Before Raspberry Pi and Arduino - Computerphile
ฝัง
- เผยแพร่เมื่อ 4 ส.ค. 2024
- Professor Brailsford with an example of what he had to do when Computer A didn't talk to Computer B.
/ computerphile
/ computer_phile
This video was filmed and edited by Sean Riley.
Computer Science at the University of Nottingham: bit.ly/nottscomputer
Computerphile is a sister project to Brady Haran's Numberphile. More at www.bradyharan.com
Watching him makes me realize how little I know.
@Kerem all info you know will be useless in 2119.
@Crossfire Punctuation use is a part of grammar.
Why would you know something so niche so far in the past, probably before your birth? You'll have your own anecdotes to tell when you're older, don't you worry.
@Crossfire Humanity is so adept at repeating the same mistakes in new contexts that the past will always be relevant.
It makes me realize how much things have changed. I have found memories of writing assembly on a PDP-1140.
The introduction on C was such a relief, though the combination of assembly and C was a perfect paring. Using C for the majority of the work with assembly were it was needed was a marriage made in heaven.
Programmers today are so far removed from the actual bits that it's difficult to see that all we are and were doing is wrangling electrons!
Which when you think of it is pure genious that the combination of logic gates can actually render a video on the screen in front of me.
A far cry from a VT-100 terminal!
Prof B. has countless interesting stories RE: tech history. I find the evolution of technology almost as interesting as the technology itself.
I remember Sibelius Academy's sheet music composer was visibly in tears, when a computer and laser printer of 1983 made the day's work in fifteen minutes.
Composer. You mean copyist?
He saw his whole career future disappear.
They could just as easily have been tears of joy from no longer having to do manual transcription. Timo Noko did state they were a composer and can you imagine anything more frustrating for a creative person than a boring job.
I think he may have been referring to a composer in the typesetting sense rather than the musical sense, but the context does make it an ambiguous term.
@@molletts That's a compositor not a composer. My father is a printer by trade and did this exact job for many years.
Thank-you Professor Brailsford for this trip down memory lane.
In 1982 I was an engineering undergraduate in Nottingham at Trent Polytechnic (as it was then). In 1985 I was offered the use of a college professional quality daisy wheel printer to print out my thesis. The printer was Centronics (parallel) only and the Sinclair QL I'd used was serial only...
I have seen this "quality driven design" kill project after project. The sales force always thinks they have to attack the highest quality competitor...ignoring the market they don't know about because there was never anything good-enough and cheap-enough to adopt. When the only people that can afford your product are a few with deep pockets, you get sucked into thinking that is the entire market. A salesman always thinks the ideal design is whatever would have made the sale to the last lead he lost.
Some very perceptive comments.Thank you! Once I saw my first boot-up page from the 1984 Postscript-based Apple Laserwriter with its "proper fonts" even at only 300 dpi , I thought to myself: " .... this changes everything!. How can Monotype and Linotype survive this?" And the answer was that they had to nearly go under, and eventually merge, in order to be reborn as mere font suppliers and a shadow of their former selves ....
The last line is a killer!!
I've said it before and I'll say it again, Prof. Brailsford is a real treasure!
he's so freakin cool :)
I could listen to Doctor Brailsford talk tech history for hours. He's forgotten more about technology than most of us have ever known.
This reminds me of when the department went from the 300 dpi printers to 1200 dpi laser printers. I was horrified at the lousy quality of the output. Turns out that the 1200 dpi printer resolution revealed the sloppiness of the PostScript programming behind the scenes. No attention to linecap ends and the like.
Was this the early days of cheap knockoff PostScript clones?
I've always heard from the previous gen that "programming PIC microcontrollers to drive any type of electronics was hellish because we only had assembler", but to hear that universities had C cross compilers back in the '80 makes me reflect...
‘Comfortably slow’. Made me laugh. The early days of computing were fun and exciting. Raspberry Pi has reignited my passion.
I remember those interfacing problems and the utility of 8-bit processors as in-line point solutions! In particular, interfacing a PDP-11 to a 12V parallel bus relay-driven SCADA system (others were trying to create computerized SCADA systems from scratch - our retro-fit approach made us tons of money). And creating DIY ADCs (analog-to-digital converters) using a few GPIO bits and an external RC integrator. So many of these "hacks" were hurriedly prototyped on wire-wrapped boards. Several of which served for over two decades in fairly nasty environments.
Then the company BlackBox came on the scene offering a ridiculous number of interface boxes that, while expensive, cost somewhat less than in-house development and were available off the shelf. And then PCs arrived with their parallel and serial ports, and the availability of easily programmed PLAs, PLDs and early CPLDs (pre-FPGA) that replaced the "jelly bean" 74-series TTL interface chips and drivers got rid of the wire-wrap boards in the wink of an eye.
Interesting stuff. I wonder if that little Z80 box is still hiding somewhere...
A friend with lots of home automation projects had a stack of BBC Micros he picked up from a classroom clearing that he used to join things together. Once, he called to say he had a new jukebox to gut, and a motorised satellite dish that needed an adaptive alignment box with keypad input. Over the weekend, he did the hardware and I coded. 6502 assembly, written on A4 in Biro and typed in... I think that was the last time ever I wrote some significant code that worked first time!
Beebs were the ultimate glue hardware, IMHO. The number of BBC Micros being used to connect things to other things must still be non-zero. Raspberry Pi was smart to use them as their spiritual ancestor.
+1 on that. I used them for a whole range of industrial and test rig jobs. Once, due to no spare being available, I wrote a control panel for a plastics extruder line in FORTH, and put it in as a temporary unit - it was still running when they scrapped the machine, years later. One reason they liked it was that the early PC based controller had to reboot from a disc if it crashed, whereas the BBC took seconds to boot from ROM.
Somewhere I still have the hardware for a cable tester I built with a BBC, it used the 1MHz bus to connect the cables and the analogue port to measure resistance. It was expandable, initially doing 32 cores. It could 'sniff out' an unlnown serial cable in seconds and then use the data to test new ones.
Lovely machines. I still have a few.
Most interesting 'back in my day' video ever. !!
So much talk about things other than the point in the title. And so enjoyable to listen to it all.
Excellent job Mr. Brailsford.
I can imagine this task required a lot of work and thinking, but one which must have turned out very statisfying as well !
At least that is what I felt listening to your story. Thank you for sharing :)
Best regards from Holland
I still remember printing out my final year project report on the uni's VAX-11/780 connected to a DEC postscript laser printer. It was magical seeing the quality of the output way back when.
As a Z80 assembly programmer back in the 80s (and still dabbling in it now), it could have been fun to write the software in assembly. Can you remember if you buffered one byte at a time from the serial port before putting it out did you receive multiple bytes from the serial before outputting them on the parallel as a burst?
This was a great story with several unexpected insights. I enjoyed it very much.
Professor Brailsford! I love to hear him talk and share all his knowledge of computer history. If I could afford it, I'd take a trip to the UK just to have a chance to meet him!
If you were around in the 80s will totally understand what he is talking about.
Back in the 80s, I worked on a Xerox 9700 laser printer.... the thing was huge, even if it only did 300dpi. :)
It could print 120 images/minute in full duplex. It had a microprocessor controller in it, but was run by a separate controller with a DEC PDP-11/34, disk pack, and 9-track tape drive. There was no direct connection to the mainframe, you would print to mag tape, then sneakernet it over. It did have some typesetting capability with its "Form Design Language" Fun Times.
When I listen to professor Brailsford, I feel like listening to my father telling me how back in his time they didn't have internet or phones, or like listening to my grandfather telling me how there was only one car and no electricity in his village during his childhood. And I can see how far technology has come in just a few generations.
To the layperson, the glass is either half full or half empty.
To the engineer the glass is twice as big as it needs to be.
Engineering: when "just enough" is perfect.
4:28 , the car analogy at work we use is, 'we just bought an exotic car, but no one knows how to drive it' .
Sage words right at the end, there! I found myself nodding in agreement in many places in this video - never mind I was in the first half of primary school when Dr Brailsford undertook this particular project, lol.
This video reminds me of projects as an electronics student in the 1990s - writing 8088 assembler and burning to EPROM so that our wire-wrapped 8088 constructions would talk to each other over RS-232 serial. Another project to build a simple graphics library for PC (DOS) -> written in assembly :)
Oh, and wiring an Amiga "ParNet" cable to network Amiga Amiga and later Amiga PC
Ahh, the Z80...my first computer was a TRS-80, complete with an audio cassette recorder for data storage and 16k of ram. Black and white monitor, which was just a rebranded Zenith TV without the tuner, and the printer used rolls of maybe 6" wide aluminized paper. Heck of a machine for its day though!
I want so many more stories from this man.
:) What fun it was - I can still smell the bromide!
Hi Julian !
Glad you are around! Need to send you a long-ish email soon before I can do the next episode (!).
: I wrote the original prototype C/Z80 interfacing software to the Omnitech typesetter.and its parallel interface. I called it PLOD (Program for Linotype-Omnitech Dialogue) But when we had to replace the Omnitech with a 202 - and get that machine ready for actual exam-paper production mode - Julian took a huge burden off my shoulders by doing a total rewrite of PLOD, for the 202 and for its own (very different) parallel interface. Thanks Julian !
It's amazing, he gave a stunning overview on hardware evolution of hardware communication in 15 minutes
One thing not mentioned is that even back then designing a board with a Z80 wasn't totally trivial. It's not like an Arduino that is already a full system. The Z80 needs a crystal or some kind of clock, a memory, a memory bus, impedance drivers for the IOs... and not least, some kind of bootstrap code for it to do something useful, unless you can map & program the PROM directly starting at address zero (which is a typical choice IIRC).
There isn't actually much choice, the Z80 will always start executing code from address zero, whether there is any code or not.
I dabble in Z80 assembler myself, so i know quite well how it works.
Amazing storyteller!
Great story, thanks for sharing it!
Thank you for such a wonderful video on a wonderful topic :-)
thank you so much for these videos.
Love this channel
I love you
The One I love you more
In the early '80s, I worked for Floating Point Systems, Inc. We had an IBM laser printer that cost $80,000. We used Runoff on Prime computers to create documents (manuals), which were published on Xerox copiers. Normal input was via IBM Mag Cards, which were written using specialized Selectric typewriters. We didn't use them. As I remember, we used a serial cable from a Prime to the printer, and it ran at 1200 baud. I don't remember if we had to convert ASCII to EBCDIC.
"Software A didn't speak with Software B"
It still happens to this day haha
*talk not speak we're talking about communication here
*_Salute_*
Thank you for your service
I remember Versatek electrostatic printers (late 70s), especially the odor of the electrostatic printing. Don't remember the resolution. I wrote a library in PDP-11 assembler that was many times faster than the vendor-supplied Fortran. The speed was necessary to meet the requirements of the contract. I think they were used to plot maps/charts. I also interfaced the PDP-11/70 to submit batch jobs to, I think, a remote CDC over either 9600 or 19200 synchronous modem (high speed in the late 70s). DEC had an interface that worked, but I had to write the software. 12-bit words with the bits in reverse order IIRC.
14:00 i got a good chuckle :) "They just didn't care about 300 dpi"
My neck of the woods - oh what we would have given for a Linotronic. I trained on the Compugraphic system. We upgraded that typesetter to the Powerview system. We eventually got a Linotype Linotronic 300 - but the ball ache of developing the film & bromide was hell. Literally (Linotype Hell). Miss it with a passion sat here at an iMac pro
Sometimes I wish I was born a generation earlier, there were so many new exciting things in development and so many revolutionary ideas to be had! And you could still know and understand everything about what you're doing, without that much layers of abstraction!
I’ve always thought the same. Born too late to see the exciting technological breakthroughs of the beginning of the computer age. And born too early to see the (hopefully) upcoming space age.
Incompatibility of machines, credit and debit cards you could only use your banks teller machines or banks or building societies that used the same technology. Even take overs of companies were stopped if the compatibility was too varied.
970 dots/inch? Hah! Our exams were usually mimeographed copies of hand written text, and we didn't know any better.
Brilliant explanation of the print industry from the 80’s to today.
Sadly the vendors did not learn from the previous generations. I recall hitting my head against the desk trying to get the early inkjet printers to use the proper fonts to print hazardous material labels. Only a search on CompuServe, blast from the past, yielded an explanation as to why I was not getting WYSIWYG.
Centronics parallel had just the one set of data bits D0-D7 but they could be kludged to be bidrectional. Bruce Eckels ran a How-To in Micro Cornucopia where he took a standard interface card and soldered an additional bit from the card's LS373 config register to the /E on the card's LS 244 output driver. By running the LS244 into tri-state mode that let the LS245 bus transceiver just read back from D0-D7. Presto, 8 bits of parallel input (albeit with a lot of parasitic capacitance and inductive crosstalk).
This man is a national treasure 😎
This was a pleasure! (even though I know nothing about it!)
I/O bridging is done more by Arduino than raspberry since raspberry is just a rebranded set top box. Before 2007, we had the mighty 68HC11.
Oh I think the PIC16F84 and its later cousins took over that job way more than a decade before 2007.
This video goes a lot further than just dev boards, very interesting lesson
Back in the 80s I remember a company called "Black Box" which specialized in making custom interfaces between equipment.
I went the other way once with a PIC chip. I took parallel data and turned it into serial DCC model train control data.
Those were the days... I did assembly and C code on a NEC v20 CPU (Intel 80286 compatible), communicating with serial ports and parallel ports, created a small kernel to run multiple threads in C and assembly an 8KB EPROM. Fun times indeed!
Dr. Brailsford takes any opportunity to talk about typesetting.
I like the way you talk. God bless you😊
Oh professor... The first laser printers were 100dpi... Then 120... Then 200... 300... 600 and finally 1200. But over time it was agreed 1200 is overkill and 300 is the normal resolution... TODAY
did the z80 circuit understand something of the IO or was it just a passthrough?
@saladnuts so that is reading in from the rs232 chip (maybe a SN75189) and output to the paralel port chip, in a loop.start: IN RS232, OUT PLI, jump start.
But he is talking about waiting loops, so there is some intelligence. My question is, did it understand the protocol or not
I was wondering that, because if it was just receiving and sending one byte at a time (including handshaking) it wouldn't have been overly complicated.
Back when MP3s became a standard for music, audiophiles were *horrified* by that! Oh no!! It's terrible!! But for the vast majority of people, it was "good enough." And now the only people who buy music on CDs are audiophiles. The rest of us download our music from iTunes, or stream it on Spotify. And it really is "good enough."
Even CDs were rubbished at the time. The gold standard was analogue - the vinyl record.
Portability is something people of my generation takes for granted.
I see Prof. Brailsford, I hit like before watching the video itself.
Ha! Remember similar problems and technologies at the Uni of Western Australia in the early 1980s.
@8m53s: Of course, Ethernet (e.g. Gb) is famously four twisted pairs in parallel, each carrying 250 Mbps in serial.
8:53
Really interesting video. However, I'd love to have subtitles to distinguish all the names and product codes.
So how much money did this very expensive printer + costs for interface + development time + daily servicing time + toner + paper save compared to the outsourcing the printing?
Professor Brailsford is such a likable guy.
I finally understand why troff exists.
I thought this was gonna be a video about PC/104 based on the title. Colour me surprised it was wayyy before even that!
I'm surprised the solution wasn't a BBC Micro. Save on hardware development by buying a device with a serial and a parallel port already built in. I accept that the Z80 had a C compiler which made things a bit easier but that C compiler was pretty limited back then. Oh, and the BBC had Forth! :-)
Well. That answers my question about whatever happened to parallel.
Nice, this type interface synthesis and conversion is nice to do using an FPGA.
I wonder if you could steer a rocket (not a space rocket, just like medium to short range) using just an arduino and it's extensions (accelerometers, etc)?
In the Z80 days, a printer (not a teletype) had a Centromics-8-bit-parallel-port. And that went over to the IBM-PC nearly until USB or Ethernet took over.
What is that CD on his desk?
Also: kitcheny curtains in an office. lol
There was a fast parallel standard: SCSI reached up to 5 Gbit/s in its last parallel incarnation (Ultra-640).
i wish i had him as my grandfather.
Is that a Don Henley CD on the desk? Actual Miles I think... Nice!
Yes - But... given the subject of this video it really should have been the Building the Perfect Beast album!
watched this and realizing that what he have learned at the end still applies today
Please add subtitles
I remember being one of the teenage computer geeks being sooo excited when Arduino first came about... now excuse me as I have a bit of an existential moment as I realize this piece of tech has been around for half my life...
I wish to be like this man one day. Sitting back in my rocking chair, telling the youngsters the story of the early days of AI and LLMs...
The need to develop interfaces for propriety hardware has been around since the start of the Industrial revolution. (Mechanical, Hydraulic, Electrical, Electronic.) But the Prof's point about designing what you need, rather than what you want, is succinct.
Typical of academic technology around that period is the use of a Z80 (2nd or 3rd gen microprocessor.) and its complicated assortment of support chips and hardware - when the real world was already using standalone microcontrollers such as the Intel 8048 family.
There were a multitude of compilers and cross-compilers in Basic, C, Pascal and ASM available.
The later i8051 family is still in use today in many upgraded and enhanced forms.
The 8051 seems to be eternally useful. I remember watching a video of a teardown of a White Van Speaker Scam AV receiver, and it included an 8051 clone in it.
About 1985 I designed and wrote the same kind of hardware/software to connect a printer to a terminal. It was easy to do that on a 8051 machine. :-)
I used the whitesmith compiler for the 6811, back in the late 1980's
Did the Z80 have a UART port, or did you bit bang it.
FWIW all EE students should be forced to replicate ports like UARTs, entirely in code (bit banged). Nothing gives you a better understanding of something thing, than build it from scratch.
Most manufacturers I've worked for want a custom board shape, full hardware and software IP ownership and low power operation. That rules out Arduino etc for production use.
Actual Miles is a great album
I was using bromide paper in 1992
Yep, there were no blogs, no online articles, no google, no youtube in fact no internet when I was doing my project. later I use to see that explorer logo and use to wonder what it will do? We had to learn everything from help files available for software and had to remember everything. Had very limited resources . very less memory to store data as well as computing speed , almost nothing compared to what we have today in our phones. All the programs were stored in five and half inch floppy disk later got three and half inch floppies, I remember how excited I was to buy my first 3 and half inch floppy. Today you cant even store 1 image from your phone to those floppy disks and I remember I stored my whole project in 1 floppy.
The good old/bad old days of printing...
So did you have Z80-compatible serial and parallel chips in that converter box, or did you have to bit-bang everything yourself?
This is not just about interfacing foreign stuff, this is about *it being way difficult even for a man with the training to do it and an entire university's resources at his disposal* back then.
"And lo! Hewlett Packard didst buildeth the LaserJet, and the people saw it was good, and it was good!"
Built like a battleship. Actually no, better than that, built like an old-school Nokia. Cheap toner, easy to find parts. Parallel port, or ethernet module, good quality fonts. I'm surprised Nottingham Uni didn't use a BBC micro for protocol conversion though, with a standard serial, parallel and additional high speed ports, plus a built-in assembler. They were brilliant machines for hacking real world systems together, the ethos very much lives on in the Raspberry Pi.
I love these videos but it feels like the camera is floating on a nearby pond. Get that camera operator a tripod!
Bash those bit!. In that realm, sometimes it's easier in assembler
Cheapo laser printers for home use and like 300 bucks are 600 dpi now, yes. But the professional grade laser printers one finds at a university will all have 1200 dpi at least, newer models might even have 2400 dpi.
Maybe a dumb question but by customers does he mean the tuition paying students taking exams? Why would they care about dpi as long as they could read it.
"paying"?
is this some kind of American joke I'm too European to understand?
He just described the disruption curve
"Oh, this is a job for a real professional.
I can feel sums of money, perhaps £10 a page."
HP might have knowledge about the printer software. The printing hardware itself came from Canon in Japan.
Editors: We get 3 times as many views if Raspberry Pi or Arduino is in the title
Prof: *mentions RPi once
Editors: *Hi fives and beers
indeed - was interesting eventually, but I did keep checking the title thinking i had clicked on the wrong clip