The curse of fake parts out of China.

[Hugo Holden]

Normally, I'm good at spotting fakes, that is if the seller actually sends me what is in the photo.

A while back I was on the hunt for some 2732 UV eproms. I found some good looking ones (in the photos at least) from a seller in China. They had large sized dies and the part appeared to be the original Fujitsu MBM2732A-25, which have a 21V programming voltage and the ones in the seller photo had a gold surround die (remember I said that the fakers couldn't bear to part with Gold?)

The reason I needed these, I was using them as a substitute for the 2332 ROM in the RM-65 video card, which is used as the character generator.

In any case, what the ebay seller actually sent me (an approx $80 purchase for 8 ROMs) : Nothing with Gold in sight. The parts were labelled MBM2732A-30 (not the -25 suffix shown in their photo) and the label looked suspicious. When I programmed these, with the correct 21V programming voltage, they malfunctioned and produced random pixel outputs.

If I programmed them at 25V they appeared to work perfectly, they were obviously the 25V programming voltage 2732 part (from somewhere else).

Also, the pins on the IC body had been "processed" in some way, the slightest bend and they broke off.

In any event when programmed at 21V they malfunctioned. One of the more mild defects from some of them (programmed correctly for the part number) is attached. At other times there were random pixels lit up on the screen.

So I have had to discard them all as junk, because clearly they are not genuine Fujitsu MBM2732A-30 parts.

It is a shame that many of the Chinese ebay sellers bring disrepute and dishonor to their spare parts industry. They had better up their game, or it will catch up with them in the end.

I finally got around to putting a plastic enclosure around my RM-65 video card and adapter board. I used 3.5mm thick phenolic for the base and 3.5 mm transparent Orange acrylic for the top. (Orange was a popular color in the late 70's, remember those burnt Orange V'Dubs).

I also decided to stick the 67 Byte initialization code for the CRTC for 22 row 72 column 525 line video on the unit's base, so when somebody finds this "video card" in 50 years, they can get it running in their AIM-65 asap. The overall size of it is oddly similar to a long IBM CGA ISA card. I also added appropriate rubber feet so the card sits horizontal when plugged onto the AIM 65.

There are a lot of bare board projects out there, you had better insulate them, or sooner or later, some debris on the bench will short something out.

I cut the panels by hand, filed the edges, then used 600 then 1000 grade paper on the bench surface, running the panel edges over those, then polished the cut edges with Novus plastic polish, with a lot of tape over the panel surfaces to avoid scratches.

 

[Timo W.]

If I programmed them at 25V they appeared to work perfectly, they were obviously the 25V programming voltage 2732 part (from somewhere else).

Where's the issue then? Just program them with 25v if they work that way.

Yes, China fakes are a big issue, but at least you got working parts.

So I have had to discard them all as junk, because clearly they are not genuine Fujitsu MBM2732A-30 parts.

Are you serious? Even while they are fully working and just need 25v instead of 21v to program, they are junk to you?

Moreover, did you rule out that your programmer is the issue? Maybe its programming voltage is out of spec.

 

[Hugo Holden]

Where's the issue then? Just program them with 25v if they work that way.

Yes, China fakes are a big issue, but at least you got working parts.


Are you serious? Even while they are fully working and just need 25v instead of 21v to program, they are junk to you?

Moreover, did you rule out that your programmer is the issue? Maybe its programming voltage is out of spec.

It is possible, but unlikely. It is the GQ-4X programmer and supported by an external supply too, not relying on the USB link. And it programs other 25V and 21 v parts normally with no issues.

Plus these IC's have have pin damage and they break off at the drop of a hat.

I know that Fujitsu would never make and label a part with a number where the programming voltage does not match the label.

Therefore, because they are parts of "unknown and questionable origin", yes exactly, I cannot accept them, I don't trust them as far as an Ant could walk in one second, and they will go in the trash where they belong.

Also, I did not get the items in the photo with the appearance or exact part number that I paid for. Wouldn't you find that unacceptable ?

Now you might want to trust them, maybe to hold the code for a nuclear reactor protocol that was running next door to your house, as Clint said: Go Ahead, make my day.

 

[ardent-blue]

I know that Fujitsu would never make and label a part with a number where the programming voltage does not match the label.

Fake parts out of China is a meme. For good [bad?] reason. It is amazing how much trouble the fakers go through to remove original printing and texture from chips that might have been scavenged from scrap, rejects, or pirated production from a Chinese factory. How long these fakes will last is anyone's guess. I am sure that each chip you received fully met the original specs.

>It is a shame that many of the Chinese ebay sellers bring disrepute and dishonor to their spare parts industry. They had better up their game, or it will catch up with them in the end.

Uh, I think that's sarcasm... [CPT Obvious here]. As long as they make money, that's all the "honor" they require...

 

[daver2]

It has also been known for someone to take a more expensive part and re-label it as a cheaper part (mainly, because then they could fulfil an order). This was a 6809 CPU. The difference being the oscillator configuration. So, even though they were the more expensive part masquerading as the cheaper part, they could not be used in their intended application (the PCB already being in existence).

A colleague of mine bought some rechargeable batteries from a Chinese supplier. The price was too good to be true... When the batteries arrived, they were purely a battery case with no insides! He got his money back...

If the parts are relabelled and you can't trust them, why use them on a project that you want. If the parts work now, will they work in a year's time or not?!

Dave

 

[Chuck(G)]

For some time, there was a spate of fake USB pen drives coming out of China. They'd report maybe 2GB, but with only 512 MB of storage inside.

 

[VERAULT]

I buy "fake" eproms from china all the time. I cant beat the price. And the word fake needs to go as they clearly ARE real eproms they have just been rebranded. They all seem to be rebranded as ST brand eproms for some reason. A simple ID check of the chip can tell you who the real manufacturer is. The only issue I run into is when they shove a non C version into the mix. I.e. a 2764 instead of the 27c64 I ordered. I always check the ID against this list: https://ctrl-alt-rees.com/2022-07-03-eprom-device-id-list.html

So whats the problem? There isnt one. I get cheap working eproms.

 

[Plasmo]

I buy them and test them immediately. Ebay has excellent refund policy. Experiences are helpful; CPLD and 6502 seldom work, but Zilog parts, TTL, CF disks, RAM are great deals. They are for hobbyist uses only and should always be treated as used parts. Always assume they are not properly handled as far as static control is concerned.

 

[VERAULT]

Always assume they are not properly handled as far as static control is concerned.

Thats a big affirmative

 

[Hugo Holden]

I buy them and test them immediately. Ebay has excellent refund policy. Experiences are helpful; CPLD and 6502 seldom work, but Zilog parts, TTL, CF disks, RAM are great deals. They are for hobbyist uses only and should always be treated as used parts. Always assume they are not properly handled as far as static control is concerned.

That is good advice. And I did not do it, it was a while before I got to try them. Though I wrote to the seller and told him that a had a bone to pick with him (explaining what that meant in our culture)

As for handling, that is a real problem. A lot of parts via the parts recycling industry are exposed to multiple traumas. One is thermal. The entire boards they were on get heated up in a big oven until all the solder melts. One other is ESD protection, they get none.

Apart from all that though, one reason I try to go for new old stock, or unused IC's, is that they have not been subjected to all manner of electrical abuse by their previous owners, because with a used part you have absolutely no idea of its history.

With a NOS part, from a reputed manufacturer (if it is a true vintage NOS part) you can be nearly 100% certain that the item is in excellent order and undamaged. This is why with these sorts of parts I trust them to not only work as advertised in the first place, but not likely fail in the future. Normally, genuine NOS ROMs are blank FF's. A couple of these ones had some data that looked incompletely erased, they had been in an eraser.

(PS: I also checked that their programming issue was not related to erasure by re-erasing them, actually that made the programming voltage issue just a little worse).

Just about every single recycled-refurbished part I have accidentally ended up with, has had some issue or other. Take these ROMs, aside from the fact they are obviously 2732's, not 2732A's as the label says, something has happened to the pins. I have been studying them to determine what:

At the junction between the thin part of the pin and the wider part, the metal has become super brittle. It is really only holding together by what appears to be a fresh layer of tin plating over the area. A small bump on the pin, it bends at a sharp angle and cracks off, after just one 45 degree bend. Testing the pin metal about 2mm from its end, it takes about 5 or 6 +/- 45 degree bends to get it to fracture like this. It almost appears like these IC's have had pin extensions attached prior to adding the fresh plating. Or perhaps the pins had been bent back and forth many times during re-furbishing and straightened in a machine last thing. Some are worse than others. So regardless of the fact I can make them work by programming at 25V, this pin issue is to me, is so hopeless, I would bin them for that reason alone, as well as not knowing their "damage history".

But, regardless what you get or decide to buy, I think the seller should send you what is in the picture. Because the seller has no clue what the beauty was in the eye of the beholder/buyer and there could be many subtle reasons why the buyer chose that exact part in the photo, in my case:

Genuine looking NOS part with gold plating.
Age appropriate date codes and correct looking manufacturer logos showing some sign of age.
Age appropriate surface oxidation on the pins (not super bright new plating)
A very large IC die (new ROMs and fakes have very tiny dies now, unlike the vintage parts).

I can usually spot a genuine old part with some detective work on a photo, but the whole thing falls on its head when the seller doesn't send what is in the photo.

 

[Bentendo64]

I think putting new pins on chips is a common thing that happens in this industry. If it looks like the pins have been tinned, there's a chance they put new pins on them.
The other reason they tin the pins, which you may be aware of, is to make the chips look "newer". I don't know if they're trying to fool people into thinking they're NOS, but it always looks pretty shoddy to me.

 

[Hugo Holden]

I think putting new pins on chips is a common thing that happens in this industry. If it looks like the pins have been tinned, there's a chance they put new pins on them.
The other reason they tin the pins, which you may be aware of, is to make the chips look "newer". I don't know if they're trying to fool people into thinking they're NOS, but it always looks pretty shoddy to me.

Yes, that is exactly right. The give-away is the shiny Tin plating. This is why one of the parameters I use, I expect to see age appropriate changes on the pins, otherwise it makes me suspicious.

My ROM problems are getting worse. A while back I stocked up on TMS2532JL types (that I did not test). These Blank ok, and most were blank. But I have 14 that fail on programming. They all have bright Tin plated pins. Of the 19 spares I have, it turns out that only 5 are genuine Fujitsu parts (that came from a different supplier in the USA and are working) the others (came from China) are recycled non programmable & useless fakes, or at best damaged used an refurbished parts, sold as new parts.

 

[Hugo Holden]

I got some excellent genuine Mitsubishi 2732 ROMs now and I was able to complete the video boards. I only had enough IC's and other parts to make 3 boards/Adapters in total. There are a lot of interesting parts, including the right angled connector that mates with the AIM-65, the only ones I could find were intended for wire wrap and I made a tool to bend the pins. And there are the interesting Euro connectors too and quite an assortment of IC's.

One of the units is now in an insulated enclosure.

Speaking of genuine vs fake parts, the most ironic thing of all, is that a transistor, probably made in its many millions as a genuine part, ended up becoming the subject of cloning with practically all Silicon waste. The 2N3055.

I think it wins the prize for the most cloned transistor. I know because I have been cutting them open for over 50 years now (and I'm not the only lunatic doing it-see attached video).

Practically all of them now, that you buy from anywhere (including major suppliers) are not the same as the original part that appeared in the early '60's from RCA. They have much smaller dies in them, typically half the linear size and 1/4 the square area, or, even smaller. So many copies and fakes, re-labelled TO-3 parts, have flooded the market over the last 30 years claiming to be a 2N3055, that you cannot trust them. The occasional genuine NOS one crops up, if you know how to spot them - there is an interesting clue.

There is a Youtube video on faked 2N3055's The thing I like about this video, at the beginning he writes 2N3055 out slowly on a whiteboard and it is as exciting as watching the Chief Brody type out "S h a r k A t t a c k" on the mechanical typewriter in the original Jaws movie !

Still, if you look at this video, there is a massive clue that the three 2N3055's were faked, or at least a modern part with a small die, without cutting them open.

Look at the large diameter blue glass insulation where the leads exit the case. This is the modern (last 20 years at least) proforma for a TO-3's base, if you see a TO-3 of any type with this, pretending to be a 2N3055 with an old date code watch out. If the insulation is smaller in diameter and black or brown, it is likely a genuine vintage part (like the photo I have attached).

Some years ago I collected some which were made by RCA for a UK military contract in the early '80's. I have used many of them up over the years (being one of my favorite transistors-- and I don't care that they are out-spec'd by modern parts, I still love them), but I just have some left now sadly. In terms of genuine 2N3055's these may well be the last of the Mohicans - photo attached.

 

[Chuck(G)]

TO-3 2N3055s were made by a lot of firms. I've got one with a 1975 date code that's labeledl "Digi-Key DK3055". Motorola made a lot of them back in the day also. How many outfits made 1N914s or 2N2222?

There's nothing wrong with a shrunk die, provided it meets power requirements. In fact, it probably has better high-frequency performance. Most of my 2N3055s are TO-220 style.

 

[Hugo Holden]

TO-3 2N3055s were made by a lot of firms. I've got one with a 1975 date code that's labeledl "Digi-Key DK3055". Motorola made a lot of them back in the day also. How many outfits made 1N914s or 2N2222?

There's nothing wrong with a shrunk die, provided it meets power requirements. In fact, it probably has better high-frequency performance. Most of my 2N3055s are TO-220 style.

There is something wrong with the smaller die. The smaller die part has a higher transition frequency than specified on the original data sheet, but that can cause HF instability for some circuits designed for the original part. So it is a bad idea to economize on the silicon and change the specs, even if it could squeeze through on the power requirement.

In the early days of course (this also applies to the design of electro-magnetic machinery, architecture/bridges/buildings) the designers were much more conservative than today and had much wider margins.

There is now an 8 pin SOIC device, where the designers claim, you can run 15A through one of its pins ! It is specmanship gone mad and many parts, if you run them near these less conservative modern latter day max specs, they simply melt.

 

[modem7]

There is now an 8 pin SOIC device, where the designers claim, you can run 15A through one of its pins !

I thought for a millisecond that someone had finally discovered a room-temperature superconductor.
One of the pins?
Not current-sharing the pins like the following SO-8 packaged device ?

 

[Hugo Holden]

The one I was referring to, which is a Hall current sensor, it suggested 30A per two pins, or 15A per pin , I can't find the exact one but some others specify 20A per two pins, or 10A per pin rms which is a lot of current for the small cross sectional area of the pin on an SOIC package, they obviously rely on some heat sinking from the pcb. If you compare the pin cross section, to what was recommended as the minimum cross section for the current carrying capacity of copper wire, back in 1960 for 20A, you will find a large difference. Lucas electrical industries in the 1960's specified a round copper conductor of no less than 2mm in diameter for 20A.

Here is one rated under the safety limiting values to 30A but it drops to 16A @ 125 deg C:

www.ti.com/lit/ds/symlink/tmcs1100.pdf?ts=1701253626652&ref_url=https%253A%252F%252Fwww.google.com%252F

Due to the paucity of solid conductors in many machines and appliances, electrical fires or failures are more common. Car behind the dash wiring fires in modern vehicles have produced some nasty car fires.

Scrimping on the conductor geometry (remember the movie Towering Inferno) is no different than scrimping on the die size in a power transistor, though at least if the die melts inside a TO-3 case, its perfectly safe, but if the junction melts and the transistors CE shorts out, it could switch on some machine that was not supposed to be continuously running, or overload a psu etc or vaporize some external wiring. Though , in the TO-3 case, the link wires from the pin to the die would normally fuse with gross overload. Though, if you look very carefully at the video, you will see that the original RCA 2N3055 had large straps that connected to the die, not thin wires like the fakes.

 

[shirsch]

I buy "fake" eproms from china all the time. I cant beat the price. And the word fake needs to go as they clearly ARE real eproms they have just been rebranded. They all seem to be rebranded as ST brand eproms for some reason. A simple ID check of the chip can tell you who the real manufacturer is. The only issue I run into is when they shove a non C version into the mix. I.e. a 2764 instead of the 27c64 I ordered. I always check the ID against this list: https://ctrl-alt-rees.com/2022-07-03-eprom-device-id-list.html

So whats the problem? There isnt one. I get cheap working eproms.

I'm curious about the EPROM id code you mention. I know that EEPROMs and other contemporary devices have this capability but I've never seen any information to suggest that legacy UV-eraseable EPROMs contain such a thing.

 

[VERAULT]

I'm curious about the EPROM id code you mention. I know that EEPROMs and other contemporary devices have this capability but I've never seen any information to suggest that legacy UV-eraseable EPROMs contain such a thing.

Im not sure what you mean. You didnt know Legacy UV eproms had ID codes? Yep they absolutely do. In fact I am constantly using this guide to determine what I have or am trying to write to: https://ctrl-alt-rees.com/2022-07-03-eprom-device-id-list.html

From my minipro I just select "read ID" and it gives me the chip ID. truly legacy eproms like a 2716 return a "FF FF" but its on the money for the more common 27C variety.

 

[shirsch]

I'm not doubting what you say, but I cannot find any information on how to read that code. I have 3 or 4 different device programmers and none of them offer that feature. Internet search turns up nothing for UV EPROM.