Sinclair ZX-81 -Inspection, Cleaning, Possible Repair

[+DrVenkman]

Had about 90 minutes this afternoon and the itch to get going on the Zeddie now that I’ve repaired the traces and installed the new SRAM socket. For the “easy 16K” mod, I cleared out the through holes for “LK2” (Link 2 -  Sinclair terminology for a jumper location) then wired that up. That puts pin 21 on a 28-pin chip onto the A10 line.  Next I soldered wires to the top ends of diodes D1, D3 and D5, (Address lines A11, A12 and A13) and soldered them to the lifted legs for pins 1, 23 and 26 of the SRAM. After cleaning the flux and any stray solder flakes off the board, I tried a power test. The result was a white screen for several seconds and then the expected K cursor. Fantastic! The several second delay on startup is normal an expanded memory computer - the OS ROM and ULA chip do a short test to determine the range of memory addresses available. A stock 1K startup is almost instant so the delay indicates that the expansion is working. I need to connect the new keyboard again and type in a little test program in BASIC to verify it 100% but all indications are good.

 

And look how ridiculously low the power draw is now with 16K compared to the original 1K SRAMs, NMOS vintage Z80 and NMOS original ULA chip. The system will run much, much cooler now even without replacing the stock 7805 (which I still mean to do eventually).

 



 



 



 

 

Very, very pleased with this little project so far.  

 

[+DrVenkman]

I connected my new keyboard temporarily today to do run a little test. Typing PRINT PEEK 16388+256*PEEK 16389 at the BASIC "K" prompt gives the top of available memory. The way things are wired currently, the first 16K is set aside for ROM space, even though the system ROM + BASIC is only 8K.

 

So everything does indeed appear to be working properly for 16K.

 

After doing some reading today, to take it to the next level and use the full 32K SRAM capacity, I need to do three more things: step one, left pin 2 of the SRAM from the socket; as it is, that location is wired to Vcc so it's tied logic-high and effectively useless. So after the pin is out of the socket, step two: tie that now-free address line on the SRAM chip to an otherwise unused address line on the mainboard. And finally, step three: run a jumper from the SRAM pin 26 (A15) over to pin 35 of the vLA81 replacement ULA chip. I understand I can leave it tied to the diode as it is and just run a connection under the board between the diode and ULA socket pin 35 but that's still a bit fuzzy. I'll figure it out, I'm sure. 

 

[+DrVenkman]

Today I had a little time in the morning so I decided to finish the 32K mod. As indicated above, I lifted pin 2 of the SRAM and connected it over to address line A14 line on diode D7, then connected pin 35 of the ULA socket over to address line A13 at diode D5. After verifying my connections, I did a power test and voila.  

 

[+x=usr(1536)]

Nicely done!

 

Were you ever able to work out what that strap on the underside of the PCB is for?  I was wondering if it may have been meant to contact RF shielding that would originally have been in the lower half of the case.

[+DrVenkman]

That strap is for grounding to the RF shield as part of the FCC certification process for North American models. UK/Europe models don't have them. On American market machines, the inside of the case top and bottom are coated with aluminized paint or something; this strap is soldered to the board at the edges, and has a part that bends up and contacts the case top as well. I keep meaning to desolder it and get rid of it - since I've done a composite mod, the machine isn't making any modulated RF, and the case will be well-shielded anyway once I get it all put back together. 

 

I have run into a secondary issue now - the machine is crashing randomly anywhere from 40 seconds to several minutes after power up. I'm not sure why - could be the vintage VR is dying or dropping out; might be a stray solder flake near a chip leg or something; or could be a dodgy socket; the stock sockets are REALLY cheap and poor-quality single-wipe sockets. I've replaced the ULA socket already but the CPU and ROM sockets are original. The SRAM had to have a new socket anyway, so that should be good. I am going to bathe the board in iso and rinse it well, then spray it with canned air; if the problem persists, I'll start by replacing the ROM socket and see if the problem persists. 

[+DrVenkman]

After work today, the first thing I did was desolder that annoying ground strap. Finally, an almost-level board to work on.  The channel select switch on US models is still soldered to the bottom edge of one end of the board near the voltage regulator; that will go too eventually - I had removed it a couple weeks ago but then was getting no picture at all from my composite mod. In fact, if I move the switch from one position to the other, my composite mod goes dark. Unfortunately, the best schematic available is from a modern UK source and it doesn't really go into how or why that might be the case - the channel select switch either grounds or feeds power to a line that runs to the RF modulator, which then causes the channel output to change [FOR MAGICAL RF REASONS I DON'T GROK].  So I need to remove the switch, figure out which two pins are connected in the "working" position, and then just add a jumper wire to those two pads on the mainboard. But until then, the switch stays.

 

The second thing I did was the first step in seeing about the flakiness I mentioned last night. At one point Sunday, I powered the system and then started pressing chips - when I got to the ROM socket, the system started flaking out and then crashed. As a temporary measure, I removed the chip and cleaned the legs with Deoxit, although they were still quite shiny and looked good. The socket wipes small and gray, weathered-looking tin. So today, I removed the old socket and replaced it with a modern gold-plated precision socket. If this doesn't cure the flakiness, I'll do the same for the CPU socket. In any case, it's nice to see that my debacle with the SRAM chip didn't portend a complete failure of my technical skills.

 

 

 

 



 



 


 

 

 

As a next step, I will replace those three 3.5mm phono jacks. Today I finally also remembered to buy some modern Traco TSR 1-2450 switching voltage regulators. One is going into this little thing, another will be going into my daily driver 7800, and I've got a third one for whatever future project might need it. The vintage 7805 inside still works and puts out 4.97V but unless I run a clip lead to it full time, I can't tell if it's starting to get unstable as it heats up, contributing to the flakiness. If it flakes for even half a second, that's enough to cause a crash. Plus, any momentary drop in supply voltage may not even be the fault of the regulator - if there's a solder flake or almost-bridge somewhere I've missed, that could cause a momentary short, pulling down the supply voltage very briefly. But anyway, those will be here in a few days and then the 7805 will get replaced. That will also have the secondary benefit of removing the need for the heat sink attached to the front of the board and will create a much cooler environment inside the confines of that tiny case.

 

[+DrVenkman]

After work this afternoon I continued tinkering with my "Zeddy of Thesus", as someone in a ZX81 Facebook referred to it.

 

The very first thing I did was remove that annoying channel select switch and jumper the location so it will work with my composite mod. I used a cut-off component leg and soldered into the appropriate pads to match the "working" switch position. 

 

 

As I was doing so and running my power test to verify the jumper was in the correct place and everything still worked, I may have gotten a clue to at least some of the instability I've been seeing. I accidentally jiggled the power connector and the screen got started to get screwy. I manipulated the power cable and sure enough the effect was repeatable. So I killed the power and checked the integrity of the connector. Ironically, of the three 3.5mm jacks on the board, the power input was the most physically intact, with only a chunk of the rear plastic housing broken - the other two had substantially more plastic cracked off and one of them had the entire bottom part cracked in two, allowing the metal contacts a LOT of movement. But just since I've been messing with it these past few weeks, the power jack has gotten looser and the plastic seems to have flexed or deformed a bit without the full support of an intact rear piece.

 

So of course, the next thing I did was replace all three. I removed most of the solder with my Hakko but the width of the contact prongs for these connectors is apparently exactly the same diameter as the through holes, and there was just enough solder adhered along the edges to create almost a fillet and they wouldn't simply fall out. So I bent up the metal enough to remove the plastic parts, then added flux and removed the contacts with a hot iron and tweezers. Even with the last bits of solder good and melted, the physical friction created by the tight fit made them a bit tough to remove, but I got them out. To ensure there was no old residual solder, I filled all the holes with fresh new solder and then cleaned them out neatly with the Hakko. That's when I realized exactly how tightly the contact tabs fit through the holes. That's good for mechanical stability but just in case, I am going to order some 3.5mm flat hex or knurled nuts to screw onto the phono plugs. When tightened down, they will add a bit of mechanical support by pressing back against the edge of the mainboard. I should have ordered them yesterday when I ordered the new switching voltage regulators - gives me an excuse to order a few more things from Digi-Key next week, I guess. 

 

 



 

 

Ignore the weird pattern in the picture - some weird interaction of the LCD of the TV and the sensor on my phone. Anyway, it's far too hot this time of year in my garage workspace to spend more time out there than I did, so after doing a simple power test to verify things still work, I'm done. Any kind of real stability tests have to wait for a break in the heat! 

[Ed in SoDak]

Excellent work Dr. Venkman! I'm enjoying following your Timex restoration project.

-Ed

[Bee]

My first computer is a ST1000 with an aftermarket full stroke keyboard.  I got mine in about the same time period.  Rather than reviving mine I transplanted my CPU in a Arduinio add on to get it back running basic again.  That led me to the Z80 membership card and my first retro plunge into CPM.  I also have a the 6502 badge.  These are kind of fun -

 

http://www.sunrise-ev.com/z80.htm

 

http://www.sunrise-ev.com/6502.htm

 

Love what you are doing on the Zeddy.

 

Thank you

[+DrVenkman]

6 hours ago, Bee said:

My first computer is a ST1000 with an aftermarket full stroke keyboard.  I got mine in about the same time period.  Rather than reviving mine I transplanted my CPU in a Arduinio add on to get it back running basic again.  That led me to the Z80 membership card and my first retro plunge into CPM.  I also have a the 6502 badge.  These are kind of fun -

 

http://www.sunrise-ev.com/z80.htm

 

http://www.sunrise-ev.com/6502.htm

 

Love what you are doing on the Zeddy.

 

Thank you

Those “membership cards” have always looked fun. Last summer, I bought these:

 

https://www.tindie.com/products/8bitforce/retroshield-6502-for-arduino-mega-2/

https://www.tindie.com/products/8bitforce/retroshield-z80-for-arduino-mega/

 

They use an Arduino Mega to provide ROM space and interface support for the CPU. They can only run at a couple hundred kilohertz but they were fun to build, cheap, and a lot of fun fooling around with them.

[Bee]

13 hours ago, DrVenkman said:

Those “membership cards” have always looked fun. Last summer, I bought these:

 

https://www.tindie.com/products/8bitforce/retroshield-6502-for-arduino-mega-2/

https://www.tindie.com/products/8bitforce/retroshield-z80-for-arduino-mega/

 

They use an Arduino Mega to provide ROM space and interface support for the CPU. They can only run at a couple hundred kilohertz but they were fun to build, cheap, and a lot of fun fooling around with them.

The Reto Z80 is what I used to verify my Z80 was in working order and execute some basic. 

 

Edited July 23, 2021 by Bee

[+x=usr(1536)]

On 7/22/2021 at 12:15 AM, Bee said:

http://www.sunrise-ev.com/z80.htm

Thank you for posting this - I am a huge fan of Gilbert Shelton, who did the cover art for that book.  Now that I know that it exists, I must hunt down a copy ?

[+DrVenkman]

Close to finishing up the internal mods of my little project Zeddy. I rebuilt the composite mod PCB by removing those (frankly useless) male header pins I used initially and soldered solid-core wire for the +5V supply and Video inputs, as well as for the Composite output to the center-pin of the RCA jack. I didn’t bother with a separate video ground connection - since I’m using the former RF modulator case, the RCA jack is riveted into the can walls. Since the can is soldered to the board’s ground plane, that’s the best ground return I could ask for the circuit. 

 

But more importantly, I finally replaced the 40 year old 7805 voltage regulator with a brand new Traco Power TSR 1-2450 switching regulator. With the system powered up, the old 7805 was providing 4.92 - 4.97 volts. That’s completely within spec but it wasn’t as consistent from measurement to measurement as I would have expected. That may have been a sign of impending failure and may also have been contributing to some of the flakiness I was seeing last weekend - there might have been some dropouts or transients that I didn’t catch. 

 

 

 



 


 

 

 

As we can see, the new regulator is rock steady at 5.02V.  I did a several minute power-on test and didn’t see any squirrelly behavior. Still too hot to do a lot of stuff in the garage and I really need to get the case restored and new keyboard installed properly for more real tests. But hey, check out that power draw compared to what I started with (about 420 milliamps on startup, dropping to 411 or so during operation). That’s about a 71% drop in power draw for the current configuration compared to stock.

 

 

 



 


 

 

[+DrVenkman]

I decided to remove the old keyboard membrane today. Based on some stuff I've seen online, I decided to heat the keyboard with my hot air station. Using the largest nozzle, I set the temp at 100℃ and medium flow and worked over the membrane for about five minutes. When I peeled off the keyboard, it literally delaminated. 

 



 

 

If that looks like cellophane film, that's because it is. Or close enough not to matter. Evidently that's either the bottom of the keyboard membrane, or the top of the adhesive ... but in any case, it came up off and left a LOT of adhesive residue. I tested it with a thumbnail and it while it came up, it would really, really difficult to get it out off mechanically this way.

 

So I went old school instead. I used naphtha (lighter fluid) poured right over the residue and scrubbed the stuff clean from the board with a paper towel. It took about 30 minutes to get it all, plus a final wipe down with isopropyl alcohol, but I got all the residue off. 

 

 

 

 

 

After I had cleaned up the case top, I noticed a small bit of plastic. Turns out one of the split screw posts picked today to break apart. Unfortunately I can't find my two-part epoxy, which is what I really want to use to fix these cracks. So for now, and purely temporarily, I tacked the pieces back together with superglue and filled in the gaps as best I can. I will get some of my favorite epoxy next week - the stuff I like for internal repairs is structure automotive body work epoxy. It's sandable, paintable, and used to repair body panels on cars. I used this stuff to repair a mounting post on the 1050 disk drive case I salvaged to build my 1088XLD and it works GREAT. I will mix some of it up, then apply it around and into the outside of each cracked screw post. Since it's an internal part, I don't even need to worry about sanding it back down or anything. And then once the screwposts are structurally sound again, when it comes time to fully reassemble the machine, I am going to fill each post hole with a small amount of E6000 craft adhesive. This was a recommendation of a friend of mine who is a PhD chemist (university research professor) who also does 3D printing semi-professionally on the side. This stuff SHOULD take the thread shape of the screws and also adhere to the inside of the broken posts, but still release if I ever need to get the machine apart again.  But that's a project for next week at the earliest. While I'm out, I need to go to a craft store and get some off-red semi-gloss enamel paint to try to match the original logo paint.

 

 

 

 

 

So anyway, after the screwpost was tacked back together, I went ahead and applied the new keyboard membrane. It looks and fits great. There's a little bit of fibrous-looking stuff along the edge of the membrane where it was cut apart at the factory but it's not really noticeable much with the naked eye. You can see it under the bright LEDs of my workbench but not in an ordinary room light.

 

 

 


 

 

 

So up next will be some epoxy for the screw posts, some craft paint for the logo, and finally, new rubber feet for the case bottom. 

 

 

 

[Ed in SoDak]

Almost there!

 

After ruining countless plastic screw posts, I learned a tip not too long ago. Someone posted to turn the screw in reverse till you feel a small clunk. That's where it aligns with the original threads. Now tighten and you'll notice how much smoother it goes in and tightens. Don't overdo the torque. Works a treat!

-Ed

[+DrVenkman]

On 7/26/2021 at 1:56 PM, Ed in SoDak said:

Almost there!

 

After ruining countless plastic screw posts, I learned a tip not too long ago. Someone posted to turn the screw in reverse till you feel a small clunk. That's where it aligns with the original threads. Now tighten and you'll notice how much smoother it goes in and tightens. Don't overdo the torque. Works a treat!

-Ed

Yep, I learned that trick years ago. After I get the epoxy (Saturday), I'll be repairing the screw post cracks - all three of the back posts were cracked to at least some degree or another, though the center post there is the only one that had any part break off completely.  Once that dries to the touch, I'm going to repaint the red "ZX81" logo on the top. After THAT dries, the entire case will get a gentle wipe with plastic restorer, then the system will be reassembled and will HOPEFULLY be good for another few decades.

[+DrVenkman]

Well .. erm. Hmm. I mean … I may or may not have just bought a Timex Sinclair TS1000 + 16K Memorex Memopack. ?

 

These little machines are addictive. ?

[Ed in SoDak]

You won't get any grief from me. I could be your enabler, lol. At least it's a compact addiction.

 

Got my first one in '84. After typing in do-nothing graphic displays titled Wow Star and such and a bunch of games from the books and magazines, I wrote a timer program for my darkroom. That became an essential tool which meant I had to have backup machines so I wouldn't be without my Timex Timer. If one died, just grab another one. I kept the program running on two TS-1000s in case one crashed mid-process. I have a 16k Memotech too!

 

My most recent fling was to buy a non-working TS-1500 and make it functional. Probably my only Timex I tried to restore as you do, keeping original appearance and function. All my others were total hack jobs. The 1500 is what the original should have been, better keyboard and 16k built-in.

-Ed

Edited July 29, 2021 by Ed in SoDak
added pic

[+DrVenkman]

3 hours ago, Ed in SoDak said:

My most recent fling was to buy a non-working TS-1500 and make it functional. Probably my only Timex I tried to restore as you do, keeping original appearance and function. All my others were total hack jobs. The 1500 is what the original should have been, better keyboard and 16k built-in.

Yeah, the 1500 is a neat little machine.  I have no idea if this new TS1000 will work or not but if not, I doubt I'll have any trouble getting it to run. Hell, even if it's a completely bare board inside, all I need to buy is a ROM. And for that matter, I can wire up an EEPROM adapter and program a new one. I have the working ULA I took out of the Zeddy, I have spare 32K SRAMs for a simple 16K internal mod, and I've got the Z80 I pulled from the Zeddy as well. If the modulator is dead or shot, I can build another little composite mod like I've done already too. Anyway, it will be fun to figure out.

 

Hopefully this will be the last of these I decide to buy. I'm afraid I'm going to want to start perusing Spectrums or something next ... What I really need to do is build a TZXduino or buy one somewhere, not buy more Zeddies.

[+DrVenkman]

eBay is a hell of an enabler. About 48 hours after I bought this, it arrived at my doorstep. I didn't even pay for fancy shipping - just straight-up USPS First Class mail. Say what you will about them, but it's hard to beat $6 to my door from Houston to the suburbs of Nashville in two days.  It's in gorgeous condition too - the tiniest few scuffs but the case and keyboard look great.  And so too does that funky 1982-era Memorex "Memopak" 16K memory expansion. I was always jealous of those as a kid - compared to the blocky, clunky 16K Sinclair expansion, the Memopak was far sexier. 

 



 

 

But here's the kicker: does it work?!?

 



 



 

 

Heck yeah! The keyboard works, the processor works, and it's pulling the expected amount of power for a stock system with NMOS chips. But does the memory expansion work too?

 

Yep. The top of addressable memory is 32,768. Subtracting the 16K set aside for ROM in the memory map, that leaves 16K visible to the computer. Fantastic.  

 

I'm not entirely sure if I want to take this one apart immediately or not. I will PROBABLY leave it stock, though I may well replace the 7805 with a Traco replacement to reduce power draw and internal heat, and I might also replace the Z80 with another CMOS Z84 to further reduce the internal heat but just replacing the 7805 will do a lot in that regard. The RF output is also fairly crummy here - lots of ghosting and smearing even through a shielded coax cable to the antenna input of the TV. I might remove the RF can and replace the internals with another composite mod like I did on the Zeddy, but not 100% sure about that. 

 

[+DrVenkman]

One of the last steps in the restoration of my Zeddy - I got some 2-part automotive epoxy today and used it to repair and reinforce the cracked screw posts. The Zeddy has five screws that hold the case halves together, and two others inside that hold the system PCB to the top of the case. Of the seven, five of them were cracked to one degree or another. So all five got liberal coatings of epoxy, then allowed to dry. The stuff is dry to touch in 15 minutes and hardens in 30. So after letting it harden all day, tonight I did a quick test fit. 

 



 

I started to worry that the Traco voltage regulator might interfere with fitting the top case and man is it close. But it juuuuuust fits. Whew. ? Traco makes the same part with a 90 degree flat mounting option, and even the base version has fairly long leads. If worse came to worse I could desolder this one and then install one of my others, leave the leads long enough to bend and let the VR have more clearance by laying it over the PCB part of the old heat sink location. But that wasn't necessary. It all fits, if only just.

The even better news is that of the five cracked screw posts, once repaired with the epoxy, four of them five hold screws just fine again. The fifth one is quite internally stripped. For now the screw is just held in place by its own friction really. Once I've got the case repainted and ready to close up semi-permanently, I will try some E6000 craft glue dripped down into the hole and then insert the screw. With luck the glue will adhere to the existing plastic and form itself around the screw  as it dries, creating new threads. So long as I don't have to over-torque it, it should work fine. And realistically, I probably won't need to get back inside the machine again ever anyway.

 

 

The very last steps will be next week - I will repaint the logo on the case, then replace the missing rubber feet. And this summer project will come to a close, I think. 

[+DrVenkman]

On a temporary break from work for a day or so, I decided to try another functionality test of my newly-upgraded and restored Zeddy. I dug around in my bin of miscellaneous cables, dongles and adapters and came up with this unholy mess: a Lightning-to-3.5mm stereo audio adapter to plug into my iPhone, and a mono 3.5mm audio cable to connect to that to the “EAR” jack of the ZX81.

 

I found a 16K game to test online (Lunar Lander) but all the games I found in the archive were so-called “.p files”. I’m not well-versed in Sinclair-speak but I gather these are generally intended to loading via something like an Arduino or PIC-based device from an SD card. Well, I have nothing like that handy, but I have plenty of things that can play .WAV audio files. Fortunately, I found a little Windows utility online that is literally drag-and-drop with no interface at all. Put the .p file somewhere handy, grab it and drop onto the program icon (which isn’t even running) and the program launches, converts the file back to .WAV, then quits transparently, all in about a second.  

 

At that point, I loaded VLC Player on my iPhone and enabled WiFI file loading, pointed my desktop PC at the server address running on the iPhone, copied the file over to my phone, and I was ready to give it a try. After some experimentation with the volume levels and the syntax of game loading (about 80% volume, command LOAD “LUNARLAN” ) I was in business. 2:52 seconds later, done!

 



 

 

This game is HARSH, man. ?

 

So yay! “Tape” loading works. That means there’s an excellent chance tape saving works too, though I need to experiment with various adapters and stuff for the mic input, plus finding a program that will save raw .WAV files. Not a high priority really, but it would be good to check for the sake of completeness. 

 

So the last step will be to repaint the logo, then install some new rubber feet. Not sure it’s worth ordering “correct” new feet from the UK but I might anyway. Eventually I will take apart the TS1000 for inspection at least, and there’s a good chance when I do that the keyboard membrane will tear. So in a couple weeks, I’ll probably buy a set of new feet, plus a TS1000-version keyboard just in case

 

[+DrVenkman]

And of course since I had the free time, I tested the TS100. It works, too.

 

RF is really bad on this machine. Ugh. 

[+DrVenkman]

Since I got program loading to work, I have discovered 3D Monster Maze. Holy crap, what a great, simple and genuinely tense little game! I sure wish I'd had a copy of this as a kid.

 

 

 

 



 



 



 


 

 

 

Well worth the ~4 minute loading time!

 

[+DrVenkman]

Well, since I have never been know for leaving well enough alone, I took apart the TS1000 yesterday.  It's the same machine internally as the ZX81, aside from the increase to a whopping 2K internal SRAM vs 1K on the ZX81. The ZX81 design underwent a couple revisions in the 2-3 years it was made. My ZX81 has a so-called "Issue 3" board - that is, the third and final board layout. The earliest Issue 1 boards (which are usually UK/non-US market versions) were hand-routed traces. There is no formal "Issue 2" board but there is ... this. It claims to be an "Issue One" board by the silkscreen, but the traces are machine-routed, Arguably, this is the "Issue 2 that never was," despite Sinclair's nomenclature. The date codes on the chips are all from around the first quarter of 1982; the date codes from my Zeddy are from the second/third quarter of 1982, which supports the idea that this is a transitional board design.

 

 

 

The obvious physical difference between the two is swapping the left-right positions of the Z80 and the ROM. The Z80 in this machine is an actual Zilog chip, not the NEC-made version that came in the Zeddy, too. The 2K SRAM chip is an actual ceramic package too, not plastic like the other IC's. The ROM chips are similar between the two but does have some partially-different markings. I know there are two or three different ROM versions but don't the difference between them off-hand.

 

 



 

 

Anyway, I opened the RF can to see if there was any explanation for the crappy video on this one compared to the Zeddy, but didn't see anything obvious. I did see that a lot of components inside had been dribbled in wax for some reason. I tested some of the wax twith a screwdriver blade to check the texture and nudged a fixed coil - somehow that inverted the colors on the board, so clearly I should've just let well enough alone, lol. Not to be deterred, I pulled the RF can, gutted it, and built a composite mod circuit as I used on the ZX81. I will complete and test that mod sometime today or Monday after work.