Lower ram

For experts to discuss very technical stuff and newbies to ask why the Spectrum they bought off ebay doesn't work.
Post Reply
dhester
Berk
Posts: 2
Joined: Sat Dec 01, 2018 11:15 pm

Lower ram

Post by dhester » Sat Dec 01, 2018 11:21 pm

I have just brought a spectrum 48K issue 3 and have an issue which I believe may be the lower ram. When I turn the Spectrum on the screen is black with a white border but there are red verticale strips.

I have read that you can "piggyback" a working ram chip to test which one is faulty. This is holding a chip on with your fingers while powering on. What I want to know is once I have found a faulty ram chip do I have to desolder it or can I just solder a working chip on top of the faulty one.
0 x

User avatar
1024MAK
Manic Miner
Posts: 480
Joined: Wed Nov 15, 2017 2:52 pm
Location: Sunny Somerset in the U.K. in Europe

Re: Lower ram

Post by 1024MAK » Sun Dec 02, 2018 1:49 am

Quick answer to your last question: I would recommend you cut the legs off the faulty / suspect chip. Then very carefully desolder the old pins. Then solder in a DIL socket. You can then plug the replacement chip into the socket.

The practice of piggybacking a spare / new chip on top of a suspect chip is something that has been around for ages. It’s reputation as a worthwhile fault finding tool has not got any better, as the results are so inconsistent.

A semiconductor chip can fail in many different ways. Starting with the simple symptoms:
Pin output driver faulty:
  • Output pin is open circuit, stuck in hi-z “tri-state” mode (where it will not drive the output), or now unable to drive high, or drive low - piggybacking will work if all pins of the piggyback chip make good contact.
  • Output pin always driving low, or short circuit to 0V/GND - piggybacking will fail and may damage the spare / new chip and will not change the fault symptoms.
  • Output pin always driving high, or short circuit to a power supply pin - piggybacking will fail and may damage the spare / new chip and will not change the fault symptoms.
Note that if the chip has suffered extensive internal damage, it is possible for output pins to short to other output pins, or to input pins as well as to 0V/GND or to power supply pins. I’ve seen this with some 74LSxxx logic chips. 4116 DRAM can also fail in such a way that the +12V or -5V supply voltages may leak out on the 5V “TTL” Interface lines (address or data pin(s)).

Input pin faulty:
  • A fault with one (or more) input pin circuitry causes one (or more) output pins to be logic high when they should be logic low - piggybacking will fail as the spare / new chip will try to drive the output pin(s) low in opposition to the failed chip. This may also damage the spare / new chip and will not change the fault symptoms.
  • A fault with one (or more) input pin circuitry causes one (or more) output pins to be logic low when they should be logic high - piggybacking will fail as the spare / new chip will try to drive the output pin(s) high in opposition to the failed chip. This may also damage the spare / new chip and will not change the fault symptoms.
In addition, there are many possible internal faults, any of which may cause the same or similar affects as a faulty input pin...

I’m not saying don’t try it. Just be aware of the problems with this method. I don’t recommend this method with rare, hard to get or expensive chips.

With DRAM chips, one thing that can be done to increase the success rate of piggybacking, is to cut the output pin(s) of the suspect DRAM chip. With the type of DRAM used in most 8 bit computers, each DRAM chip only has one output pin. If you cut the pin in the middle and leave a clean break, but with enough leg / pin left so that if needed, you can join both bits back together with solder. On the chip that will be piggy on top, bend out the output pin(s) into a curve, so that these will not touch the bit of pin(s) on the suspect chip, but will touch the bit of pin(s) still in the PCB. You may have to slight bend the pin(s) still mounted in the PCB out and away from the suspect chip a bit. A piece of 90g or 100g paper cut to size can also help as an insulator.

From the symptoms you describe, it does look like the Z80 CPU, the ULA and the ROM are working, but the Spectrum is stalling during the RAM check.

However, it may not be just a single DRAM chip that is faulty. It could be one, or two, or all of them. All of them will fail to operate correctly if the on board DC/DC converter/inverter circuit fails to operate correctly or it is faulty. Have you tested the -5V and the +12V supplies to the 4116 DRAM chips?

If all the supply voltages are correct, the quickest easiest way to find out which DRAM chips have failed, is to use a ROM based RAM tester. In most cases, this will tell you which DRAM chips are faulty.

Mark
0 x

dhester
Berk
Posts: 2
Joined: Sat Dec 01, 2018 11:15 pm

Re: Lower ram

Post by dhester » Sun Dec 02, 2018 10:30 am

Mark

Thanks for your answer, my initaal thought was desoldering the chips and putting in dip sockets, but was just trying to find a "quick fix"i.e lazy fix.

I have a temperature controlled soldering iron and a manual solder pump, think I might invest in one of those vacuum desoldering guns.

I think from doing a bit of research, i.e googling, that it is the ram, I have a few spare Z80's lying around as I have bulit a couple of RC2014 kits, and I am also thinking about changing the rom for an eprom so that could fix that if ihey are the problem. From what I have read getting ULA chips can be difficult so I hope it's not that.

Thanks for the informative answer.

David
0 x

User avatar
1024MAK
Manic Miner
Posts: 480
Joined: Wed Nov 15, 2017 2:52 pm
Location: Sunny Somerset in the U.K. in Europe

Re: Lower ram

Post by 1024MAK » Mon Dec 03, 2018 4:23 am

Keep in mind that a EPROM has a single /CS line, but the Sinclair ROM chips have two /CS lines. So either a logic IC is needed, or use two signal diodes and a resistor to form a wired logic gate.

This is a very good ROM based RAM testing program ;-)

Mark
0 x

Post Reply