With Don Lancaster’s first, primitive ASCII encoder built, it was only natural that I wanted to test it out.  Wiring up the keyboard is going to be an eye-crossing affair, and before I start going down the rabbit hole of ‘Why isn’t it generating the proper character?’ with that attached, it’d pay to make sure the encoder is working correctly on its own.

Thanks to my online friend Chuck, it was suggested that I could power the thing, which expected a 10V power supply (or 12V if you changed a diode, and that is what will end up happening as the TVT power supplies 12V), with either a 9V battery or 9V supply.  9V would be just enough to give the thing some life.

Excitedly I soldered in leads for positive and ground, and then used one of those cheapo Radio Shack multi adapters, with no head on.  After figuring out which pin socket was positive and negative, I hooked up!

Now what we have to do is simulate keypresses.  I’ll try to explain this as best I can within the limits of my own understanding.    Take a look at this page from the ASCII encoder article here:

Looking at the table in the bottom left corner, as an example, the comma (,) – the ASCII code for comma is 010-1100.  You read from bit 7 down to bit 1 (most significant to least).  I don’t know why.  Don’t ask. 🙂

Anyway, for our purposes, to generate the desired comma keypress, we need to look at the parts placement guide in the upper right of that same page.  Now, this thing isn’t well marked, but essentially comma belongs to a group of symbol keys.  You’d have to understand how to read a schematic to figure that out in this case, but I had help.  The first pin we need to short for comma is the pin marked /,?.

Now the second bit is tricky and I’ll admit, I have no idea why it’s done this way.  But essentially if you look at the lower pins for the larger connector on that parts placement guide, you’ll see they are all three digit combinations of 1s and 0s.  Those correspond to the last three bits of the binary code.  In the comma’s case, that is 100.  So what that means is, to generate a  comma, our keyboard matrix must be set up in a way that one side of the comma key contacts the /,? pin of the encoder, and the other contacts the pin marked 100.  Since my keyboard isn’t ready yet, I’m just going to use a short piece of wire and.. er.. short those two pins.  Then I’m going to use a logic probe to detect how the bits are set with comma pressed:

Voila!

To orient yourself, I am moving my logic probe from pin 1 down through pin 7.  Since low on the logic probe is 0, and high is 1, that produces: 0011010  or reversed: 010-1100.  Exactly what we expected!

Now truthfully, the video above was taken after some major tweaking.  I made some serious rookie mistakes, including accidentally installing 2N5139 transistors where the 2N5129 should be.  They are not the same!  Luckily I didn’t blow anything up.  Once I got that booboo fixed, it was really just a matter of nailing down some accidental solder bridges.

The last hangup was a diode at D13.    This is a Class A example of some of the difficulties faced by hobbyists: errors!  Looking at the parts placement diagram from the article above, you’d think D13 belongs right between X-Z and H-0 on the far right side of the encoder.  WRONG.  It’s supposed to be right between D25 and D20.  You can even see the pad for it.  Ugggggh.  This kind of thing fills the inexperienced hobbyist with dread, because this is actually a very simple device.  The TVT is not.  I do have the advantage of having a full page of corrections included in the instructions I’m working from, so there’s that, but who knows what else was missed!  A good rule of thumb: when in doubt, check the schematic!  Unless that’s wrong too.  Or.. you can’t read a schematic..

Wow, that got addictive.

Thought I’d just add some of the resistors, a bit at a time, solder in… here and there when I had a few minutes.  But no, once I started placing components it became really addictive.  I couldn’t stop.  I got them all in in about half an hour!

I went line by line from the parts list, installing all resistors of one value first, then another.  Then diodes.   I shelled out for vintage on those too — they were easy enough to find thanks to my old standby on ebay, acpsurplus!

Transistors are all in, disc cap is in.  Man, this is starting to look like a thing!

But I gotta be careful.  The traces are tiny and packed close together.  Twice I’ve accidentally bridged them.  And that’s what I’ve caught, there could be others!!  Have to look really, really closely.

Pretty much all there except for jumper wires.  Time for a beauty shot!

And there you have it folks — one pseudo-vintage Don Lancaster-designed brute force ASCII encoder!

Now to test!

So yeah, here I was thinking I had it all covered and was starting to build my ASCII encoder.  I had bought a slew of resistors of different values from Ebay on the cheap.  I think I paid $20 for 30+ different values.  But of course, I wasn’t thinking ‘vintage’.  And on installing the first one on my ASCII encoder, I realized right away I didn’t like the look of it.

The whole point of this TV Typewriter project is to produce something that is as close to 1973 specs as possible.  Now of course, it’s probably not going to be possible to get everything vintage.  At least not right away, but it is something to try for and I had simply gotten lazy with the resistors, figuring they were such a minor part nobody would notice.

Unfortunately once I installed that first one, I noticed.  The electric blue colour just looked wrong in context.  I mean, it wouldn’t have been entirely inconsistent to use some new ones.  After all, after 40 years, stuff breaks.  I could simply say ‘well by now, some of these would have had to have been changed out’.  But I had gone to all that trouble getting correct vintage ICs.  If there was the possibility of getting period correct resistors (and diodes), I wanted to do it.

Turns out ebay has huge stocks of vintage electronic components.  And soon I had hit the jackpot with an auction that offered vintage Allen Bradley carbon composite resistors of virtually every value.  I was able to do a one stop shop and get them all!

They just look so much more correct.

Now, I was given some warning about these.  Resistor technology has advanced a lot in the 4 decades since this project came out.   Old carbon composite resistors had a tendency to ‘wander’ over time and were not as precise as today’s carbon film.  I guess as I embark on building the encoder in earnest, I’ll have to check each of these with my multimeter, and hope any wandering happens years from now, and not while this novice hobbyist is trying to get the thing working!   But yeah, cross fingers, on to building we go!

We don’t tend to give keyboards a second thought here in 2016.  They are ubiquitous and cheap.  You can get a brand new one for as little as $5.00.

Not so in 1973.  Keyboards in 1973 were up there cost wise.  Home computing was virtually non existent.  If you were a computing enthusiast or hobbyist, you faced a pretty steep bill for a new keyboard.  You could, of course, pick up a surplus unit.  These were discarded keyboards, typically from card punch machines.  But there was a problem — these typically used old teletype or punch card coding standards like Baudot or EBCDIC.  With ASCII taking over, you had your work cut out for you if you wanted to convert to this relatively recent standard.

Enter Don Lancaster.

Don preached electronics on the cheap from the pulpit of hobbyist magazines like Radio Electronics, and in February of 1973 he published an article in that magazine describing a cost effective, if tedious, means of building one’s own keyboard.

I don’t know how much of a hit the keyboard project was.  It was a serious job, including hand winding springs and precise measurements and drilling of components.  Having built the beast though, to make it truly useful one needed an encoder.  And that was Don’s next project, in April of that same year.

The device itself was about four inches by four inches, and was intended to bolt onto the Low Cost Keyboard.  Don would ultimately improve on this design later on with emphasis on reducing size and complexity.  But for my purposes, this is the encoder he used with his first TVT, and that’s what we’re going to build.

Since I have no experience building electronics from the ground up, I figured the encoder was a good starter project.  The TVT itself involves much more complicated circuitry and a power supply, which is kind of intimidating.  I need to get comfortable and learn on something I can afford to accidentally destroy, with minimal risk of death.  This I can build and hook up to a bench power supply and hopefully just test with a logic probe.  Plus, it’s going to be really cool resurrecting a project that was conceived 2 years before I joined this world.

Essentially the goal with an ASCII encoder is to be able to press a key and have it generate a 7 or 8 bit code that corresponds to the letter, number or function assigned under the ASCII standard.  A code table is in the bottom left of this page from the article:

Your keyboard keys are arranged in a matrix.  When you press a given key, it causes certain logic gates on the encoder to operate and enables or disables certain bits, from 1 to 7.  If for example we press the P key, the encoder will cause bits 5 and 7 to go high (a logical 1) while all others remain low (a logical 0).  The resulting code – 101-0000 is passed from the encoder to whatever device you have connected, and based on the ASCII standard that device knows to generate a P.

In a modern keyboard, all of this work is typically done by a single IC.  But in 1973 such ICs weren’t widely available, so Don’s design was what he called ‘brute force’, using fairly basic logic and 3 Motorola hex inverter ICs to set the individual bits.  The encoder required a 10V, 25ma power supply but this could be modified to 12 or even 5 volts by changing one of the diodes.

You’ll recall I etched this board already as part of my first run of TVT boards.  To do this I simply lifted the artwork from a PDF of the article, flipped it, cleaned it up, printed it on magazine paper and then ironed it onto the board.  My vinegar/peroxide etchant made quick work of this little board.

Apparently a few of these were available commercially, at least according to the article.  It doesn’t look like they had any silkscreening for parts placement, but I decided to keep it easy by printing off the parts placement and transferring it to the non-copper side of the board once it was etched.

Assembling the required parts was not hard.  In fact, I just did a few ebay purchases and within a couple of weeks had all the resistors and bits I needed.  I even found NOS 1972 vintage Motorola MC9818P ICs!

That said, once I had it all in front of me I realized I wasn’t too happy with the resistors and electrolytic capacitors on offer.  They look too modern!  And small!  The 100uf 10v caps on Don’s original were fairly big metal can type caps.  Today’s equivalents are about 1/2 to 1/3 the size.  I installed them anyway for now — they’ll work for testing purposes.   The old metal can style seem to be unavailable anywhere at the moment.  You can see in the upper right corner just how small it is compared to the footprint the ‘silkscreen’ expects it to be.

I also used one of my modern, Chinese made, carbon film resistors.  I didn’t like it.  It looks too modern.  I think I’ll do some more research and see if I can hunt down some ‘correct’ vintage resistors.

Getting those first few parts installed though really changes things.  This is going from a dusty historical footnote to a living, breathing device!

Having totally missed out on an original set of TV Typewriter boards months ago, I had to proceed with Plan A, which was to make my own.  And that was fine — truth be told, I was kind of looking forward to it.  I like making stuff, so this was right up my alley.  And plus, now I at least knew what the originals looked like.  The prototype presumably would lack some features but probably otherwise be the same.

The ‘stock’ TV Typewriter consists of four boards: the mainboard (called ‘mainframe’, in the parlance of the day), timing, cursor and memory.  The stackable, modular nature of the design meant other features could be added, and in fact in future articles new features would be added, including additional memory and even a UART board to turn the TVT into a (somewhat) useful serial terminal.

The first difficult bit of business was getting the right PCB stock.  I really wanted the thing to have a vintage look so I could show it off to people.  Most vintage boards have this green tinge to them (they call it ‘natural’).   But the modern PCB stock on offer today is typically brown or other colors that were not as common or available at all back then.  Searching through vendors, most of them for some strange reason wouldn’t show the backsides of their boards or even note the PCB color.  I guess to modern PCB fabbers that doesn’t matter.

Now, I did find a site that offered PCBs that looked pretty damn close to the original ‘natural’ color I wanted:

http://www.americanmicroinc.com/g10-fr4/

So I got in touch.  They were very friendly and responsive, and offered to send a sample sheet.  I wanted to produce enough boards to build two TV Typewriters, one a replica of the prototype and the other a replica of the Radio Electronics cover unit.  They informed me that to make enough single sided copper clad and cut to size would cost $400USD.  Yike.  But, if that got me the result I wanted, money was no object.  I had tried a few shots in the dark on ebay but the PCBs that arrived (2 months later) from Hong Kong were brown.

When the sample sheet arrived it was double sided, and since I lacked etchant at that point I had no way to tell what color the PCB underneath was.  Meanwhile, I found one of the rare ebay PCB vendors willing to offer info on color:

http://www.ebay.com/itm/150846166149

Again part of the problem here is we’re dealing with photographs.  From my experience in railroad modeling, I knew how much a photograph could distort things.  Matching color off a photograph?  Yikes.  But, I did have vintage boards from my computer collection from the same era, so I could check against those.   Regardless of what I bought, it was going to be a crapshoot.  At least in this case, it wasn’t going to be $400.

Now to order some etchant and wait another 2 months for these boards to arrive (everything from China takes forever, it seems).

Oh man.. that was close.

I had been warned in embarking on this project that I would likely have a *very* difficult time finding certain parts, especially in the correct vintage that I wanted (1972-1973).  As it turns out, I had gotten a little smug and thought I had it all when in fact, I was missing the one thing there was no substitute for — the shift registers.

One of the things that set the original TV Typewriter apart from later variants was its use of shift register memory.   To be frank, I’m not an expert on these things and don’t know the exact differences between shift register memory and the kind of RAM we use now, but suffice to say shift registers went back quite a ways before the TV Typewriter and were already well on their way to being obsolete for TVT purposes when Don’s project came out.  This is one of the key reasons why the TV Typewriter II project came out shortly afterwards – easier to acquire parts and far better 2102 static RAMs.   In the forums discussing my project with others, I was warned that these had been hard to acquire even in 1973, and now would be almost, if not outright, impossible.

Undeterred and being a newbie to the world of vintage ICs, I looked at the parts list for the TVT and discovered I needed ‘2524 shift registers – Signetics’.  Okay.   I tossed that into Ebay’s search and came up with these:

Hah!  I thought — this is easy!  The dates were newer than I wanted, but I had them, right?  Perfect!  This is going to be a snap!  Sure, I won’t have the date codes all correct, but I’ll have what I need to get it up and running anyway and can look for closer parts matches later.

Anyway, I kept on acquiring chips here and there assuming my most difficult find was over.  But one day on the forums I made a horrifying discovery — that little G after the S in SG2524 actually meant something.  A poster had been incredulous that I’d found the elusive shift registers, so to prove it I posted a photo.  Turns out, despite the similar part number these were entirely different chips!  And looking at the artwork for the Memory board confirmed it — the shift registers I needed were 8 pin ICs.. these were 16!  Augh!  Turns out, these were ‘signal modulators’.  I needed shift registers, and the part number for those was 2524V.

A desperate search commenced.  Months went by with.. nothing.  Every part site I searched said ‘Nope’.  Nothing on ebay.  Nothing anywhere.  I was apparently looking for crazy rare unobtanium.  I was despondent.

Then one day, I got lucky.  I found a site that said yes, we have them!  6500 of them, actually!  2524V, 8 pin.  Yes!

Sort of.

I don’t know if you’ve ever experienced these ‘surplus parts’ websites.  They’re a bit like those ads you see in the paper offering you business opportunities.  Make $500 a day like me !  Write for details!. Only the thing they are promising is never quite delivered.  Unlike the guy that promises you a life of leisure, many of them don’t even respond to the requests you make of them for quotes.   I’d stumbled onto a few sites that purported to be able to locate 2524V chips.  They’re all the same — they have a ‘Request Quote’ button you press, and then you put in your particulars and what you’re looking for.  And yeah, it’s just goes into a void.  Once in a while one of these sites will be bothered enough to respond — usually just a trite four or five word message: Nope.. don’t have it, sorry.

So I was suspicious.  Knowing how rare these ICs are, I really doubted anyone had 65 of them lying around, let alone 6500.  But like I said, I was lucky.  This one not only responded but in the affirmative: yes, we have them.  I checked part numbers twice with them, confirmed the pins, every detail matched.  The only negative was they were 1976 rather than 1972-1973.  But I was okay with that.  Something was better than nothing, and mid-70s was definitely better than 80s.  I ordered 20; to have enough to build two TVTs (each needs 6) and have spares.

2 weeks and $100US later a dimunitive little package arrived in my post box.  And there they were!

I couldn’t believe it!  I had the unobtanium in my hands!  I wanted to shout it from the rooftops, but they probably would have just locked me up.  That’s the problem with this hobby: not enough of us around to appreciate finds like this.   Hot damn!

Anyway, the project is a go!  And in between waiting for the little 8 pin angels to arrive, I found steadily more and more (correct) ICs online, using what I’d learned about triple checking part numbers.

With ICs almost all in hand, it was time to turn towards the most important part of the whole business — making some boards!