The RCA TV/monitor is a hot chassis design, so to electrically isolate it from the computer, Radio Shack added an opto-isolator to its video input, powered by 5 volts DC coming through the TRS-80's DIN video cable. So to use it with another video source, you'd need to connect it to an external 5 VDC power supply.
Yikes, I knew of the adapter board, but not why, and the fact the monitor was live chassis (I had a later monitor, not this one). I hate live chassis, they are out to literally kill you - especially where I live with 240V power.
Adrian... regarding that "odd" transistor configuration fed from the LM723 linear regulator chip... it's a Sziklai Pair Darlington configuration. Just like a regular Darlington Pair, a Sziklai Pair provides much higher gain than any single bipolar transistor, but with the added benefit of being able to pull the output voltage within about 0.6V of the input supply rather than being subject to the 1.2V double emitter-base drop in a regular Darlington pair. What you are looking at is basically a discrete LDO (low-dropout) regulator circuit. The 68 -ohm resistor from emitter to base on the power transistor improves the way the collector of the smaller transistor controls the base drive of the power one and allows it to turn off quicker. It also ensures that if the 723 isn't driving the Sziklai Pair then the base of the power transistor is hard clamped to its emitter, ensuring it is fully off. In operation you can think of the two transistors in that circuit as a single very high gain NPN pass transistor. In addition to linear power supplies, the Sziklai Pair is also often seen in Class B power amplifiers since it allows you to use two identical (NPN) power transistors on the output driven on the low side by a lower-power NPN and on the high side by a PNP. This overcomes the problem of finding well matched NPN-PNP high-power complementary pairs. This method in commonly known as a Quasi-complementary output stage. en.wikipedia.org/wiki/Sziklai_pair
I had this computer in 1979 that's why I'm watching. But OMG how does anyone know about all this stuff. I'm impressed. I didn't have the expansion and saved everything on a radio shack tape recorder. There was no software so you had to enter everything yourself. A fantastic magazine called something like TRS80 Micro came packed with pages of basic stuff. You'd type it in and spend the night troubleshooting your typing errors. That's how you learned basic, by troubleshooting a program you typed in. Wild west days for sure. Lotsa fun.
If you dig out the TRS8BIT newsletter from September 2017 and Christmas 2017 there's a good rundown of the lowercase mods and the different character generators. Ian Mavric sells (via ebay) a pre-built "Gendon3" character generator based on an EPROM which if you make a bunch of trace cuts gives you 3-line descenders. It should be possible to build a similar PCB adapter to replace the Radio Shack/Motorola character generator ROMs. i'm sure I've seen such a thing done in the past but can't find a link. I'm glad my own TRS-80 Model 1 was from May 1980 and they had integrated a lot of fixes. Still had to mod it for lowercase but it only needed a trace cut and double stacking the RAM chip but everything else just worked... Thanks for the TRS-80 videos, always nice to see.
Ian Maverick is THE MAN when it comes to TRS-80 mods and repair. Bought quite a lot from him in my process including new power switches, character generators and similar. The one I have going now had a strange custom character generator EPROM from the 80s that is very similar to the one Ian offers with descenders but had a few glyphs/symbols changed out.
I wish those TRS8BIT magazines were more searchable - as a newbie to the TRS-80, finding troubleshooting info is very difficult without hounding the right people who know what random tome the answers are in. TRS8BIT really needs to be converted to HTML.
@@SockyNoob the issues are all downloadable to be sure, but only in PDFs by year, and with no searchable table of contents. Info on the TRS-80, to this newbie, seems to be indexed very poorly.
I was wondering about this - I would assume that "character generator" is nothing but a ROM... Maybe it's a fast rom, but it sure seemed to me there should be a way to put a modern EEPROM on a board and build new ones!
It's so cool to see one of these running. That was the first computer I ever used, back in 1980. Taught myself programming on it when I was 15. Two years later, started my IT career which lasted many decades. And it all started with one of these.
It was also the first computer I used in 1980, but I was 10, it was in my 5th grade classroom. It's what I first learned BASIC on. Now I can buy something the size of a stick of gum 100x as powerful for $5 How times have changed. My career? well, using computers is my life. I'm just not good at getting hired. So much time in front of the computer, I failed to develop the social skills necessary to actually obtain employment. Now, whole new generations have taken over and I don't even bother against the new competition. Kinda failed in life. Now I have carpal and back problems from being in front of a computer so long and can't do a thing. And it all started with one of these.
Most of the DIY lowercase mod instructions specifically mention the issue with Level II BASIC as one of the primary reasons for the switch. The official lowercase kit that Tandy/Radio Shack sold included the character generator that allowed incompatible software to work correctly. There was no way to buy the character generator separately, so the DIYers were stuck with a switch. The issue is still fundamentally with the software (like L2 BASIC) that incorrectly use ASCII code 0-31 instead of the proper 64-95. The replacement character generator was just a hack/workaround for that buggy software, by echoing characters 64-95 to 0-31 (at the cost of the 32 characters that were previously at 0-31).
" The official lowercase kit that Tandy/Radio Shack sold included the character generator that allowed incompatible software to work correctly. " Back in the day I worked tech support for Packard Bell, later Reveal and Newcom, the latter 2 sold peripherals. 90% of the time if something didn't work it was a Tandy, and there was going to be no getting it to work. Those things were terrible.
The later TRS-80s which were produced actually came with the newer character generator already installed, which is probably why most of the docs, etc, that Adrian found didn't mention it. It was only the earlier-produced Model Is which actually had this problem and required swapping out the character generator. I remember at the time we had two TRS-80s in the house, and did the 8-bit conversion on both of them. One (mine) worked fine with no changes required to the character generator. The other (my father's) required fixing the character generator as well (luckily, we were prepared for the eventuality and as I recall we managed to find a third party who was selling "clones" of the newer CG IC (they're really just a PROM chip) which we were able to use). Also, the reason the old character generator also had such funky lower-case characters was because at least some of them didn't contain enough memory to actually encode the full character space, so it couldn't do descenders. This also meant that it didn't encode the lower portion of the graphic characters either, so if you tried to run anything that used the TRS-80's "pixel" graphics (which was the whole reason for the 8-bit mod), you would get blank horizontal bands in the display, so you really wanted the newer character generator chip anyway. (Interestingly, the version shown here does appear to encode the full graphic characters, but without descenders, which seems odd..) As I remember, by the time they came out with the Model III, they had (finally!) fixed the BASIC ROMs to use the correct character set, so the Model III character generators actually had the original special characters in the lower set, and could display some characters that the Model I couldn't as a result (I have vague recollections of a friend of mine putting a Model III ROM and character generator into his Model I so he could make it display everything the Model IIIs could, but I don't remember if there were other complications as a result)...
Loud applause for sticking with it, Adrian! Delighted that you got the Model 1 working. I've really enjoyed your TRS-80 series, especially the Model II, which is my favorite. Keep it up!
I can appreciate how frustrating a video like this must be to make, but make no mistake, to me as a viewer who knows more or less nothing about these machines, it is anything but frustrating to watch. It's absolutely fascinating, and I'm sure that if the fix had been more straight forward of you by chance found the problem at once, the videos would have been much less instructive. Even if I don't know the particular platform you're working on, I can appreciate good troubleshooting and learn something from it. Thank you for sharing your struggles instead of just cutting to the chase!
Could not resist a good Model 1 "Trash 80" story here. Rewind back to Christmas of 1977 and the Model 1 was feverishly being pushed out for Christmas. I had 2 children, one 11 and the younger 9. Grammy gave us $350 to buy a color TV for our deprived children who watched Yogi Bear on an old B&W TV. I was a hardware tech for AT&T maintaining PDP-11s for AT&T and wanted to get them something more meaningful so I ponied up another $350 and bought a TRS-80 Model 1. I have no idea of what the present value of that $700 is currently. Christmas morning came and the children opened up the boxes, loaded up a Pong-like game from the cassette recorder (after 3 or more tries). Much amazement followed, but soon the fun was interrupted when the computer got too hot and failed. They played into the night alternately putting the computer in the freezer and reloading the game until the next failure. Well, the next morning I pulled out the circuit board and was horrified to see virtually all of the wave connections were cold soldered with solder balls everywhere. Took me several hours to cleanup and resolder all of the connections. Worked like a charm afterwards. My 11 year old learned Basic and wrote a Yahtzee game. Looking for something to challenge him, I purchased a machine language compiler (still on cassette) and turned him loose on writing several little programs to generate Morse Code and such. He then rewrote the Yahtzee program in Z80 machine code. It was blazingly fast. He went to MIT on a work/study masters scholarship in Computer Science from Texas Instruments and still works in the industry. That old TRS-80 taught him perseverance and rewarded him with a rewarding lifelong career.
This is a great story. My family also bought me a Model I in about 1978 and I learned everything I could about it. It also stimulated my love of computers and inspired me to study computer science and eventually do a PhD, which has kept me gainfully employed since then. I'm very grateful to my parents for giving into my badgering to buy me the computer as I'm sure your son is to you too.
the way those 2 transistors are connected is known as a sziklai pair, gives high gain like darlington but uses 2 different polarity transistors and gives lower base drive to output voltage drop
Live this video, one of my faves; the reason being that this is the real world for most of us doing repairs... twists and turns and always short of one part. Many thanks.
But what is awesome is to watch you work the problem. And take me back down memory lane. So sorry I couldn't come up to Elmhurst this weekend (I'm only a couple hours away ). Hopefully next year!
The collector connection on the PNP power transistor was definitely the fault causing the smaller transistor to get hot and the large one to stay cold. It's fairly easy to explain why this is the case. The configuration of those two transistors with one PNP and the other NPN is called a "compound Darlington" transistor, this has some advantages over a regular Darlington configuration with both NPN, including reduced "dropout voltage" or minimum input to output differential, which is useful in a voltage regulator. With the collector disconnected nearly all the current will flow through the emitter to base junction of the PNP transistor (acting as a diode) and only a very small amount through the resistor. The full combined current will then flow through the collector to emitter junction of the NPN transistor, which is doing all the work, this is why it gets so hot. Even though the PNP transistor is still carrying the full current through its emitter to base junction the voltage drop is only 0.6 volts so the power dissipation is 8.6 times less than it would normally be when most of the current is going through its emitter and collector with a 5.2 volt drop, since power dissipation is voltage x current. The demotion of the PNP transistor to a diode also means you lose the gain of the PNP transistor, this reduced gain depending on the gain in the IC will cause the voltage regulation to be worse so it will tend to sag more under load. When you reconnect the collector the current through the collector/emitter of the NPN transistor will drop by the ratio of the hFE (gain) of the PNP transistor - data sheets suggest 15 to 200 for hFE. This massive drop in current is why the small NPN transistor no longer gets hot. How good is the emitter/base junction of a power transistor as a diode when the collector is left flailing in the wind ? Pretty good actually. IE (maximum emitter current, eg base + collector) of that transistor is 17 amps, but IB maximum base current is still 5 amps. So the PNP transistor would survive this fault just fine if the total current drawn was less than 5 amps, and because the voltage drop is only 0.6 volts it would be a long way below its maximum power dissipation! If there is enough gain in the IC to compensate for the loss of gain of the PNP transistor (or the following circuitry isn't too fussy about the voltage regulation being a bit worse than normal) the only real symptom of this fault would be that small transistor getting hot but apparently surviving. It could have been working like this for years! The phenomenon of an oxidised connection that measures OK with a multimeter but goes more or less open circuit or very high resistance as soon as you try to pass any real current through it is real and actually happens a lot. I see you got caught out with that on the ribbon cable on one of the previous TRS 80 videos as well. While it tests good with a meter there is in reality only the tiniest whisker of a microscopic connection in amongst all the oxidation, as soon as any current flows this microscopic contact point disintegrates and the connection is lost. Modern digital multimeters only pass microamps in ohms mode and even in diode test mode only a few milliamps, this can give very misleading results in situations where there is severe corrosion or there is only a point/whisker connection occurring due to a mechanical problem like a loose screw. You had both corrosion and a loose screw here...
@Arnold Rimmer not quite. The IC is doing the regulating. At no time was the base of the NPN handling much current, it was all going through emitter-collector. The NPN is more like a pre-amp for the output transistor.
@Arnold Rimmer it's no different to using a Darlington transistor. It would not amplify the ripple current as the voltage is sensed after the pass transistors. The transistor arrangement shown is a Sziklai pair, which basically makes a very high gain NPN transistor, without the double emitter drop of a regular Darlington. It is the same arrangement as used in modern LDO linear regulators
I so very much enjoyed this. My high-school buddy got one of these for his birthday in 1977. It seemed to us like a mainframe, which was probably down to the color scheme and serious-looking text display and professional-looking blocky graphics. It looked like it belonged in a glass room. "No kids stuff here, boys, this is a REAL computer." AND, it was programmable. My buddy programmed some of the rules to our table-top Avalon Hill war games so that it handled all die rolls and turn management. It was like Billion Dollar Brain only at this guy's house from school. Boy, did I ever want one but $600 seemed as far away from me as the Andromeda Galaxy. Wouldn't you know, after college I actually did end up programming mainframes (real ones) -- a job which I still do to this day. Watching this repair series completed a beautiful circle.
I love it when I see other folk have exactly the same frustrations I do sometimes, nice to know I'm not alone when poring over a schematic trying to work out what might be causing problems a, b and c. As for penny pinching Rat Shack just imagine what the PET may have looked like if Tramiel had let them have their way instead of rejecting them and allowing Chuck Peddle to design it instead. Excellent stuff as always!
The character set issue is why many 3rd party mods had a switch to disable the 8th bit, or else some other special circuitry to make the video memory not 8-bit clean, and why the tandy mod had a replacement char generator. Without that, you'd have the garbage until you loaded the lowercase driver. The ROMs (all versions I ever saw) for some reason cleared the bit for alpha chars resulting in the wrong char set. I patched that out (decades ago!) and programmed my own EPROMs. Worked great.
One wonders why in the world they did that. Was there some kind of miscommunication with Microsoft which ended up with MS coding the output routines as if the alphanumerics were literally 6 bit with the letters shifted down to 0-31 instead of there being that fake-bit-6 circuitry? (There is an official 6 bit ASCII subset with that layout... I think it's used for some naval/marine purpose, not usually computers, but it is out there.)
As others have noted, there is a fuse inside the power block. As an ex-TRS-80 tech, I can vouch for the fact that that fuse can pop pretty easily. I remember the Tandy TS bulletin that described how to carefully crack them open and replace the fuses. (Towel, vice, etc...) It will blow! (I ran my own computer repair shop thru the '80s mostly working on TRS and Atari computers, as well as a few S-100 and industrial/CNC devices...) Fun to watch these vids! Keep on hackin'! Stu
Adrian. Thanks for the video. Takes me back to my teenager days when Radio Shack was tops in my book. I Owned a TRS80 Model III which was my first computer. Then I owned a Model 1000 which I Played the heck out of. Thanks for showing all your mistakes. I like to see those. even though this was really furstrting for you, it really was a teaching moment when dealing with the EProms.
Yeah your frustration is evident… But as I've said before it's often the failures that teach us the most… This showed a lot of excellent best practices… Thanks for sharing this!
RS used at least three different character generators from Motorola. The third one also had the raised lowercase and funny characters, but the lowercase 'a' was shifted up two lines. Many years ago I made a crazy outboard circuit to shift the specific lowercase letters down by two lines. (it used a 7483 adder which ran warm all the time!) The video driver code for some reason went out of its way to force the uppercase to the 00-1F range, and I'm surprised that wasn't fixed in the 1.3 ROM. I still think the patch I suggested in the last video comments will probably work with the funny chargen: 0471: FE403808 D640FE20 -> 00000000 0000FE60 And I like that external floppy drive case you rigged up. An IMPORTANT WARNING: do not turn the computer off with a disk in the drive and the door closed! If you see the drive select flash, it probably also pushed a glitch through the drive heads, zapping whatever sector is under the head. I lost a few sectors of data that way back in the day.
I was in the first computer lab in my school district in 79/80. We had a row of Wang programmable (model 452 I think), access to the county mainframe to run Fortran and Cobol code and a TRS-80 with 8kb of memory and a cassette backup to run our MS-Basic work. Later my parents got a TRS-80 Model 16 for their business that I would run experimental programs on when they weren't using it. I learned to debug and simplify my code on the Model 16 when I wrote a graphical Tie-Fighter shooter game that went from about 250 lines of code to its final version at 95 lines. Nice to see someone keeping the old tech alive. Thanks.
Not frustrating, deeply satisfying. If it was easy, anybody could do it! The sense of accomplishment when you plow through multiple problems cannot be beat.
My first computer was a TRS80 Model 1 with the expansion pack you showed, disk drive and matching printer. I remember the frustrating problem my Model 1 eventually developed. The connection between the computer and the expansion pack went flaky which caused random, spontaneous reboots - usually right in the middle of programming. Nevertheless, the Model 1 set me off on an IT career that's still going 34 years later.
Good job Adrian. In the future, many of these early computer models will be rare or non-operational. You'll be one of the monks that has any knowledge them.
No matter how good you are, there are always difficult problems to solve. I like watching this, too, because I think about how many places someone can inject doubt into the process. It's a problem with troubleshooting, and doing it for other people. When you're doing it for yourself, you know where you're at. When you're doing it for someone else, doubt can be injected into the process at different points to try to steal a customer, or steal credibility.
This is the computer I learned on. I ran newdos80 ver 2. Radio shack double density kit installed. All was ok when I ran trsdos 2.7. Newdos sometimes would skip writing a sector. I applied all the patches so newdos would work. I used my sytem at work and it became a big problem in dbl density. I called apparant. Spent 2 hours on the phone. They told me that it was in the write driver and they will fix it. A month later got a package in the mail. A new copy came. It didn’t fix it. I ended changing the 1793 and the problem was still there. Took the expansion interface to a computer shop. The put ferric cores around my cables and that fixed it. But it drove me crazy.
My very first computer was a TRS-80 Model 1 with cassette input got it I think in 1976. The manager a Radio Shack would routinely give me copies of programs. My friend Mark at the same time had a Commodore Pet. When I saw his graphics I was jealous. From there I moved up to an S-100 bus CPM MACHINE. Followed by a Commodore Vic-20, major upgrade to a C-64. Loved that machine. Had my first printer & modem. Actually used local C-64 based BBS. Then a TeleVideo Luggable KayPro Clone with CPM DUAL 5.25” Disk Drives, an external Modem & Daisy Wheel Printer! Then I went to work for a small engineering company. Had my first PC DOS Clone with an NEC Z-80 chip and an actual 20 mb hard disk, green mono monitor. Had a hardware based RTTY and CW modem for my ham rig! Oh the early glory days. At that same company we used the Tandy 40 character LCD kind of early laptops. Used it with a scanner wand to read barcode.
I must have got lucky, I never had any issues with my TRS-80. It work every time I hooked it up as time went by. Drug it to school & back, Electronics school, no issues.
Adrian, these videos are good to watch and follow a long with you on your diagnosis/debugging. You are too hard on yourself about them. Don’t be afraid to publish these!
You can program 21V or 25V Eproms with the TL866-II! Al you need is to isolate/ cut the Programpin between Eprom and the TL866-II and use a Benchpowersupply and inject the correct Volatege external to the Eprom. You need to dissable the Pincheck in the TL866 Software and can program with the standart Epromprofiles of with Programmer.
Regarding Adrian's concern about a lack of any fuse protecting the analog inputs to the voltage regulator, I think there actually is a protection fuse buried inside the black radio shack power brick. Some other videos show people cracking open the black plastic enclosure to do repairs and replacements on the voltage step-down transformers, which might also reveal a fuse buried inside too.
Hi Adrian. At 4:36, you don’t need to lift the resistor; just set the DMM for regular resistance, then measure the B-E junction. If you get 68 ohms, the transistor is either open, or has a high resistance in that direction (switch the leads, and try again.) If you get 0.1 ohms or less, then the B-E junction is shorted as you already tested the 68 ohm resistor (you DID test the resistor, right?), and if you get something like 40-65 ohms or so, then the B-E junction is probably good, as resistors in parallel divide (the 68 ohm resistor, and the (effective) resistance of the B-E junction as seen by the DMM.) HTH! 😌
Regular resistance setting on a DMM wouldn't be adequate as they typically don't provide sufficient voltage/current to turn on a diode. You'd likely read the transistor junction as open when it is really OK. You would likely detect a short though, but then that's what the low ohm resistor was effectively giving as a false reading.
Fun stuff! I was able to find some info on that character generator. The Sams ComputerFacts list it with both 8046670 and MCM6670. Using the MCM number it wasn't hard to turn up a datasheet. There were a couple versions of that chip - the 6670 was mask-programmed and the 6674 which came preprogrammed. The datasheet includes the font in the preprogrammed version and guess what - it's just what you were seeing with the different control chars and non-descender lowercase characters. So looks like RS started off with the off-the-shelf version and later around to programming their own custom layout. That datasheet is interesting as it includes info on ordering your custom layout with the proper punchcard or paper tape formats.
I like your EMS t-shirt. Back in the day I always used EMS floppies because I never had problems with them, and the labels looked cool! I seem to recall they weren't all that expensive either.
You should get a thermal camera. There are some FLIR ones for just a couple/few hundred $. Can record the video too. Can easily look at the whole board w/o having to touch.
You can even get them in a phone. We use a Blackview BV6600 Pro at work that has one built in, runs Android, and has an 8000-odd mAh battery. Makes diagnosis a hell of a lot easier!
these vids where things go -way wonky, actually make me feel better! i’m pretty good at problem solving- and to see you having issues… makes me think… if these projects were easy… then they’d be no fun! i love learning, and failures are how we learn! thank you for these difficult projects!
This really brings back memories, I was in my highschool's first ever computer class. We have 4 of the TRS-80 complete with cassette recorders in 1979-1980. The expansion block was HUGE when it came out, and my senior year the principal popped for a hard drive!
Thanks for bringing back some early geeky memories. I worked at Radio Shack when I was 15 and got a Model 1, then a Model 3. Wrote Z80 assembler to copy system floppies without knowing what I was doing! It started my career in computer science, BSc and MSc, 22 years at IBM.
19:00 I would have tried replacing it with a 74HC chip, before building an adapter for a HEF/4000 series IC. The reason for using 74C (or 4000) was often either the high input impedance or the symmetric voltage swing. The more modern HC variant got that as well (like most CMOS) but with the same pinout as all other 74-series ICs.
@@herrbonk3635 Yeah, they specifically use CMOS in that part of the design because they want both the reliably close-to-the-rail outputs *and* the higher "high" input threshold. The system that's used to adjust the horizontal and vertical position is basically analog; there's a trimmer resistor and a capacitor, and the rate at which the capacitor charges enough to activate the input of the next CMOS stage is what sets the delay. Using a TTL compatible HCT instead of an HC would change the time constant they designed for. The only concern I see with HC is it is a lot *faster* than C in gate propagation delay, but I don't think that'd matter much here; for horizontal hold you're trimming within a few microseconds, losing a few hundred nanoseconds of gate delay should be easy enough to compensate for.
Proper descenders were the later character generator ROM. The moved all the character up to give space for the descenders. The model 1 uses a lot of memory mapped IO (except for port #$FF), but the model 3 moved a lot of the IO to ports so its not just the keyboard that changed.
Also WRT to the ROMs. You could just replace the entire level 2 board with say a single 27C128 and a bit of select logic and put all the ROM images in one part.
There's a great replacement power supply PCB which uses modern transformers and sits inside the expansion chassis. Ian Mavric (at least used to) sell kits for them. He also sells/sold lower case mod kits including replacement character generators. You should look him up and maybe check out all of the stuff he's done for the TRS-80 over the years.
I had a Model 1 way back in the early 80's I had a Lobo LX-80 expansion interface for it but never had floppy drives. The Lobo interface was originally made for their model 1 compatible computer but worked fine on original Model 1.
Great video Adrian. Like a lot of others, I did the lowercase mod to my Model I back in the day and I also remember changing the character generator to get descenders. I had forgotten about the problem with booting into Level II but it's coming back to me now. I still have the machine and documents in storage and you've inspired me to dig it out and try and work out exactly what mod it was that I installed.
Hey, I owned one of them and I loved it. Of course, typing in the instructions for your program every time was a tad of work but fun to as it helped you learn about programming and how the computer actually functioned. There was only basic on this one and it was built in. It cost right at $500 too and was a lot back then.
It has been so nostalgic to see you repair a TRS-80 model I, and it makes me want to go into my basement and dig mine out to see if I can get it running. I bought mine on May 26, 1979. It got lightning-fried about a year later and just about everything got fried...a neat trick considering it was unplugged at the time. So I don't know what date the board is, but it was repaired...the tech said "every circuit was fried". It doesn't have the lower case mod, nor does it have a keypad, but I have an E/I and a floppy drive (I thought I had two, but I think that's just my 30+ year old memory). The last time I did, I got that garbage screen...I forgot that it was normal with an E/I attached, and forgot about the break key thing. Now I REALLY want to try it. The only real "mod" I ever did on it was replaciing hte short keyboard cable with a 6 ft. ribbon cable so I could put the keyboard on my lap while the computer was on my desk (bare board all hooked up). Thanks again for the memories!
It's fun to see some frustration, Adrian - it's a more realistic view into the process! Given the frustration, at the end of the video when you said "if you didn't like it, you know what to do" - it made me realize for the first time another meaning for that phrase. Made me laugh out loud! 😂
If you did encouter an issue with a 74Cxx series chip, you might try substituting a 74HCxx or 74HCTxx, which are much more common series today. Most likely the 74HCxx would be the closest approximation. The 74HCT series is supposed to have TTL-compatible input levels.
I started following you for the lightbulb videos. The old machine repairs are fun but I'd be happy if you branched out too. I could listen to you talk about anything.
I have vague memories of doing a lower case mod (ok, really my high school comp sci teacher did the mod) where I got to keep the alternate characters and had to live with the non descender lower case characters. I could use them in my own programming by poking them into video memory as I attempted to write games. My model I was one I bought as Radio Shack was clearing them out as a 4k machine and we later upgraded to 16k with aftermarket RAM. It might have even been a level I machine which we upgraded to Level II at some point. I do remember having to load a key debounce routine for a period of time (not sure how that worked). I may also have had to load a patch for the lower case to work properly with that alternate character set ROM or maybe we got a patched Level II ROM set. It was a long time ago and a lot of that stuff was pushed out of my memory but maybe my minimal recollections will jog other people's memory on such things. I believe my teacher had opted for the true descender character ROM when he did his own Model I so the different lower case mods was definitely a thing.
Adrian, I had a TRS-80 Mod 1 when they were new. So it was fun to watch you to bring yours back to life. Though I don't have the machine anymore, I did find some original software documentation, and disks in a storage locker. I have TRSDOS, Scripsit word processor, UOLISP Lisp interpreter. As I continue to unpack, I may find a Pascal compiler with IDE, and maybe an APL interpreter. If you are interested in them, I can send them to you for free. They might help you create a software collection around the machine you rehabilitated. Let me know if you want them in a reply. I think it is great that you are reconstructing history of early PC's by rehabbing machines of that era.
What memories this video brought back! I had a Model I 16k that I modded with an LNW expansion interface. I don’t remember if I did the lower-case mod in it - I remember all the issues and I think you were limited to a 5x8 matrix for the character generator which didn’t allow descenders. I’m pretty sure I waited until I got my Model III which had 8 bit and a 5x9 matrix CG. But I do remember being envious of the IBM PC’s serif typeface for the screen font, so I set out to build a mod for the CG. It was really just a PROM that I replaced with an EPROM which required an 18 pin to 24 pin adapter I cobbled up, so I was able to have a custom character generator. I wrote a little editor in BASIC that let you design your own characters; outputting a binary file to load into the EPROM. I thought this might be a good magazine article, so I wrote one up and sent it to 80 Micro, the premiere TRS-80 magazine at the time; but they rejected it saying someone else had beat me to it. I submitted it to another startup magazine who accepted it, sent me a $25 binder; and then promptly went out of business. I don’t think I ever saw any article about custom CG’s; but by then I had moved on to building a Taiwanese PC-XT clone from a bare PC board. I jumped ship from Tandy to IBM and never really looked back.
Apologies if this has been said in the other 300+ comments - Ian Mavric may have original character generator chips that are the “new” character types. I had a faulty one of mine replaced a few years ago.
A few thoughts on the regulator... 1. Thermal paste for power transistors often needs to be explicitly nonconductive, which is not necessarily the case for the kinds used for PC hardware, where voltages are generally minuscule but heat transfer is a major concern. Thermal conductivity tends to be tied to electrical conductivity, so it is not unusual to see metals being added to enhance performance (e.g. the classic Arctic Silver 5, or even going liquid metal). For power supplies and audio amps, you want the plain white silicone goop. 2. The regulator pass transistor topology is a CFP or Sziklai pair. It's got a lower voltage drop than a Darlington and actually sports inherent feedback. Now being a loop of two common emitter circuits, it isn't exactly the fastest thing, but that isn't too much of an issue if there's a healthy amount of current flowing, as in a regulator pass transistor or a Class A amplifier output.
Also, Noctua NT-H1 has an operating temperature range of only -50 to 110°C, which is fine for a CPU but perhaps a bit low for a power transistor. (NT-H2 goes up to 200°C, in case I didn't read the part number on the video correctly, and would go well with the absolute maximum junction temperature of 200°C for a 2N6594.) Both NT-H1 and NT-H2 thermal paste compounds are not electrically conductive. To be clear, perhaps you could mention that it needs to be zinc oxide thermal paste (which could have a silicone carrier) rather than just any plain white silicone goop.
First thing to debug electronics is always check power supply. And a multimeter is never enough to check a computer voltage rail. ⚡️give it to an oscilloscope to check for glitch. Even a switching power supply can show an ok voltage on a multimeter. But under a scope you will see the spikes that can create havoc on a logic machine.
The regulator circuit would send all the current through the little transistor if the collector on the big one got disconnected. The big one would act as just a diode, bypassing the resistor and feeding current straight to the little one, which would then drop all the voltage itself.
The 723 regulator IC was quite commonly used in the 70s and early 80s, and available from many sources - MC1723 from Motorola, uA723 from Unitrode/TI/Fairchild/NEC, and LM723 from National Semicondutor. It's still made today and a lot of small cheap bench power supply projects use it
The MCM6670 character generator is unobtanium but it is just a ROM. 7 bit ASCII + 3 bits for the row, and 5 bits output. A 1K or larger EPROM can be substituted as long as you have a socket adapter.
Boy does this video take me back.. I bought the first TRS80 Model 1 Level 1 4K sold at the Yuma AZ store for $795. I was in the Army, stationed at the Army Yuma Proving Grounds at the time, and had everybody in the barracks looking over my shoulder while I played with my new toy.
We had a model1 48k, a bunch of mods, and 2 floppy drives. We ran trs-dos and new-dos. My dad did something that allowed him to read disks from the model 1 on a ibm pc. I also remember a game called death star. It was so big, my dad and and friend of his too all the instructions out of it. It was on a cassette because it ran from basic.
In the early 80s I responded to a classifieds ad selling a TRS-80 Model I for $25 (The ad was selling it for parts). It included the expansion module and two 5.25" disk drives. My dad drove me to the lady's house thinking it could be a nice project to try and fix it. I get there and while my dad was talking to the lady I was looking over the machine. Then I opened my big mouth. :p Hey the ribbon cable from the cpu (keyboard) to the expansion module was plugged in upside-down (I never understood why it wasn't keyed). I flipped it and the machine revved right up! The lady was mildly irratated, laughed and said I could have it anyway. I used the heck out of that machine!
The thing I hated about the Model 1 was the Interface was microphonic. Cat bumped into the thing and it would lock up. The buffered cable the eventually released helped, but it would still lock up randomly if you put your coffee cup down too hard on the desk. I ultimately had mine tricked out with two disk drives.
Whew. Glad you figured that out. I am looking forward to the rest of the C64 batch to be repaired. Keep up the good work. Oh and thanks for the heads up on the 2532s. I have TMS2532JL EPROMs I don't think they are A variants and I do have the data sheets so hopefully I will remember to make sure the GX4 has the correct voltage.
I read somewhere once that Radio Shack expected this to be a very hardcore-hobbyist machine and that they'd sell like 5,000 units total ever, which really explains so much about this monstrosity to me. By the time they were at the Model III they knew they were in the computer business for the long haul (and they legitimately were until Taiwan ate their lunch in the late 1990s) and everything was much less stupid.
If you pause at 21:21, you can decode what it says by looking up each letter (below the "wrong set" letters). Here's what it says: MEMORY SIZE? RADIO SHACK LEVEL II BASIC READY
It's so odd that they constantly designed for the capability to mod, but didn't make it easy for customers OR service technicians. I watch a video of one technician who had to call the Tandy central design center to ask how an upgrade was supposed to be done. It was a little thing whatever it was, but a screw that needed to be removed was buried deep and required a LONG screwdriver.
Loved this video. In the early '80s I bought my first computer, a second hand model 1 that had a lowercase mod - but mine had the switch, which made it usable when running without an operating system. A few months ago, I decided to write a model 1 emulator. I found the font data sheets and created a font ROM file from the data sheet. My emulator behaved exactly the same as your model 1. All the output was in the strange glyphs. I had to modify the screen driver to force bit-6 on to get a legible output. I assumed that it was the level II ROMs output writing ASCII 0-31 instead of ASCII 65-96 to display uppercase characters. Now I know that the font ROM should have been upgraded when adding the extra bit of video RAM. I guess I will have to update my ROM font file to reflect what a properly upgraded machine should look like and un-patch my screen driver.
I had a TRS80, before they had to specify it as a Model I. I don't know if they penny-pinched across the board, but they definitely did in order to hit the entry-price level, and I'm glad for it. I was in 9th grade and got my mom to co-sign a loan for me so I could get one. :). Anyway, the character generator circuitry: there was a $99 upgrade to get "Level II BASIC", and that *included* an 'expanded character set' that gave you the lower case letters. That's how they pitched it in marketing materials. This was around the time that a RAM upgrade was 8 separate ICs for 16K, and was > $100 from a 3rd party.
Loved the video Adrian thanks! An idea for the keyboard cable: how about grabbing a bunch of breadboard jumper wires and using a bunch of male to female on the motherboard side, a similar bunch but male to male on the keyboard side and wrap the loose ends with tape to make them stay together as one connector. Depending on the length needed you could get away with cutting a single bunch in two parts if the breadboard cables are long enough. The color coding would make it easy to figure out how to connect them or you can even switch one position (maybe ground) from male to female to key it if needed. Hope this helps!
As a kid, I borrowed a TRS-80 Model 1 and taught myself how to program it, until the color version was released, then all summer long rode my bicycle every day 7 miles through 2 towns to the Radio Shack that gave me permission to learn on their demo system with cassette drive (to save/load program data).
I did the lower case mod to my Model 1 a zillion years ago (like in the 70's!); it has the original L2 basic rom set. I know my CG ROM had the greek characters, and I did not change either the BASIC or CG ROMs.... but: the lower case mod I did (from memory, my TRS-80 is in storage) had a small switch on the bottom of the case. I'm pretty sure the switch intercepted the extra data line created by the new VRAM chip and either plumbed it through or held it low (or high), so you could effectively turn the LC mod on and off. NEWDOS/80 and other software could deal with the lower case mod, and BASIC loaded from the disk was also happy with it, so I left the switch in "enable" mode all the time unless for some reason I wanted to go back to cassette basic. Someday I'll haul the machine out of storage and see if it still works, and use the diagnostic ROM if it doesn't! Great video series, brings back memories for me.
Thanks for the memories! I had an early Model 1 with expansion, cassette and dual floppies. I did the lowercase mod and got the strange characters. I assumed it was my poor workmanship. Radio Shack would not undo my mod, but replaced my main board, for a price, of course. I was back to uppercase, but at least I had a working computer after that. After watching your video, I think my mod was correct, but I probably had the earlier character chip. :-) Many enjoyable hours with that great first computer.
The RCA TV/monitor is a hot chassis design, so to electrically isolate it from the computer, Radio Shack added an opto-isolator to its video input, powered by 5 volts DC coming through the TRS-80's DIN video cable. So to use it with another video source, you'd need to connect it to an external 5 VDC power supply.
Yikes, I knew of the adapter board, but not why, and the fact the monitor was live chassis (I had a later monitor, not this one). I hate live chassis, they are out to literally kill you - especially where I live with 240V power.
@@paulstubbs7678 Yikes
Adrian... regarding that "odd" transistor configuration fed from the LM723 linear regulator chip... it's a Sziklai Pair Darlington configuration. Just like a regular Darlington Pair, a Sziklai Pair provides much higher gain than any single bipolar transistor, but with the added benefit of being able to pull the output voltage within about 0.6V of the input supply rather than being subject to the 1.2V double emitter-base drop in a regular Darlington pair. What you are looking at is basically a discrete LDO (low-dropout) regulator circuit. The 68 -ohm resistor from emitter to base on the power transistor improves the way the collector of the smaller transistor controls the base drive of the power one and allows it to turn off quicker. It also ensures that if the 723 isn't driving the Sziklai Pair then the base of the power transistor is hard clamped to its emitter, ensuring it is fully off. In operation you can think of the two transistors in that circuit as a single very high gain NPN pass transistor.
In addition to linear power supplies, the Sziklai Pair is also often seen in Class B power amplifiers since it allows you to use two identical (NPN) power transistors on the output driven on the low side by a lower-power NPN and on the high side by a PNP. This overcomes the problem of finding well matched NPN-PNP high-power complementary pairs. This method in commonly known as a Quasi-complementary output stage.
en.wikipedia.org/wiki/Sziklai_pair
Its nice to see 2 fellow amateur radio operators also share my passion for retro computers. A big 73 to IZ8DWF and KI3V from TA2KQ
I had this computer in 1979 that's why I'm watching. But OMG how does anyone know about all this stuff. I'm impressed. I didn't have the expansion and saved everything on a radio shack tape recorder. There was no software so you had to enter everything yourself. A fantastic magazine called something like TRS80 Micro came packed with pages of basic stuff. You'd type it in and spend the night troubleshooting your typing errors. That's how you learned basic, by troubleshooting a program you typed in. Wild west days for sure. Lotsa fun.
If you dig out the TRS8BIT newsletter from September 2017 and Christmas 2017 there's a good rundown of the lowercase mods and the different character generators. Ian Mavric sells (via ebay) a pre-built "Gendon3" character generator based on an EPROM which if you make a bunch of trace cuts gives you 3-line descenders. It should be possible to build a similar PCB adapter to replace the Radio Shack/Motorola character generator ROMs. i'm sure I've seen such a thing done in the past but can't find a link. I'm glad my own TRS-80 Model 1 was from May 1980 and they had integrated a lot of fixes. Still had to mod it for lowercase but it only needed a trace cut and double stacking the RAM chip but everything else just worked... Thanks for the TRS-80 videos, always nice to see.
Ian Maverick is THE MAN when it comes to TRS-80 mods and repair. Bought quite a lot from him in my process including new power switches, character generators and similar. The one I have going now had a strange custom character generator EPROM from the 80s that is very similar to the one Ian offers with descenders but had a few glyphs/symbols changed out.
I wish those TRS8BIT magazines were more searchable - as a newbie to the TRS-80, finding troubleshooting info is very difficult without hounding the right people who know what random tome the answers are in. TRS8BIT really needs to be converted to HTML.
@@natethefighter they really don't have a website or articles online? That's strange
@@SockyNoob the issues are all downloadable to be sure, but only in PDFs by year, and with no searchable table of contents. Info on the TRS-80, to this newbie, seems to be indexed very poorly.
I was wondering about this - I would assume that "character generator" is nothing but a ROM... Maybe it's a fast rom, but it sure seemed to me there should be a way to put a modern EEPROM on a board and build new ones!
It's so cool to see one of these running. That was the first computer I ever used, back in 1980. Taught myself programming on it when I was 15. Two years later, started my IT career which lasted many decades. And it all started with one of these.
It was also the first computer I used in 1980, but I was 10, it was in my 5th grade classroom. It's what I first learned BASIC on.
Now I can buy something the size of a stick of gum 100x as powerful for $5
How times have changed.
My career? well, using computers is my life. I'm just not good at getting hired. So much time in front of the computer, I failed to develop the social skills necessary to actually obtain employment. Now, whole new generations have taken over and I don't even bother against the new competition. Kinda failed in life. Now I have carpal and back problems from being in front of a computer so long and can't do a thing.
And it all started with one of these.
Most of the DIY lowercase mod instructions specifically mention the issue with Level II BASIC as one of the primary reasons for the switch. The official lowercase kit that Tandy/Radio Shack sold included the character generator that allowed incompatible software to work correctly. There was no way to buy the character generator separately, so the DIYers were stuck with a switch. The issue is still fundamentally with the software (like L2 BASIC) that incorrectly use ASCII code 0-31 instead of the proper 64-95. The replacement character generator was just a hack/workaround for that buggy software, by echoing characters 64-95 to 0-31 (at the cost of the 32 characters that were previously at 0-31).
" The official lowercase kit that Tandy/Radio Shack sold included the character generator that allowed incompatible software to work correctly. "
Back in the day I worked tech support for Packard Bell, later Reveal and Newcom, the latter 2 sold peripherals. 90% of the time if something didn't work it was a Tandy, and there was going to be no getting it to work. Those things were terrible.
The later TRS-80s which were produced actually came with the newer character generator already installed, which is probably why most of the docs, etc, that Adrian found didn't mention it. It was only the earlier-produced Model Is which actually had this problem and required swapping out the character generator.
I remember at the time we had two TRS-80s in the house, and did the 8-bit conversion on both of them. One (mine) worked fine with no changes required to the character generator. The other (my father's) required fixing the character generator as well (luckily, we were prepared for the eventuality and as I recall we managed to find a third party who was selling "clones" of the newer CG IC (they're really just a PROM chip) which we were able to use).
Also, the reason the old character generator also had such funky lower-case characters was because at least some of them didn't contain enough memory to actually encode the full character space, so it couldn't do descenders. This also meant that it didn't encode the lower portion of the graphic characters either, so if you tried to run anything that used the TRS-80's "pixel" graphics (which was the whole reason for the 8-bit mod), you would get blank horizontal bands in the display, so you really wanted the newer character generator chip anyway. (Interestingly, the version shown here does appear to encode the full graphic characters, but without descenders, which seems odd..)
As I remember, by the time they came out with the Model III, they had (finally!) fixed the BASIC ROMs to use the correct character set, so the Model III character generators actually had the original special characters in the lower set, and could display some characters that the Model I couldn't as a result (I have vague recollections of a friend of mine putting a Model III ROM and character generator into his Model I so he could make it display everything the Model IIIs could, but I don't remember if there were other complications as a result)...
Loud applause for sticking with it, Adrian! Delighted that you got the Model 1 working. I've really enjoyed your TRS-80 series, especially the Model II, which is my favorite. Keep it up!
I can appreciate how frustrating a video like this must be to make, but make no mistake, to me as a viewer who knows more or less nothing about these machines, it is anything but frustrating to watch.
It's absolutely fascinating, and I'm sure that if the fix had been more straight forward of you by chance found the problem at once, the videos would have been much less instructive.
Even if I don't know the particular platform you're working on, I can appreciate good troubleshooting and learn something from it. Thank you for sharing your struggles instead of just cutting to the chase!
Adrian calls it 'frustrating to watch', I call it a 'a mid episode cliffhanger.'
Could not resist a good Model 1 "Trash 80" story here. Rewind back to Christmas of 1977 and the Model 1 was feverishly being pushed out for Christmas. I had 2 children, one 11 and the younger 9.
Grammy gave us $350 to buy a color TV for our deprived children who watched Yogi Bear on an old B&W TV. I was a hardware tech for AT&T maintaining PDP-11s for AT&T and wanted to get them something more meaningful so I ponied up another $350 and bought a TRS-80 Model 1. I have no idea of what the present value of that $700 is currently.
Christmas morning came and the children opened up the boxes, loaded up a Pong-like game from the cassette recorder (after 3 or more tries). Much amazement followed, but soon the fun was interrupted when the computer got too hot and failed. They played into the night alternately putting the computer in the freezer and reloading the game until the next failure.
Well, the next morning I pulled out the circuit board and was horrified to see virtually all of the wave connections were cold soldered with solder balls everywhere. Took me several hours to cleanup and resolder all of the connections. Worked like a charm afterwards.
My 11 year old learned Basic and wrote a Yahtzee game. Looking for something to challenge him, I purchased a machine language compiler (still on cassette) and turned him loose on writing several little programs to generate Morse Code and such. He then rewrote the Yahtzee program in Z80 machine code. It was blazingly fast.
He went to MIT on a work/study masters scholarship in Computer Science from Texas Instruments and still works in the industry.
That old TRS-80 taught him perseverance and rewarded him with a rewarding lifelong career.
This is a great story. My family also bought me a Model I in about 1978 and I learned everything I could about it. It also stimulated my love of computers and inspired me to study computer science and eventually do a PhD, which has kept me gainfully employed since then. I'm very grateful to my parents for giving into my badgering to buy me the computer as I'm sure your son is to you too.
the way those 2 transistors are connected is known as a sziklai pair, gives high gain like darlington but uses 2 different polarity transistors and gives lower base drive to output voltage drop
Complementary Darlington is another term.
The March 2019 episode of the TRS-80 Trash Talk podcast had an interview with the Model I designer, Steve Leninger. Definitely worth a listen.
Live this video, one of my faves; the reason being that this is the real world for most of us doing repairs... twists and turns and always short of one part. Many thanks.
It was fun to watch you poke around on the insides of the trs80 model 1. It has been almost 40 years since I did that.
Sometimes the best thing is to walk away, not overthink it. Clear your mind and new approaches will pop in your head when you least expect it
But what is awesome is to watch you work the problem. And take me back down memory lane. So sorry I couldn't come up to Elmhurst this weekend (I'm only a couple hours away ). Hopefully next year!
The collector connection on the PNP power transistor was definitely the fault causing the smaller transistor to get hot and the large one to stay cold. It's fairly easy to explain why this is the case.
The configuration of those two transistors with one PNP and the other NPN is called a "compound Darlington" transistor, this has some advantages over a regular Darlington configuration with both NPN, including reduced "dropout voltage" or minimum input to output differential, which is useful in a voltage regulator.
With the collector disconnected nearly all the current will flow through the emitter to base junction of the PNP transistor (acting as a diode) and only a very small amount through the resistor. The full combined current will then flow through the collector to emitter junction of the NPN transistor, which is doing all the work, this is why it gets so hot.
Even though the PNP transistor is still carrying the full current through its emitter to base junction the voltage drop is only 0.6 volts so the power dissipation is 8.6 times less than it would normally be when most of the current is going through its emitter and collector with a 5.2 volt drop, since power dissipation is voltage x current.
The demotion of the PNP transistor to a diode also means you lose the gain of the PNP transistor, this reduced gain depending on the gain in the IC will cause the voltage regulation to be worse so it will tend to sag more under load.
When you reconnect the collector the current through the collector/emitter of the NPN transistor will drop by the ratio of the hFE (gain) of the PNP transistor - data sheets suggest 15 to 200 for hFE. This massive drop in current is why the small NPN transistor no longer gets hot.
How good is the emitter/base junction of a power transistor as a diode when the collector is left flailing in the wind ? Pretty good actually. IE (maximum emitter current, eg base + collector) of that transistor is 17 amps, but IB maximum base current is still 5 amps. So the PNP transistor would survive this fault just fine if the total current drawn was less than 5 amps, and because the voltage drop is only 0.6 volts it would be a long way below its maximum power dissipation!
If there is enough gain in the IC to compensate for the loss of gain of the PNP transistor (or the following circuitry isn't too fussy about the voltage regulation being a bit worse than normal) the only real symptom of this fault would be that small transistor getting hot but apparently surviving. It could have been working like this for years!
The phenomenon of an oxidised connection that measures OK with a multimeter but goes more or less open circuit or very high resistance as soon as you try to pass any real current through it is real and actually happens a lot. I see you got caught out with that on the ribbon cable on one of the previous TRS 80 videos as well. While it tests good with a meter there is in reality only the tiniest whisker of a microscopic connection in amongst all the oxidation, as soon as any current flows this microscopic contact point disintegrates and the connection is lost.
Modern digital multimeters only pass microamps in ohms mode and even in diode test mode only a few milliamps, this can give very misleading results in situations where there is severe corrosion or there is only a point/whisker connection occurring due to a mechanical problem like a loose screw. You had both corrosion and a loose screw here...
Very nice description of the circuit! Thanks for this.
@Arnold Rimmer not quite.
The IC is doing the regulating. At no time was the base of the NPN handling much current, it was all going through emitter-collector.
The NPN is more like a pre-amp for the output transistor.
@Arnold Rimmer it's no different to using a Darlington transistor. It would not amplify the ripple current as the voltage is sensed after the pass transistors.
The transistor arrangement shown is a Sziklai pair, which basically makes a very high gain NPN transistor, without the double emitter drop of a regular Darlington. It is the same arrangement as used in modern LDO linear regulators
I so very much enjoyed this. My high-school buddy got one of these for his birthday in 1977. It seemed to us like a mainframe, which was probably down to the color scheme and serious-looking text display and professional-looking blocky graphics. It looked like it belonged in a glass room. "No kids stuff here, boys, this is a REAL computer." AND, it was programmable. My buddy programmed some of the rules to our table-top Avalon Hill war games so that it handled all die rolls and turn management. It was like Billion Dollar Brain only at this guy's house from school. Boy, did I ever want one but $600 seemed as far away from me as the Andromeda Galaxy. Wouldn't you know, after college I actually did end up programming mainframes (real ones) -- a job which I still do to this day. Watching this repair series completed a beautiful circle.
I love it when I see other folk have exactly the same frustrations I do sometimes, nice to know I'm not alone when poring over a schematic trying to work out what might be causing problems a, b and c. As for penny pinching Rat Shack just imagine what the PET may have looked like if Tramiel had let them have their way instead of rejecting them and allowing Chuck Peddle to design it instead. Excellent stuff as always!
Wait, Commodore was going to partner with Radio Shack originally? This story just keeps getting wilder.
Not understanding the Tramiel comment. Thought he was known to penny pinch as well, no?
The character set issue is why many 3rd party mods had a switch to disable the 8th bit, or else some other special circuitry to make the video memory not 8-bit clean, and why the tandy mod had a replacement char generator. Without that, you'd have the garbage until you loaded the lowercase driver.
The ROMs (all versions I ever saw) for some reason cleared the bit for alpha chars resulting in the wrong char set. I patched that out (decades ago!) and programmed my own EPROMs. Worked great.
One wonders why in the world they did that. Was there some kind of miscommunication with Microsoft which ended up with MS coding the output routines as if the alphanumerics were literally 6 bit with the letters shifted down to 0-31 instead of there being that fake-bit-6 circuitry? (There is an official 6 bit ASCII subset with that layout... I think it's used for some naval/marine purpose, not usually computers, but it is out there.)
I really like your split-screen editing including all the relevant information visible at once!
As others have noted, there is a fuse inside the power block. As an ex-TRS-80 tech, I can vouch for the fact that that fuse can pop pretty easily. I remember the Tandy TS bulletin that described how to carefully crack them open and replace the fuses. (Towel, vice, etc...) It will blow! (I ran my own computer repair shop thru the '80s mostly working on TRS and Atari computers, as well as a few S-100 and industrial/CNC devices...) Fun to watch these vids! Keep on hackin'! Stu
Adrian. Thanks for the video. Takes me back to my teenager days when Radio Shack was tops in my book. I Owned a TRS80 Model III which was my first computer. Then I owned a Model 1000 which I Played the heck out of. Thanks for showing all your mistakes. I like to see those. even though this was really furstrting for you, it really was a teaching moment when dealing with the EProms.
Yeah your frustration is evident… But as I've said before it's often the failures that teach us the most… This showed a lot of excellent best practices… Thanks for sharing this!
RS used at least three different character generators from Motorola. The third one also had the raised lowercase and funny characters, but the lowercase 'a' was shifted up two lines. Many years ago I made a crazy outboard circuit to shift the specific lowercase letters down by two lines. (it used a 7483 adder which ran warm all the time!)
The video driver code for some reason went out of its way to force the uppercase to the 00-1F range, and I'm surprised that wasn't fixed in the 1.3 ROM. I still think the patch I suggested in the last video comments will probably work with the funny chargen:
0471: FE403808 D640FE20 -> 00000000 0000FE60
And I like that external floppy drive case you rigged up.
An IMPORTANT WARNING: do not turn the computer off with a disk in the drive and the door closed! If you see the drive select flash, it probably also pushed a glitch through the drive heads, zapping whatever sector is under the head. I lost a few sectors of data that way back in the day.
I was in the first computer lab in my school district in 79/80. We had a row of Wang programmable (model 452 I think), access to the county mainframe to run Fortran and Cobol code and a TRS-80 with 8kb of memory and a cassette backup to run our MS-Basic work.
Later my parents got a TRS-80 Model 16 for their business that I would run experimental programs on when they weren't using it. I learned to debug and simplify my code on the Model 16 when I wrote a graphical Tie-Fighter shooter game that went from about 250 lines of code to its final version at 95 lines.
Nice to see someone keeping the old tech alive. Thanks.
Not frustrating, deeply satisfying. If it was easy, anybody could do it! The sense of accomplishment when you plow through multiple problems cannot be beat.
Nothing like a nice long repair caper to cap off the long weekend. 😌 Keep up the good work!
Repair caper makes it sound like a heist repair movie. Ocean's TRS-80.
@@silmarian lmao
My first computer was a TRS80 Model 1 with the expansion pack you showed, disk drive and matching printer. I remember the frustrating problem my Model 1 eventually developed. The connection between the computer and the expansion pack went flaky which caused random, spontaneous reboots - usually right in the middle of programming. Nevertheless, the Model 1 set me off on an IT career that's still going 34 years later.
39:27 - Far from being frustrating to watch Adrian, it's fascinating to see how you determined what the fault was and how to fix it.
Good job Adrian. In the future, many of these early computer models will be rare or non-operational. You'll be one of the monks that has any knowledge them.
@Mr Guru Sure, if you have the time. Most of us don't.
No matter how good you are, there are always difficult problems to solve. I like watching this, too, because I think about how many places someone can inject doubt into the process. It's a problem with troubleshooting, and doing it for other people. When you're doing it for yourself, you know where you're at. When you're doing it for someone else, doubt can be injected into the process at different points to try to steal a customer, or steal credibility.
This is the computer I learned on. I ran newdos80 ver 2. Radio shack double density kit installed. All was ok when I ran trsdos 2.7. Newdos sometimes would skip writing a sector. I applied all the patches so newdos would work. I used my sytem at work and it became a big problem in dbl density. I called apparant. Spent 2 hours on the phone. They told me that it was in the write driver and they will fix it. A month later got a package in the mail. A new copy came. It didn’t fix it. I ended changing the 1793 and the problem was still there. Took the expansion interface to a computer shop. The put ferric cores around my cables and that fixed it. But it drove me crazy.
My very first computer was a TRS-80 Model 1 with cassette input got it I think in 1976. The manager a Radio Shack would routinely give me copies of programs. My friend Mark at the same time had a Commodore Pet. When I saw his graphics I was jealous. From there I moved up to an S-100 bus CPM MACHINE. Followed by a Commodore Vic-20, major upgrade to a C-64. Loved that machine. Had my first printer & modem. Actually used local C-64 based BBS. Then a TeleVideo Luggable KayPro Clone with CPM DUAL 5.25” Disk Drives, an external Modem & Daisy Wheel Printer! Then I went to work for a small engineering company. Had my first PC DOS Clone with an NEC Z-80 chip and an actual 20 mb hard disk, green mono monitor. Had a hardware based RTTY and CW modem for my ham rig! Oh the early glory days. At that same company we used the Tandy 40 character LCD kind of early laptops. Used it with a scanner wand to read barcode.
I must have got lucky, I never had any issues with my TRS-80. It work every time I hooked it up as time went by. Drug it to school & back, Electronics school, no issues.
Adrian, these videos are good to watch and follow a long with you on your diagnosis/debugging. You are too hard on yourself about them. Don’t be afraid to publish these!
You can program 21V or 25V Eproms with the TL866-II! Al you need is to isolate/ cut the Programpin between Eprom and the TL866-II and use a Benchpowersupply and inject the correct Volatege external to the Eprom.
You need to dissable the Pincheck in the TL866 Software and can program with the standart Epromprofiles of with Programmer.
Can't say for 25v eproms but 21v ones program just fine at 18v but the cycle time needs to be increased
Regarding Adrian's concern about a lack of any fuse protecting the analog inputs to the voltage regulator, I think there actually is a protection fuse buried inside the black radio shack power brick. Some other videos show people cracking open the black plastic enclosure to do repairs and replacements on the voltage step-down transformers, which might also reveal a fuse buried inside too.
Hi Adrian. At 4:36, you don’t need to lift the resistor; just set the DMM for regular resistance, then measure the B-E junction. If you get 68 ohms, the transistor is either open, or has a high resistance in that direction (switch the leads, and try again.) If you get 0.1 ohms or less, then the B-E junction is shorted as you already tested the 68 ohm resistor (you DID test the resistor, right?), and if you get something like 40-65 ohms or so, then the B-E junction is probably good, as resistors in parallel divide (the 68 ohm resistor, and the (effective) resistance of the B-E junction as seen by the DMM.) HTH! 😌
Regular resistance setting on a DMM wouldn't be adequate as they typically don't provide sufficient voltage/current to turn on a diode. You'd likely read the transistor junction as open when it is really OK.
You would likely detect a short though, but then that's what the low ohm resistor was effectively giving as a false reading.
Not only is your video informative, but your commentators, those that know the TRS-80 line of computers, have some great insight!
Taking a moment (and pausing at 21:41) you can use the printout to decode the output. The message reads:
MEMORY SIZE?
RADIO SHACK LEVEL II BASIC
READY
Fun stuff! I was able to find some info on that character generator. The Sams ComputerFacts list it with both 8046670 and MCM6670. Using the MCM number it wasn't hard to turn up a datasheet. There were a couple versions of that chip - the 6670 was mask-programmed and the 6674 which came preprogrammed. The datasheet includes the font in the preprogrammed version and guess what - it's just what you were seeing with the different control chars and non-descender lowercase characters. So looks like RS started off with the off-the-shelf version and later around to programming their own custom layout. That datasheet is interesting as it includes info on ordering your custom layout with the proper punchcard or paper tape formats.
I like your EMS t-shirt. Back in the day I always used EMS floppies because I never had problems with them, and the labels looked cool! I seem to recall they weren't all that expensive either.
You should get a thermal camera. There are some FLIR ones for just a couple/few hundred $. Can record the video too. Can easily look at the whole board w/o having to touch.
You can even get them in a phone. We use a Blackview BV6600 Pro at work that has one built in, runs Android, and has an 8000-odd mAh battery. Makes diagnosis a hell of a lot easier!
@@cleanycloth yep exactly. I have a flir ONE PRO and does the trick!
these vids where things go -way wonky, actually make me feel better! i’m pretty good at problem solving- and to see you having issues… makes me think… if these projects were easy… then they’d be no fun! i love learning, and failures are how we learn! thank you for these difficult projects!
This really brings back memories, I was in my highschool's first ever computer class. We have 4 of the TRS-80 complete with cassette recorders in 1979-1980. The expansion block was HUGE when it came out, and my senior year the principal popped for a hard drive!
Thanks for bringing back some early geeky memories. I worked at Radio Shack when I was 15 and got a Model 1, then a Model 3. Wrote Z80 assembler to copy system floppies without knowing what I was doing! It started my career in computer science, BSc and MSc, 22 years at IBM.
19:00 I would have tried replacing it with a 74HC chip, before building an adapter for a HEF/4000 series IC. The reason for using 74C (or 4000) was often either the high input impedance or the symmetric voltage swing. The more modern HC variant got that as well (like most CMOS) but with the same pinout as all other 74-series ICs.
yep, that would seem to be the obvious thing, or even HCT as the outputs are the same, just input levels compatible to standard/LS TTL
just looked at a datasheet for the C types and they have CMOS level inputs so a HC type would likely work
@@andygozzo72 Not so sure they wanted TTL input levels (or output) in that design though, if you look at the schematic.
@@herrbonk3635 Yeah, they specifically use CMOS in that part of the design because they want both the reliably close-to-the-rail outputs *and* the higher "high" input threshold. The system that's used to adjust the horizontal and vertical position is basically analog; there's a trimmer resistor and a capacitor, and the rate at which the capacitor charges enough to activate the input of the next CMOS stage is what sets the delay. Using a TTL compatible HCT instead of an HC would change the time constant they designed for.
The only concern I see with HC is it is a lot *faster* than C in gate propagation delay, but I don't think that'd matter much here; for horizontal hold you're trimming within a few microseconds, losing a few hundred nanoseconds of gate delay should be easy enough to compensate for.
Wonderful job on that Trash-80. :) I love these machines.
I love videos like this. When you had FIVE chips fail in sequence, I KNEW it was PEBKAC.
Proper descenders were the later character generator ROM. The moved all the character up to give space for the descenders. The model 1 uses a lot of memory mapped IO (except for port #$FF), but the model 3 moved a lot of the IO to ports so its not just the keyboard that changed.
Also WRT to the ROMs. You could just replace the entire level 2 board with say a single 27C128 and a bit of select logic and put all the ROM images in one part.
There's a great replacement power supply PCB which uses modern transformers and sits inside the expansion chassis. Ian Mavric (at least used to) sell kits for them. He also sells/sold lower case mod kits including replacement character generators. You should look him up and maybe check out all of the stuff he's done for the TRS-80 over the years.
You could probably adapt the system to run from a Meanwell power supply as well
I had a Model 1 way back in the early 80's I had a Lobo LX-80 expansion interface for it but never had floppy drives. The Lobo interface was originally made for their model 1 compatible computer but worked fine on original Model 1.
Great video Adrian. Like a lot of others, I did the lowercase mod to my Model I back in the day and I also remember changing the character generator to get descenders. I had forgotten about the problem with booting into Level II but it's coming back to me now. I still have the machine and documents in storage and you've inspired me to dig it out and try and work out exactly what mod it was that I installed.
Hey, I owned one of them and I loved it. Of course, typing in the instructions for your program every time was a tad of work but fun to as it helped you learn about programming and how the computer actually functioned. There was only basic on this one and it was built in. It cost right at $500 too and was a lot back then.
It has been so nostalgic to see you repair a TRS-80 model I, and it makes me want to go into my basement and dig mine out to see if I can get it running. I bought mine on May 26, 1979. It got lightning-fried about a year later and just about everything got fried...a neat trick considering it was unplugged at the time. So I don't know what date the board is, but it was repaired...the tech said "every circuit was fried". It doesn't have the lower case mod, nor does it have a keypad, but I have an E/I and a floppy drive (I thought I had two, but I think that's just my 30+ year old memory). The last time I did, I got that garbage screen...I forgot that it was normal with an E/I attached, and forgot about the break key thing. Now I REALLY want to try it. The only real "mod" I ever did on it was replaciing hte short keyboard cable with a 6 ft. ribbon cable so I could put the keyboard on my lap while the computer was on my desk (bare board all hooked up). Thanks again for the memories!
Can't tell you how much I enjoyed this. My first computer was a Model 1. I modded it constantly as new options came to market.
It's fun to see some frustration, Adrian - it's a more realistic view into the process!
Given the frustration, at the end of the video when you said "if you didn't like it, you know what to do" - it made me realize for the first time another meaning for that phrase. Made me laugh out loud! 😂
If you did encouter an issue with a 74Cxx series chip, you might try substituting a 74HCxx or 74HCTxx, which are much more common series today. Most likely the 74HCxx would be the closest approximation. The 74HCT series is supposed to have TTL-compatible input levels.
I started following you for the lightbulb videos. The old machine repairs are fun but I'd be happy if you branched out too. I could listen to you talk about anything.
You survived it! I was starting to get worried but you still have your mental health. 😃
I have vague memories of doing a lower case mod (ok, really my high school comp sci teacher did the mod) where I got to keep the alternate characters and had to live with the non descender lower case characters. I could use them in my own programming by poking them into video memory as I attempted to write games.
My model I was one I bought as Radio Shack was clearing them out as a 4k machine and we later upgraded to 16k with aftermarket RAM. It might have even been a level I machine which we upgraded to Level II at some point.
I do remember having to load a key debounce routine for a period of time (not sure how that worked).
I may also have had to load a patch for the lower case to work properly with that alternate character set ROM or maybe we got a patched Level II ROM set. It was a long time ago and a lot of that stuff was pushed out of my memory but maybe my minimal recollections will jog other people's memory on such things.
I believe my teacher had opted for the true descender character ROM when he did his own Model I so the different lower case mods was definitely a thing.
Adrian, I had a TRS-80 Mod 1 when they were new. So it was fun to watch you to bring yours back to life. Though I don't have the machine anymore, I did find some original software documentation, and disks in a storage locker. I have TRSDOS, Scripsit word processor, UOLISP Lisp interpreter. As I continue to unpack, I may find a Pascal compiler with IDE, and maybe an APL interpreter. If you are interested in them, I can send them to you for free. They might help you create a software collection around the machine you rehabilitated. Let me know if you want them in a reply. I think it is great that you are reconstructing history of early PC's by rehabbing machines of that era.
The two transistors are set up as a Darlington configuration. the smaller transistor is supplying enough gain to operate the larger transistor.
This is not a darlington pair. It's sziklai pair because the transistors has opposite polarisation. PNP and NPN.
Darligtons has same type of bjt.
Nice shirt, I haven't seen the elephant memory logo since the 80s.
I really enjoy your videos AdrIan, I've watched most of them. I hope you keep on doing youtube content for a long time!
What memories this video brought back! I had a Model I 16k that I modded with an LNW expansion interface. I don’t remember if I did the lower-case mod in it - I remember all the issues and I think you were limited to a 5x8 matrix for the character generator which didn’t allow descenders. I’m pretty sure I waited until I got my Model III which had 8 bit and a 5x9 matrix CG. But I do remember being envious of the IBM PC’s serif typeface for the screen font, so I set out to build a mod for the CG. It was really just a PROM that I replaced with an EPROM which required an 18 pin to 24 pin adapter I cobbled up, so I was able to have a custom character generator. I wrote a little editor in BASIC that let you design your own characters; outputting a binary file to load into the EPROM.
I thought this might be a good magazine article, so I wrote one up and sent it to 80 Micro, the premiere TRS-80 magazine at the time; but they rejected it saying someone else had beat me to it. I submitted it to another startup magazine who accepted it, sent me a $25 binder; and then promptly went out of business. I don’t think I ever saw any article about custom CG’s; but by then I had moved on to building a Taiwanese PC-XT clone from a bare PC board. I jumped ship from Tandy to IBM and never really looked back.
Nice to see this old tech bring fixed,sadly when it's gone ,it's gone.I have a trs80 model 1 pocket pc that works yah !
Apologies if this has been said in the other 300+ comments - Ian Mavric may have original character generator chips that are the “new” character types. I had a faulty one of mine replaced a few years ago.
A few thoughts on the regulator...
1. Thermal paste for power transistors often needs to be explicitly nonconductive, which is not necessarily the case for the kinds used for PC hardware, where voltages are generally minuscule but heat transfer is a major concern. Thermal conductivity tends to be tied to electrical conductivity, so it is not unusual to see metals being added to enhance performance (e.g. the classic Arctic Silver 5, or even going liquid metal). For power supplies and audio amps, you want the plain white silicone goop.
2. The regulator pass transistor topology is a CFP or Sziklai pair. It's got a lower voltage drop than a Darlington and actually sports inherent feedback. Now being a loop of two common emitter circuits, it isn't exactly the fastest thing, but that isn't too much of an issue if there's a healthy amount of current flowing, as in a regulator pass transistor or a Class A amplifier output.
Also, Noctua NT-H1 has an operating temperature range of only -50 to 110°C, which is fine for a CPU but perhaps a bit low for a power transistor. (NT-H2 goes up to 200°C, in case I didn't read the part number on the video correctly, and would go well with the absolute maximum junction temperature of 200°C for a 2N6594.) Both NT-H1 and NT-H2 thermal paste compounds are not electrically conductive. To be clear, perhaps you could mention that it needs to be zinc oxide thermal paste (which could have a silicone carrier) rather than just any plain white silicone goop.
It was great meeting you today at VCF Midwest. Enjoy the Malort!
First thing to debug electronics is always check power supply. And a multimeter is never enough to check a computer voltage rail. ⚡️give it to an oscilloscope to check for glitch. Even a switching power supply can show an ok voltage on a multimeter. But under a scope you will see the spikes that can create havoc on a logic machine.
The regulator circuit would send all the current through the little transistor if the collector on the big one got disconnected. The big one would act as just a diode, bypassing the resistor and feeding current straight to the little one, which would then drop all the voltage itself.
Oh man. All those setbacks. But you still managed to make it work again. Way to go Adrian!
The 723 regulator IC was quite commonly used in the 70s and early 80s, and available from many sources - MC1723 from Motorola, uA723 from Unitrode/TI/Fairchild/NEC, and LM723 from National Semicondutor. It's still made today and a lot of small cheap bench power supply projects use it
The MCM6670 character generator is unobtanium but it is just a ROM.
7 bit ASCII + 3 bits for the row, and 5 bits output.
A 1K or larger EPROM can be substituted as long as you have a socket adapter.
If you're really clever you can add an R8 row select line and have perfect lowercase descenders.
Great work! Look forward to each Saturday so I can unwind watching the great content. Many thanks!
Boy does this video take me back.. I bought the first TRS80 Model 1 Level 1 4K sold at the Yuma AZ store for $795. I was in the Army, stationed at the Army Yuma Proving Grounds at the time, and had everybody in the barracks looking over my shoulder while I played with my new toy.
We had a model1 48k, a bunch of mods, and 2 floppy drives. We ran trs-dos and new-dos. My dad did something that allowed him to read disks from the model 1 on a ibm pc. I also remember a game called death star. It was so big, my dad and and friend of his too all the instructions out of it. It was on a cassette because it ran from basic.
Heh- just when I got back from seeing you in person at VCFMW this video was in my feed. Both your video today and LGR's yesterday. :D
Perseverance brings fascinating videos! Thanks for another excellent series on a machine that wasn't so popular over here in the UK.
In the early 80s I responded to a classifieds ad selling a TRS-80 Model I for $25 (The ad was selling it for parts). It included the expansion module and two 5.25" disk drives. My dad drove me to the lady's house thinking it could be a nice project to try and fix it. I get there and while my dad was talking to the lady I was looking over the machine. Then I opened my big mouth. :p Hey the ribbon cable from the cpu (keyboard) to the expansion module was plugged in upside-down (I never understood why it wasn't keyed). I flipped it and the machine revved right up! The lady was mildly irratated, laughed and said I could have it anyway. I used the heck out of that machine!
The thing I hated about the Model 1 was the Interface was microphonic. Cat bumped into the thing and it would lock up. The buffered cable the eventually released helped, but it would still lock up randomly if you put your coffee cup down too hard on the desk. I ultimately had mine tricked out with two disk drives.
Whew. Glad you figured that out. I am looking forward to the rest of the C64 batch to be repaired. Keep up the good work. Oh and thanks for the heads up on the 2532s. I have TMS2532JL EPROMs I don't think they are A variants and I do have the data sheets so hopefully I will remember to make sure the GX4 has the correct voltage.
My first computer! bought it, the exspander, two 5" drives for 35$ at a pawn shop. My mom bought me a basic book and I programmed a few games....
I read somewhere once that Radio Shack expected this to be a very hardcore-hobbyist machine and that they'd sell like 5,000 units total ever, which really explains so much about this monstrosity to me. By the time they were at the Model III they knew they were in the computer business for the long haul (and they legitimately were until Taiwan ate their lunch in the late 1990s) and everything was much less stupid.
If you pause at 21:21, you can decode what it says by looking up each letter (below the "wrong set" letters). Here's what it says:
MEMORY SIZE?
RADIO SHACK LEVEL II BASIC
READY
Always nice to get the backstory of an item. :)
Nice work! I Really need to get myself one of those desoldering guns.
Nice to see you got the Model 1 working. I love the elephant memory shirt too.
It's so odd that they constantly designed for the capability to mod, but didn't make it easy for customers OR service technicians. I watch a video of one technician who had to call the Tandy central design center to ask how an upgrade was supposed to be done. It was a little thing whatever it was, but a screw that needed to be removed was buried deep and required a LONG screwdriver.
Loved this video.
In the early '80s I bought my first computer, a second hand model 1 that had a lowercase mod - but mine had the switch, which made it usable when running without an operating system.
A few months ago, I decided to write a model 1 emulator. I found the font data sheets and created a font ROM file from the data sheet. My emulator behaved exactly the same as your model 1.
All the output was in the strange glyphs. I had to modify the screen driver to force bit-6 on to get a legible output. I assumed that it was the level II ROMs output writing ASCII 0-31 instead of ASCII 65-96 to display uppercase characters. Now I know that the font ROM should have been upgraded when adding the extra bit of video RAM.
I guess I will have to update my ROM font file to reflect what a properly upgraded machine should look like and un-patch my screen driver.
Nearly an hour of Adrian fixing a TRS-80. YES PLEASE!
I had a TRS80, before they had to specify it as a Model I. I don't know if they penny-pinched across the board, but they definitely did in order to hit the entry-price level, and I'm glad for it. I was in 9th grade and got my mom to co-sign a loan for me so I could get one. :). Anyway, the character generator circuitry: there was a $99 upgrade to get "Level II BASIC", and that *included* an 'expanded character set' that gave you the lower case letters. That's how they pitched it in marketing materials. This was around the time that a RAM upgrade was 8 separate ICs for 16K, and was > $100 from a 3rd party.
You made it again! congratulations.
Loved the video Adrian thanks! An idea for the keyboard cable: how about grabbing a bunch of breadboard jumper wires and using a bunch of male to female on the motherboard side, a similar bunch but male to male on the keyboard side and wrap the loose ends with tape to make them stay together as one connector. Depending on the length needed you could get away with cutting a single bunch in two parts if the breadboard cables are long enough. The color coding would make it easy to figure out how to connect them or you can even switch one position (maybe ground) from male to female to key it if needed. Hope this helps!
Love these videos of real struggles of fixing machines
As a kid, I borrowed a TRS-80 Model 1 and taught myself how to program it, until the color version was released, then all summer long rode my bicycle every day 7 miles through 2 towns to the Radio Shack that gave me permission to learn on their demo system with cassette drive (to save/load program data).
I did the lower case mod to my Model 1 a zillion years ago (like in the 70's!); it has the original L2 basic rom set. I know my CG ROM had the greek characters, and I did not change either the BASIC or CG ROMs.... but: the lower case mod I did (from memory, my TRS-80 is in storage) had a small switch on the bottom of the case. I'm pretty sure the switch intercepted the extra data line created by the new VRAM chip and either plumbed it through or held it low (or high), so you could effectively turn the LC mod on and off. NEWDOS/80 and other software could deal with the lower case mod, and BASIC loaded from the disk was also happy with it, so I left the switch in "enable" mode all the time unless for some reason I wanted to go back to cassette basic. Someday I'll haul the machine out of storage and see if it still works, and use the diagnostic ROM if it doesn't! Great video series, brings back memories for me.
Thanks for the memories! I had an early Model 1 with expansion, cassette and dual floppies. I did the lowercase mod and got the strange characters. I assumed it was my poor workmanship. Radio Shack would not undo my mod, but replaced my main board, for a price, of course. I was back to uppercase, but at least I had a working computer after that. After watching your video, I think my mod was correct, but I probably had the earlier character chip. :-) Many enjoyable hours with that great first computer.