Crystal oscillator on 464

[llopis]

I have a 464 that's not showing any video on startup. I've quickly narrowed it down to the GA not receiving any signal from the crystal oscillator. But here is where I start having some questions.


This is an early PCB Z70100 with the following setup:




This is the signal generated from the crystal:


And this is the signal generated from the crystal from a working board (but a later version):


The one in the faulty CPC seems like a much weaker signal, but since the circuit is quite different, maybe that's normal. Maybe someone could confirm that for me.


Anyway, the problem happens at the NAND gate with inputs 11 and 12. Those inputs are like the screenshot above, so pin 11 (and 9 and 10) should be the negated version of that, but it's 0V instead. I checked and I don't see a short to ground (from the capacitor perhaps), so I'm guessing the ZN 7400E is defective. Does that sound right?

[Audronic]

@llopis


Unplug the 40007 or the 40010 and measure the Clock again.
The gate array may be loading the output of the Oscillator ??


Good luck.   Ray

[llopis]

Unplug the 40007 or the 40010 and measure the Clock again.
The gate array may be loading the output of the Oscillator ??

I forgot to mention I already did that.


Also, the first thing I thought was that the oscillator was bad, so I swapped it with the working one from the other board, but the signal was the same (and then I realized that it was probably OK, although I'm not 100% sure since I don't have another board of the same type to test it).

[Bryce]

Connect pins 12/13 to the 5V rail and see if you get 5V on pin 8. Then connect pins 12/13 to GND and see if you get around zero volts on pin 8. If either test fails, the 7400 is dodgy. If it passes, then possibly the crystal is bad.

Bryce.

[llopis]

Connect pins 12/13 to the 5V rail and see if you get 5V on pin 8. Then connect pins 12/13 to GND and see if you get around zero volts on pin 8. If either test fails, the 7400 is dodgy. If it passes, then possibly the crystal is bad.

I know the crystal is fine because I swapped it with a working one.


I was tempted to do that test, but I wasn't sure if it was OK to do that without disconnecting the 7400 first (would it fry the crystal for example putting 5V on pins 12 and 13?). Or do you mean disconnect it first, put it on the breadboard, and do the tests there?

[Bryce]

No, you can do the test with the 7400 in place and without removing anything else. It won't damage anything (as long as you don't leave it connected for a long time). However, if you've already determined that it's not the GA or the crystal, then the 7400 is the only thing left. It's extremely unlikely that any of the discrete parts have failed, so the only other things it can be is a bad connection or the 7400.

Bryce.

[llopis]

No, you can do the test with the 7400 in place and without removing anything else. It won't damage anything (as long as you don't leave it connected for a long time). However, if you've already determined that it's not the GA or the crystal, then the 7400 is the only thing left. It's extremely unlikely that any of the discrete parts have failed, so the only other things it can be is a bad connection or the 7400.

Good to know. Thanks! I already ordered a couple of 7400s, but I'll run a test this evening to make sure.

[llopis]

Yes, there's something funky going on about the 7400. I pulled it out (since I was planning on replacing it anyway), so testing it on a breadboard was very easy.


All the gates seem fine, except for 12-13, which seems to give 1V when I put in 5V on both. So I'm hoping that was it. We'll see when the replacement arrives.

[Bryce]

Exactly 1V? The TTL level for a zero is anything up to 0.8V, so it's very close to what it should be.

Bryce.

[llopis]

Exactly 1V? The TTL level for a zero is anything up to 0.8V, so it's very close to what it should be.

The gate on 12-13 was outputting 0.9V as low, and the gate 9-10 was outputting 2V as low! The other two were outputting pretty much 0V. Highs were always correct everywhere (~4.95V).

[Bryce]

Ok, that's definitely dodgy.

Bryce.

[llopis]

BTW, it was the 7400. I put a new one and we have a heartbeat.


Now I need to figure out why I get a gray screen with border (typical RAM fault), but X-Mem with the RAM Test doesn't work. Hmmm....

[Bryce]

Do you know the history of the machine? It sounds like it got a whack of an over-voltage or reversed polarity supply. I have never once seen a CPC with that 7400 damaged.


Bryce.

[llopis]

Do you know the history of the machine? It sounds like it got a whack of an over-voltage or reversed polarity supply. I have never once seen a CPC with that 7400 damaged.

No, unfortunately I don't. I end up buying broken/non-tested machines online for the fun of repairing them, restoring them, and selling them. Unfortunately that means I don't know the history behind most of them, which is a shame.


I sometimes wonder that, and I have theories about some of the machines, but that's about it. For example, recently I restored a Spectrum +2 in really horrible shape https://twitter.com/noel_llopis/status/1058371445259157505 It had paint on it, scribbles, and was all covered in some kind of fine dust. It dawned on me that the dust was chalk, and so it all made sense: abandoned computer in the back of the classroom eventually gets covered in paint and kids drawing on it with markers.


Anyway, back to this one, I suspect there's something more than just the usual blown RAM, so I'll post here any findings. Maybe we can do some detective work and piece together what happened.

[Bryce]

Well the fact that the 7400 is now doing its stuff, means that there is no dead short in any chip. However, the fact that the RAMtest didn't work suggested that something more than the RAM is damaged. I'm not sure exactly which chips the RAMTest needs to run, maybe @gerald  can help you with that one. Then I'd check them first.


Bryce.

[llopis]

Well the fact that the 7400 is now doing its stuff, means that there is no dead short in any chip. However, the fact that the RAMtest didn't work suggested that something more than the RAM is damaged. I'm not sure exactly which chips the RAMTest needs to run, maybe @gerald  can help you with that one. Then I'd check them first.

It's interesting that a 464 that I repaired just a few days ago had somewhat similar problems: gray video square when turning it on, but X-Mem and the RAM Test didn't work at all (either black or gray square). Some of the RAM chips got *really* hot really quickly, so they were probably shorting out. I replaced 3 of them, and then X-Mem worked and pointed to a fourth bad RAM chip. I replaced that one and everything worked.


So maybe not enough power was reaching the X-Mem device to run because of the shorts? (No idea, just speculating--I know Spectrum add-ons have power issues sometimes).


In this particular case no chip gets particularly hot, so it won't be that easy.

[Bryce]

Yes, a low impedance short could pull the 5V rail low enough to stop many things working, but if the rail on this device is at 5V then that's not the issue this time. I assume you've checked that the 5V rail really is around 5V?

Bryce.

[llopis]

Yes, a low impedance short could pull the 5V rail low enough to stop many things working, but if the rail on this device is at 5V then that's not the issue this time. I assume you've checked that the 5V rail really is around 5V?

Haha... such a basic check that I didn't do it because I was so excited to see the clock heartbeat.


I just checked and it's rather low. On the AY chip I only get 4V. But it's also only drawing 0.7A from the power supply, so there isn't a crazy short somewhere or I would have noticed that maxing out.


Still, 4V is low enough that I should look for the cause of that, right?

[Bryce]

Most TTL level ICs stop working just below 4.75V so many chips aren't functioning because of that at the moment. If you have a decent accurate meter with good resolution, you could check around the PCB to see if you can locate the issue. The voltage right at the input jack should be much closer to 5V. The lowest voltage you find will be very close to where the broken chip is. If the voltage at the input is also as low as 4V, I'd check that the switch and power socket are clean.


Bryce.

[gerald]

4V is  on the low side for a working CPC ...
Most of the time, if the ramtest does not start, you should check the 5V and for hot ram devices.Sometimes it's also the power switch that need a bit of cleaning.
As for the requirement of RAM test, only the Z80 / CRTC / GA need to work. However if the memory bus on the Z80 side is stuck by a faulty ROM or PPI/FDC it will not start.

[llopis]

Upon further investigation, 4 of the RAM ICs were fried. The Z80 is fine.


The DC jack was somewhat corroded, so I cleaned it up and resoldered it. Between that and replacing the bad RAM ICs, voltages are now around 4.5V, which is what I'm seeing in other 464s, so that's probably fine.


Still just getting the gray square (with a few bits of garbage in the middle) and the RAM Test isn't working from X-Mem. I suspect the 40010 GA is a goner.


I have a working 464 PCB of the kind that takes both types of GAs. So just for fun I put the 40010 GA in that one, and it didn't even give me a gray square, which adds weight to my theory. Hopefully I'll get a system with a 40010 one of these days to test it out.

[Audronic]

@llopis


The pinouts for the 40007 and the 40010 are different.
Do you have sockets on the board for Both types ??


Ray

[Bryce]

Upon further investigation, 4 of the RAM ICs were fried. The Z80 is fine.


The DC jack was somewhat corroded, so I cleaned it up and resoldered it. Between that and replacing the bad RAM ICs, voltages are now around 4.5V, which is what I'm seeing in other 464s, so that's probably fine.


Still just getting the gray square (with a few bits of garbage in the middle) and the RAM Test isn't working from X-Mem. I suspect the 40010 GA is a goner.


I have a working 464 PCB of the kind that takes both types of GAs. So just for fun I put the 40010 GA in that one, and it didn't even give me a gray square, which adds weight to my theory. Hopefully I'll get a system with a 40010 one of these days to test it out.

Don't put a 40010 in a 40007 socket. I haven't gone through each pin, but there's most likely some signals clashing due to the different pinout and it may damage something.

Bryce.

[llopis]

The pinouts for the 40007 and the 40010 are different.
Do you have sockets on the board for Both types ??

Yes, I know, don't worry :-) It's the board that it's prepared for both types. It was a Z70200 MC0008D board.I had never tried it, but I'm assuming it accepts both types of GA (in this case it didn't, but I'm guessing it's the bad GA).

[llopis]

One update and one interesting related fact.


1. I replaced the GA with another 40007 and it started working almost fine. I get a blue screen and the words "Read" and "rror" in multiple places on the screen. I suspect the CRTC is bad.


2. In the board that I got the new GA from, I noticed this.


This is the 7400 that was dead in the other board, and here is someone trying to do something with it. I wonder if they are a bigger source of trouble than I suspected. That's adding an extra capacitor from pins 9,10,11 (the input to the final stage that will generate the clock) to ground. There's one already there C130, so they felt the need for another one. What exactly is that doing? Dampening the signal down a little bit? I don't have a really good intuitive understanding of that set up.


(Yes, the board looks a bit wet. It just came out of a nice hot shower :-) )