Philips MSX2 NMS 8250 Faulty RAM: TestRAM shows only 64KB's of (128 KB's expected) mapper memory. Any hint?

Page 3/3
1 | 2 |

By friguron

Resident (35)

friguron's picture

19-03-2017, 13:22

sd_snatcher wrote:

About the DRAM chips: if only IC135 is defective (most plausible problem), then replace it with a DRAM chip of the very same maker and speed rating as the existing IC136. The speed rating is the number after the dash in the chip code.

Original MSX-2 NMS 8250 memory chips don't show any speed rating number written on them (just a raw TMM41464 ). VRAM chips OTOH do show AP-12 speed rating. Someone said some messages before something like: "8250 main RAM chips are rated 120 ns", so as far as I get AP-10 or AP-12 chips, I should be ok.

sd_snatcher wrote:

If you can't find an exact match, then you'll have to replace both IC135 and IC136. In this case, it doesn't matter which 41464 you'll get. Just check its datasheet to be sure that it has the same pinout of those that are on the board.

THAT's an amazing piece of info, thanks.

I have to say as I don't have the proper specs of the original memory IC's of my 8250, I'll get them from Toshiba datasheet and see what I can deduct from them... And anyway, as you said above, the safest approach should be getting some new memory chips, all of them being exactly the same one, and then replacing IC135 and IC136.

So, NEC D41464 C-10 chips should do the trick then? I ask for this specific model because apparently they're by far the most frequent ones you can find on the internet.

sd_snatcher wrote:

And it's highly recommended to solder turned-pin sockets so you can easily replace the chips in the future.

Turned-pin sockets are those with round holes. They are more expensive but also much better than the basic model with square holes. After you finish soldering, spray a good electronic contact cleaner in the sockets to clean any residues and assure a good connection, then insert the new chips.

I already have some sockets for the day I'll do the final desoldering/resoldering operation. They're not "rounded hole" ones, they're "normal" ones... I'll try to look for some rounded hole ones somewhere, but that shouldn't be such a big problem (compared with a proper desoldering of the old chip).

Thanks anyway, your answer is full of many useful info.

By sd_snatcher

Prophet (2295)

sd_snatcher's picture

19-03-2017, 13:55


compared with a proper desoldering of the old chip

Since you're not a professional in this art of soldering/desoldering (no offense here), I cannot emphasize enough that the best approach is to cut the old chips pins and remove them one by one. This reduces the risk of damaging the PCB to near zero.

To cut the pins of the chip, another tip is to use a the correct cutting plier for this job, specific for electronic circuits. Like this one:

I wouldn't recommend at all the use of an inadequate cutting plier like the more usual model pictured below. They're too "fat" for the job, and will mostly just chew the pins and cause too much tension on the tracks of the PCB, causing fractures.

By Jipe

Paragon (1191)

Jipe's picture

19-03-2017, 13:59

sorry for my very bad english

yes the 1st and the 2nd test valid the first bank of 64k is in IC133 IC134

the 3rd test is for unvalid the Nyyrikki post


If we disconnect R127 (CAS2) we are kind of disabling ic136 and ic135... If computer boots, lower ram is inside ic134 + ic 133 isn't it?

By friguron

Resident (35)

friguron's picture

19-03-2017, 14:00

Sure, when I face the chip replacing, I'll cut their pins as cleanly as I can with these tools.
I understand the risk of damaging the PCB and that's something I couldn't afford. I already decided about this some days ago.

Thanks anyway.

By friguron

Resident (35)

friguron's picture

19-03-2017, 22:57

Hi everyone,

I've just measured my NMS 8250 with my multimeter (I don't have access to my MSX2 frequently, just some days in the week, today was one of these days), and indeed, IC135 is the faulty chip for sure:

1.- Two pins are not perfect silver color, but slightly brownish.
2.- White serography on the faulty chip is kind of erased around these 2 pins.
3.- Definitive evidence: I've used a multimeter in diode mode, and measuring these two pins against many of the other pins of the same IC, measurement returns "open circuit". I've done the same test with these 2 pins on IC133, IC134 and IC136, and measurement wasn't "open circuit", but something like 500, 600, 700, etc...

Conclusion: I'm happy knowing the exact cause of my faulty memory, so I'll finally order around 5 or 6 41464 chips, some DIL slots, and as soon as I receive them I'll perform the substitution, just cutting legs, soldering slots, and inserting chips onto these new slots.
I feel bad for IC136, currently working, as I'll probably sacrifice it for a good cause Sad
I feel I'm breaking some precious electronics from the past "without reason".

Until I receive the new chips, thanks everyone.

Greetings from Barcelona.

PS: How can a random memory chip become broken in that way? Electrically all of them are connected to the very same VCC, but only this one got somehow burnt: Why can it happen? May I expect similar behaviour with the new working one in the future?

PS2: RAM and VRAM chips on my 8250 are exactly the same ones, and ALL of them have the AP-12 suffix on the chip description (unlike official Philips document, where the AP-12 suffix was only printed onto the VRAM mem chips, but not onto mapped RAM ones).

By friguron

Resident (35)

friguron's picture

20-03-2017, 10:14

BTW, I have the chance to get these 2 DRAM chips in Barcelona:

A kind forum member (hi @xavirompe :) might sell them to me if I decide they're valid...
Apparently, they're from a Panasonic MSX2... The only drawback might be they're rated 150ns instead of 120ns which is the speed of the original ones my NMS-8250 has...

If I replace both DRAM modules (IC135 and IC136) with these two 150ns pieces, should I have any problem?

I understand 150ns is a safe speed for a non-overclocked Z80 like mine...


Page 3/3
1 | 2 |
My MSX profile