Jump to content
IGNORED

Sinclair ZX-81 -Inspection, Cleaning, Possible Repair


Recommended Posts

So like many a proto-nerd in the early 80's, my first "real" computer was a Sinclair ZX-81. No, not the later co-branded Timex Sinclair version, but an actual honest-to-Sir Clive Sinclair. The summer I turned 13 in 1981, I saved up money from parents and grandparents and allowance cash for a couple months and responded to a Sinclair ad in a computer magazine and sent away for a ZX-81 and 16K memory expansion pack.

 

zx81_7.thumb.jpg.14f3f10399e585271ce8d05b39c70417.jpg

 

Recently, I picked up a "not tested" example on eBay that seemed relatively intact, and it also included the same 16K Sinclair memory expansion. When I got it, I dug up a 9VDC center-positive PSU with a 3.5mm plug (one of my Atari VCS/2600 PSUs) and tested it. The machine produces a picture but it's VERY RF-noisy and I don't see a Sinclair BASIC cursor that I think I should see - just a messy, noisy white. There was no keyboard response either. So ... this may end up being just a cosmetic clean up, depending on what I eventually find.

 

 

 


 

IMG_0317.thumb.jpg.2c1033d31bfe043e298ccf147b42590f.jpg

 

IMG_0318.thumb.jpg.81f4374141bed0dcc7c59b378abbd792.jpg
 

 

 

Today, I gave the external a once-over, cleaning off a few decades of light dirt and probably nicotine stains. Cosmetically, there are a few scratches and the red paint is mostly worn off the embossed ZX81 logos on the computer and memory pack, but it's not too bad. Three of the four rubber feet were missing. When I removed the one remaining foot to expose what I thought would be a final screw, I found I was mistaken. So it's apparent someone has been inside this one before. That became even more apparent when I discovered that three of the screw holes are stripped out. Well, a problem for another day.

 

 

 


IMG_0338.thumb.jpg.5f0433a1414207cb04b96ad456c94e3f.jpg

 

IMG_0339.thumb.jpg.4475e008112e5456f1a9fdb4461e0ba0.jpg
 

 

 

Inside however, the board is relatively clean and appears original and unworked on. The board layout is relatively simple and straightforward, as you would expect from a Z80 SBC with only 2K of on-board RAM. There is a Ferranti ULA chip, the NEC Z80 processor, a Sinclair ROM, a couple small RAM chips, and some support logic. However, the keyboard membrane ribbon has the apparent texture of paper and whoever was inside this machine before wasn't careful taking it apart. The way the top of the case is basically heat-press fit over top of the membrane and RF shield make me think it rather unlikely I will get it apart very easily. However, if the machine is repairable electronically, I can likely bodge up a replacement for the PCB edge connector on the main board and then solder wires to the individual leads to the keyboard matrix. Assuming the key connectors inside the membrane actually work (big assumption), it would work.

 

 

 


IMG_0349.thumb.jpg.14e58a26e2ac5dbcc5fa8130e0607fdd.jpg

 

IMG_0346.thumb.jpg.71bba73af21b042e9b58622a57d19456.jpg

 

IMG_0347.thumb.jpg.38653aca54fe2355c2bb268e02a346fd.jpg

 IMG_0348.thumb.jpg.7deb84f71d63c5d13588e3f8e81804a2.jpg

IMG_0350.thumb.jpg.02ae300c0f8934428743db86428ac9cc.jpg

 

IMG_0351.thumb.jpg.f257635f7be06cb5ff728a0c1874c7ac.jpg

 

IMG_0352.thumb.jpg.756b09a27a24587e325c771da3e8e321.jpg

 

IMG_0353.thumb.jpg.34b045f346c0989f001b8a5b47ce4b36.jpg

 

 

The underside of the board is a cool red finish - very stylish and quite out of character for Sir Clive to have spent money on red solder mask that was never likely to see the light of day, but it does look good. It has this weird, RF shield/grounding strap/board spacer thing soldered across the back side though, and the pins of the soldered chips were left long around the area. Very odd, cheap-looking detail.

 

 

 


IMG_0344.thumb.jpg.9f66259bb2ff6eb3593cc43f20f1591b.jpg

 

IMG_0345.thumb.jpg.ce1887cb8f8473f2ec47f24e776216b1.jpg

 

IMG_0354.thumb.jpg.713d445880b59312b91c091aa1f990a1.jpg

 

IMG_0355.thumb.jpg.1f630ad286b7df304b7779d165753303.jpg
 

 

 

I also took apart the external memory pack, and all I can think of is that the engineers who designed this must have been a) super inexperienced; b) super cheap labor; or c) drunk. WTF is there an RF coil on a memory expansion? Why does a measly 16K DRAM board need all these electrolytic caps? Why is the whole board, top and bottom, dotted with so much flux residue? There's a spot on the back that looks a lot like re-work but when you clean off the flux, nope. Factory original and all the chips match in brand and date code, so who knows?

 

 

 


IMG_0333.thumb.jpg.5a9bcd31567c51e55e1cfa8620e78e46.jpg

 

IMG_0335.thumb.jpg.22343ccdf78d40bd052c1d3904ebf17f.jpg

 

IMG_0336.thumb.jpg.bf9fce144bcab8e7652867141474231d.jpg

 

IMG_0337.thumb.jpg.0a08d20bc36c5b3e9897c0f98a8b6025.jpg

 

IMG_0334.thumb.jpg.806b14c52fcf246b2ec696acb962dd62.jpg
 

 

 

Anyway, this will be a low-priority, low-stress restoration/maybe repair project as I have time and motivation over the coming months. I will never really use this even if I get it to boot, so even if I get it running and can't get the keyboard working, that would be fine. Mostly, having this machine is about knowing that I have an example of the very first computer I ever owned. It'll still be a fun conversation piece in my game room someday and that's really enough. Anything else will be gravy.

 

Edited by DrVenkman
Formatting
  • Like 5
Link to comment
Share on other sites

A quick Google reinforces what I thought I had already read over the years about the failure modes, and it's likely that the Ferranti ULA (Uncommitted Logic Array) chip has died. Originals are unobtanium for people not sitting on a stack of NOS machines or a closet full they hoarded in the 80's to salvage, but fortunately nerds are clever and creative. I've ordered this from a maker in New Zealand. We'll see if that gets me up and running.  If not, there's not a whole lot else to go bad on this board. There's an oscillator, a Z80 processor, a ROM and some RAM.

 

https://www.tindie.com/products/charlieingley/vla81-zx81-ula-replacement/

 

I'm going to connect a power supply to it this weekend and verify the 7805 is good, check for a clock signal and power to all the right places on the board, maybe give that nasty 16K board a bath in warm soapy water followed by an alcohol rinse, then wait for the replacement ULA to arrive in a few weeks from the other side of the world. 

Link to comment
Share on other sites

Heh. Clever nerds abound the world over. Plenty of other folks have run into the stupid membrane keyboard cable tearing and some chaps over in the UK have me covered. Brand new-made replacement keyboard membranes with a much better quality plastic. One will be on its way to me shortly.

 

https://www.sellmyretro.com/offer/details/Brand-New-Sinclair-ZX81-Keyboard-Membrane-(inbuilt-keyboard)-2529

 

But once I get this thing up and running, dear god someone stop me from trying to load programs off an actual cassette tape again. :)

Link to comment
Share on other sites

8 hours ago, DrVenkman said:

then wait for the replacement ULA to arrive in a few weeks from the other side of the world. 

The seller has already sent me shipping information and an NZ Post tracking number. Should take 3-5 weeks to get here, but it's no biggie. It'll take a week or two for the replacement keyboard membrane too, I'm sure.

Link to comment
Share on other sites

More restoration/repair fun with the Zeddy (Sinclair ZX-81). realized yesterday that the 3.5mm phono jacks used for power input and cassette I/O are REALLY loose and worn out. So today I tightened them by manually pressing down the metal contacts in the plastic housings, which helped a bit. I also rigged up a ghetto 3.5mm power connector so I could run the thing on my bench PSU. You can use an Atari 2600-type PSU, which what I did for my test, but the Sinclair originals were rated at 1.2A, versus an Atari standard 500mA. The Atari PSU worked to verify it would power on the other day but in no way could supply enough to run the 16K external RAM pack.

 

 

 


IMG_0373.thumb.jpg.e361fb50e2fe87eee5322c2ba38a7835.jpg

 

IMG_0374.thumb.jpg.6d31e188493121c9c0984501c9319696.jpg
 

 

 

So I rigged up this adapter, verified the output of the 7805 (4.96V solid) and the Vss and GND of each main chip (all right at 4.94 - 4.95). Then I decided to scope the CLK signal (ULA pin 14 and the video out (ULA pin 16), as well as the clock input to the Z80 (pin 6, IIRC). I seemed to have good voltages and signals and looking at the video signal with my scope free-running, I could even capture what looked like the black "K" cursor of the little Zeddie. Huh. The system SEEMED to be essentially working.

 

 

 


IMG_0368.thumb.jpg.0a3ceb98e89bc93c10f6d2ae64b9bd9a.jpg

 

IMG_0369.thumb.jpg.358866716e6a31efad0afbaed1adffbf.jpg

 

IMG_0366.thumb.jpg.05bd37545cd63a15744babb73fd38193.jpg
 

 

 

So I connected my RF cable and turned on the TV. Shazam! I have no idea why I wasn't getting a visible image the other day except that maybe the power jack was too loose to provide enough solid current to run properly. BTW, I set my bench supply to 9.0V, 1.2A maximum. At boot, the little board only pulls 0.430A initially, dropping to .405 or so after running for a few minutes. The little no-name electrolytic caps need to be replaced so I will probably do that later, maybe today. I also need to rig up a little transistor/resistor composite mod after I've had a day or two to look up good instructions. I've seen people do them right inside the existing RF modulator can and use that same RCA jack for output.

 

 

 


IMG_0370.thumb.jpg.97edd01a7f8901ec4ca039b6366c678f.jpg
 

 

 

I also need to order 3 new 3.5mm mono phono jacks to replace the ones on the board - they're all almost completely worn out at this point. Unfortunately, the best modern equivalents of the original are generally only available from EU and UK sellers, and they're all basically out of stock until the fall. So I will have to put that part of the restoration on the back burner. 

 

 

 


IMG_0372.thumb.jpg.e2cf17ed91e26fa26b670bd869e74556.jpg
 

 

But the last thing I did was dig out the remaining shards of the degraded keyboard membrane plastic connector from the press-fit edge connectors. When my new keyboard arrives from the UK in a week or two, I will clean those connectors really well with Deoxit before I install the new keyboard. They are plain tin and clearly need some TLC.

 

 

 


IMG_0371.thumb.jpg.b0f51ab64806d2a1f27358fc93130e73.jpg
 

 

Link to comment
Share on other sites

I decided to see if the noise in the video was due to the crappy RCA cable I was using for the RF and it sure was. I dug out one of the slender vintage coax cables from one of my AV-modded Atari machines and nearly all that RF noise is gone. There’s still a little bit of sparking nonsense there but once I do a composite mod on this little thing, it should be super sharp and clean. 
 

 


7FB17DA3-D9B5-42B7-BA76-BCB71DEE9AC6.thumb.jpeg.8fa9f79cd86ba5c01b534fc29e886362.jpeg


 

Link to comment
Share on other sites

Now that I have a relatively clean RF signal and know the thing basically works, I went out after work to play with it using my oscilloscope. I also cleaned the PCB edge connector, but they are plain tin covered in decades of grime. I simply could not get the system to boot with the RAM pack installed. Of course, I still don't know if the RAM pack itself works, so I'm really just fooling around. :)

 

I mostly just played around capturing the video signal generated by the ULA chip before it goes into the RF modulator. I first managed to capture a single line down low in a single frame with some of the black pixels for the inverse-video K cursor. Then I let the scope free run for a second or two and scrolled manually through the captured data until I found an entire 10-11 scanlines' worth of data to show the whole cursor, one video line after another. Fun way to help "see" analog video in a different way.

 



IMG_0393.thumb.jpg.0e37011a7eed3556b109faf334b66dd3.jpg

 

IMG_0392.thumb.jpg.d340371502121de925793e055d4ee82a.jpg

 

IMG_0388.thumb.jpg.1866235a9449785c35b4f249cc90c519.jpg

Link to comment
Share on other sites

On 6/26/2021 at 12:58 PM, DrVenkman said:

So like many a proto-nerd in the early 80's, my first "real" computer was a Sinclair ZX-81. No, not the later co-branded Timex Sinclair version, but an actual honest-to-Sir Clive Sinclair. The summer I turned 13 in 1981, I saved up money from parents and grandparents and allowance cash for a couple months and responded to a Sinclair ad in a computer magazine and sent away for a ZX-81 and 16K memory expansion pack.

Very cool the ZX81 was a fantastic machine for learning to program efficiently in BASIC and asm. The kids who got the C64 didn't get these same advantages and were more likely to learn to write sprawling code or not code at all and just play games :) 

 

The semi graphics games look fantastically retro at 64x48 and I'm amazed by this surreal graphics driver that reprograms the ULA to render close approximations of 8-bit pattern combinations from ROM to generate a hi-res display:

Hires Pacman on a real ZX81 - YouTube

 

  • Like 1
Link to comment
Share on other sites

So after another frenetic workday, I decided to recap the Zeddie tonight, and do a Composite mod. The recap went fine - there are only two electrolytics on the board so it's not a big deal.  But I did discover why not to trust no-name 80's caps in Sinclair machines - my meter can measure capacitance. Both of them are WAAAY out of spec - the 1uF cap measured over 1.22uF, and the 22uF cap measured at almost 30.5uF. They've been replaced with modern no-names that are at least in-spec. 

 

 

 


IMG_0396.thumb.jpg.05ca714368d7666c09d0c80d0b9303ec.jpg

 

IMG_0397.thumb.jpg.c7592e4293906524ffca3ec14b517b2e.jpg

 

IMG_0398.thumb.jpg.b30877041b81bc78f19e3a21fbd434ed.jpg

 

IMG_0399.thumb.jpg.47e4b500665d4953043a89c1cc165d03.jpg
 

 

 

But the Composite mod isn't working quite right, and along the way, the well-worn tip of my Hakko FR301 decided to self-destruct. For real - see the pics. It happened when I wasn't having luck figuring out the issue with the Composite mod and decided to remove the RF can entirely so I could pull out the original board, just be sure nothing on that board was connected where I couldn't see or might otherwise be interfering with my mod. Oh, well. I've used this thing a LOT in the last three years. I'll have a replacement this weekend.

 

 

 


IMG_0403.thumb.jpg.514563c7a7ba9b04f93c6f6e945d4113.jpg

 

IMG_0402.thumb.jpg.73f3f758228271302be01bc8abcb4ba3.jpg
 

 

 

Not sure what the heck is up with the mod here - I've duplicated a simple mod I've seen others perform successfully, and even cut it all apart and rebuilt it from scratch but no change. Might post over on ZX Spectrum World to see if others have any ideas. It's a BC337 NPN transistor with the collector connected to +5V, the base connected to the low-amplitude composite signal coming in from the ULA, and the emitter connected to the video connector. Should be easy-peasy. Hmm. 

 

 

 


IMG_0401.jpg
 

 

Link to comment
Share on other sites

Some more research on the Sinclair forums indicate that my ULA chip version doesn’t produce a good “back porch” to the composite video signal and results like I’m getting are expected from a simple composite mod. I have a modern CPLD replacement coming (the vLA81) but it’s coming from New Zealand and may well not show up until August. Until then, found a design for a little breadboard circuit that can be fabricated and fitted inside the existing RF can if I remove the existing PCB. I have all the necessary components except the suggested transistor, which I should be able to get this weekend.

 

D289E391-8371-40E3-970E-A683CB655CAE.thumb.jpeg.a0d32202e43bdd7de6ccde3691d7d4d2.jpeg

 

So, once I get the RF can off the board, I will build the new composite mod circuit and see how it works.

  • Like 1
Link to comment
Share on other sites

Well, yesterday I fabricated the video mod above and nada. Bupkis. Zilch. No sign of video at all. I did see my bench supply pulling the usual ~410 milliamps with the system idling, so I didn't think it was dead, but having removed the RF can, I couldn't see any video to be sure. Rather than beat my head against the problem, I decided to step away from the issue and come back to it today. 

 

After sleeping on it, I went back out to the workbench today and started examining my Zeddie and the video mod from first principles. First thing, before I nuked the RF modulator, the machine booted up and appeared to run normally, albeit without a working keyboard. After removing the modulator, I couldn't really be sure the little box was still working. 


So ... first principles. I plugged it into my bench PSU and then scoped the clock signal on both the ULA and Z80 and verified that much. Then rather than randomly checking address and data lines, I scoped pin 16 of the ULA, which is the raw, unbuffered video signal. I saw what looked like a good video signal, and if I put my scope into free-run and then scrolled through the waveforms, I found the "K" cursor pixels encoded along the scanlines near the bottom of a frame. So that told me the system is still working and I hadn't killed it accidentally with my work.


Next, I went back to my completed little video mod board and compared the schematic with what I'd built - so far it all looked great. So then I checked continuity on all the points I was using for connecting my mod when testing. Ground? Check. +5V? Check. Video? Video? Video? Bueller? Anyone? Nothing. So then - finally! - I looked at the actual traces on the board. Turns out the place I was looking at for video, where the RF modulator gets its signal - isn't directly connected to the video output of the ULA as I thought. Instead, it's connected through a couple of resistors and then through a diode to ground, probably for over-voltage protection against static discharge..


So now that I found a good, direct connection to the ULA, I tacked my video pickup line to the chip-side leg of a diode, then connected THAT signal to my mod. Bam. Victory. ? 


The signal is a little over-driven but I can figure out what's up with that. I might need a 75 ohm resistor on the output to the video jack, or perhaps I can play around with the value of the grounding resistor on the emitter leg of the transistor (currently 100 ohms). Anyway, it works. Once I get the signal level dialed in, I will figure out how to pack it all into the existing RF can, and then mount the can back to the machine.  Unfortunately, the little Dupont headers I installed on the mod board won't fit properly - with wires connected, things will be too tall for the lid of the can to close, and maybe too tall for the entire machine to close. I will have to tack-solder the connections to the pins directly or something.  But those details can wait. I still want to upgrade the internal memory plus replace the existing ULA with the vLA81 replacement. That new chip will allow for a full 32K internal upgrade (the most the machine can support without a bunch of new internal logic) and supposedly has a better, more standard video output. If that output works as advertised, I might be able to remove the new mod entirely and just wire the jack directly to the new ULA. And I will also swap out the 7805 and the three phono jacks whenever I can find some replacements in stock.

 

 

IMG_0449.thumb.jpg.4923c27785057dd83b3dc4901c1bc2aa.jpg

 

IMG_0450.thumb.jpg.a3ee4b388552d4eeb1b7fee8a1c25924.jpg

 

IMG_0452.thumb.jpg.8b1729ea34cf94cc87bc61b2dba43365.jpg

 

IMG_0453.thumb.jpg.a811ee2fd314d0ea6422e0c373fee4bc.jpg

 

IMG_0454.thumb.jpg.62a5ff896eeeeadbecc7b1bafbe543e1.jpg

Link to comment
Share on other sites

Fixed the ghosting issue I was getting! After about half a pint and a couple hours to let my subconscious work, I decided to double-check the values for the resistors on the +5V line feeding the transistor collector and the resistor connecting the emitter/video out to GND. Sure enough, the one on the collector is supposed to be 10 ohm; I had 100 ohm resistors on both. D’oh! Swapped out the incorrect value and problem solved. 
 

 


B4B14801-622D-464B-8B12-3E800333D967.thumb.jpeg.3e89d8861397f9ec3352ec88aaea8dda.jpeg
[/spoiler]

  • Like 1
Link to comment
Share on other sites

The new keyboard membrane arrived today all the way from the UK, only a little over a week after I ordered and despite the long US federal holiday weekend. The quality is excellent in both texture, apparent durability/sturdiness and the quality of the printing. Best of all, the plastic ribbon connector is MUCH higher quality than the original. Interestingly, there are a few minor differences in key labels ("Rubout" on the replacement keyboard versus "Delete" on my the original). I attribute that to the US-market machine I have here, while the seller of these new-made replacements is from the UK. No matter. It will look great on the machine once I finish working on it and get the case properly cleaned, repainted and reassembled.

 



IMG_0475.thumb.jpg.0ac1c41fec6d4367e00644c6caccc075.jpg

 

IMG_0474.thumb.jpg.db3512c3405d2ca0d3dea8d372fc6629.jpg

 

IMG_0476.thumb.jpg.b24143c9832a5cb5b50319542c6bf553.jpg

Link to comment
Share on other sites

I got a few new chips today from Mouser. Picking up a little trick from some YouTube videos, I replaced the original 1982 NEC-made Z80 in my Zeddie with a modern ZiLOG CMOS replacement. Note the big drop in power - the initial startup/in-rush current dropped from about 460 milliamps down to 360 milliamps, and the steady-state idle current drops from about 411 milliamps to about 350 milliamps. The new CPU also runs at darn-near ambient temp - the chip doesn't heat up at all to the touch, at least under idle. By contrast, the NEC original gets reasonably warm. Note that this new chip is less than a year old as I post tonight.

 

 

 


IMG_0479.thumb.jpg.3bbb1a98bb061a787d49151a99b6476c.jpg

 

IMG_0482.thumb.jpg.99d0d46ae44e71add5a11673b6959535.jpg

 

IMG_0483.thumb.jpg.4c5b8698227dc92fca770cfb98fa4d3a.jpg
 

 

 

EDIT: Looking at the current draw after posting, I realized that prior to the chip swap, the idle current draw went up after I did the Composite mod compared to what it was before hand. That's a little bit surprising but I have added a high-speed transistor and several new resistors into the machine. But at the same time I've also removed the old RF modulator. Maybe the RF coil and other modulator-bits naturally draw less power. I mean, I guess they must, right? I have objective evidence via the current being pulled by my bench supply. 

 

Anyway, I also got in some 32K SRAM chips. One of these will go into the Zeddie this weekend or next week. The upgrade to 16K is pretty easy - unsolder the two original 1024 x 4 bit original SRAMs, clear out the helpfully-provided through holes for a new 28-pin socket, install the socket and new chip, then wire in about 3 or 4 wires to some existing diodes. Utilizing the full 32K will require another wire plus some extra address decode logic, OR that new vLA81 chip I have coming already to replace the existing vintage ULA chip. The new chip is CPLD and has the extra address logic coded in. So when that chip arrives from New Zealand in a couple more weeks, I should be able to just swap it in, add the extra wire and then have access to a full 32K internal memory. The vLA81 will also run a lot cooler than the old NMOS-based original and should draw a lot less current too. 

 

 

 


IMG_0485.thumb.jpg.b36863602874746ff893d07c3cfa8edc.jpg

 

IMG_0486.thumb.jpg.8709f36d929e0d324f4b6b27ea68312d.jpg
 

 

Link to comment
Share on other sites

Well, lookie what arrived today all the way from literally the opposite side of the planet (New Zealand), only 13 days after I ordered. This will be installed in my little Zeddie this weekend, replacing the original aging ULA chip. It will enable a proper, correct Composite video output with the expected "back porch" to the Composite signal, and will allow expansion of the system to a full 32K internal memory with one of the new 32K SRAM chips I received the other day, without the need for additional address decoding logic otherwise required.

 



IMG_0489.thumb.jpg.1b6fbc06d29c776616b227c297000428.jpg

 

 

  • Like 1
Link to comment
Share on other sites

This morning’s update on the Zeddie - I received the new replacement ULA chip from NZ last night, the vLA81. I removed it from the antistatic bag and the foam protective layer and found that several of those cylindrical gold pins had gotten bent, probably when I removed the foam without paying close enough attention. These things are a SERIOUS PITA when they bend: they’re fragile, hard to straighten out, and very, very apt to simply breaking off at the bend if that happens. 

 



94BAA8DA-9BF1-4CCC-B430-0379BACCEBD6.thumb.jpeg.d88b60487b56ba4e1e80f0d25564c6d8.jpeg

However, when I removed the old ULA chip, I realized the existing socket was a terrible-quality single-wipe and was not in great shape. Fortunately, I have DIP-40 machine pin sockets in my parts stock. So I decided to remove the existing crappy socket and use the machine pin socket instead.  I veeeeerrry slowy and carefully got the bent pins more or less into shape, then fitted the new vLA81 into the machine pin socket and proceeded to remove the existing socket. 

 

Amusingly - or not - several socket legs literally broke off on the first first pass through with my Hakko. When I removed the plastic shell, a few of the wipes remained, falling out of the old, degraded plastic of socket almost like it wasn’t there. A drop of flux on each one, a touch of the iron and they all lifted out with tweezers. Next I soldered in the new socket, vLA81 already installed. I know it’s not good practice to do this but having gotten the chip into the socket without breaking any of those bent pins, I have no intention of removing it again!

 



8EBFDF4A-F6DA-4457-A2A8-29E8B4D6DFB5.thumb.jpeg.f3e05fe61e7ffc8311c497981b7445d5.jpeg

 

F33287EA-C9AE-491D-A859-23F8400F4F88.thumb.jpeg.37d2d9a0f54ee1de8d3a8db78526b493.jpeg

And now that it’s installed, note two things: first, the perfect, beautiful composite image; and second, the power draw of the machine has dropped a further 100 milliamps! This is 39% less power - and thus less heat! - the machine was drawing before I swapped the NMOS Z80 chip for the modern CMOS version, and now the vintage Ferranti-made ULA for the CPLD replacement.

 



F3B09FAB-8B63-45DA-9457-7FE2F1826A77.thumb.jpeg.fc701d5cc59b925cad10f9a2a38dc324.jpeg

 

C673C35F-50B2-434E-8EB6-4536731E65C6.thumb.jpeg.4b71ab98c8514ce60863405509e72d53.jpeg

 

8C59EDD9-6E67-479D-9B02-60D507BB15FB.thumb.jpeg.e6805b8e1cf35b441a04e75d59d41b78.jpeg

 

Next up will be replacement of the two existing 512-bit SRAMs for a new 32K SRAM; combined with the vLA’s extra address decode logic compared to the original, this will effectively max out the machine’s available memory and is twice what is needed for most vintage software for the machine.


After I get that done, I will replace the three worn and damaged phono sockets, swap out the 7805 for a new Traco replacement and then start work on the case - removal of the old keyboard, cleaning and repainting of the lettering, and installation of the new keyboard. I have the new sockets inbound from the UK, and I will order the new switching regulator next week. 

Link to comment
Share on other sites

I was killing time early this afternoon with some restless energy. We had to drive about 40 minutes to the airport in the late afternoon to pick up our son who flew in across the country to see us for the first time in about 18 months due to COVID. So I found myself trying to spend about an hour doing something besides spinning my wheels until we had to leave. 

 

I decided to go out into the garage and do two things. First, since I have the new vLA81 installed and working, there's no need for the "breadboard-esque composite wiring I've been using for tests, so I needed to finish the install and put it into the modulator can, and then get the modulator can soldered back to the board; and second, see if the new keyboard I ordered from the UK is working.

 

Installing my little protoboard into the can was easy - I built the composite circuit to fit a piece of board I had cut specifically to fit the inside of the can. However, when I fabbed the circuit I had intended to connect +5V, GND and Video to the circuit via the DuPont pins I installed on the protoboard. Unfortunately, with female DuPont connectors, the wires stick up too high to fit into the height of the can. D'oh! So instead I just soldered the wires directly to the male pins, then soldered a Composite output wire from the output pin to the center pin of the existing RF can RCA jack previously used for the RF modulator. Once all that was done, I tucked the wires down into the can and closed it up. After testing that everything worked, I soldered the can back to the PCB. 

 

 

 


IMG_0508.thumb.jpg.a6d5486a85a88e5150ea3f3df082410d.jpg
 

 

 

Then I tested my new keyboard. After installing the new membrane connectors into the board connectors, I was able to type in a simple BASIC program. That went fine and the keyboard works great. Even better, the computer itself runs the program as expected so it too is working just fine. :)

 

 

 


IMG_0505.thumb.jpg.c0295b3410ff0e6a9913b1ab9a029ad7.jpg

IMG_0507.thumb.jpg.2c624e3811cf2b3e6d6e8a1e247bced1.jpg

IMG_0506.thumb.jpg.771227bffd4fc2ae0a0e955ebd155254.jpg

 

 

I am really quite pleased with the combination of the nice, clean signal from the new vLA81 chip combined with the Composite mod board I built. 

 

 

 


IMG_0510.thumb.jpg.d029361d07c5aa3208e5e0be0ebd48b1.jpg

 

  • Like 1
Link to comment
Share on other sites

Late to the party, it seems, but this is very interesting to me.  While I've been aware of the Timex / Sinclair tie-up since I was a kid, reading that Sinclair sold the ZX81 under their own name in the US is news to me - I'd always thought that the TS1000 was as close as they came in that regard.

 

Any recommendations for history resources?  While I'd like to learn as much as possible about the Sinclair history behind the machine, I'm also interested in any technical differences between it, a PAL ZX81, and the TS1000.

  • Like 1
Link to comment
Share on other sites

52 minutes ago, x=usr(1536) said:

Late to the party, it seems, but this is very interesting to me.  While I've been aware of the Timex / Sinclair tie-up since I was a kid, reading that Sinclair sold the ZX81 under their own name in the US is news to me - I'd always thought that the TS1000 was as close as they came in that regard.

 

Any recommendations for history resources?  While I'd like to learn as much as possible about the Sinclair history behind the machine, I'm also interested in any technical differences between it, a PAL ZX81, and the TS1000.

I think the TS1000 is identical to an NTSC/US model ZX81 with the exception that the TS1000 came with 2K of RAM internally versus the ZX81's 1K. Other than that, they should be the same as any ZX81. There are two ZX81 PCB board revisions, the Issue 1 (pale green with hand-routed traces) and the Issue 3 (red PCB and machine-routed traces). There are also a couple different ULA chip versions - the earlier one like mine doesn't produce a "back porch" in the composite signal, so many modern TV's can't sync to the black level properly; the later ones create a proper "back porch" in the video. All three main ULA revisions run hot and the chips are prone to age/heat related failure. There are also a couple ROM revisions, mostly to fix some bugs in the BASIC interpreter I gather.

 

Really, the systems are pretty simple even for 8-bit computers.  The best resource for info about them I've found are the Sinclair ZX World Forums. They're not terribly active by AtariAge standards, but there's a LOT of great stuff there, including the Composite circuit schematic I built to use in this machine. 

  • Like 1
Link to comment
Share on other sites

@DrVenkman: thanks for that; the clarification and additional detail is appreciated.  You've prompted me to do some additional digging while I'm stuck at work overnight ?

 

Although I never had any Sinclair machines growing up, there was at least one PAL ZX81 in the extended family, and a couple of other relatives moved on to the ZX Spectrum at different times.  I was in Atariland at the time, so really didn't pay a ton of attention to what was being done with either the Spectrum or the ZX8* machines.

 

Having said that, they certainly are machines that I have a definite admiration for, however.  While they may have been built to a (low) price, it's absolutely staggering what people were and still are able to do with and get out of them - particularly given some of the limitations of the architecture.  That's not to knock the architecture: it is what it is, and very clever in a lot of ways.  Its simplicity is, in many ways, its greatest strength, and goes some distance towards explaining the tremendous influence it (and the later Spectrum series) had on computing in a rather large portion of the world.

  • Like 1
Link to comment
Share on other sites

I totally agree. The "Zeddie" was my only Sinclair machine and the only one I hold any nostalgia for. But when I started seeing adds for the later Spectrum models, that got my attention and I wanted one, even though I never got one. The interest in the "Speccy" led to some interest in the VIC-20 (which I also never owned) and ultimately to the Atari 400/800, which ultimately won out. I sold my ZX81 to help fund the purchase of my Atari 400 and never looked back until recent years. 

 

I would still like to ultimately find a Speccy and a VIC-20 to explore those roads not taken back in early/mid 1982, but for now, I'm getting a lot of satisfaction in restoring and upgrading this little ZX81 and seeing where it leads. 

  • Like 1
Link to comment
Share on other sites

6 hours ago, wierd_w said:

This is just idle curiosity-- Is the ZX81 capable of artifact color?

 

Since this is NTSC, manipulation of the pulse timing could trick even modern monitors into thinking there is a color burst signal.

I don’t think so, at least with stock hardware. The video signal is generated by ULA chip, which I don’t think is capable of manipulating the signal like that. I guess conceivably someone could create something like the vLA81 chip I put into mine that is ALSO capable of additional video modes, then a modified ROM to take advantage of those features. But at that point you might as well track down a Spectrum, I would think.

Link to comment
Share on other sites

So … it’s been a week … 

 

Last Sunday before it got too hot, I went out to my garage workbench and set out to remove the two SRAMs. Given the ease with which I have done many similar removals on other makes of boards, and how straightforward it was to remove the ULA socket for the vLA81 replacement, I simply added some fresh solder and used my Hakko to clear out the holes. I did my usual “click test” with a small screwdriver to verify that the chip legs were free, then started wiggling the first chip free. And here’s where things went awry. 

 

Turns out the legs were indeed free, but … some of the fresh solder I added to aid with desoldering had evidently flowed up UNDER the chips and adhered to the uncoated traces on the top half of the board. When I removed the first chip, I lifted two traces and pads. Damn it. Quite frustrated, for the second chip I didn’t even try to lift the chip off the board. Instead, I cut off each leg at the chip shoulder as close to the chip package as possible. This shows very clearly what I mean about the solder having flowed up and under the chips. At that point, irritated with both myself and the absurdly cheap way these PCBs were made, I set everything aside for a few days. I figured I was likely to have to bodge those damaged traces too, so I ordered some 30ga Kynar wire and put the whole thing out of my mind for a few days.

 

 


4C301286-6FAE-4C0F-9334-E4FE006178B6.thumb.jpeg.0f0d5f54cf2316f4510578f082669a07.jpeg

 

C6771BD8-505B-45AC-8D1D-91A758B4B74E.thumb.jpeg.a01546b710e5c688ebec1ffb635a4bfa.jpeg
 

 

 

Yesterday I went out to the garage and reassessed things. Fortunately, after trimming back the torn traces, clearing out the pins from the second SRAM socket and verifying those pads and traces, and giving the board a very basic clean up to remove the worst of the carnage, things looked much better. I then printed off a copy of the schematic and called it a day.



B300BA63-238B-4957-8FC4-732FA1A04719.thumb.jpeg.ed58d7676bef2c76de32e4333bff769b.jpeg

 

4DE6EA47-5375-4FC0-9A57-84F10D0C6D30.thumb.jpeg.6984052d6da111e36077322f5472137d.jpeg

 

Today, armed with the schematic, I traced out the connections from those damaged pads and verified that one of them very definitely needed to be replaced. The missing pad and trace for pin 2 of the now-removed SRAM runs over to a via, and then under the board, there’s a trace down to a diode on the address line. The missing pad and trace for pin 3 of that same SRAM runs over to a via, and under the board there’s a trace to something else but I can’t quite follow the trace, and blinding checking for continuity around the board didn’t help. Since I have the Kynar wire now, I simply bodged both traces just in case. I could have done these repairs under the board where they didn’t show, but since I’m not ever planning to take this machine back to stock, and frankly NOTHING important about it will be stock anyway except the ROM and case plastics, I didn’t bother. The Kynar is so thin it fits under the new SRAM socket with no problem. The next step will be installing the new SRAM with a few legs bent out, and some jumper wires run to diodes under the chip for the necessary address lines for an internal 16K upgrade. A full 32K upgrade will require one more address line jumper over to the new vLA81 chip, but I plan on doing that last.

 

 


2BAE07E7-A556-474A-B825-2E43F81CE1D1.thumb.jpeg.ad0d1b750d81e226613d89ade91a2e06.jpeg

 

5C921882-E977-41C9-8A3F-8F6F7254FEC3.thumb.jpeg.5907adbc42b0f42e78c12327b3e2ff8c.jpeg

 

F1C2A672-DD81-49BA-8EFF-2F8E662636DC.thumb.jpeg.f379372fbc1fdb205bc32ad06aaece50.jpeg
 

 

 

During the past week I also received a set of new 3.5mm phono jack PCB mount connectors to replace the cracked and broken originals. These will be installed after the memory upgrade is complete and verified working. I find it interesting that these new connectors appear physically identical to the 40 year old originals, minus the age related wear and breakage of course. And still made in England too.

 

 


DA6A7374-EC5A-45F7-B1DA-EF1624F4F837.thumb.jpeg.795e32b843abbefa47304f139fc9e577.jpeg

 

C3F4BF7A-D765-4B6F-BCB9-03250D41EB44.thumb.jpeg.c53bbea47f82afbe6a3d302dfc352b79.jpeg
 

 

  • Like 1
Link to comment
Share on other sites

So as a post-script to the SRAM removal/repair bodge story, as a lesson to anyone who does this in the future and for anyone who has never worked on a Sinclair before, if I ever do this again, here’s what to do differently. First, if you are intent on saving the existing SRAMs, probably don’t bother adding any fresh solder - it will just flow up and under the chips and stick to those traces. Instead, add liquid or gel flux liberally and THEN desolder each pin. Second, if the chips don’t essentially fall out of the board, don’t try desoldering again. Tape off adjacent componts and traces with Kapton or generic polyamide tape and go to gentle heat, maybe low-flow 250C air and work over the area, using tweezers or a small jeweler’s screwdriver to verify the legs are free. That will be slower but since you haven’t added any solder to flow up under the chip and stick, it’s less likely to damage anything. 

 

Alternately, the safest way to save the PCB is simply cut the chips out at the shoulders, then desolder each pin with flux, a hot iron and tweezers, then use a solder sucker, wick, or desoldering iron to clear each hole. However, given the trace runs used, it’s not even necessary to clear each hole. Just remove each leg and call it a day.

  • Like 1
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...