I have an Apple IIe motherboard I am restoring. 6 of the 8 4164 DRAM chips have what ShadowTron Blog refers to as "pin rot". Meaning the pins are literally black, crusty and falling apart. I might be able to salvage them...maybe. I ordered some replacements off eBay but they were expensive.

Anyway, I got to thinking. A few years ago I ordered a DRAM replacement from a guy on YouTube that is an expert on the ZX Spectrum. His board is a modern SRAM replacement card that plugs into the ZX Spectrum via one of the DRAM sockets. This board handles the /CAS, /RAS, etc. and brings the ZX Spectrum up to 48K. I can't remember if the ZX Spectrum uses 4164's or something similar.

Many systems that use that type of RAM probably run in the 70-150ns range. Which makes me wonder why there haven't been more replacement cards for these systems. I guess the demand isn't high enough.

Seems like a small board with a CPLD and SRAM chip would be a good replacement option for these systems. In the case of the Apple IIe, you would have plenty of room to put a replacement board in.

Have you seen any such boards for these old systems?

_________________
Cat; the other white meat.

The original 16 and 48k Spectrums would usually have one bank of 4116s from 16 to 32k, and one bank half working 4164s to give the other 32k on the 48k system.
I'm not sure what the later model and Amstrad versions used.

IIRC, the ZX Spectrum used the 4164 that is 64K but half of it was bad. But I don't think they did that scheme for the entire lifespan of the Speccy.

_________________
Cat; the other white meat.

It should be pretty easy to make a CMOS SRAM solution with a UM61512AK and a couple of 8-bit latches.

_________________
Bill

Later ZX Spectrums (128K) models used sixteen x 1 bit by 64 bit DRAM chips (giving a total of 128K bytes).
Or four x 4 bit by 64 bit DRAM chips (giving a total of 128K bytes) in the case of the +2A/+2B/+3/+3B machines.

Wherever the DRAM inputs and outputs are connected together, in most cases, a SRAM chip can be used. If you are not able/willing to remove the address multiplexers, then a latch will be needed to hold half the address.

Mark

The Commodore 64 uses 4164s. A C64 board stripped of its VIC-II and SID chips ought to be cheaply available.

Looking carefully at the 4164 datasheet, there are some use-cases described that would be fairly complex to replicate around an SRAM chip. Demultiplexing the address pins would be the easy part; the whole concept of the timing is different, in that the /RAS and /WE lines are edge-triggered with respect to address and data inputs, whereas SRAMS normally expect level-triggered control signals and stable inputs.

It also looks to me as though the output is expected to be held at its original value if a "hidden refresh" (/RAS strobe while /CAS still low) is performed or a "read-modify-write" (/WE strobe while /CAS still low) occurs. So you can't simply assume that latching the high half of the address on every /RAS fall and hard-wiring the SRAM's DQ line(s) to the data I/O pins is appropriate - it might work for some replacements in which those two features go unused, but may result in shorts and malfunctions in other cases.

What's more, the 4164 is a single-bit slice of RAM where most modern SRAMs are 8 bits wide, so the logic would either need to be repeated 8 times, or patch wires extended to the D/Q pins of the other 7 sockets from a master daughterboard fitted to one.

None of this is impossible, of course; it just requires careful implementation. Probably a GAL16V8 and three 8-bit latches could be made to adapt a 64Kx8b SRAM to replace 8x 4164s with some patch wiring.

Other options may include finding directly 4164-compatible chips, not all of which are marked "4164", as this chart illustrates: http://www.minuszerodegrees.net/memory/4164.htm A newer FPM or EDO DRAM chip could also be used with the caveat that these tend to have shared I/O pins with separate /WE and /OE lines, so the 4164's behaviour of holding the original output valid during a late-write cycle would require latching to emulate.

Regarding the SRAM replacement...

That's something I'd like to do one day but not anytime soon.

FYI, this is the exact board I bought a few years ago:
http://zx.zigg.net/URR/

On one of the RAM chips in one of my IIe's, I was able to literally scrape the black gunk off the pins. So I may have salvaged one of them. The others might be possible.

One chip, a mux IIRC, was so bad that one of the pins "wiggle" when you touch it. So it's about to fall off. But I found a new replacement for it on Mouser.

_________________
Cat; the other white meat.

Speaking of ‘black gunk’, one problem with some older DIL chips, is that the manufacturers used silver in the alloy of the pins. Over time this will tarnish. The final colour being black. This is normally mostly on the surface. I use a glass fibre cleaning ‘pencil’ to clean it off. If the chip is in a real bad way, you will also see some pitting of the legs as well.

Mark

I need to get me one of those pencils. I've been using normal pencil erasers to clean NES carts.

I thought about dropping a dab of solder on the really bad pins (after I clean the gunk off). Just to fill in those pits. If I can still buy new stock for the chips, I don't even bother. I'd rather pay the $0.40 to replace it. Unfortunately, there are no newly produced 4164's any more.

_________________
Cat; the other white meat.

There are lots of vendors on fleabay selling NOS or nicely refurbished 4164s and equivalents.

I believe the old 'eastern block' countries may still even be making them.

_________________
Bill