low memory fault test

amateus
Posts: 24
Joined: Wed Mar 21, 2018 2:28 pm

low memory fault test

Post by amateus »

Hi all,

I have a 48k issue 4b with a missing / low -5v. I have replaced tr4 and tr5 and replace all caps, still to now avail. I checked the ics and ic11 is a lot more warmer then the others.

Assuming there is an ic with a short that it's pulling down -5v, and assuming this is the only problem, if we take it out (by cutting the pins or by actually desoldering) and measure voltages in another ic, -5v should read normal again correct?

Thanks,
António

User avatar
1024MAK
Posts: 2989
Joined: Mon Sep 26, 2011 10:56 am
Location: Looking forward to summer in Somerset, UK...

Re: low memory fault test

Post by 1024MAK »

If IC11 is faulty, then yes cutting it out may remove the overload on the -5V supply.

Before you do that, what screen display do you get? Post a photo here.
What power rail voltages do you get:
  • Input 9V (nominal) voltage?
  • +5V voltage?
  • +12V voltage?
  • -5V voltage?
Can you also take a photo of your board showing the corner where the speaker is mounted please, and post it here.

Mark

amateus
Posts: 24
Joined: Wed Mar 21, 2018 2:28 pm

Re: low memory fault test

Post by amateus »

Ok thank you for clarifying my doubt.

So some background. This is my teen spectrum, a issue 4b, never had any problem back in the days, so never needed repair. The only intervention was a somewhat comp mod with a switch which allowed to use a tv or a monitor.

Initially didn't have power, so I changed the voltage regulator. Change tr4 and tr5 and all electrolytic capacitors. Also changed D19.

I now have 12v and 5v present, but -0.76v on -5v rail. IC11 is warmer than the others, not so hot too burn (I never let it running long to allow), but warmer.

The image is not static, and I can hear some buzzes from the speaker.

Pictures
104431907_2688499871421046_6003480824944677776_o.jpg
105503339_300711650971802_7405458491612208222_n.jpg
Thank you,
António

User avatar
1024MAK
Posts: 2989
Joined: Mon Sep 26, 2011 10:56 am
Location: Looking forward to summer in Somerset, UK...

Re: low memory fault test

Post by 1024MAK »

What voltage do you get at the junction of D11 anode / D16 cathode / C46 negative lead?

What voltage do you get across C79?

Is it possible that the buzzing noise that you hear is coming from the coil (small transformer) rather than the speaker? Often the coils buzz. If it is the coil buzzing, that’s normal, and is a good sign.

Mark

amateus
Posts: 24
Joined: Wed Mar 21, 2018 2:28 pm

Re: low memory fault test

Post by amateus »

Hi Mark,

Thank you for the help. Definitely the speaker. It's not so much as a buzz, but more as the sound of the keys repeating quickly.

Anyway, I measured the voltages several times and it change a lot (didn't use to) around those.

From ground (faceplate) to (assuming I have identified all correctly - see image):

D11: -6.5 to -6.0 (I did record -4.7v yesterday, for example)
d16: ~10v
c46 negative: -6.2 to -6.0
c79: ~14v

7805 reads 5v and 9.7v

So not really sure what to make of it :) Should I try replace D11 and D16?
If so, can you please confirm me references:
D11 - 1N4148
D16 - my manual says issue 4b does not use it which is wrong, I guess?

Thank you once again,
António
Attachments
105025991_560363191257255_9121529690862870399_n.jpg
Last edited by amateus on Mon Jun 29, 2020 4:33 pm, edited 1 time in total.

amateus
Posts: 24
Joined: Wed Mar 21, 2018 2:28 pm

Re: low memory fault test

Post by amateus »

Also, if it helps, I did a continuity test between c79 neg and -5v but gives me resistance, around 1.3k. I got 0 c79 neg to r79.

Is this ok? shouldn't it be 0 as well?

Thanks

User avatar
1024MAK
Posts: 2989
Joined: Mon Sep 26, 2011 10:56 am
Location: Looking forward to summer in Somerset, UK...

Re: low memory fault test

Post by 1024MAK »

Last year I put a page about the ZX Spectrum DC-DC converter power supply circuitry on my blog page here. Your issue 4B should already be using the same or a very similar version. Note that some components may be numbered differently.
amateus wrote:
Mon Jun 29, 2020 11:15 am
Definitely the speaker. It's not so much as a buzz, but more as the sound of the keys repeating quickly.
So the Z80 is most likely executing garbage then.
amateus wrote:
Mon Jun 29, 2020 11:15 am
D11: -6.5 to -6.0 (I did record -4.7v yesterday, for example)
d16: ~10v
c46 negative: -6.2 to -6.0
c79: ~14v
So those results are okay (apart from the voltage on C79, the other voltages will vary, as you are measuring half wave rectified AC).
amateus wrote:
Mon Jun 29, 2020 11:15 am
7805 reads 5v and 9.7v
Both okay.
amateus wrote:
Mon Jun 29, 2020 11:15 am
So not really sure what to make of it :) Should I try replace D11 and D16?
If so, can you please confirm me references:
D11 - 1N4148
D16 - my manual says issue 4b does not use it which is wrong, I guess?
No, I think the diodes are okay.

With the power OFF measure the resistance of resistor R79. It should be a 2.2kΩ resistor, but due to other components in the circuit, it may measure less than that value. If it tests between 1kΩ and 2.2kΩ, it’s likely to be okay.

In which case, the next step is to cut out the suspect DRAM chip (IC11).

Mark

amateus
Posts: 24
Joined: Wed Mar 21, 2018 2:28 pm

Re: low memory fault test

Post by amateus »

Thank you Mark,

R79 (D12 on my board) has a 2.15kΩ.

I guess that IC11 is going out (and in a good time, as I've just received the sockets and 4116 drams I ordered).

I'll cut IC11 and measure voltage again.

I'll keep you posted.

António

amateus
Posts: 24
Joined: Wed Mar 21, 2018 2:28 pm

Re: low memory fault test

Post by amateus »

And 5(!!) IC's after, I have -4.97v. IC 6, 10, 11 12 and 13. One after another becoming hot as I took the previous Uff

I'll take sometime to desolder the pins and to solder the sockets, as this is a work I've never done and I want to do it with a lot of patience, testing continuity along the way. My worst fear: destroying a track :?

Anyway, thank you so much Mark for your help on testing the board.

I'll keep you posted on the outcome.

António

User avatar
1024MAK
Posts: 2989
Joined: Mon Sep 26, 2011 10:56 am
Location: Looking forward to summer in Somerset, UK...

Re: low memory fault test

Post by 1024MAK »

I’m not surprised. The -5V supply is vital to the health of the 4116 (or equivalent) DRAM chips. If it is missing or is well out of specification while the +12V supply is present, these chips can self destruct :(

My recommendation is to cut each leg of each of the defective chips as close to the plastic body as possible.
Then once all the legs have been cut, remove the chip.

Then you can heat up each pin one at a time, once the solder melts, remove the pin with long nose pliers. Only pull very gently, if you need force, not all the solder has melted yet and you risk damaging the tracks or pulling the through-hole plating out with the pin. Once all the pins have been removed, clear out the remaining solder from the holes using a solder sucker / solder pump or solder braid. Some extra liquid flux can help. If you don’t have any, try applying some new 60/40 (tin/lead) solder, then try again to clear the hole.

Mark

Post Reply