Fake 82C55 or genuine?

[migry]

If this is posted in the wrong place, I am sure that the mods will move it!

I am starting to new project. I want to build a PC-XT circuit based on the 8088 and original Intel chipset parts. Now I recently found a YouTube channel where someone had done just that.

First job it to gather all the parts. BTW I decided to omit the DMA chip, as I don't think it is used in the original XT. Perhaps it is needed for the floppy controller??? So I turned to the usual sources. Since I was looking for old retro ICs, I had to search around. I ended up buying 4 different types of devices from the same seller, one I have used previously. The AMD 8088 looks genuine as does the Mitsubishi 8253. Date codes appear reasonable and printing is ink. The 82C55 look like they have been remarked, and the NOS 8253s have scratch marks on the top (seller says this was likely caused by "dust" in the tube!).

Over the past couple of years I have been burnt by fake re-marked chips sourced either directly or indirectly from China. This I am sure is a well known problem.

So the 82C55 PPI chips. They were sold in twos, supposedly NOS by Harris. I examined them closely when they arrived, and I am suspicious as to whether they are genuine. The markings are laser etched. This is something I have seen on the fake parts (PLCC MC68000) which I had bought, which had been sanded down and then laser etched. I guess this allows slower devices such as CPUs to be marked as if they operated at a faster speed. Who knows what these Chinese fakers are thinking?

For the 82C55s The texture of the plastic on the bottom is not the same as on the top, the bottom is very smooth and shiny. When I check other 40 pin devices in my junk box, I find parts with all sorts of textures, but top and bottom are always the same.

I googled so see if I could find a genuine Harris part. Here is what I found. My gut feeling is that it looks genuine.

​
Here is the NOS part.



I spoke on the phone to the seller and he swears blindly that it is a genuine NOS and came from a tube marked Harris.

Has anyone any other examples of the Harris parts?

I'm not too familiar with Harris and have no other examples of their ICs.

I have now been trying to test the device. Using a multimeter I have confirmed that the GND and VCC pins are in the correct position. This is done by looking for the input/output clamping diodes. This is strong evidence that these are 8255s as GND and VCC are not in the corners. I have used an Arduino to stimulate the device and it is definitely a 8255. I am able to count on one of the ports using LEDs as a monitor. I simply haven't tried the other ports.

When the ports are input I used a multimeter on the current range to short the input to either GND or VCC. When shorted to GND I was seeing around 400uA of current. Since this is a CMOS part I was expecting more or less zero. Looking at the datasheet I am a bit confused as to what this current ought to be. When shorted to VCC measured current is zero. My concern is that sometimes remarked devices have NMOS devices remarked as CMOS. FWIW I was able to use IPA to remove the MC68HC000-16 marking to reveal the original MC68000-8 markings of a part bought a year or so ago. So either the fakers/re-claimers don't know or care. I am going to try to measure input threshold and Vol/Iol and Voh/Ioh to see if I can tell if it really is CMOS.

It's a lot of effort, and perhaps I am being paranoid, but having been burnt with my 68000 purchases (and other ICs) I want to make sure that the parts are genuine and work.

Thank you for any feedback.

 

[Eudimorphodon]

Re: omitting the DMA chip: all IBM 5150/60s and most clones thereof have DMA both to run the floppy controller and to provide refresh for the DRAM memory. If you use SRAM (or some other method to provide refresh, see: PCjr, Tandy 1000) then you don’t necessarily need it, but you may need a patched BIOS if you intend to use a vintage one.

 

[Chuck(G)]

If it's really a Harris chip, I don't doubt for a second that it's CMOS. That was Harris' specialty. As to it really being a Harris 82C55, well, who knows?

 

[migry]

Re: omitting the DMA chip: all IBM 5150/60s and most clones thereof have DMA both to run the floppy controller and to provide refresh for the DRAM memory. If you use SRAM (or some other method to provide refresh, see: PCjr, Tandy 1000) then you don’t necessarily need it, but you may need a patched BIOS if you intend to use a vintage one.

Yes, I intend to use SRAM. So no need for refresh.

Although I haven't thought it through very much, I have found BIOS sources and I don't mind modifying the code if I were to use polled floppy I/O. In fact I might even try to use an Arduino (teensy) to behave like a floppy or even hard disk, and implement some simple hardware interface and a simple software protocol. Reason is i suspect writing raw code to access the floppy disk controller IC might not be exactly simple. Having said that I do have some uPD765 ICs which I think are compatible with the ones used in the XT. Again I admit that I haven't investigated what floppy support is in the available BIOS code which I found.

Annoyingly I threw out all my old ISA motherboards and ISA cards quite some time ago, something which I now regret!

 

[migry]

... As to it really being a Harris 82C55, well, who knows?

Well I'm hoping that someone visiting this forum might know! Although I dip in and out on an irregular basis, there is clearly a lot of expertise and experience to count upon, and I can't think of anywhere better.

 

[Hugo Holden]

Hi,

I have been spotting fake IC's and transistors for nearly 2 decades now. The clues that an IC is a fake, of which there are a number, point with some probability, but never 100%.
But I can say this, in your case, I'd happily bet $1000 that your IC is fake, by that I mean a total fake, some other IC type, not even a compatible part.

The reason is the package has been though the conditioning process to remove the original markings, leaving it with that satin or micro sand blasted look. The fakers found out that customers were checking white ink markings with solvents (fresh ink is easily dissolved compared to old ink) so they moved to laser engraving. So an IC with this finish and laser markings is a near definite fake, until proven otherwise, and it would be very very low probability it was genuine.

As you have noted, many of the older IC's have a glossy surface finish.

To avoid vintage IC fakes I do the following:

Avoid any IC's with date codes post mid 1990's if 70's/80's vintage or year 2000 if the original parts was 90's vintage.
Avoid IC's with that flat finish & laser engraving.
Go for a part from a major manufacturer with a definitely correct version of their logo.
Look for signs of storage and aging on the pins, and IC body, suggesting the part is genuine & old.
(If the part is 80's vintage and super shiny pins and fresh new look, it raises a suspicion)
Go for IC's in ceramic packages (if can find) the fakers rarely attempt to re-label ceramic parts.
Usually, if the part has gold plated pins, rarely faked as the fakers are Gold scavengers.
Buy a part inside the USA, UK, Germany, Europe....because, most fakes are coming out of China.
Some IC packages are unique, for example Motorola and Hitachi TTL IC's have rounded corners on the package. They are hard to fake.
Buy from a seller you know and have experience with, who have real stock in their inventory.

The length the fakers are going to is astounding. They have a process now where transistors with short legs pulled from pcb's have the legs cut diagonally. New long lead wires are welded on, and then they are tinned over to conceal the weld. Sometimes they fracture here though and you can see the welds under magnification. For transistors, look for fine buffing marks on the body, suggesting the original markings were buffed off.

If you order a transistor from some places, they look in the equivalence manual and find a part they think will work, & re-condition its cosmetics and re-label it.

One time, I ordered some rare Teledyne jFets in a TO-5 case. One of the transistor manuals the fakers had, contained an error and listed the part number as an NPN transistor, so they sent "refurbished" NPN transistors with a very tidy label of the jFet's number on them. So I sent them an email saying they shouldn't believe everything they read in their transistor equivalence manual.

One piece of advice, don't buy the IC's unless the seller has really good photos of them and you can see exactly what you are getting, inspecting the fine details. Never buy something with a generic or no photo.

Of course the person you got your IC from might be totally innocent and did not know they had been supplied with fakes. Many sellers are agents or on-sellers and do not have actual inventory, they are not really stocking suppliers, they have downloaded massive internet parts lists, when somebody orders a part they attempt to find it elsewhere. If the sellers inventory contains the part number CY9C6264, you know they don't keep real parts themselves. Why ? because this memory IC's pdf data sheet circulated the internet, but the part itself was never manufactured & doesn't physically exist. The story behind this interesting Magnetoresistive RAM is explained on pg 20 of this article:

https://www.worldphaco.com/uploads/T...THE_DS1225.pdf

I just looked on ebay and saw some Toshiba TMP82C55 AP-2 , 3 for $8.99, probably genuine and inside the USA, 100% + feedback seller. I think these are ok.

The thing about fake parts, once they get into supply chains, it can get hard to trace back to where they originated and justice is never served. One horrifying consequence is already re-labelled and working compatible fakes have got into military supply chains and avionics supply chains, but often the devices are under-rated, smaller transistor dies etc and prone to failure , this sort of thing could kill a jet engine's control system etc or cause other chaos. So the faking of semiconductors I think is a very serious issue.

 

[Chuck(G)]

If you're not wedded to DIP, the PLCC versions are newer and faster--and in many cases, cheaper. Unlikely candidates for counterfeiting also. I think the Chinese fakers tend to concentrate on older hard-to-find DIP parts.

 

[daver2]

The other thing to do is to get an accurate vernier caliper (digital ones are quite cheap now) and to accurately measure the dimensions of the device and compare those measurements with the data sheet. Obviously, you have to purchase the device first though...

If you want a laugh - I have seen examples of where someone has re-marked a more expensive chip to be a cheaper chip (in this case it was a 6809 CPU). They could have sold the chip (as a genuine pull) for more money than they sold it for remarked - and it pi55ed off the purchaser because it didn't work... An x-ray of the device indicated is was a genuine 6809 device - just of a different type. Go figure!

Dave

 

[migry]

If you're not wedded to DIP, the PLCC versions are newer and faster--and in many cases, cheaper. Unlikely candidates for counterfeiting also. I think the Chinese fakers tend to concentrate on older hard-to-find DIP parts.

I like to use the plug-in breadboards for development, so DIL/DIP is my goto package, however a year or so ago I bought 20 off PLCC MC68000 (from a Chinese seller) because 1) they were 20MHz and 2) they were very cheap (big red flag - yes I know!). At that price I thought it worth the risk.

They all had the exact same markings, purportedly indicating that they came from the same batch. All were 100% non functional when tested in my homebrew dev board.. It appeared like they had been sanded on the top and then laser etched. My gut feeling (pity I don't know how to find this out) is that at the time these CPUs were manufactured, it is likely Motorola didn't use laser etching for marking. I've only seen inked Moto devices, but then again this is only a small sample.

Of course the smoking gun (for these PLCC 68000's) was when the packages were turned over to reveal the bottom..

1. the colour of the plastic was slightly different, there been lighter and darker bottoms
2. the size and position of the pin 1 dimple was missing or in a different position or was a different depth
3. the plastic moulding markings (the larger circular indents) varied
4. some had ink markings and some didn't

Of course the PLCC (and DIP) pins prevent any sanding or fakery on the bottom side.

If I get around to it I will make a video for the usual internet sites.

--migry

 

[migry]

Hi,

I have been spotting fake IC's and transistors for nearly 2 decades now. ... .

Dear Hugo many thanks for your details and very pertinent advice.

I am being a little cautious about identifying the seller, as this is a case of "he said" "she said". I would like to discuss the exchange over the phone with the seller, but probably best I don't, except to say they are firm is saying these are genuine devices.

I did source the devices from what appeared to be a legitimate seller in the EU. In the listing the picture was not clear enough to see that the markings were by laser. I avoided any sellers with Chinese links on purpose, but to be fair I have bought what appear to be 100% legitimate devices (more 68000's) from Chinese sellers, but as has been said, many are simply re-sellers and probably have no idea what they are selling and have likely never even looked at the devices. They are just out to make a quick buck, which given the low prices mean that they have to sell an awful lot. I bought 4 different types of PC-XT related devices from the same seller. The CPU looks 100% genuine. I have different "concerns" about another type of device.

BTW to re-iterate, the device being tested using an Arduino in my bread board, confirms that it is a 8255 family device. It is operational, but not yet 100% tested. My concern about the likely re-marking is that the genuine type and speed information has gone. It should be CMOS, but 400uA of IIL seem too high. I do have a CMOS Mitsubishi device which I will test and compare.

Of course since these devices are no longer manufactured, we have to turn to the usual web sites, US, EU or even Chinese. Often there is not much choice. Having been burnt several times it is a case of "let the buyer beware".

BTW I've also heard stories of "fake" devices getting into the military supply chain.

 

[daver2]

>>> Of course since these devices are no longer manufactured.

To all intents and purposes 'yes' for general use.

I work in an industry where the paperwork and traceability is just as important as the device itself. Then, I turn to a company that have stocks of devices and the associated paperwork. If they don't have stocks of the devices themselves, they have stocks of (say) the Intel silicon wafers and can encapsulate the devices. Unfortunately, this makes an 82C55 USD 55 each - but they are guaranteed and traceable back to Intel manufacture...

A few years ago I purchased 8085, 8155 and 8255 devices (to name but a few) and got Intel 2716 silicon encapsulated to make 'new' and tested EPROMs (in 2016).

Within industrial circles, allowing fake and substandard devices into the supply chain generally occurs because someone somewhere wants to make a fast buck or places orders based upon the cheapest supplier. Mistake!

Dave

 

[jlang]

Easy way to tell if its a CMOS part is measure ICC (chip held in reset)
NMOS ICC 120MA.
CMOS ICC 10uA.

joe

 

[Chuck(G)]

No longer manufactured?

Not according to Renesas, who lists it in their "active" catalog. A quick check via Mouser shows it in stock.

 

[daver2]

Good spot Chuck. I had forgotten about the rad hard and and space environment CMOS stuff.

We looked at those for our job as well - but couldn't use the CMOS parts in the end (for a variety of reasons).

Dave

 

[kdr]

Then, I turn to a company that have stocks of devices and the associated paperwork. If they don't have stocks of the devices themselves, they have stocks of (say) the Intel silicon wafers and can encapsulate the devices. Unfortunately, this makes an 82C55 USD 55 each - but they are guaranteed and traceable back to Intel manufacture...

A few years ago I purchased 8085, 8155 and 8255 devices (to name but a few) and got Intel 2716 silicon encapsulated to make 'new' and tested EPROMs (in 2016).

Would that be Rochester Electronics?

They always seem to show up in the Digikey searches for all sorts of now-obsolete chips, and I've always wondered how much more trustworthy they'd be as compared to the other grey-market sources for old chips...

 

[Chuck(G)]

You might try these folks. Lots of NOS stuff.

 

[Hugo Holden]

While it would be good to get genuine old stock Intel IC's, I have had a lot of luck over many years with Japanese IC's, both Hitachi & Toshiba & Mitubishi. (the same applies to Japanese semiconductors). There also appears to be less fakery of Japanase IC's and transistors. I have a very large collection of TTL's, and working with these at least, I have never found a faulty Hitachi TTL IC, though for the first time recently I found a faulty Hitachi TTL IC, but the only time ever. Generally, I find the Japanese to be unrelenting perfectionists with their electronics and their component quality (which I admire), and they rarely if ever over spec a part. The Japanese don't do fakes, because it is dishonorable.

I'm not sure about Japanese cmos, but I would guess they are just as good. So I would go for these, they have to be a bargain, the seller has sold quite a few and everyone left good feedback, so it means, likely, they tested them and they work !

https://www.ebay.com/itm/39310322089...EAAOSwDpVgCKIl


(Some of the radiation hardened ceramic body vintage space TTL IC's I think are very impressive, I could imagine them in a satellite returning to Earth, partially burning up on re-entry, smashing into the side of a rocky cliff with debris flying everywhere, rolling down onto a highway and being run over by a Mac truck. Then pulling one of those TTL's out of the remains, and it would still be working. Maybe I'm just a romantic over TTL's, they do also have a very high ESD damage resistance, I have posted this before, it shows how good TTL's are in this respect, the scale is not linear as level 9 has about 4 times the energy of level 5)

 

[Chuck(G)]

The point is that there's no need to go to China to get these things. Almost every major house made them and there are still plenty out there.

 

[daver2]

>>> Would that be Rochester Electronics?

Yes it would...

Rochester only stock items that have been purchased from the OCM (Original Component Manufacturer). For example, for the components that we purchased, we asked them to supply the evidence of where they got them from - which they were able to do. This (in our case) was a 'bill of transfer/sale' from Texas Instruments / Intel to Rochester Electronics with the lot and date codes etc.

They don't purchase (or supply) components from third parties.

The components all come with Certificates of Conformance etc.

We have also toured their facility and examined how they store the devices etc.

We have 'tested out' their component track and trace capability. We gave them a Rochester Electronics part and they were able to tell us who had purchased a component with this date and lot code and how many they had purchased. That was good news - because it wasn't part of the batch of components that I had bought! The PCB assembler had purchased some of their own components - and some of these had made it onto one of our prototype boards.

As I have stated in my previous post, they also have silicon wafers and can make ICs from those. I had 2716 EPROMS and 8274 Intel MPSC devices encapsulated.

I would have absolutely no hesitation in recommending Rochester Electronics - but they are expensive and have a minimum order requirement. If you are in business, and require 'genuine' components, I would have no hesitation in recommending them.

Dave

 

[Eudimorphodon]

BTW to re-iterate, the device being tested using an Arduino in my bread board, confirms that it is a 8255 family device. It is operational, but not yet 100% tested. My concern about the likely re-marking is that the genuine type and speed information has gone. It should be CMOS, but 400uA of IIL seem too high. I do have a CMOS Mitsubishi device which I will test and compare.

Not knowing anything about your test setup or equipment, are you sure you’re measuring input leakage current and not bus hold up/down current? The datasheet for the part says the max for that is 400uA.

How much did you pay for these? I mean… I get wanting to know you got your money’s worth, but it sounds like the parts are at least 8255s, and honesty I can’t imagine these really being worth a faker’s time. CMOS variants are almost certainly less valuable than vintage TTL ones and it’s not as if the 82c55 is a particularly rare part. If you were buying a thousand of them for a production run maybe, but for a hobby project… ? Would it even matter in this case if it were the TTL variant?