Black screen CPC464 seems to be failing more as i diagnose it

[ClintonGoodfellow]

Hi all,

I'm new here and hoping to get some advice as I'm at a loss with this machine.

The machine was in poor shape and was displaying a black screen so i started to look about the board.

Apart from being very grubby, all chips were receiving 4.9v and could detect a clock from the crystal, the xtal input on the ga and on the z80.

No other signals were present on the z80. It was running hot so swapped it out and got the missing signals. Still black screen.

I couldn't detect a clock on pin 40 of the unpopulated 40007 location or pin 21 on the crtc. Decided to take a break as there was more to do.

Came back a couple days later and now  i cant pick up any clock signals at all. Nothing on the crystal, ga, z80? Im not sure whats happened.

Any advice on where to look would be fantastic.

Cheers

[ClintonGoodfellow]

Forgot to add that its a z70200 board with a 40010 gate array fitted.

[SerErris]

Do you have any chance to do a logic test on IC125?

This is a 74HCU04 Hex Inverter. This chip is a unbuffered CMOS version of the 74LS04, but it is key for the clock signal that it is CMOS and unbuffered in this application.
If this chip is toast - then you will not get anything on the clock. I am not sure if you can measure the crystal if the inverter is toast. The CMOS and the first inverter gate are forming an oscillator, that is then cleaned by the second inverter gate. So esentially the inverter is the osciallator that is driving the crystal and it is then stabilizing the oscillator to a certain frequency.

It is very difficult to test it in system, as the lines are all connected. But if you desolder it, you would be able to put it in a breadboard, power it and then test every input if the output works to expectations, meaning - if you ground input, output shall be high, if you put +5V to input the output should be low.

As long as you cannot see a clear good clock on the second inverter gate output, there is no hope for the rest of the system. Everything is coming from this clock to live.

Here is a good article on clock designs that derive the clock via an unbuffered inverter (like the one we have in the CPC).
https://www.ti.com/lit/an/szza043/szza043.pdf?ts=1729102058232&ref_url=https%253A%252F%252Fwww.google.com%252F

Maybe just replace the inverter ...
However the Toshiba TC74HCU04AP is pretty expensive - and I am not sure if any other unbuffered inverter will do the job the same. Any other brand of the same 74HCU04 is 30-80 cents.

[McArti0]

It is very difficult to test it in system

Testing it is very easy because the beginning of the oscillator does not have any current source. Please do not scare yourself or others. 

IC125 pin1 to ground/Vcc and test Pin2 (should be Vcc/ground cros pin1) and test Pin4 (should be ground/Vcc like set pin1)

[SerErris]

I am just not sure if that will test anything. Is is a oscillator with a feedback loop. You are actually influencing the feedback loop and therefore I am not sure on the output. That has nothing to do with "damage" or fear, but about questionable result and insight.

However if there is no reliable oscillation on Output of Gate 1 - Pin2 of IC125 (that should be measurable with an Oscilloscope), then there is only those root causes possible:

1. Gate is dead or not working in specs (e.g. maybe much longer gate delay).
2. Crystal is dead.
3. Some traces on the mainboard are broken or corroded.
4. Solder joints are broken, dry - so maybe intermittent connection?
5. Capacitors (C130, C131) or resistor R143 broken.

That should be pretty much extensive on what is wrong with the clock.

[ClintonGoodfellow]

Thank you for all these responses.

I do have a scope, albeit a hantek usb one which can be a bit of a pain to use.

I'll check those components, traces and pin2 of ic125 and report back.

I have breadboards, so pulling the chip to test further as suggested is also possible.

[ClintonGoodfellow]

Hi.

Got a couple mins spare time to test a bit further.

On powering on for the first time i was receiving a clock signal off pin2, but over the course of a minute it dwindled away to just a roughly 2v high signal.

Both caps were removed and seem fine. R143 was removed and has a value of 10.09M ohm.

With the cap removed both sides of c131 were showing gnd. Traced through to the xtal, removed it and one of the two pins is shorted to the metal housing, that cant be right surely?

[SerErris]

I am not sure of what crystal is in the later 464s, which you have. But if it is still the HC-18/U housing (which was pretty much standard), then there should be no connection between the housing and the two pins. They shall be isolated. However the case should be connected to ground. Probably it is tied down with a lead over the crystal?

You cannot view this attachment.

A HC-18/U looks like this:

It is compatible to the new HC-49/U - which just have thinner leads. But that should not be any issue.

The Resistance of R143 is correct (10M in the schematics), Caps typically do not break in this pf caps. They are normally ceramic and will not break of change its capacitance.

What about the 74xx04 ?

Can you also check the power supply on IC125? Is that draining somehow (getting lower over time?).

If you have removed all the other parts (e.g. crystal, caps and Resistor) you can apply low or high to pin 1 and measure pin 2. Is it giving a stable output (and the correct one?)

It might be just right to put everything into the breadboard and check if it works there with bench power (+5v), so that you can eliminate everything else that could influence it.

Looking for a replacement of the crystal, this would match, but would also require the replacement of C130,C131.
https://www.reichelt.com/ie/en/shop/product/tht_crystal_16_000000_mhz_-30_ppm_cl_32pf_hc-49u-1605

I just asked chatGPT on that question, and here is the short summary, which you will need to test with an OSC to see if the outcome is as expected (e.g. 16Mhz clock):

Questions:
 This is a clock signal oscillator for the amstrad cpc, with 16Mhz clock signal. It feeds the output of pin2 of the 74HCU04 into another inverter to make a clean clock signal. Can you recalculate the values of C130 and C131 if the load capacity of the crystal is 32pF and the series Resistance is 20?

Answer:
Conclusion

• Capacitors C130 and C131 should be 54 pF each, but 56 pF is a practical and readily available value.
• Ensure to verify the circuit with an oscilloscope to fine-tune the values if necessary, considering the impact of stray capacitances and the layout of your PCB.

This setup should ensure the crystal sees the correct load capacitance of 32 pF, resulting in stable oscillation at the intended 16 MHz frequency.

The result above does assume 5pf stray capacity from the PCB, which still might be off and I cannot even calculated it.

Or other way around the load capacitance of the crystal used by Amstrad will be

The actual issue is, that we cannot find out the real values of the old crystal if we do not exactly know what manufacturer and model it was. If you can fotograph it, we can potentially find that information. Not sure if there is any vendor marking on it.

Also find attached the whole calculation from ChatGPT for our nice little circuit.

Okay, and if you replace the 74xx04 with a new CD74HCU04E (PDIP) from TI and the crystal above, you would need two 56pF capacitors. To replace the circuit. Those two Cs bring you pretty close 33pF load capacitance of the requested 32pF capacitance and that should work here.

R143 will be still 10M as in the original. You also should look for very low tolerance capacitors here (5% or lower).

Also if you use another crystal, you will need to recalculate C as outlined in the PDF.

[ClintonGoodfellow]

Really appreciate you taking the time to reply, there's a lot to digest for a learner like me. I haven't been back out just yet to look over these suggestions but should do very soon.

There is a definite connection between one of the pins and the metal housing of the crystal. I got it under the scope and could see rust around the where the isolated pin goes into the housing. Cleaned all this off but unfortunately the short remains. I suspect the crystal is toast. I do have a picture of it which I will attach.

I also have already removed ic125 as I wanted to see underneath it to view the condition of the traces. I'll be belling these out to confirm they're sound and I'll get that chip put in a breadboard and follow your guide in testing it.

I aim to get back to you asap. Thanks again 

[SerErris]

I tried to figure out, what the maker of the quartz was and what series it is to understand what the load capacitance and series resistance would be. Unfortunately to no avail.

It looks like CITIZEN to me (which hat CTZ markings) however I cannot find anything on the internet wich has CTZ and 56 in it. It is for sure an HC-18/U, so any HC-49/U or S would fit.

https://www.mouser.com/ProductDetail/ECS/ECS-160-20-4X?qs=ObzGnTthPd%252BlRw5FTf6P8A%3D%3D

This one would do it, and it would better fit to capacitors ...

You would need two 30pF capacitors like those:
https://www.mouser.com/ProductDetail/KEMET/C325C300JAG5TA?qs=sGAEpiMZZMsh%252B1woXyUXj1p5ReH8PUn7pCzRCPFSDXI%3D

Lower voltage would do it, however this is the cheapest at 5% tolerance.

Or you can take this one with 1% tolerance. The voltage rating ist 25V which is still plenty.
https://www.mouser.com/ProductDetail/KEMET/C315C300F3G5TA?qs=h3%2Fj8evtlm2tyDJUu3qLkQ%3D%3D

If you get one of those:
https://www.mouser.com/ProductDetail/Tusonix-CTS/538-011-D-9-35LF?qs=0eVsa72UW5ANzVGWgUkM4g%3D%3D
you can adjust the capacitance of one of the capacitors (C130) ...

However you would need to adjust the legs, as it seems to be 0.1 inch too wide.

[ClintonGoodfellow]

I have got some time this afternoon for some further shenanigans.

The traces from IC125 to the next significant component all check out, confirms traces and via's are good.

I got the chip onto a breadboard, applied VCC and GND and tested the gates. All six worked as expected inverting the input to the output pin. I think this chip is sound, which is good news. Took the opportunity to put a socket in.

I think the next stage is to order the above crystal and the matching caps to see how we get on.

Not knowing the history of the computer is a pain as we're going in blind, hoping for a result in the end but its been good learning more about the workings and design.

If there is anything else you think I should check, let me know.

Thanks again.

 

[SerErris]

For now let's focus on getting a stable clock, from there we tackle one thing after the other.

[ClintonGoodfellow]

https://www.mouser.com/ProductDetail/ECS/ECS-160-20-4X?qs=ObzGnTthPd%252BlRw5FTf6P8A%3D%3D

https://www.mouser.com/ProductDetail/KEMET/C315C300F3G5TA?qs=h3%2Fj8evtlm2tyDJUu3qLkQ%3D%3D

Ordered the above items, one crystal and a pair of the 1% tolerance caps for c130 and c131. Is there anything else I need to change or do before I install them?

[SerErris]

Test them in breadboard first (with 1M resistor) to see if it works.

[Bryce]

Test them in breadboard first (with 1M resistor) to see if it works.

The chances of a new one being broken is as good as zero. Modern crystals are pretty robust.

Bryce.

[SerErris]

It is not about broken, but about the frequency right - so the chip + the crystal + the two caps + the resistor. 

I would just put that into a breadboard and check the frequency before soldering everything in and then finding out - not working, or wrong frequency etc.

[Bryce]

Why would the frequency be wrong on a component bought from a reliable supplier?
The capacitor he has linked to is the wrong value for the crystal, but the crystal itself should be fine.

Bryce.

[SerErris]

Yes, that is true. however the capacitors influence the frequency. And if that is not correctly calculated (e.g. I made a mistake) than you have to desolder the caps again and fit correct ones. 

Also the NAND gate is still not out of the equation. It might work if you put on a simple voltage (0 or +5v), but that does not mean that it does its work as an Oscillator.

Of cause you can just solder it. And probably it will work. However testing it out of circuit would make any changes easier ;-)

[ClintonGoodfellow]

Never worry these things happen!

How would I go about calculating the values for the two caps? I'm going to try getting them from eBay as I don't have anything else to get from mouser at the moment.

I got a couple of those crystals just incase something has damaged the other one.

[Bryce]

Never worry these things happen!

How would I go about calculating the values for the two caps? I'm going to try getting them from eBay as I don't have anything else to get from mouser at the moment.

I got a couple of those crystals just incase something has damaged the other one.

You don't need to calculate it. They tell you right on the first page and it's also what the "20" in the part number means.

Bryce.

[ClintonGoodfellow]

Ah, I did see that, but wondered it if meant something else! Might have to just pay the mouser postage as the price for low-% tolerance caps is very expensive on eBay. Best I can find there is 10% tolerance for a reasonable price. The KEMET 1% ones above are coming in at £7ea!

[Bryce]

The capacitor value doesn't need to be exact. Even a simple ceramic cap with +/- 20% is fine for this application.

Bryce.

[ClintonGoodfellow]

https://www.ebay.co.uk/itm/234095672334?mkcid=16&mkevt=1&mkrid=711-127632-2357-0&ssspo=zKp1MezwR1C&sssrc=4429486&ssuid=umekACLEQHK&var=533727471807&widget_ver=artemis&media=COPY

Would these do?

[Bryce]

Yup.

Bryce.