What kind of code is on 1980's, 1990's scope's/etc ?

[engrguy42]

Wow, another stroll down memory lane...

Z80 reminds me of when I built the Heathkit/Zenith H89 (or was it a Z89?) with a Zilog Z80 CPU. None of this Windows rubbish. Good old Assembly language and Basic.

I'll raise you the Tektronix terminals from the 70's and 80's. No CPU; all discrete logic.

Oh really?? You want a piece of this?? Huh?

Here we go...

IBM 360 mainframe with SLT (Solid Logic Technology...okay, I had to look that part up)

Punch card readers and writers. Fortran statements, one line per card. You punch one line at a time to a card, then stack them all together in this 2ft long cardboard box, then walk them into the campus computer building and some poor slob loads them into this reader that pulls one card at a time and reads it into memory. And you get a big folded paper output from a monster line printer the next day that says you screwed up a comma in card #443 and you have to do it all over again and wait another day.

And don't even get me started on paper tape punches...

[WattsThat]

I'll raise you the Tektronix terminals from the 70's and 80's. No CPU; all discrete logic.

And the HP9100A RPN calculator in 1968. Not one digital device.

Except, of course, for the DTL and ferrite memory.
http://www.vintagecalculators.com/html/hewlett-packard_hp_9100a.html

Well the “DTL” was built with discreet transistors... yeah, I guess that’s a digital device... with enough base current. I guess I’ve got to be far more specific with my language for all the hair splitters out there. Jeez.

engrguy42: Long live unit record equipment. My favorite? The IBM 2560 multi-function card machine. AKA the MFCM, also known as the mother F’er card machine when it destroyed half of your deck of homework FORTRAN 5081’s in a few seconds.

[joeqsmith]

My Hitachi analog scope used a Z80.   My first real DSO is a LeCroy 7200.  It has a VME chassis and uses 68000 architecture.   They used PSOS.   

https://en.wikipedia.org/wiki/PSOS_(real-time_operating_system)

I later bought a 7200A that is PC based using an Intel processor.  That system also uses PSOS.   I had given mine away but then a friend of mine sent me another.   It needed a little work.  You can see what it looks like here. 

[engrguy42]

Apologies to the OP for hijacking his thread....I promise I'll behave after this...

I recall coming out the the computer building one time and I saw a box of some poor guy's hundreds of Fortran cards lying all scattered in the snow. He was just standing there, almost in tears.

All I could say was "wow...dude...that sucks".

I can't imagine having to re-sort that mess, or worse yet re-punch them.

[oPossum]

The Heathkit 4850 was a DSO with a Z80. Could use a CRO for a display.
(Attachment Link)

Never knew they did that one, thanks. It must have right before they went under. One on eBay right now for $80. Was it a complete kit or partially assembled like many of their last products were with the high density digital stuff factory built?

It was available as a kit or assembled. There was also a version without the front panel for a little lower cost. It required a computer to operate. The one with the font panel could be used with a CRO for display and/or computer connection.

I got one when Heathkit was exiting the kit business in the early 90's. The best feature is the powerup front panel and relay test. It's a craptacular DSO.

[tggzzz]

I'll raise you the Tektronix terminals from the 70's and 80's. No CPU; all discrete logic.

And the HP9100A RPN calculator in 1968. Not one digital device.

Except, of course, for the DTL and ferrite memory.
http://www.vintagecalculators.com/html/hewlett-packard_hp_9100a.html

Well the “DTL” was built with discreet transistors... yeah, I guess that’s a digital device... with enough base current. I guess I’ve got to be far more specific with my language for all the hair splitters out there. Jeez.

Except, of course, that DTL was the basis for the first large-volume ICs, the Signetics SE100 series and Fairchild 930 series
http://www.calcuseum.com/SCRAPBOOK/BONUS/72524/1.htm

Ferrite memory is also digital

[David Hess]

"1980's, 1990's" covers a lot of years but there wouldn't really be many CPUs to choose from.

Most of them are probably either Z80 or 6800 (1980s)  or 68000 (1990s).

I doubt Intel would have been inside any of them, although there's bound to be somebody who knows different.

Nope. The Tektronix 2030 has an 8080 processor inside.

2030?  Were you thinking of a different model number?

Anyway, the Tek 2246 series scopes use an 8088 CPU, which is why they have a snappier UI than the much more expensive 2465 series, which use a lowly 6800.

The 7854 mainframe uses a TMS9900 CPU, oddly enough.

The 7854 is pretty weird and the changes Tektronix made in the following DSOs show that they learned a lot from it.

The Tektronix 468 used an 8085, the 2230 used an 8088, the 2232 used an 80188, and the 2440 series DSOs used a 6805 and 68809.  Often custom processors for dedicated tasks like display list rendering and waveform processing are also present.

Tektronix had separate design teams for low and high end products but perhaps counter intuitively, the low end products like the 22xx series of 4-channel oscilloscopes often seem more advanced than their contemporary higher end products like the 24xx of series.  There does not seem to have been much sharing between the design teams.

[chris_leyson]

HP3585 spectrum analyser used the HP9825 16-bit processor. HP9825 desktop computer software was written in assembler and there are complete listings in some of the HP9825 related patents. Also, the HP9825 has almost identical architecture/instruction set to an early HP2000 series mini computer, HP2114 I think.

[tggzzz]

I'll raise you the Tektronix terminals from the 70's and 80's. No CPU; all discrete logic.

And the HP9100A RPN calculator in 1968. Not one digital device.

Except, of course, for the DTL and ferrite memory.
http://www.vintagecalculators.com/html/hewlett-packard_hp_9100a.html

Well the “DTL” was built with discreet transistors... yeah, I guess that’s a digital device... with enough base current. I guess I’ve got to be far more specific with my language for all the hair splitters out there. Jeez.

engrguy42: Long live unit record equipment. My favorite? The IBM 2560 multi-function card machine. AKA the MFCM, also known as the mother F’er card machine when it destroyed half of your deck of homework FORTRAN 5081’s in a few seconds.

There is also discrete transistor logic that is neither DTL nor RTL - but based on magnetic cores. It was in the first computer that I used, an Elliott 803.
http://www.cs.man.ac.uk/CCS/res/res03.htm#e


And if you want to be obscure, consider that I have a Tektronix test instrument that does divide-by-5 using only 3 transistors. It is discussed in several places on this forum, notably https://www.eevblog.com/forum/repair/tektronix-184-restoration/msg2823978/#msg2823978 and intermittently in the monster TEA thread.

Even earlier digital memories were made from liquid mercury and CRTs

[shakalnokturn]

https://www.youtube.com/embed/p7SkE5pERtA

[PaulAm]

Except, of course, for the DTL and ferrite memory.
http://www.vintagecalculators.com/html/hewlett-packard_hp_9100a.html

I actually have one of these and it is an incredibly awesome piece of engineering.  The program ROM is a board with a zillion diodes.  The ferrite is for user program memory and allows program retention over power cycles.  There are a few details available on the programming effort to squeeze transcendental functions into it.  Absolutely amazing.

On some early instruments, HP used a proprietary processor.  It's in the 3325a, some of the early LCR meters, some GPIB boards and perhaps some others.  It has a very limited instruction set and memory map and had to be programmed in assembly.   As a software guy I am impressed by what they managed to squeeze out of it.

[Fungus]

I actually have one of these and it is an incredibly awesome piece of engineering.  The program ROM is a board with a zillion diodes.

A lot of early calculators reduced those diodes into a single chip but they were still hard-wired-logic on the inside, very far from being what most people would class as a microprocessor.

eg. I've got a green-screen Casio calculator where you can see the digits on the screen glitching when you press keys and 'clocking' when it computes results. The logic gates inside the ALU are hard wired to the output segments in the display. The screen reflect what's going on inside the chip.

(model is CASIO Personal 1)

[tggzzz]

Except, of course, for the DTL and ferrite memory.
http://www.vintagecalculators.com/html/hewlett-packard_hp_9100a.html

I actually have one of these and it is an incredibly awesome piece of engineering.  The program ROM is a board with a zillion diodes.  The ferrite is for user program memory and allows program retention over power cycles.  There are a few details available on the programming effort to squeeze transcendental functions into it.  Absolutely amazing.

On some early instruments, HP used a proprietary processor.  It's in the 3325a, some of the early LCR meters, some GPIB boards and perhaps some others.  It has a very limited instruction set and memory map and had to be programmed in assembly.   As a software guy I am impressed by what they managed to squeeze out of it.

Assembler programming was the norm in 70s and first half of the 80s, as were miniscule memories. Consider, for example:  http://files.righto.com/calculator/sinclair_scientific_simulator.html

[TomS_]

]

Yeah, no kidding...I've often wondered why virtually every bit of software has to go to the users to find all the bugs in seemingly never ending alpha/beta/whatever cycles, but hardware (for the most part) just seems to work.

Go figure.

Users won't be able to run the buggy software without working hardware. 😂

[Howardlong]

Tektronix even used Smalltalk - what a bad idea.

That was a prime example of tail-wagging-the-dog development if ever I saw it.

It’s a trend that continues to this day, trendy developers with something to prove, ignoring all the risks in favour of getting a new line on their resume.

[tggzzz]

Tektronix even used Smalltalk - what a bad idea.

That was a prime example of tail-wagging-the-dog development if ever I saw it.

I can't comment on the Tek example, but that certainly wasn't true with the HP examples I saw. They worked well, partly due to a careful partitioning between hardware and software, and partly due to an understanding of what the software could and couldn't achieve.

It’s a trend that continues to this day, trendy developers with something to prove, ignoring all the risks in favour of getting a new line on their resume.

The trick is to be able to work out which new technology offer something new and beneficial, as opposed to mere trends offering something new and different.

It takes experience and wisdom to be able to do that, which conflicts with many HR-droids views of what makes a good developer.

[Gyro]

The 7854 mainframe uses a TMS9900 CPU, oddly enough.

The TMS9900 was a fun CPU choice. It stored all of it's general purpose registers in RAM, so that Subroutines and Interrupt routines could adopt a completely separate register set just by changing the pointer. It also had no Stack in the traditional sense, it just stored the PC for a return.

[miken]

[Disclosure: I work for Keysight Technologies, formerly Agilent/HP. Opinions are my own.]

The firmware for the HP 8568A was written in something colloquially known as WHEELGOL, with calls into assembly routines. The name comes from Lynn Wheelwright, whose Master's Thesis was an ALGOL compiler built on BYU's TREE-META system. I'm not sure how far forward it was carried, but I suspect all the 68K-based 856x models were on that framework.

There's a little article in the June 1978 HP Journal but not many details. The '68A and '66A initially used a custom HP microprocessor, and then the '66B got the 68K.