Issue 4A issues

balford · June 2013

Hi all,

I've been trying to get the Issue 4A board that came DOA with the case I purchased for my Frankenspectrum to work.

I've recapped and socketed all chips, and verified that the lower RAM IC sockets IC6-IC13 are getting correct voltages. However I get the screen in the following video when powering up:

https://www.youtube.com/watch?v=RkjpvkFEKXw

I understand this is indicative of a Speccy with no lower RAM present at all, however all the 4116 chips are in their sockets when this test is performed.

Removing the ROM modifies the pattern to a more uniform herring bone pattern without the periodic vertical stripes, and removing the Z80 results in a bright white paper (as does holding the Z80 in reset). I've tried two other Z80's with the same result, and the ULA works fine in my Issue 3.

I've also removed the upper RAM chips and the associated logic in IC23-IC26, so this should be behaving as an unexpanded 16K for the moment.

I've got some 4116's on order, but does the fact that I see the above pattern mean that all the existing chips are toast? Any other suggestions for fault finding this one?

1024MAK · June 2013

The video does not show the classic faulty or no "lower" 16k 4116 DRAM image.
This looks to me like there is a problem with either the data bus, or the address bus.

Remove all socketed chips and use a meter to test for continuity and also for short circuits. When beginning the hunt for short circuits test each pin of each RAM chip socket to the pins next to it, then to all other pins on the same socket (but keep in mind the data in and data out pins should be linked).

Also there are known to be some errors in the Sinclair schematic diagrams :cry:
So write down anything that does not seem right...

Which chips were removed and replaced with sockets other than the 4116 DRAM chips?

Mark

balford · June 2013

1024MAK wrote: »

Which chips were removed and replaced with sockets other than the 4116 DRAM chips?

All of them :smile:

My intent was to get this working as a test bed just in case any other Speccys in my collection suffered problems, hence the socketing.

I've desoldered the sockets and visually checked for shorts, but I'll try your suggestions next.

Thanks!
B

1024MAK · June 2013

balford wrote: »

All of them :smile:

My intent was to get this working as a test bed just in case any other Speccys in my collection suffered problems, hence the socketing.

I've desoldered the sockets and visually checked for shorts, but I'll try your suggestions next.

Thanks!
B

Hmm, the idea is to find the problem using the meter. It is possible that in desoldering some sockets, that you have disturbed the offending solder splash. Keep in mind that some solder splashes are very fine whiskers and may not be easily visible to the eye..

Mark

balford · June 2013

Hmmm. I must have done something like this, as pin 9 on all 4116 sockets is shorted to pin 4 (RAS). Since this is also shorted to pin 35 on the ULA, I'm guessing this isn't good :sad:

1024MAK · June 2013

Not sure if you have these, but if not here are the links...

ftp://ftp.worldofspectrum.org/pub/sinclair/technical-docs/ZXSpectrumIssue4ab-Schematics.gif
ftp://ftp.worldofspectrum.org/pub/sinclair/technical-docs/ZXSpectrumIssue4a-Layout.gif

Mark

balford · June 2013

Thanks Mark, that's been really useful :-)

I've located and eliminated a couple of shorts, and I now have a machine that's attempting to reset, but by the colours when I hold it in reset when power is applied (two pixel width bar in magenta on black), I reckon IC6 and IC7 have had it.

More when I get the replacement 4116's... :D

Edit: Just so I know in the future, what's best practice for desoldering chips so as to avoid (as much as possible) solder splashes? I used a B&Q heat gun on the low setting to remove the IC's, then cleared the pins using a solder sucker.

Cheers,
B

1024MAK · June 2013

Solder splashes can result from both desoldering and soldering/resoldering operations.

The best aid to help prevent solder splashes is for the PCB to have a really good solder mask on both sides of the PCB... But this adds to the cost, so Sinclair did not do this.

If I ever find a bomb proof preventive solution, I'll let you know...:p

In the meantime, it comes down to taping off / protecting areas where you are not working.

For soldering, using a good temperature controlled iron helps, as the tip does not over heat (which can cause the solder and flux to spit).

Regularly clean the solder-sucker, as these have a nasty trick of dropping solder "dust".

After you finish desoldering or soldering, use an old toothbrush and clean the board with isopropyl alcohol (this helps to remove the flux). Then carefully check with a magnifying glass. Finally test with a meter on the resistance range.

At start-up, the border and centre screen have random colours. One of the first things the Z80 CPU does is to set the border to white. This is before it accesses any of the RAM.
What happens next depends on the working state of the RAM. The ROM program then attempts to perform a simple RAM test. This causes the centre screen to change to all black, followed by thin red vertical lines being drawn on the black screen.
If the CPU finds some RAM is good (and remember, the ROM starts the CPU testing the RAM from the top {address 65535} down), it will clear the centre screen to white and display the copyright message.

Mark

DEATH · June 2013

You'll get a screen display like that sometimes when the ROM is buggered too, or isn't getting power etc etc.

balford · June 2013

So just to update this one...

I had some 4116 chips on order from the states, which haven't arrived yet. Being the impatient <expletive> I am, I modded a spare 4164 from an old +2 to act as a test chip:

I swapped this around the lower RAM sockets until I got a sane screen output powering on with the machine held in reset.

Releasing reset, and... WTF

Looking closely at the bottom left you'll see a couple of corrupt screen bytes - these sometimes appear, sometimes not.

Anyway, next check just in case is to reswap the ULA from my Issue 3, and guess what.. IT'S ALIVE

I must have messed up the ULA check initially, as I was sure I'd swapped it, but now it's pretty solid - this fault follows the ULA over to the Issue 3 board.

So the damage here is one 4116 and a 6C001E-7 ULA.

Parts are on order, hopefully we can call this one done shortly. Need to find it a case, since I evicted it from its original home to build my Frankenspectrum.

B

balford · July 2013

Just to put this one to bed, one new 4116 and a replacement ULA later and this is now fully functional.

Anyone got a Spectrum+ case for sale to give this board a (new) home?

balford · July 2013

Bump. Sourced a rubber key case for this board so it's no longer naked, we're not quite done yet though as I have a noise issue on composite out:

As can be seen, this is present on the paper area only and doesn't move or flicker much with CPU activity or reset. It's most prominent on yellow paper, but can be seen regardless of the colour.

The board has had all capacitors replaced, 7805 and coil swapped out, and TR1/TR2/TR4/TR5 swapped for spares (which didn't change anything, but BC547 NPN parts I tried in place of TR1/TR2 made text display quite blurred).

Any ideas as to what to try next before I dive headlong into the schematics?

Thanks,
B

DEATH · July 2013

TR1 and TR2 - BC547. Do you get this screen effect with the original ZTX transistors also?

Check those transistors - are they 547's or 547B's? The B suffix has a higher gain - and the BC549 has a low noise figure, so try them also if you get the chance.

You will get that pattern on an unmodified Spectrum, and it's perfectly normal - in fact I believe Spectaculator emulates that very effect.

balford · July 2013

Hi Death,

The screen display shown is with ZTX313 spares in place of TR1 and TR2. The BC547C's I tried led to a blurred display (only other generic NPN's I had to hand).

This is my only composite modded machine though, so it's quite possible I'm not used to the clarity that reveals these artifacts :-P

Thanks for the advice, much appreciated!
B

DEATH · July 2013

By the way, to avoid confusion: The ZTX313 has a gain of around 40. If you get a chance to try BC549's, let me know how you get on.

roko · July 2013

My guess would be a still existing problem with the decoupling of the power lines +5, -5, and +12 Volts. In relation with the power consumption of the 4116ses.

balford · July 2013

Death: Tried BC549C's which gave a slightly (only very slightly though) better picture than the ZTX313's in situ.

Roko: That's my thinking too. I swapped out all the 4116's with no change. I'd need a scope to troubleshoot this further, which I don't have, so I can live with it for the moment :smile:

1024MAK · July 2013

I have seen the faint vertical lines on the screen on 16k and 48k Spectrums. I'm sure I have seen comments from people speculating that the core problem is slight fluctuations of the power rails in the ULA (a form of ground bounce maybe).

Mark

balford · July 2013

Problem solved! (or at least masked...)

I came across this thread which describes similar interference following replacement of lower RAM with modified 4164's, and how replacing C5-C8 with higher capacitance parts removed much of the noise.

I'm using period correct 4116's, however on a whim I changed out C5 to C8 for 1uf caps, and here's the result:

Still a small hint of noise, and of course the underlying source of the ripple hasn't been addressed, but as a quick fix it'll do nicely :smile:

B

Issue 4A issues

Comments