Spectrum Computing

By connecting +5V to pin 5 on IC23, you are overriding the signal from elsewhere in the circuit.This point was deliberately chosen, as the existing signal comes via a resistor (R70) which limits the current flow so that the logic gate that feeds resistor R70 is not damaged. Pin 5 is an input to a logic gate.

With a +5V feed applied to pin 5 on IC23, it causes the output of the logic gate in IC23 to output a logic 1 (logic high). This is a control line to all of the 'upper' RAM chips (called "/CAS"). With this control line held at logic 1 (high), it means that all of the 'upper' RAM chips will ignore the Z80 CPU and will not respond. Hence they are disabled.

In most (but not all) RAM chip failure modes, telling a faulty 'upper' RAM chip to ignore the CPU means the ZX Spectrum will start up thinking it is a 16k model.

I see the thread has moved on since I started writing this

Mark

rps wrote: ↑Mon Jun 18, 2018 1:04 pm Ok, so you think lower ram possibly. Does this mean you can rule out upper ram? Or could it still be upper ram?

Also, if I was to get parts to make repairs and see where I get with it. Is there ways to test a chip when you remove it from the board?

Thanks
RPS

At this stage, it looks like the ROM, the CPU and the ULA are all at least partially working. Most likely suspects remain any of the RAM chips. So far no RAM has been eliminated.

The only easy way to test a suspect chip, is either with specialist test gear, or in a known good working ZX Spectrum.

Can you please post a photo of the wire that you soldered across IC23 please.

Mark

rps wrote: ↑Mon Jun 18, 2018 1:04 pm Ok, so you think lower ram possibly. Does this mean you can rule out upper ram? Or could it still be upper ram?

Also, if I was to get parts to make repairs and see where I get with it. Is there ways to test a chip when you remove it from the board?

No, you can’t rule out upper RAM completely. Then again, since the screen is part of the lower RAM, and it looks like the screen isn’t corrupted, it might not be lower RAM either. I just suggest a method of eliminating potential causes one by one. Technically (albeit, very unlikely) the problem might be with the ROM itself, or, possibly one of the address or data lines.

There’s no easy way of testing ICs outside a circuit, and each IC would require a special test rig in most cases. Sometimes, it’s advantageous to have a spare test “known-good” board with all ICs socketed. Then you can take individual ICs out of the board you’re woking on and putting them in the test board. Conversely, you can take out ICs from the “known-good” board and use them in the board under investigation, thus ruling out chips one by one.

That just covers the ICs. The fault may be with some of the passives (capacitors may affect timing), but diagnosing it would definitely require an oscilloscope.

1024MAK wrote: ↑Mon Jun 18, 2018 2:23 pm I see the thread has moved on since I started writing this

Bah. Tell me about it! We need to coordinate better.

Hi,
Here is the image of IC23 and a wire from pin 5 to 14 as mentioned. Sorry it’s not a great image, I cracked the glass on my camera lens...

So when I powered on this morning, no red lines. Turned off and plugged in again and got red verticle lines. I dont think it knows what is going on!

Thanks
RPS

Interesting. I don’t recall any 48K Spectrums that came with a Zilog’s Z80 originally. They were usually either a NEC or an SGS. Does it look that the board has been worked on before to you? (Any soldering job look different, say, visible flux residue, etc.?)

Although not as common as the NEC D780C-1 CPU (a Z80 clone), Sinclair did use Zilog Z80A CPU's in both ZX81's and ZX Spectrum's. And going by the date codes and wave soldered joints, these are the original chips that were fitted.

@rps do you have any expansions? If yes, please list them.

Mark

Hi,
No expansions to my knowledge and I can’t see any solder work that doesn’t look original either.

I will take some more photos for you...

Here’s a few photos. If you want anything specific let me know...

1024MAK wrote: ↑Tue Jun 19, 2018 11:12 am Sinclair did use Zilog Z80A CPU's in both ZX81's and ZX Spectrum's.

ZX 81s yes, 128Ks virtually exclusively, but for 16/48K it was a rare spectacle. Hence, my suspicion. But the board does look like it’s all original parts. On the first pic, the soldering job on the CPU had a suspicious shine to it, though.

Would be great to see the other side of the board, too.

Still not convinced the CPU or the lower RAM quite have all their marbles.

Here is the cpu closer. In real life it all looks original to me.

If you want a better photo, I can try get one for you?

Thanks

Sent the wrong photo!!!

Yes, looks legit.

I should have asked this sooner, but are all the voltages within range? (They likely are, but still.)

Erm, I have no idea about that, sorry can’t answer that question.

My friend if bringing my volt meter back tonight so will that help with anything?

Thanks

rps wrote: ↑Tue Jun 19, 2018 3:58 pm My friend if bringing my volt meter back tonight so will that help with anything?

Oh, absolutely.

Hi,
Is there something in particular that I can check, such as chip and what pins?

I am guessing that is a huge question?
thabks
RPS

rps wrote: ↑Tue Jun 19, 2018 8:06 pm Hi,
Is there something in particular that I can check, such as chip and what pins?

I’d start with the lower RAM ICs (IC6–13). These take three different voltages: +5V, –5V, and +12V. The ground is on pin 16, pin 1 is –5V, pin 9 is +5V, and pin 8 is +12V. Just be careful not to short anything.

Then check the voltages on the upper RAM (IC15–IC22). These just use the +5V rail. Pin 16 is ground, pin 8 is +5V.

IC25 and 26 have the same power pin arrangement.

Next, check the gates IC23 and IC24: pin 7 is ground, pin 14 is +5V.

This is just to make sure each IC is getting power.

Post your results here.

On the lower ram they come in about half what they should be...

rps wrote: ↑Tue Jun 19, 2018 9:50 pm On the lower ram they come in about half what they should be...

Really? Are you sure? Even the +5V rail? How about the rest of the ICs I mentioned, then? Surely, the +5V rail must be close to spec; otherwise the machine wouldn’t function at all.

Test the voltage on the emitter of TR5 to ground (any ground). The emitter should be pin 1 (leftmost pin, looking from the flat side).

I’m getting 4.9, 12 and -4.7 on ic6 to 13
4.9V on ic 15 to 22
5V on ic 23 to 26
The regulator is getting to 70C after a few minutes and the voltages are dropping to half

rps wrote: ↑Tue Jun 19, 2018 10:47 pm The regulator is getting to 70C after a few minutes and the voltages are dropping to half

All voltages? If so, then maybe you should replace the regulator (they’re dirt-cheap and sourced easily). What PSU are you using, by the way?

If it’s not the voltage regulator, it might be either TR5/TR4, or one of the ICs overloading the output. Another possible cause is C74 (very, very unlikely, though).

It’s the original I think...

rps wrote: ↑Tue Jun 19, 2018 10:57 pm It’s the original I think...

Do you have another PSU? Preferably a known-good, 9V 2A (or more) switch-mode kind?

In any case, I’d also test the current the computer is drawing by connecting your multimeter (in ammeter mode) with the PSU’s output. Ideally, you’ll also want to monitor the voltage (if you have another multimeter).

I have neither another power supply or another meter.
My friend says he may be able to but it would take a week or more until he can.