Brand new r2 48kB, not quite working

For experts to discuss very technical stuff and newbies to ask why the Spectrum they bought off ebay doesn't work.
chequered flag
Drutt
Posts: 31
Joined: Sat Mar 10, 2018 7:22 am

Re: Brand new r2 48kB, not quite working

Post by chequered flag »

Keep the original power supply as a novelty item for display. When the ZX Spectrum was introduced in Sweden back in 1983 the Sinclair power supply had to be replaced to pass electrical noise regulations.
User avatar
1024MAK
Bugaboo
Posts: 3104
Joined: Wed Nov 15, 2017 2:52 pm
Location: Sunny Somerset in the U.K. in Europe

Re: Brand new r2 48kB, not quite working

Post by 1024MAK »

The external Sinclair "PSU" is a normal transformer - bridge rectifier - smoothing capacitor arrangement.
The 1.4A type should be used for 48k ZX Spectrum and ZX Spectrum + machines. Although some 0.7A and 1.2A (designed for the ZX81) were apparently supplied with early machines.

Typical failures are (in no particular order):
  • Output cable failed (copper conductor broken, but insulation appears fine).
  • Broken PCB tracks or broken transformer legs/terminals (some designs)
  • Blown thermal fuse (mounted under the transformer or inside it)
  • Damaged/blown rectifier diode(s)
  • Life expired electrolytic capacitor
Sinclair used a number of companies to produce these, so there are a number of variations. From the outside, they look the same, or very similar (PSU for the 16k and 48k models). A new case was used for the ZX Spectrum + model, but internally the arrangements are the same.

Most use four individual diodes. Most use a single 4700uF 16V electrolytic capacitor. But some use two x 2200uF.

Any with broken tracks, broken transformer legs/terminals, or blown thermal fuse should be scrapped. If a diode has failed and badly burned the board, again, scrap it. Otherwise repair is straightforward. Renew any damaged or suspect diodes, renew the capacitor(s) with a new 4700uF 25V or 35V part. 105 degree type being preferred.

Mark
:!: Standby alert :!:
“There are four lights!”
Step up to red alert. Sir, are you absolutely sure? It does mean changing the bulb :dance
Looking forward to summer later in the year.
hjalfi
Drutt
Posts: 28
Joined: Wed Jul 18, 2018 11:58 am

Re: Brand new r2 48kB, not quite working

Post by hjalfi »

Nothing looks particularly explodey in the power supply, but it generates about half the voltage it should; it's got two 1000uF capacitors and a single 2200uF one, all in parallel across the output lines. Wow. It's been a while since I've seen a PSU quite this primitive.

Naturally, my parts drawers have precisely one fewer than the required number of capacitors to recap the whole lot today: it's that 2200uF one in the PSU. Everything else I've got (including the one for the composite mod).

New membrane and TRACO switching regulator on order. Thanks!

Additional questions; see https://photos.app.goo.gl/n1CDAmenxYHrSU338:

- there's an extra 4.7uF capacitor (of a different make) spider patched between the positive end of C43 to R58. Original or aftermarket?

- on the east side of the board next to KB2 there's a couple of what look suspiciously like configuration links, labelled N / H / I. They're connected underneath according to the white lines. Any idea what these are?

- so far I've removed the keyboard twice and plugged it in once; the connectors feel really grim and don't grip the membrane very well. I'm hoping the replacement will be a bit stiffer, or at least less likely to fall apart if I bend it too much. Is it worth trying to clean or service the connectors to make removing the keyboard easier?
chequered flag
Drutt
Posts: 31
Joined: Sat Mar 10, 2018 7:22 am

Re: Brand new r2 48kB, not quite working

Post by chequered flag »

- there's an extra 4.7uF capacitor (of a different make) spider patched between the positive end of C43 to R58. Original or aftermarket?
The radial capacitor soldered to C34 is a recommended service modification. Maybe original from factory.

http://blog.retroleum.co.uk/electronics ... -spectrum/

- on the east side of the board next to KB2 there's a couple of what look suspiciously like configuration links, labelled N / H / I. They're connected underneath according to the white lines. Any idea what these are?
ROM select links. N-N for NEC ROM. Matches your NEC ROM in the picture. H-H for Hitachi ROM

http://blog.retroleum.co.uk/electronics ... -an-eprom/
- so far I've removed the keyboard twice and plugged it in once; the connectors feel really grim and don't grip the membrane very well. I'm hoping the replacement will be a bit stiffer, or at least less likely to fall apart if I bend it too much. Is it worth trying to clean or service the connectors to make removing the keyboard easier?
Gentle clean. If the new membrane doesn't grip you can make it thicker on the non conducting side. The old membrane may have a sticky transparent piece of plastic for that purpose that you can reuse.

https://www.youtube.com/watch?v=76gyKWRI7dY

https://www.youtube.com/watch?v=-PJLPEI8RW8

https://www.youtube.com/watch?v=O61Yy00GHEc

https://www.youtube.com/watch?v=Q9cgCAZti98
User avatar
1024MAK
Bugaboo
Posts: 3104
Joined: Wed Nov 15, 2017 2:52 pm
Location: Sunny Somerset in the U.K. in Europe

Re: Brand new r2 48kB, not quite working

Post by 1024MAK »

It's well worth reading the service manual, here is a link to a HTTP version

Mark
:!: Standby alert :!:
“There are four lights!”
Step up to red alert. Sir, are you absolutely sure? It does mean changing the bulb :dance
Looking forward to summer later in the year.
hjalfi
Drutt
Posts: 28
Joined: Wed Jul 18, 2018 11:58 am

Re: Brand new r2 48kB, not quite working

Post by hjalfi »

Status update: caps replaced, though I cocked up and I might need to replace some again --- I'll find out when I try to put reassemble later; replacement membrane is here and I've disassembled (and cleaned, ugh) the old keyboard; TRACO regulator should arrive tomorrow. Kudos to RWAP for delivering to Switzerland amazingly quickly.

Query: the tape holding the old faceplate on appears to have metastasized into a foul, yellow, chunky mucous-like substance, except mucous looks more appetising. The big problem is that it's lumpy, so when I stick the faceplate on it probably won't lie flush. I'd like to clean it off and use new tape. What's safe to use on the case without risk of melting it? Will isopropyl alcohol work? I'd be very hesitant about using acetone of white spirit.
chequered flag
Drutt
Posts: 31
Joined: Sat Mar 10, 2018 7:22 am

Re: Brand new r2 48kB, not quite working

Post by chequered flag »

I bathe my plastic cases in warm water and mild washing-up liquid. Not too hot, don't want to melt the plastic. The heat makes the glue soft and then I scrape it off with a small screwdriver. Any scratches I make will be under the new double sided tape.
hjalfi
Drutt
Posts: 28
Joined: Wed Jul 18, 2018 11:58 am

Re: Brand new r2 48kB, not quite working

Post by hjalfi »

I did actually use a water bath to get the faceplate off (it wasn't like the old membrane could get any more broken!); I'll have another go. I have a soft plastic spudger that'd be ideal for scraping. Apparently IPA isn't a good idea because it can discolour plastics.

I can't find out what the plastic case is made of. I assume it's something very, very cheap. Anyone know?
User avatar
Seven.FFF
Manic Miner
Posts: 735
Joined: Sat Nov 25, 2017 10:50 pm
Location: USA

Re: Brand new r2 48kB, not quite working

Post by Seven.FFF »

I've also used a steam cleaner to soften that kind of gunk.
Robin Verhagen-Guest
SevenFFF / Threetwosevensixseven / colonel32
NXtel NXTP ESP Update ESP Reset CSpect Plugins
User avatar
1024MAK
Bugaboo
Posts: 3104
Joined: Wed Nov 15, 2017 2:52 pm
Location: Sunny Somerset in the U.K. in Europe

Re: Brand new r2 48kB, not quite working

Post by 1024MAK »

hjalfi wrote: Tue Jul 24, 2018 1:56 pmI can't find out what the plastic case is made of. I assume it's something very, very cheap. Anyone know?
Note: this applies to both ZX81 membranes (which are stuck on the case) and the glue (or double sided sticky tape) used on rubber keyed ZX Spectrums. No other machines.

The black plastic case is ABS. Normal IPA solvent does not damage ABS plastic as far as I know (been using IPA on ABS plastic for many years). But IPA alone may not help a great deal with the glue.

When I do it, I use a hair dryer and a wide flat bladed screwdriver at a shallow angle. I then wipe the screwdriver in paper towel. Then scrape again... Only once I have got as much as possible off with this method, do I use IPA. In my experience, the IPA softens the glue, but does not fully dissolve it. So you have to wipe it off. As soon as the IPA evaporates, the glue goes back to being glue!

Mark
:!: Standby alert :!:
“There are four lights!”
Step up to red alert. Sir, are you absolutely sure? It does mean changing the bulb :dance
Looking forward to summer later in the year.
hjalfi
Drutt
Posts: 28
Joined: Wed Jul 18, 2018 11:58 am

Re: Brand new r2 48kB, not quite working

Post by hjalfi »

Hot water and a iFixit-clone stainless steel spudging tool worked reasonably well --- I managed to get most of the lumps off. It'll lie flat now, so I think I'll leave it at that, as it's good enough.

Man, that glue is repulsive. I actually tried a nylon pot scourer thing, and the nylon came off on the glue.

Thanks!

(Still waiting for the regulator to arrive.)
chequered flag
Drutt
Posts: 31
Joined: Sat Mar 10, 2018 7:22 am

Re: Brand new r2 48kB, not quite working

Post by chequered flag »

hjalfi
Drutt
Posts: 28
Joined: Wed Jul 18, 2018 11:58 am

Re: Brand new r2 48kB, not quite working

Post by hjalfi »

Woohoo!

Image

I own a really very nice Tektronix 7603 oscilloscope weighing 13kg, so calibrating the video was a cinch --- thanks for the links; I'd never have thought to do it otherwise.

However...

I have a mini composite monitor, which won't even admit there's a signal there; a big TV with composite in, which will show a picture, but very reluctantly, with fadeouts, glitches, and it can't see the colour burst so everything's a mosaic of black-and-white; a composite-to-HDMI converter, which will show colour, but has a lot of trouble locking onto the image and it rolls and glitches; and a ultracheap composite capture device, which shows a rock solid image and nice, rich, and wrong colours. The image above came from that. You can see that the white is grey, and that the bright0/bright1 bands are... weird.

Is this just digital TVs not liking composite, or is there something I can do? When working on the board, I added a 100uF capacitor between the composite video signal and output pin (as recommended).

(I have a lot of this on video; I'm editing.)
chequered flag
Drutt
Posts: 31
Joined: Sat Mar 10, 2018 7:22 am

Re: Brand new r2 48kB, not quite working

Post by chequered flag »

Is this just digital TVs not liking composite, or is there something I can do?
I have three different digital TVs. Newer and higher resolution gives better picture.

There are some adjustments to do on the TV.

All auto detect is bad. Lock it to PAL video signal if possible.

Disable all filters that are supposed to improve the picture. These filters get confused by the dot crawl that is always present. These filters will also introduce a delay that is sometimes noticeable when you play games. Select "game mode" if possible.

Increase brightness until bright white becomes white. Normal white should be grayish.

Increase contrast until you get sharp edges between the color bars.
hjalfi
Drutt
Posts: 28
Joined: Wed Jul 18, 2018 11:58 am

Re: Brand new r2 48kB, not quite working

Post by hjalfi »

Way ahead of you own the TV settings, sorry! Image 'enhancement' filters drive me nuts and I always try to turn them off...

But there's something more than that going on. I'm trying to source a real composite CRT and see if that's any better. I hear that the ZX81 omitted various important parts of the video signal for simplicity; did the Spectrum?

In other news, I have produced a massive 2h30m video of me working on the board. I'm not expecting anyone to actually watch this, but:

https://www.youtube.com/watch?v=24WUYppOExA

Useful bits are at about 2:20:20 where I have the output video hooked up to an oscilloscope, and 2:28:00 where I have it connected up to a monitor (via a HDMI upconverter) and you can see sync problems.
chequered flag
Drutt
Posts: 31
Joined: Sat Mar 10, 2018 7:22 am

Re: Brand new r2 48kB, not quite working

Post by chequered flag »

No, Spectrum is better than ZX81 video signal. I have better colors from my issue two both on CRT and digital TVs.

Maybe try to replace LM1889N Video Encoder IC, reasonable to try.
https://www.retroleum.co.uk/zx-spectrum-chips

Replace ULA, not so reasonable. It seem to do its job here.

Maybe bypass internal video all together and add external "Graphics Card" analog RGB SCART
http://www.fruitcake.plus.com/Sinclair/ ... erface.htm
There is a 3D printed case for SPECTRA now. I have it.

Or go digital HDMI
https://www.bytedelight.com/?product_cat=zx-hd
chequered flag
Drutt
Posts: 31
Joined: Sat Mar 10, 2018 7:22 am

Re: Brand new r2 48kB, not quite working

Post by chequered flag »

I watched the video now.
If the output turns blue, your TV now thinks the Spectrum is outputting a SECAM signal, so reverse the adjustment bit by bit.
Try to adjust to a more yellow kind of white. Like cream.

What if I can’t achieve a good video output?

Sometimes even given the best equipment available, the colour tinge cannot be removed from the signal. In this case, the following can be tried:

Recap the machine (this is recommended anyway, but faulty caps can cause noise on the video output)
If you have a frequency meter or a multimeter that can do basic frequency measurement, adjust TC2 to 4.433619MHz or as close as possible (probe pin 17 of the LM1889).
Installing a 4K7 resistor between pin 14 of the LM1889 and ground
Replacing the LM1889 chip.
hjalfi
Drutt
Posts: 28
Joined: Wed Jul 18, 2018 11:58 am

Re: Brand new r2 48kB, not quite working

Post by hjalfi »

Well, I fired it up today on the capture card with the intend of typing in my colourbars program and tuning it; and was met with black. There was signal there, but the capture card wasn't showing an image. The machine still works and still shows (glitchy) images on the other devices. So whatever's causing my other digital devices to be temperamental is now causing the capture card to be temperamental.

I don't have any way to directly measure frequency, but assuming my $16000 (2015 dollars) oscilloscope is accurate, a scanline is dead-on 64us, which should be right:

Image

I tried adjusting the pots with it connected to one of the glitchy monitors and it was a thoroughly frustrating experience. I think the monitor's automatic decoding is defeating my ability to make small changes and see the results. I also get the impression that the waveform varies over time.

My symptoms match those described here:

https://www.reddit.com ZZZ /r/zxspectrum/comments/5t1neb/heeelllppp_composite_mod_problems/ (ZZZ inserted to prevent automatic preview)

...which someone pins down to impedance mismatch issues, but there isn't a suggested solution. I can see the sync pulse on the scope and it looks nice and clean, with only a little bit of ringing. I've gone through the video diagnostics section of the service manual to the best of my ability and I can see all the various thingies twiddling as suggested, although I'm not sure about pin 13 --- the signal changes when I attach the scope.

I think I've found a source for a CRT, so I'll try that next, unless someone has a recommendation. If it is an impedance issue, maybe the transistor mod would help?
User avatar
Ast A. Moore
Rick Dangerous
Posts: 2640
Joined: Mon Nov 13, 2017 3:16 pm

Re: Brand new r2 48kB, not quite working

Post by Ast A. Moore »

This may not be related to your problem directly, but it’s about an Issue 2 machine, and the guy does show a few oscillograms and explains them. Might be of some use:

https://www.youtube.com/watch?v=KKFOxLqHl1s
Every man should plant a tree, build a house, and write a ZX Spectrum game.

Author of A Yankee in Iraq, a 50 fps shoot-’em-up—the first game to utilize the floating bus on the +2A/+3,
and zasm Z80 Assembler syntax highlighter.
User avatar
1024MAK
Bugaboo
Posts: 3104
Joined: Wed Nov 15, 2017 2:52 pm
Location: Sunny Somerset in the U.K. in Europe

Re: Brand new r2 48kB, not quite working

Post by 1024MAK »

With reference to the comments about the ZX81 video signal. The ZX80 was only designed with a monochrome (black and white) TV set in mind. The ZX80 does not generate the part of the video signal known as the 'back porch'. A monochrome TV will work without this part of the video signal being to the specification. The first and second versions of the ULAs used in ZX81s also don't generate the 'back porch' part of the signal. But the last (third) version of the ZX81 ULA does generate the correct 'back porch' part of the signal.

The ULA in the ZX Spectrum is far more complex than the ZX81 ULA, hence not only does it generate the 'back porch' part of the signal, but it actually generates the video signal on it's own (you will still get a picture with the Z80 CPU removed from the board). Unlike the ZX80 and the ZX81 where the Z80 CPU is fooled by the ULA in getting video data from the memory for it.

However, these computers don't produce a proper TV specification interlaced TV signal. Instead they repeat the same field every time (instead of an even field followed by an odd field making one interlaced frame).

The LM1889 is a colour encoder, it only generates the colour signal. Colour being added to the monochrome signal. Without the LM1889, you would get a monochrome black/grey/white picture.

There are FOUR ways to composite mod a 16k/48k/+ ZX Spectrum. These are listed from least complex (and cheapest) to most complex. as you go further down the list, so the resulting signal 'should' be more compatible with more TVs and monitors.
  1. Just use a piece of wire.
  2. Use an electrolytic capacitor, typical values are 100uF 16V
  3. Use a NPN transistor in emitter follower configuration, followed by an electrolytic capacitor (100uF, with a 100nF paralleled)
  4. Use the configuration used in the 128k machines.
Mark
:!: Standby alert :!:
“There are four lights!”
Step up to red alert. Sir, are you absolutely sure? It does mean changing the bulb :dance
Looking forward to summer later in the year.
hjalfi
Drutt
Posts: 28
Joined: Wed Jul 18, 2018 11:58 am

Re: Brand new r2 48kB, not quite working

Post by hjalfi »

Thanks for the links; all useful. The transistor mod is simple enough I may just try it, and see what happens --- it can't make things worse.

It's still really weird that there's no sign in the scope of anything wrong. The timing looks fine (64us per scanline). The voltages look fine (1V from the bottom of the sync pulse to the white level), which suggests the monitor's producing the right loading. Twiddling the pots causes the monitor (actually, the composite-to-HDMI upscaler in this setup) does, in fact, have the right impedance. The waveform shape looks about right, too. (Turns out that what I thought was interference noise in the video is, in fact, perfectly correct signal modulation.)

One really interesting thing is that my big TV shows the image in black-and-white, with mosaic patterns instead of colours. That suggests it's not seeing the colour burst. So it's probably worth me checking that the colour modulation is actually happening at 4.43MHz. I see now that the colour clock is generated from a different crystal; is there a chance that has failed?
User avatar
Ast A. Moore
Rick Dangerous
Posts: 2640
Joined: Mon Nov 13, 2017 3:16 pm

Re: Brand new r2 48kB, not quite working

Post by Ast A. Moore »

1024MAK wrote: Fri Jul 27, 2018 10:42 am
  • Use the configuration used in the 128k machines.
Um, that would be quite a complicated piece of circuitry. The TEA2000 needs digital RGB signals, and the 48K ULA only provides the YUV output. Converting that to RGB isn’t particularly easy. I’ve seen a schematic for converting YPbPr to RGB using four LT6552 video difference amps—I think it could be adapted for YUV signals. You’ll still need a sync, though.

EDIT: Oh, found another schematic that generates the sync signal:

Image

You could take the existing Y, R-Y, and B-Y signals from the Speccy and thus greatly simplify the circuit above.

Then again, you might as well use these signals to drive a SCART input directly. :lol:
Every man should plant a tree, build a house, and write a ZX Spectrum game.

Author of A Yankee in Iraq, a 50 fps shoot-’em-up—the first game to utilize the floating bus on the +2A/+3,
and zasm Z80 Assembler syntax highlighter.
hjalfi
Drutt
Posts: 28
Joined: Wed Jul 18, 2018 11:58 am

Re: Brand new r2 48kB, not quite working

Post by hjalfi »

Or maybe, and bear with me, I know this is a long shot... maybe I could just connect the composite decoupling capacitor the right way round?

Okay, so I also added the transistor mod, so maybe that made a difference too, but all the monitors I've tried this with are quite happy now. (Although the instructions I found, at http://www.projectavr.com/spectrum-comp ... nd-refurb/, said to use a PNP transistor. I picked a 'universal PNP' from the parts drawer. Sadly I couldn't find a 100R resistor or another 100uF capacitor (certainly wasn't going to reuse the old one), so I picked the closest I had at 120R and 47uF, reasoning that this is fuzzy analogue stuff; it seems fine.)

Image

That image is after calibrating the colours. The white's not very bright and the yellow's kinda sickly.

Anyway, thanks for all the help. I've learnt a tonne and I have a nice apparently-functioning Spectrum. Cassette recorder's on order, and now I think of it I should recap the joystick interface.

I also have a ZX81 on order. That'll need recapping too; maybe I should do that one in ceramics.
User avatar
Seven.FFF
Manic Miner
Posts: 735
Joined: Sat Nov 25, 2017 10:50 pm
Location: USA

Re: Brand new r2 48kB, not quite working

Post by Seven.FFF »

Good stuff!

For the ZX81, you could try doing the front porch mod: https://youtu.be/1irH3KuGyl0
Robin Verhagen-Guest
SevenFFF / Threetwosevensixseven / colonel32
NXtel NXTP ESP Update ESP Reset CSpect Plugins
chequered flag
Drutt
Posts: 31
Joined: Sat Mar 10, 2018 7:22 am

Re: Brand new r2 48kB, not quite working

Post by chequered flag »

That link is now collected. Really nice.
http://www.projectavr.com/spectrum-comp ... nd-refurb/

Did you see the "Sinclair Spectrum Composite Mod PCB" that fits inside the RF box.
http://www.projectavr.com/esp-01-esp-03 ... ing-board/
Post Reply