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.
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/
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: Sun Jul 29, 2018 6:19 pmThe 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.
The 'white' is not supposed to be a bright white (unless the BRIGHT attribute is set). The normal white should be more like a light grey.

I'd be supprised if there are any electrolytic capacitors in a joystick interface. Unless it gives problems, there is no need to change the capacitors in a joystick interface.

The main reason for replacing the capacitors in a ZX Spectrum is to improve the reliability, as a failure of an electrolytic capacitor in the DC-DC converter section (or a capacitor supplied by, or supplying it) can cause it to fail. The loss of the -5V rail then may cause damage to the 4116 (or equivalent) DRAM chips. Also the inside of the ZX Spectrum 16k / 48k machines gets rather hot, and increased temperature decreases the life of electrolytic capacitors.

With the ZX81, there are only two electrolytic capacitors, one to smooth the "+9V" rail and one in the /RESET circuit. Both are non-critical in that a failure of the capacitor will not cause any semiconductors to become damaged. So no need to routinely replace them. Just replace on failure. Not that these cause many problems anyway.

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 »

My nice new Sony PVM has arrived!

The Spectrum displays in mono on it. It displays in colour on two out of my three composite-capable LCDs (the third is a big-screen TV and just shows black).

I was kinda expecting a real CRT to do better than a modern LCD. Is this behaviour symptomatic of anything?

(And yes, I have verified that the Sony does actually work with a colour signal!)
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 »

Odd. Is the Sony a multistandard monitor? If yes, does it have manual or automatic switching? And if automatic, is there a way to force it to PAL I standard? (that's I as in aye!). Or maybe just the PAL 50Hz setting.

Some multistandard monitors and TVs sometimes get confused by the Spectrums video output.

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 »

It does PAL and SECAM, with a switch. It also has lights to let you know whether it detects colour. Interestingly, if I route the video through an CV-to-HDMI-to-CV cover I get nice vibrant colour and the PAL light lights up. Connected directly I get mono and no light.

This suggests that the monitor's not detecting the colour burst, right? That implies that my dubious transistor mod is dubious. That would explain the skewed colour on one of my LCDs, actually.

I used the circuit from http://www.projectavr.com/spectrum-comp ... nd-refurb/:

Image

...although I had to make some substitutions: the resistor's a 120R, the transistor is an off-the-shelf 'general purpose' PNP (which the article said should be fine). Do any of these values actually matter?
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: Thu Aug 23, 2018 7:15 am ...although I had to make some substitutions: the resistor's a 120R, the transistor is an off-the-shelf 'general purpose' PNP (which the article said should be fine). Do any of these values actually matter?
Well... maybe...

In theory, a transistor rated for HF (high frequency) signals should be used rather than a general purpose one. A composite video signal has frequencies between a few tens of Hz through to 6MHz.

The larger the value of capacitor, the greater the frequency range it will pass. However, electrolytic capacitors become slightly inductive at higher frequencies. So 100nF ceramic types are sometimes wired in parallel with them. Also larger values of capacitor allow more current to flow.

The value of the resistor should really be determined in order for the circuit to have a 75 ohm output impedance. 100 ohms is not too bad, 120 ohms is a bit further away...

But I'm not absolutely convinced making any changes would make much difference. As most of these 'buffers' designs were put together by experimentation.

It could be that your monitor does not like the non-standard video signal from the ZX Spectrum, or it could be that parts of the signal it is receiving are marginal, hence dropping from colour to monochrome.

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 »

I really like it when I find a problem that turns out to be caused by me doing something really dumb --- because it's usually very easy to fix.

Turns out the Spectrum was just poorly calibrated, and five seconds' work with a screwdriver fixed it.

It looks really nice on my Trinitron PVM monitor, and a quick colourbar program shows that they're all the right colour, as well. Now I just need to find some good games...

Thank you very much for the assist!
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 »

Excellent news :D

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.
Contact Sam Cruise
Drutt
Posts: 15
Joined: Sat Dec 16, 2017 5:00 pm

Re: Brand new r2 48kB, not quite working

Post by Contact Sam Cruise »

1024MAK wrote: Wed Jul 18, 2018 5:34 pm Oh, and it's normal for the heatsink attached to the 7805 voltage regulator to run hot. As long as the nut and the bolt holding the heatsink to the 7805 is tight, the 7805 will be okay. If the 7805 gets too hot, it will reduce it's output, and then you will know about it. It is however worthwhile putting a very thin smear of heatsink compound on the 7805 metal tab where the heatsink makes contact to aid heat transfer. Also resolder it's legs.

The ULA also runs hot. Some 4116 DRAM chips run warm. Everything else (with the lid off) should run cool or barely warm.

Mark

I've been going through this forum trying to find a solution to my issue - this thread seemed to come the closest to it.

I have gotten a refurbished issue 4B 48K Spectrum that starts up fine but then loses the display image after a couple of minutes. The case above the power supply feels very hot when it shuts down. I opened it up and it still has the original 7805 voltage regulator. I'm using a modern 9V power supply that works well with a Spectrum+. I'm also using the ZX-HD add-on for HDMI output - this also works fine with the Spectrum+ but the display is lost after a few minutes with the 48K Spectrum.

I also notice that there's no heatsink in the case - is this normal? I had thought that the heatsink could be eliminated if using an upgraded voltage regulator, but my Spectrum has the original 7805.

Could it be that the 7805 is overheating and reducing the output (as above), and then the ZX-HD no longer has enough power to drive the HDMI display? As mentioned before, the ZX-HD, TV, and power supply all work fine with the Spectrum+ I have.

Thanks for any 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 »

Contact Sam Cruise wrote: Sun Nov 24, 2019 10:45 pm I have gotten a refurbished issue 4B 48K Spectrum that starts up fine but then loses the display image after a couple of minutes. The case above the power supply feels very hot when it shuts down. . . . it still has the original 7805 voltage regulator. . . . I also notice that there's no heatsink in the case - is this normal?
The 7805 definitely needs a heatsink, especially if you’re running additional hardware (although yours probably uses the 9V directly from the PSU). But, yeah, you don’t want to run it without a heatsink. Even a small slab of aluminum will be better than nothing. As a test, remove the heatsink from your working Spectrum and put it into the 4B. See if it operates as normal. They rarely burn out when properly heat sunk, but in your case, I’d also be looking into replacing the 7805.
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 »

The drop-in switch mode regulator has worked like a dream for me --- no heat sink, no heat, no problems. Strongly recommended!
Contact Sam Cruise
Drutt
Posts: 15
Joined: Sat Dec 16, 2017 5:00 pm

Re: Brand new r2 48kB, not quite working

Post by Contact Sam Cruise »

Ast A. Moore wrote: Sun Nov 24, 2019 11:42 pm
The 7805 definitely needs a heatsink, especially if you’re running additional hardware (although yours probably uses the 9V directly from the PSU). But, yeah, you don’t want to run it without a heatsink. Even a small slab of aluminum will be better than nothing. As a test, remove the heatsink from your working Spectrum and put it into the 4B. See if it operates as normal. They rarely burn out when properly heat sunk, but in your case, I’d also be looking into replacing the 7805.
Thank you. I will try the other heatsink. I don't own a soldering iron, is there anything else I'd need on top of that to replace the 7805?
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 »

Contact Sam Cruise wrote: Mon Nov 25, 2019 12:30 am I don't own a soldering iron, is there anything else I'd need on top of that to replace the 7805?
Solder, flux, and basic desoldering/soldering skills. The latter is of particular importance, given that you’re going to be working on an old PCB, which poses a higher risk of damaging/lifting the traces.
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 »

When removing or fitting a heatsink, don’t try to use the 7805 voltage regulator chip as a mechanical fastening. Always use a suitable spanner to hold the nut and a suitable screwdriver to turn the machine screw. Otherwise you risk damaging the regulator chip, it’s legs or the PCB tracks/pads.

No flat insulators or insulated bushes are needed on the 7805 and heatsink in the ZX Spectrum. But do make sure that the metal heatsink does not touch (and short out) any other conductors or components (other than the large area of copper track under the 7805 voltage regulator chip).

The machine screw, washer and nut should be done up tightly so that the heatsink makes good firm contact with the metal tab of the 7805.

The idea is that heat can freely flow from the tab of the 7805 to the heatsink with no air gaps.

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 »

If you're new to soldering, I'd definitely recommend practicing on something other than the Spectrum first. It's pretty easy to work on, with huge joints and tracks the size of your thumb, but Sinclair definitely skimped on the materials and the board does have a bit of tendency to fall apart while working on it. I got lucky.

You will need: a simple 25W soldering iron, a solder sucker (https://www.banggood.com/Manual-Solder- ... 29699.html), some leaded solder (don't get the lead-free stuff, it's noticeably harder to work with and the total amount of lead you'll be using is negligible), a set of small needle-nosed pliers and wire cutters, an old chopping board or cutting mat to work on because you will be dribbling molten metal around the place, and Big Clive's soldering tutorial video: https://www.youtube.com/watch?v=VtJ5gngTm-A If you've got three hands you will be at a significant advantage.

For desoldering just find some random old junk like a radio or TV (new stuff has much smaller joints, get something old) and try removing components. After all, you can't break it if it's already broken.
Contact Sam Cruise
Drutt
Posts: 15
Joined: Sat Dec 16, 2017 5:00 pm

Re: Brand new r2 48kB, not quite working

Post by Contact Sam Cruise »

Thanks for these replies and for all the advice on installing a heat sink and replacing the 7805. Very much appreciated!

I think I’ll try with a heatsink first and will replace the regulator if this doesn’t help.

While I wait for those parts to come, I won’t fire up this Spectrum - especially as it doesn’t have a heatsink in it right now. Does it make sense to you that it would power on okay, then the regulator would shut down when it hit a certain temperature, causing the edge connector to fail (which my ZX-HD is plugged into)? When I let it cool down and power up again, it starts up fine until the regulator gets insanely hot, then the HDMI picture shuts down again.

Interestingly, the picture isn’t lost if I just use composite output, even without the heatsink. Maybe it would just take longer for this to happen?

Does this make sense?
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 »

Contact Sam Cruise wrote: Mon Nov 25, 2019 8:41 pm Interestingly, the picture isn’t lost if I just use composite output, even without the heatsink. Maybe it would just take longer for this to happen?
It’s quite possible the ZX-HD is drawing extra power from the 5V and 12V rails, which are all generated by the DC-DC converter circuitry after the 7805. (Well, the 5V rail is supplied by the 7805 directly.) Other than that, it’s all just pure speculation.

I’ve never heard of a 7805 “self-regulating” itself down upon overheating. It either outputs a stable 5V or it overheats and dies. Note that a heatsink is necessary if the output current draw is over 40–50 mA. The Spectrum itself draws about 600 mA (or more). Any additional peripherals will increase that requirement. So, it’s best not to experiment without a proper heatsink.
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.
Contact Sam Cruise
Drutt
Posts: 15
Joined: Sat Dec 16, 2017 5:00 pm

Re: Brand new r2 48kB, not quite working

Post by Contact Sam Cruise »

Thanks again! I will try again with a heatsink.
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 »

7805 voltage regulator chips are part of a wider range of series pass voltage regulator chips.

Most include over temperature protection. When they get too hot, they reduce the output voltage to reduce the current demand by the load (the circuitry powered by the output terminal). The actual output voltage is not specified when this happens, as there are a number of variables involved. But it is often a fairly low voltage and (for a 5V 7805 type) will be well below the minimum acceptable value of 4.75V. By reducing the output current, the chip can reduce the power that it has to dissipate as heat. Hence reducing the temperature rise.

The 7805 provides the +5V power supply rail. In a 16k, 48k or plus model, everything except parts of the video circuitry requires the +5V rail. The specification for a +5V rail in a computer like a ZX Spectrum is it has to be between 4.75V and 5.25V. Although some circuitry may kinda work at lower voltages, this is outside the specifications.

There is also a DC/DC converter circuit that converts the nominal 9V input to a +12V rail and a -5V rail. The DC/DC converter circuitry uses the +5V as a stable reference voltage in order to correctly control the voltage on the +12V and -5V rails. So if the +5V rail voltage drops, the voltage on the +12V rail will also fall, and the voltage on the -5V can also go out of specification.

The +12V rail is used by some of the video circuitry and both the +12V and the -5V rails are used the first 16k bytes of RAM chips (the ‘lower DRAM’ is formed from 4116 or equivalent DRAM chips).

The 4116 or equivalent DRAM chips get very unhappy if the -5V is not present and if there is still a high enough voltage on the +12V rail, they can self destruct.

So, to sum up, if no heatsink is fitted, don’t power the machine up. In the same way, you don’t run an internal combustion engine (that is normally water cooled) without a radiator connected.

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.
Post Reply