16 Ko Ram pack common failure ?

[amauget]

Dear All,

I have two Sinclair 16 Ko Ram Packs which are out of order.
When connected to my zeddy, the screen is fuzzy or blank.
Here are samples of what I get :

Fuzzy screen.jpg (49.89 KiB) Viewed 8667 times

Fuzzy screen 2.jpg (56.47 KiB) Viewed 8666 times

Are there known failures with this particular hardware ?
Is there something to look at with an oscilloscope ?
I already unsoldered all the RAM chips from one board and tested them into another computer, there are all good. I have also checked the voltages. They are okay.
I'm concerned by the ULA chip. I imagine there is a way to replace it with LS chips...
I've seen I can replace all this stuff by using only two chips but I love to keep originals things in working condition.
I'm waiting for your wise advice

By the way, the boards are issue 3 models, as pictured below.

RAM Pack Issue 3 - comp side.jpg (120.5 KiB) Viewed 8666 times

[Paul]

I Also have several broken 16K RAM Packs.
I replaced them by something different. But I can fully understand that you prefer a repair.
I would make a modern SD-RAM on a PCB fit into the original Case.
But the repair of the original might also be possible.
Does anyone have the schematics of issue 3 RAMBoards?
The issue 1 (i think) can be found here:
http://www.retroisle.com/sinclair/zx81/ ... matics.php
Could be helpful for comparison.
As the ULA is 20 Pin, maybe (if power is on ordinary pins) it could be replaced by a GAL 20V10?

[Paul]

OOPS,
I forgot to ask:
before replacing any chips, they don't break that often,
are your voltages ok?
There should be -5 V at Pin 1 of the RAM chips, +12V at pin 8, +5V at pin 9 and Ground on pin 16. With some tolerance.
These funny voltage generators from siclair often dont do what they are supposed to do.
Also, they rely on the original PSU, regulated PSU don't have enough noise to generate the +12V and -5V.
Maybe you can measure and post your measured voltages?
And while you are measuring, could you give us the pin layout of the ula?
What pin of the ULA is connected to what pin of PSU, RAM, Connector or not connected at all?
Thanks
Paul

[RetroTechie]

Are there known failures with this particular hardware ?

Yes... old age.

These 16K*1 bit RAM chips (aka 4116 type) are prone to failure these days wherever you find them. And they weren't very reliable to begin with, largely because of needing 3 supply voltages (to be applied & removed in specific order) where most memory chips use only a single supply voltage. Just look at how many ZX Spectrums have broken down because of this exact reason. If the RAM pack still works those chips are probably doing what they're supposed to, but that doesn't mean their electrical behavior is the same as on day 1.

Those supply voltages are generated using a switching voltage converter, in true Sinclair fashion built as simple & with the cheapest components possible. And which includes electrolytic capacitors - also components that don't age well.

If you want to keep things in original state as much as possible (but fix this problem), I'd start with replacing those electrolytic capacitors. Almost missed this RAM pack has an ULA in it - I wouldn't worry too much about this chip.

I've seen I can replace all this stuff by using only two chips (..)

No, just 1 (a 32K*8 SRAM), if you want the basic 16K RAM. The extra logic comes into play when you want more than 16KB RAM (like: use all of that 32KB SRAM chip). The great thing about internal RAM upgrade is you do away with the unreliable edge connector issues (beside the RAM pack itself, and keeping that connector free).

[amauget]

Here are the voltages.
- 1st RAM pack : -5,09 V / 11,95 V / 4,85 V
- 2nd RAM pack : -5,32 / 13,03 V / 4,89 V

The pin layout of the ULA is in progress.

[1024MAK]

Here are the voltages.
- 1st RAM pack : -5,09 V / 11,95 V / 4,85 V
- 2nd RAM pack : -5,32 / 13,03 V / 4,89 V

These are within spec.
It is still a good idea to replace all the electrolytic capacitors to ensure the ripple is low.
Ripple on a supply line to a digital chip can cause problems.

Mark

[amauget]

A small update : all caps have been replaced.
The problem is still there.

[1024MAK]

If you are sure that in one of your units, all the 4116 DRAM chips are okay, that leaves a open circuit connection (dry joint or broken track) or one of the two logic chips (74LS157 and the ULA) may be faulty.

The ZX-Panda 16k RAM pack from Stonechip Electronics that I have uses 8 x Mostek MK4217N-3 (their 4116N equivalent 16k x 1 bit DRAM chip), 2 x 74LS157N, 1 x 74LS00N and 1 x 74LS02N.

Edit: I've just had a look at another non-working (non-Sinclair) RAM pack that I have. This one uses the usual 4116 DRAMs, 2 x 74LS157N, 1 x 74LS00N and 1 other chip to generate the extra supply voltages.

So if the ULA is faulty, it could be replaced with LS logic chips.

I have a non-working Sinclair 16k RAM pack somewhere, I will have a look for it later.

Meanwhile if you have a 'scope or a logic probe check the address lines going to the 74LS157 and the ULA, the multiplexed address lines between the 74LS157 / ULA and the 4116 DRAM chips, and the data lines going to the 4116 DRAM chips for "stuck" bits (lines that are always at a logic low or high level when they should be changing).

Mark

[1024MAK]

Here is a partial circuit diagram (schematic) of my issue 3 Sinclair 16k RAM.

The voltage generator circuitry is not shown.
The address lines from the edge connector go to the 74LS157 and the ULA. Then the multiplexed address lines from the 74LS157 and ULA go to the 4116 DRAM chips. Each data line from the edge connector goes to an individual 4116 DRAM chip (pins 2 & 14).

Mark

[amauget]

Thanks Mark, you beat me ! I had just finished the ULA pinout

The good news is I have found a bad connection on one of my RAM pack. It need a little soldering.
For the second, I have to use my scope.