National Semi. NMC27C33 EPROM Info ? (Being sold as Fake (ST) M2732A-2F1 etc)

[ortek_service]

Yesterday, I tried to program some (marked as) ST M2732A-2F1 'Fast' (21V Programming) (200ns access-time) Windowed Ceramic package EPROM's (Date-coded 2013 Week 22?) - on a Dataman 48Pro+ for ChrisOddy (after he'd just bought these from China via an online auction site and found they were giving over-current errors etc in his TL866 programmer, so managed to get a refund).

I found these all kept failing the programming on the Dataman 48Pro+ at address 0. However, I discovered that location 800h was getting programmed, instead of location 0! But the 'Intelligent' programming algorithm was reading contents of location 0 after trying to program it, before moving onto next location, so stopped after a single byte readback verify check on this (Although strangely its displayed buffer address kept counting rapidly upto random 0Bxxh / 0Cxxh locations, even though failure error message gave address =0000h)

So, thinking that the upper address bit seemed to be stuck-high when programming, I then tried programming just 800-FFFh, and this all worked! - But readback then showed 000-7FFh was also programmed with the same contents as the upper-2KB 'half'.

This led me to think they may actually be 2716, that had been re-marked, so I tried setting device to a 'Generic 2716' (Discovering for the first time, that Generic in Device Operation allows you to change the programming voltages & Timings - Which I never realised you could do with these later Dataman Pro (Elnec) ones, only that you could often change these on the original Dataman 48(LV/XP) 'Advantech' software).
However, it didn't seem to program at all as a 2716.

I then had the idea to look through the window with a high magnification 'Biological' >1000x lens on a Digital Microscope I'd recently bought-in. And after being able to focus on various odd numbers around the edges of the die, I then found the National Semi logo further inwards from one edge along with '1989' on an edge. Before suddenly spotting 'NMC27C33(E/F?)'
So this confirmed it wasn't really an ST 2732A(-2F1)! - although they look professionally-marked as such.


Googling for the mystery 'NMC27C33' only finds a die-shot image (also attached a copy, that will have been compressed a lot) - also from a counterfeit ST M2732A on a 3yrs ago Tweet by the famous IC Reverse-Engineer Ken Shirriff: https://twitter.com/kenshirriff/stat...22111377661952
- Where someone has replied that they do actually program OK as a NSC27C32B (12.75Vpp), on their 'Needhams' IC-programmer. So trying to program these at the marked 21Vpp may well cause problems.

So I was wondering if anyone has ever seen a genuine NMC27C33, marked as such on the package?
Or any Info on it? - As it seems very elusive, with no published datasheets?
(I've only ever seen 27C513, that were a paged-Mode? version of the more-common 27(C)512 - So I thought it might be a similar idea with these)

And if you're buy any 2732's - particularly ST M2732A's from China etc. via online marketplaces where many M2732A's are being listed, then it seems you need to be very wary of these fakes that have been around for many years. And these don't all have the same date codes / speed grades etc. even though they look to have been quite professionally re-marked, as no obvious signs of being 'black-topped' or re-labelled (Although underside, where there's often some markings, looked like it had been polished a bit to remove these).
But you'd think there must be a finite supply of the NMC27C33 originals, which presumably were an abandoned preliminary product?

[ortek_service]

Originally Posted by ortek_service

>> ... found the National Semi logo further inwards from one edge along with '1989' on an edge. Before spotting 'NMC27C33(E/F?[/B])'. So this confirmed it wasn't really an ST 2732A(-2F1)! - although they look professionally-marked as such.
Googling for the mystery 'NMC27C33' only finds a die-shot image also from a counterfeit ST M2732A on a 3yrs ago Tweet by the famous IC Reverse-Engineer Ken Shirriff https://twitter.com/kenshirriff/stat...22111377661952
- Where someone has replied that they do actually program OK as a NSC27C32B (12.75Vpp), on their 'Needhams' IC-programmer. So trying to program these at the marked 21Vpp may well cause problems.
>>
>>

Well I tried earlier today to reprogram one of these 'NMC27C33' (marked incorrectly as an ST M2732A-2F1), on a (Elnec-designed) Dataman 48Pro+, but there wasn't any NSC27... devices listed amongst its supported ones
However, it did support the 'NMC27C32B', and when I tried this it programmed one straight away with no-errors"
(With a fixed AAh byte fill at least).

I couldn't seem to find what Vpp the Dataman 48Pro+ was using, as control software doesn't seem to show it anywhere!
- but I assume it was using 12.75V that the B suffix seems to indicate on 2716/2732's (With A suffix on these usually for 21V and no suffix normally for the full original Vpp voltage).

So I'm wondering if putting 21Vpp on it is rather messing-up its address pin levels / decoding logic etc. causing it to write to a different location than what it reads back with Vpp switched off? But fortunately they do seem to have survived having a Vpp so much higher, being applied.

Originally Posted by ortek_service

>>
And if you buy any 2732's - particularly ST M2732A's from China etc. via online marketplaces very wary of these fakes that have been around for many years.
>>
But you'd think there must be a finite supply of the NMC27C33 originals, which presumably were an abandoned preliminary product?

Well as it appears these can be programmed, once you set the right Vpp etc, then they aren't quite as 'fake' as might first seem.
- So not sure why they didn't keep them marked as a National Semi part, and just label them as being an NMC27C32B
(or NSC27C32B), with 12.75Vpp, as should just have worked OK as that and no-one would think they were fake.
(Plus it seems access speeds on NMC27Cxx parts are always 200 or 150ns, so -2 for 200ns seem to be reasonably expected)

Although it's still a bit of a mystery why the die is marked with seemingly non-existent NMC27C33(E?)