Sharp MZ-80K help needed on repair

[pudwink]

This is what I am getting with the ram tester

[pudwink]

been playing with it this morning I either get the ? screen most times or slightly corrupt screen or just plain black screen.

[TonyDuell]

I would have thought that the fact the RAM tester program runs and produces a sensible display implies that :

The Z80 CPU is running, and can execute machine code programs

The Z80 can write to the video memory

The video memory is basically working

The video display hardware can read out the video RAM and display it correctly

(The video monitor is working too)

That means that quite a bit of the machine is working correctly, which is good.

What happens if you switch off, pull out one of the RAM chips and power up again. Do you get the same display or does the '?' change to something else?

[Timbucus]

Looking at the above I agree with Sirius it looks like huge swathes of the machine are working fine... I think we can perhaps suspect the address decoding circuitry - I will try and explore specifics later when I get back home and can look at the circuit again.

[pudwink]

Taking out a ram chip made no difference still got ? I did once get this display as shown in the pic which I am guessing what the icons should be showing but I dont think it would be a reliable state of the memory.

[SiriusHardware]

It seems we all agree there isn't much wrong with the machine now.

Interpreting that display - it looks like a very good and useful utility actually - we can see from the 'key', bottom left, that it should indicate a blocked-in circle or 'dot' where the RAM device is good (=OK), and an 'X' where the RAM is bad (=KO).

Instead it's showing '?' which I'll take a guess means 'not found', ie, none of the locations within that device returned a valid value when written to and read back from.

Given that the original RAM, all of it, was subjected to +19V, it would be entirely possible for all of those devices to be damaged so I would be interested to know what happens if some of the recently purchased RAMs are swapped in instead, if that has not already happened.

However there is also the possibility that there is a damaged buffer sitting between the CPU and the system RAM, and that is making all of the RAM look faulty. I haven't looked yet to see if there is such a buffer, just going to have a look at that now. It's also possible that there is a problem with the RAM refresh circuitry or perhaps as Tim suggested, the address decoder

I'm also still interested to know if the RAM test ROM runs correctly every time if the machine is reset by shorting the reset switch contacts rather than by powering it off and on.

Edit: Crossed with pud's last post. Maybe there is one, or several dud RAMs which are messing up the rest. Try methodically replacing one of the existing RAMS at a time to see if you can consistently get that O/X display instead of the '?'s.

[SiriusHardware]

There is a buffer (IC50) between the RAM data outputs Do0-Do7 and the CPU's D0-D7 bus but the outputs of the monitor ROM are also on the Do0-Do7 bus, so the fact that the machine is (mostly) able to run code from ROM suggests that data from both RAM and ROM is passing through this IC OK.

[pudwink]

All the ram chips in the machine fully passed the dram tester I have unless some of the sockets are dodgy I don't know if we get a consensus you think it's the ram then I will replace all the sockets which will be a pain. Unless in the short term someone suggests something else to try first.

[TonyDuell]

You have a 'scope, I believe. I'd look for 'activity' -- square waves or pulses -- on the pins of the DRAM chips. I am guessing these are 4116-like devices (sorry, I don't have the MZ80K service manual to hand, I have an MZ80B...) in which case :

Power (check these with your multimeter):
Pin 1 : -5V
Pin 8 : +12V
Pin 9 : +5V
Pin 16 : Ground

Address inputs, there are 7 of this on pins 5,6,7,10,11,12,13. All should be showing changing logic states

Data In : pin2 and Data Out : pin 14 should also almost certainly be changing

RAS/ : pin 4 should be changing state, it would even be used for refeshing the RAM even if the CPU is not trying to access it

CAS/ : pin 15 and Write Enable : pin 3 could be high all the time if the machine is not accessing that part of memory, or if it is not writing to memory. I would certainly expect them to do something after a reset, it may be worth wiring a push-to-make switch to the reset switch 'holes' on the PCB so you can reset the machine easily.

[SiriusHardware]

Tony, the service manual (quite a decent manual and a good quality scan) was posted in #3 of this thread.

pudwink: What Tony said. Try scoping all of those pins on every one of the RAMs. You could also turn the power off and patiently use your meter to check that every RAM IC pin is connected (zero ohms) to its associated top side PCB pad - this may avoid you having to replace any or all of the sockets. It would be tedious, but not as tedious or expensive as replacing all the sockets on the off chance that one or more of them might possibly be faulty.

If you do want to replace the sockets (and we are not suggesting you do yet) then the way which has been suggested here before is to snip out the cross spars on the original sockets so that each socket becomes two single inline sockets, then rock those gently back and forth from side to side until all the pins snap, then desolder the pins / clear the holes individually. Note this method should NOT be used to remove turned-pin sockets, but I don't think they will be turned-pin types in something this old.

[Mark1960]

Try removing all the ram, check what you get on the display.

Then fit only one ram after testing it first in your ram tester.

You could also try that single ram in multiple positions to check it gives the same result.

If this seems to show you have a good ram chip, try again with different ram chips, only one fitted, until you have enough to try a full bank.

[pudwink]

With the ram test eprom I got it through the sharp MZ Facebook group and apparently the guy who gave it me says (the upper left digit should progress from step 0 to 3 then reload he says it seems to be stuck and could be an issue with VSYNC polling) whatever that means.

[SiriusHardware]

"Polling" in general means a method whereby the system continually looks at a signal to see what state it is in and responds accordingly.

Vsync is the signal which occurs at the start of every video frame and your advisor is suggesting that the RAM test program uses the Vysnc signal for timing / synchronisation purposes and can not proceed because it is not 'seeing' the vsync signal.

In hardware terms it looks like the MZ80K uses port pin PC7 (pin 10) of IC5, the 8255 IC, to read the vertical blanking signal, V-BLANK, so with the RAM test ROM running, try scoping IC5 pin 10 to see if you have a waveform there. I think it will be mainly 'high' with short 'low' pulses so you may have to 'tune around' with the Time / Div control in order to see them.

On this particular waveform you may get a better 'picture' if you locate the rotary trigger threshold control, upper right on the front panel, and pull it outwards towards you - that tells it to trigger on the falling edges of signals rather than the rising edges.

It might also be worth looking at the V-GATE signal on IC5 pin 14.

[pudwink]

I have hit a problem I took the board out to wire in a reset button all done correctly put the board back in and got the lines on the screen back.I redid the eprom and now all I get is this.I have removed the reset switch as well no difference.

[pudwink]

I took the board out again I have replaced the M-Rom socket again just in case and reflowed the Z80 socket just incase pulling the power supply in and out (its quite tight)might of cracked a joint and turned it back on got my ram-test screen again thought I had sorted it but turned it off and on again and back to the screen in my last post.
I have put the reset switch in again but dont think the reset is working if I press it sometimes it blanks the screen press it again and screen comes back with different garbled screens.I also put the oscilloscope on pin 26 of the Z80 and it just remains high pressing reset does nothing.

[pudwink]

Just checked IC5 pin 10 photo 1 pin 14 photo 2 IC 5 is not in the best condition and I had to replace 2 of the legs because of corrosion.

[Timbucus]

That VGate does not look healthy on pin 14 - it should be +5v or 0v not at 1v - it ends on IC24 pin 1 - so perhaps also worth checking the signal on Pin 3 of IC24 - in that state I cannot see how it will act at all well - unless I have missed something. That gate after the inverter on ICT21 ensures the blanking signal is gated correctly on the three input NAND IC17 but, we seem to have working video - so this fault condition could account for the Black video as it is always blanked.

What model EPROM programmer do you have? If it is a TL866 then it has logic gate test ability so removing IC24 and testing would not hurt. While it is out scope Pin 14 again to ensure it is the full 5v - there may be other points it terminates on I have missed.

[pudwink]

Pin 3 of IC 24 shows as per picture I have a GQ 4x4 programmer but as I will destroy the chip getting it out as its not socketed and I have no replacements will have to leave it in.
Do you think it is worth ordering a replacement NEC D8255AC-5 (IC5) to see if it makes a difference as I said I had to replace two legs because of corrosion.

[Timbucus]

That as I suspected is the reason if you had a black screen at the time. It is hard to say if it is the 8255 not pulling it up to 5v or IC24 pulling it down to 1v. If the 8255 is socketed then substituting it could be worthwhile but, don't leave it too long if you find it only has the 1v on pin 14.

As the first step of removing IC24 you would plan on snipping the pins so if you snipped pin 1 near the top and bent it out - it should return the signal to full 5/0 movement and indicate if a new 8255 was a good idea. It would be possible to blob solder to restore pin 1 to test the new 8255 fully then - a little destructive but, reversible.

[SiriusHardware]

Revised version of Tim's suggestion above: If the 8255 is in a socket which it sounds like it is, remove the 8255 and carefully bend pin (14) out at an angle so that when you plug the chip back into the socket, the pin hangs out over the edge of the socket so that it is not connected to anything.

Then try measuring the waveform on 8255 pin 14 again - it it still only going between 0V-1V or is it now going between 0V and ~5V?

Your other problem - half the screen showing scrambled characters instead of the correct ones - seems like a different new issue with the video display system, specifically,

-The system is displaying the contents of the screen RAM correctly but is no longer able to write data to the first half of the screen memory.

or

-The system is showing the second half of the screen memory correctly but is no longer able to show the first half of the screen memory correctly.

You will need to fix this problem before fixing any other because you need a fully working display in order to run diagnostics on the other part.

The introduction of this new half scrambled display fault sounds like a case of cause and effect - what exactly did you do just before this fault appeared, did you replace the CPU socket or the MROM socket? If so, you need to do a close inspection of what you did and if necessary meter out all the tracks going from whichever device's socket you replaced just before the fault appeared.