I sell it on Ebay? ;-)

ST

[RupertGo]

For those who enjoy reliving the Coding Wars, the BBC ran a short series last month on Radio 4 about the cultural history of various programming languages called Codes That Changed The World. They're all available here http://www.bbc.co.uk/programmes/b05qqhqp/broadcasts/2015/04, I think indefinitely, with no region block.

(oblig. disclaimer: I was involved in making these)

[Wim_L]

Pascal was a terrible language. Fundamentally broken: http://www.lysator.liu.se/c/bwk-on-pascal.html

Pascal was specifically designed for teaching. It intentionally lacked features that caused inexperienced new programmers to develop hard to debug problems. It should be judged in that context. It was neither terrible nor fundamentally broken. It should be judged in terms of how well it met its original intended purpose.

Kernighans comments are about how Pascal is unsuitable as a systems programming language are perfectly well considered. And in doing so he entirely avoids highly opinionated value judgements like "terrible" and "fundamentally broken".

Also take into account that the criticism was of standard Pascal at the time. The proprietary extended Pascal languages (like Turbo Pascal) had ways to get past some of the limitations of standard Pascal, at the cost of becoming harder to port to other compilers.

[mcs_5]

See J.R. Stockton's page for more details and patches that might help.

There are also TSRs that "fix" the problem without patching executables or recompiling the code. I use r200fix written by Veit Kannegieser (available here: http://hobbes.nmsu.edu/download/pub/dos/r200fix.zip).

My PCB software, accounting software and probably other stuff has the r200 problem...

[vkirov]

Are there any analogs for those LSI chips? I realy want to build this thing just for hobby purposes.

[PKTKS]

I have seen Dave's  project...

And I do have some half dozens schematics and boards
from that time period too.

Mostly are obsolete but they are part of my stuff.

I have already found the FREE AUTOTRAX version released from ALTIUM.
Fine it works good enough under DOSBox.

But I can not find the companion SCHEDIT  set with libraries so to view
my schematics too in original format.

If anyone can help by pointing any  loadable usable version...

That would be very helpful to have my stuff back.

Thank you very much  any help welcome.
Paul