ZX Spectrum 128k toastrack horizontal bars at power up

[Tiger]

On the 128K it is possible that TR4 (ZTX650 better ZTX651) and TR5 (ZTX213 or BC557) make this kind of fault. C111 and C112 are in this 12V (a.c.) rail, so I think the -12V / +12V isn't stable enough ...

[kbjunkie]

Hey Tiger,

Thanks for the info - I had replaced TR4 previously (as it was dead) which got the +/-12v back again - and both voltages appear to be just fine now (+12.02 and -11.99).

I'm not understanding about C111 and C112 - those capacitors appear to be in a part of the schematic unrelated to TR4/TR5 -

They look to be in a part of the schematic related to the AY-3 and serial driver (1488)?

Thanks,

KbJ.

[Tiger]

The connection 12 V a.c. is at the collector of TR4, there it is -12V a.c.. In the schematic it is only "separated" not "unrelated" and the "-" is lost in your separated schematic part. What's about the -5V at D19? Possible that you have to replace this 5.1V Zenerdiode - best check all diodes in this line. Is the measurement of the 12V just fine, when the fault appears? I would think this power drops if there are the horizontal bars ...

By the way - most ZX schematics in the www have little faults. The reason is that the originals are "handmade" and don't have the best readable quality so in the web-published schematics is often something "lost in translation".

[1024MAK]

The machine should still work if the line driver (IC33) does not get the -12V supply. So lift one end of resistor R137. Then power it up and see what happens.

The -5V and -12V supplies are not used for anything important.

Mark

[kbjunkie]

I've checked all of the diodes in that area out of circuit and all read fine. The 5.1 Zener has -5.02v on it when the machine is running.

I lifted one side of R137 and the machine still does the same thing. From cold, it will start up, white border, black centre and then drop into 48k basic. If I wait about 10-20 seconds and then hit reset, I get the boot menu and all appears good. No matter how many times I hit reset during that 'warm up' period, I still get either the drop to 48k basic, or the 128 menu appearing but not fully drawn..

Looking at Brendon's service manual (which isn't finished I guess) there's a section with voltages to check. On page 28 it says that there should be -12v (+/-3v) on Pin 1 of U34 - I get 0v.

It's a weird issue.. feels like a power issue or a cap issue but all voltages are good (+5.01 and +11.89 measured just now).

@Tiger - the 12v line, measured from a cold start (when the problem starts) is +11.89 volts and doesn't change at all, even when the machine starts to work again. +5v line is the same.

Thanks for the help guys

KbJ

[1024MAK]

On the toast rack schematic that I’m looking at, the chips are referenced with IC not U.

If you mean IC34 (1489), again, this chip is not critical. The machine should run with it missing (that is not an instruction to remove it). As this is just a receiving level shifter, it only needs a single +5V supply (which is on pin 14). Pin 1 is one of the input pins.

Or do you mean IC33? IC33 (1488), if being used, it should have a +12V supply on pin 14 and a -12V supply on pin 1. Again, this chip is not critical. The machine should run with it missing (that is not an instruction to remove it).

This is a weird one. Apart from the video output/colour encoder section, and the line driver for the RS232/AUX/Keypad ports, all the rest of the digital circuitry runs off only the +5V supply.

It may be a poor (as in intermittent) connection somewhere. These can be affected by both heat and by vibration. Either a poor contact in a chip socket, a dry solder joint, cracked/broken PCB track, or maybe a solder splash that is intermittently shorting between adjacent conductors.

The “48K” ROM area, when the Z80 starts executing from address 0x0000 (that is, the Z80 address), the ROM code instructions do the following: set the border to white, then it starts a memory check (designed only to find out how much RAM is fitted, not a full test). This test causes the ink and paper area to be filled to a black colour. Then very briefly, you may see vertical thin red lines.

If an error is found, it’s supposed to just get stuck in a loop, resulting in a “frozen” machine. But as it’s not a proper test, if there is a problem, anything can happen next. Including what appears to be a reset (but is just the Z80 executing the start up code again).

Note that the code in the “128 BASIC” ROM is different, obviously. But if a problem occurs when the “48K” ROM is paged in, as this ROM does not have a clue about the bank switching, the machine may stay in “48K” ‘mode’.

Mark

[Tiger]

Ok - the first thought of mine was that there is a faulty RAM. But your discharging experiments with C111 und C112 confused me so that I focused it to the inverter stage TR4/TR5.
As 1024MAK described IC 33/34 didn't make any sense for video signal - they are for the RS232 function.
Because your power measurements are ok, I would concentrate to check all the connections and contacts in the RAM area as 1024MAK told. Maybe there is more than the 2 replaced RAM IC faulty ...

[kbjunkie]

Hey guys..

Right, solved the issue..

I think I was being too focussed on those 2 caps.. I think that was a red herring all the time..

I checked out the address and data lines from the CPU with the scope and they looked a bit funky.. I know that that's normally the case with the Spectrum, but.. These waveforms did look especially bad. I could see that both busses had activity but they looked worse than the first time I'd looked at them a couple of weeks ago.

I took the ROM back out, stuck it in my programmer and verified it against a download of the 128 rom. It verified. I did that a few times and every time it verified fine.

I left it out of the Spectrum and stuck in Brendans ROM with my diagboard. Whilst it was running, I checked the busses again and they looked MUCH better..

I burned a copy of the ROM onto a 27C256 and popped it into the socket.. and guess what.. All working now. Issue gone.

I popped the old ROM back in and the busses were all over the place again.. although I did notice that they got better as the machine warmed up. I guess I have a ROM that's failing?

Anyhow - I really appreciate both of you for the help and suggestions you provided!

Cheers,

KbJ

[Tiger]

[1024MAK]

What type of ROM is it? Maybe post a picture of it

Is it a mask ROM chip (plastic DIL/DIP package, no quartz window)?

Or is it a EPROM (ceramic DIL/DIP package with a quartz window)?

Or is it a One Time Programable (OTP) EPROM plastic DIL/DIP package, no quartz window, but has a similar part number to a EPROM?

Mask ROMs are normally very, very good in terms of reliability.

EPROMs (including OTP) are internally actually analogue technology, so have multiple failure modes. As they age, their response time may get worse (a programmer may not pick this up), and/or they may forget some data (sometimes called bit rot). Because they are analogue, if they are borderline in terms of working, it’s certainly possible that they can become sensitive to temperature.

Anyway, glad we helped, and well done on fixing it

Mark