Topics

Fake 1802 processors


Martin
 

https://photos.app.goo.gl/bmdTPNQQ6tUh4F9d6

You can see which processor is fake. This is a typical cheap eBay part with the "standard" date code of 027. These parts fail self test in my ancient Fluke troubleshooter. Somehome, I have an 1802 pod for it. 

The others came from a U.S. supplier. 

I may try some of the more genuine looking parts from China.


joshbensadon
 

The tops all have that sanded look, not polished.  Be sure to give those sellers negative feedback, if ever anyone deserved it, it would be someone who messes with 1802 history!



From: cosmacelf@groups.io <cosmacelf@groups.io> on behalf of Martin <martin.reynolds@...>
Sent: Saturday, August 8, 2020 8:55 PM
To: cosmacelf@groups.io <cosmacelf@groups.io>
Subject: [cosmacelf] Fake 1802 processors
 
https://photos.app.goo.gl/bmdTPNQQ6tUh4F9d6

You can see which processor is fake. This is a typical cheap eBay part with the "standard" date code of 027. These parts fail self test in my ancient Fluke troubleshooter. Somehome, I have an 1802 pod for it. 

The others came from a U.S. supplier. 

I may try some of the more genuine looking parts from China.


ian may
 

I agree with Josh, they all look like fakes to me. As well as the surface finish, look at the right hand edges where bits of epoxy have been broken off by the sanding that has been done to remove the original markings. If there are any markings left on the bottom it might be possible to work out who made them, but you might only find scratch marks showing where the markings once were.

There are a few eBay sellers with 1802s classified as USED. Since there is no incentive to disguise these as new, the original markings are usually intact. It is still possible to get a bad part, but at least it may not have had all the handling required to grind off the original markings, apply a layer of some kind in an attempt to fill the sanding marks and then apply a new set of marks.

Fakes can also be detected because the markings are too good, especially with a bunch of them in the one picture. The original manufacturers marks can typically be at an angle, at a different height on one chip compared to another or even have letters at different angles (typically RCA ones).

Cheers, Ian.


Robert
 

There are tons of remarked chips and outright fakes on eBay. Some of the worst are the 6821 PIA chips. I hate seeing remarked chips and even if they work are worth far less. I really wish they would stop it. If I ever get remarked chips from eBay again I’ll file a claim. That may be the only way to get them to stop.

Thing is I’d pay more for good used pulls. At least I’d know what I’m getting, what brand, what model 6820, 6821, 68B21, etc. With relabeled chips you just don’t know.

When restoring on boards I prefer to match type, brand, similar date codes, and even markings. Relabeled chips ruins that option.

Robert

-----Original Message-----
From: "ian may via groups.io" <fps16xn3=yahoo.com@groups.io>
Sent: Sunday, August 9, 2020 8:30am
To: cosmacelf@groups.io
Subject: Re: [cosmacelf] Fake 1802 processors

I agree with Josh, they all look like fakes to me. As well as the surface finish, look at the right hand edges where bits of epoxy have been broken off by the sanding that has been done to remove the original markings. If there are any markings left on the bottom it might be possible to work out who made them, but you might only find scratch marks showing where the markings once were.

There are a few eBay sellers with 1802s classified as USED. Since there is no incentive to disguise these as new, the original markings are usually intact. It is still possible to get a bad part, but at least it may not have had all the handling required to grind off the original markings, apply a layer of some kind in an attempt to fill the sanding marks and then apply a new set of marks.

Fakes can also be detected because the markings are too good, especially with a bunch of them in the one picture. The original manufacturers marks can typically be at an angle, at a different height on one chip compared to another or even have letters at different angles (typically RCA ones).

Cheers, Ian.


Magnus Kesselmark
 

What's iside the remarked/fake chips? Bad processors or "anything"???


I have stocked up some 10 CPUs, of which a few are from AliExpress (but wich are wich?? I have them marked, actually.). And there's a shop over here in Sweden that sells 1802s and I think theyu really are NOS, like the 2SJ50/2SK135 that are sold (at high prices).

/Magnus


Den 2020-08-09 kl. 15:33, skrev Robert:

There are tons of remarked chips and outright fakes on eBay. Some of the worst are the 6821 PIA chips. I hate seeing remarked chips and even if they work are worth far less. I really wish they would stop it. If I ever get remarked chips from eBay again I’ll file a claim. That may be the only way to get them to stop.

Thing is I’d pay more for good used pulls. At least I’d know what I’m getting, what brand, what model 6820, 6821, 68B21, etc. With relabeled chips you just don’t know.

When restoring on boards I prefer to match type, brand, similar date codes, and even markings. Relabeled chips ruins that option.

Robert

-----Original Message-----
From: "ian may via groups.io" <fps16xn3=yahoo.com@groups.io>
Sent: Sunday, August 9, 2020 8:30am
To: cosmacelf@groups.io
Subject: Re: [cosmacelf] Fake 1802 processors

I agree with Josh, they all look like fakes to me. As well as the surface finish, look at the right hand edges where bits of epoxy have been broken off by the sanding that has been done to remove the original markings. If there are any markings left on the bottom it might be possible to work out who made them, but you might only find scratch marks showing where the markings once were.

There are a few eBay sellers with 1802s classified as USED. Since there is no incentive to disguise these as new, the original markings are usually intact. It is still possible to get a bad part, but at least it may not have had all the handling required to grind off the original markings, apply a layer of some kind in an attempt to fill the sanding marks and then apply a new set of marks.

Fakes can also be detected because the markings are too good, especially with a bunch of them in the one picture. The original manufacturers marks can typically be at an angle, at a different height on one chip compared to another or even have letters at different angles (typically RCA ones).

Cheers, Ian.






--
Detta e-postmeddelande har sökts igenom efter virus med antivirusprogram från Avast.
https://www.avast.com/antivirus


Lee Hart
 

Robert wrote:
There are tons of remarked chips and outright fakes on eBay. Some of the worst are the 6821 PIA chips. I hate seeing remarked chips and even if they work are worth far less. I really wish they would stop it. If I ever get remarked chips from eBay again I’ll file a claim. That may be the only way to get them to stop.

Thing is I’d pay more for good used pulls. At least I’d know what I’m getting, what brand, what model 6820, 6821, 68B21, etc. With relabeled chips you just don’t know.

When restoring on boards I prefer to match type, brand, similar date codes, and even markings. Relabeled chips ruins that option.
I agree; these certainly look like "blacktopped" chips that have been sanded, painted black, and re-marked with new numbers. They may be good or bad; they may be 1802's or something else entirely!

When I built the 6502 badges for the VCFMW show, I bought hundreds of Chinese "Rockwell R65C02P4" chips. The results were all over the map. Some were NOS (New Old Stock) parts that still had their original markings -- these looked like parts that had been in sockets, so it was easy to salvage them.

But most of them had the original markings sanded off, and new marking printed on. Most were good; but you never knew. One batch might be 100% good, and the next ALL BAD! I had to test each and every chip myself to produce a reliable product.

The situation is becoming the same with the 1802. It used to be I could buy them from reputable distributors; but they are all out (except for the high-priced shops that specialize in single-quantity repair parts).

Then I bought from surplus dealers and auctions where the parts were NOS. These are largely drying up, too. The Chinese re-marked chips have basically taken over the market.

At present, BG Micro still has a stock of real RCA CDP1802ACE NOS chips. Rochester Electronics also has a good stock of NOS chips, but they have a minimum $250 order.

The 1802 is still in production, but only produced to order. When I asked (a few years ago), it was something like a 1000-piece minimum order for around $4000 for Intersil to make some in the plastic DIP40 package.

Lee Hart

--
A designer knows he has achieved perfection not when there is
nothing left to add, but when there is nothing left to take away.
-- Antoine de Saint Exupery
--
Lee Hart, 814 8th Ave N, Sartell MN 56377, www.sunrise-ev.com


Martin
 

OK I dragged out the tester and went through all the parts, notes attached to the pictures. 

Summary: all probably contain 1802 silicon, indicated by what looks like factory markings on the underside. 
My laser marked Intersil processor (I think this is the one that came with the unit) shares some back side markings with the remarked, failing devices from China. 
The "new" devices from my US source have no back side markings and suspicious differences in packaging (mold marks) for the same date code. 




On Sun, Aug 9, 2020 at 6:50 AM Magnus Kesselmark <tubes@...> wrote:
What's iside the remarked/fake chips? Bad processors or "anything"???


I have stocked up some 10 CPUs, of which a few are from AliExpress (but
wich are wich?? I have them marked, actually.). And there's a shop over
here in Sweden that sells 1802s and I think theyu really are NOS, like
the 2SJ50/2SK135 that are sold (at high prices).

/Magnus


Den 2020-08-09 kl. 15:33, skrev Robert:
> There are tons of remarked chips and outright fakes on eBay.  Some of the worst are the 6821 PIA chips.  I hate seeing remarked chips and even if they work are worth far less.  I really wish they would stop it.  If I ever get remarked chips from eBay again I’ll file a claim.  That may be the only way to get them to stop.
>
> Thing is I’d pay more for good used pulls.  At least I’d know what I’m getting, what brand, what model 6820, 6821, 68B21, etc.  With relabeled chips you just don’t know.
>
> When restoring on boards I prefer to match type, brand, similar date codes, and even markings.  Relabeled chips ruins that option.
>
> Robert
>
> -----Original Message-----
> From: "ian may via groups.io" <fps16xn3=yahoo.com@groups.io>
> Sent: Sunday, August 9, 2020 8:30am
> To: cosmacelf@groups.io
> Subject: Re: [cosmacelf] Fake 1802 processors
>
> I agree with Josh, they all look like fakes to me. As well as the surface finish, look at the right hand edges where bits of epoxy have been broken off by the sanding that has been done to remove the original markings. If there are any markings left on the bottom it might be possible to work out who made them, but you might only find scratch marks showing where the markings once were.
>
> There are a few eBay sellers with 1802s classified as USED. Since there is no incentive to disguise these as new, the original markings are usually intact. It is still possible to get a bad part, but at least it may not have had all the handling required to grind off the original markings, apply a layer of some kind in an attempt to fill the sanding marks and then apply a new set of marks.
>
> Fakes can also be detected because the markings are too good, especially with a bunch of them in the one picture. The original manufacturers marks can typically be at an angle, at a different height on one chip compared to another or even have letters at different angles (typically RCA ones).
>
> Cheers, Ian.
>
>
>
>
>
>
>
>

--
Detta e-postmeddelande har sökts igenom efter virus med antivirusprogram från Avast.
https://www.avast.com/antivirus





Martin
 

Thanks, Lee - this is helpful. My evidence is that these are likely all 1802s (because of the back side markings matching those on my Intersil CPU), but for all we know they've been recycled in a fairly brutal fashion with no handling precautions, then quite effectively beautified as long as they have all pins. I doubt that they are tested, because  the fallout rate is low enough (20%) that they can outsource the testing to, well, you...

I don't think the Fluke system that I use is particularly sophisticated, it is all TTL inside. So the test part just doesn't work if it fails, it isn't some corner condition being exposed.

Then, I am sure that these parts are vulnerable to handling electrostatic discharge, which to your point makes them  unreliable in the long term.. 

On Sun, Aug 9, 2020 at 10:19 AM Lee Hart <leeahart@...> wrote:
Robert wrote:
> There are tons of remarked chips and outright fakes on eBay.  Some of the worst are the 6821 PIA chips.  I hate seeing remarked chips and even if they work are worth far less.  I really wish they would stop it.  If I ever get remarked chips from eBay again I’ll file a claim.  That may be the only way to get them to stop.
>
> Thing is I’d pay more for good used pulls.  At least I’d know what I’m getting, what brand, what model 6820, 6821, 68B21, etc.  With relabeled chips you just don’t know.
>
> When restoring on boards I prefer to match type, brand, similar date codes, and even markings.  Relabeled chips ruins that option.

I agree; these certainly look like "blacktopped" chips that have been
sanded, painted black, and re-marked with new numbers. They may be good
or bad; they may be 1802's or something else entirely!

When I built the 6502 badges for the VCFMW show, I bought hundreds of
Chinese "Rockwell R65C02P4" chips. The results were all over the map.
Some were NOS (New Old Stock) parts that still had their original
markings -- these looked like parts that had been in sockets, so it was
easy to salvage them.

But most of them had the original markings sanded off, and new marking
printed on. Most were good; but you never knew. One batch might be 100%
good, and the next ALL BAD! I had to test each and every chip myself to
produce a reliable product.

The situation is becoming the same with the 1802. It used to be I could
buy them from reputable distributors; but they are all out (except for
the high-priced shops that specialize in single-quantity repair parts).

Then I bought from surplus dealers and auctions where the parts were
NOS. These are largely drying up, too. The Chinese re-marked chips have
basically taken over the market.

At present, BG Micro still has a stock of real RCA CDP1802ACE NOS chips.
Rochester Electronics also has a good stock of NOS chips, but they have
a minimum $250 order.

The 1802 is still in production, but only produced to order. When I
asked (a few years ago), it was something like a 1000-piece minimum
order for around $4000 for Intersil to make some in the plastic DIP40
package.

Lee Hart

--
A designer knows he has achieved perfection not when there is
nothing left to add, but when there is nothing left to take away.
         -- Antoine de Saint Exupery
--
Lee Hart, 814 8th Ave N, Sartell MN 56377, www.sunrise-ev.com




Magnus Kesselmark
 

My collection of 1802:s The ones with a yellow sticker are my Chinese, AliEXpress chips.

Also note the ICL7217 chips; also bought from AliX. I guess they are pulled from some surplus. The third IC seems to have had a sticker taped to it.

http://www.pastisch.se/HF/1802_1.jpg

http://www.pastisch.se/HF/7217.jpg


/Magnus



Den 2020-08-09 kl. 19:35, skrev Martin:
Thanks, Lee - this is helpful. My evidence is that these are likely all 1802s (because of the back side markings matching those on my Intersil CPU), but for all we know they've been recycled in a fairly brutal fashion with no handling precautions, then quite effectively beautified as long as they have all pins. I doubt that they are tested, because  the fallout rate is low enough (20%) that they can outsource the testing to, well, you...

I don't think the Fluke system that I use is particularly sophisticated, it is all TTL inside. So the test part just doesn't work if it fails, it isn't some corner condition being exposed.

Then, I am sure that these parts are vulnerable to handling electrostatic discharge, which to your point makes them  unreliable in the long term.. 

On Sun, Aug 9, 2020 at 10:19 AM Lee Hart <leeahart@...> wrote:
Robert wrote:
> There are tons of remarked chips and outright fakes on eBay.  Some of the worst are the 6821 PIA chips.  I hate seeing remarked chips and even if they work are worth far less.  I really wish they would stop it.  If I ever get remarked chips from eBay again I’ll file a claim.  That may be the only way to get them to stop.
>
> Thing is I’d pay more for good used pulls.  At least I’d know what I’m getting, what brand, what model 6820, 6821, 68B21, etc.  With relabeled chips you just don’t know.
>
> When restoring on boards I prefer to match type, brand, similar date codes, and even markings.  Relabeled chips ruins that option.

I agree; these certainly look like "blacktopped" chips that have been
sanded, painted black, and re-marked with new numbers. They may be good
or bad; they may be 1802's or something else entirely!

When I built the 6502 badges for the VCFMW show, I bought hundreds of
Chinese "Rockwell R65C02P4" chips. The results were all over the map.
Some were NOS (New Old Stock) parts that still had their original
markings -- these looked like parts that had been in sockets, so it was
easy to salvage them.

But most of them had the original markings sanded off, and new marking
printed on. Most were good; but you never knew. One batch might be 100%
good, and the next ALL BAD! I had to test each and every chip myself to
produce a reliable product.

The situation is becoming the same with the 1802. It used to be I could
buy them from reputable distributors; but they are all out (except for
the high-priced shops that specialize in single-quantity repair parts).

Then I bought from surplus dealers and auctions where the parts were
NOS. These are largely drying up, too. The Chinese re-marked chips have
basically taken over the market.

At present, BG Micro still has a stock of real RCA CDP1802ACE NOS chips.
Rochester Electronics also has a good stock of NOS chips, but they have
a minimum $250 order.

The 1802 is still in production, but only produced to order. When I
asked (a few years ago), it was something like a 1000-piece minimum
order for around $4000 for Intersil to make some in the plastic DIP40
package.

Lee Hart

--
A designer knows he has achieved perfection not when there is
nothing left to add, but when there is nothing left to take away.
         -- Antoine de Saint Exupery
--
Lee Hart, 814 8th Ave N, Sartell MN 56377, www.sunrise-ev.com






Avast logo

Detta e-postmeddelande har sökts igenom efter virus med antivirusprogram från Avast.
www.avast.com



coolcash3030
 

You also have to be wary of ordering cheap or even not so cheap chips on Aliexpress....many are pulls and refurbished....most sellers are honest about that but I did hear of some outright fakes being sold by some AE sellers....of course these pulls are a gamble ... I recently got some 2764s 10 for $10 .... 9 tested good and 1 doa. My main beef with the AE chip sellers is the packaging....Saran wrap! I gave some negative feedback on the packing....I know these pulls were probably abused but let's give these chips a fighting chance.


Lee Hart
 

Martin wrote:
Thanks, Lee - this is helpful. My evidence is that these are likely all
1802s (because of the back side markings matching those on my Intersil
CPU), but for all we know they've been recycled in a fairly brutal
fashion with no handling precautions, then quite effectively beautified
as long as they have all pins. I doubt that they are tested, because
the fallout rate is low enough (20%) that they can outsource the testing
to, well, you...
Exactly. People like an honest bargain. But it's dishonest to hide or lie about the provenence of these parts (where they came from, what's been done to them).

The old "Poly Paks" surplus company (remember them?) was famous for buying "floor sweepings" scrap parts, bagging them up, and selling them in bulk for absurdly low prices. Their advertising was (nominally) honest: Things like...

- We sell good junk
- No chance to test
- Buy 'em by the barrel
- Only a little out of spec
- Memory chips, 99% good bits
- Blemished 7-seg display
- How good? Who knows!
- Unmarked resistors, capacitors, diodes, etc, 100 for $1.98
- Used parts; their loss is your gain

I don't think the Fluke system that I use is particularly
sophisticated, it is all TTL inside. So the test part just doesn't work
if it fails, it isn't some corner condition being exposed.
I had a Fluke 9000, though with the Z80 probe. For those who don't know it, it's a clever piece of equipment. The pod has a set of analog switches and a ROM. You unplug the CPU, plug the pod in instead, and plug the CPU into a socket on the pod. A (serial?) cable connects the pod to the main unit, which is like a big desk calculator.

Testing a CPU chip is hard. Testing some random circuit connected to it is even harder. So the Fluke uses a technique called "signature analysis". In effect, it's like connecting headphones successively to each pin, and listening for the "noise". You compare the noise to a working systems "by ear", and decide whether it sounds normal or not.

The Fluke uses the ROM in the pod to run a known program, and captures the bitstream "signature" on each pin with a shift register. It displays the result as a 4-digit hex number, like a CRCC check. Any incorrect or misplaced bits will change the signature.

Then, I am sure that these parts are vulnerable to handling
electrostatic discharge, which to your point makes them unreliable in
the long term.
I test my 1802's in a Membership Card, hooked up to a bench supply that meters the voltage and current. An 1802 with a blown input generally draws excessive supply current. I also run a simple program that tests the memory and I/O.

Lee

--
A designer knows he has achieved perfection not when there is
nothing left to add, but when there is nothing left to take away.
-- Antoine de Saint Exupery
--
Lee Hart, 814 8th Ave N, Sartell MN 56377, www.sunrise-ev.com


Martin
 

A little off topic, but here is my  work from years ago fixing up the big brother of the 9010. 

I think I got it working. The links to "artfromny" don't work, as he seems to have changed his business. His pages do seem to be in the archive. 

Looks like there is also a Tektronix device - 8001 / 8002A 
Found a picture of a later unit. 
Hard to tell what board is being debugged, but look at tke keypad and displays tacked on to the side of the main design. 

 



 


On Sun, Aug 9, 2020 at 12:35 PM Lee Hart <leeahart@...> wrote:
Martin wrote:
> Thanks, Lee - this is helpful. My evidence is that these are likely all
> 1802s (because of the back side markings matching those on my Intersil
> CPU), but for all we know they've been recycled in a fairly brutal
> fashion with no handling precautions, then quite effectively beautified
> as long as they have all pins. I doubt that they are tested, because
> the fallout rate is low enough (20%) that they can outsource the testing
> to, well, you...

Exactly. People like an honest bargain. But it's dishonest to hide or
lie about the provenence of these parts (where they came from, what's
been done to them).

The old "Poly Paks" surplus company (remember them?) was famous for
buying "floor sweepings" scrap parts, bagging them up, and selling them
in bulk for absurdly low prices. Their advertising was (nominally)
honest: Things like...

- We sell good junk
- No chance to test
- Buy 'em by the barrel
- Only a little out of spec
- Memory chips, 99% good bits
- Blemished 7-seg display
- How good? Who knows!
- Unmarked resistors, capacitors, diodes, etc, 100 for $1.98
- Used parts; their loss is your gain

> I don't think the Fluke system that I use is particularly
> sophisticated, it is all TTL inside. So the test part just doesn't work
> if it fails, it isn't some corner condition being exposed.

I had a Fluke 9000, though with the Z80 probe. For those who don't know
it, it's a clever piece of equipment. The pod has a set of analog
switches and a ROM. You unplug the CPU, plug the pod in instead, and
plug the CPU into a socket on the pod. A (serial?) cable connects the
pod to the main unit, which is like a big desk calculator.

Testing a CPU chip is hard. Testing some random circuit connected to it
is even harder. So the Fluke uses a technique called "signature
analysis". In effect, it's like connecting headphones successively to
each pin, and listening for the "noise". You compare the noise to a
working systems "by ear", and decide whether it sounds normal or not.

The Fluke uses the ROM in the pod to run a known program, and captures
the bitstream "signature" on each pin with a shift register. It displays
the result as a 4-digit hex number, like a CRCC check. Any incorrect or
misplaced bits will change the signature.

> Then, I am sure that these parts are vulnerable to handling
> electrostatic discharge, which to your point makes them  unreliable in
> the long term.

I test my 1802's in a Membership Card, hooked up to a bench supply that
meters the voltage and current. An 1802 with a blown input generally
draws excessive supply current. I also run a simple program that tests
the memory and I/O.

Lee

--
A designer knows he has achieved perfection not when there is
nothing left to add, but when there is nothing left to take away.
         -- Antoine de Saint Exupery
--
Lee Hart, 814 8th Ave N, Sartell MN 56377, www.sunrise-ev.com




Robert
 

I still have my Fluke 9010 and I think I have a 1802 pod for it. Never used that one. I rarely use it but it has helped on a couple 6808 and Z80 repairs. Nice piece of gear that I need to make room on the bench to use more.

-----Original Message-----
From: "Martin" <martin.reynolds@...>
Sent: Sunday, August 9, 2020 4:28pm
To: cosmacelf@groups.io
Subject: Re: [cosmacelf] Fake 1802 processors

A little off topic, but here is my work from years ago fixing up the big
brother of the 9010.
http://www.iobium.com/9100a%20tester/9100atester.htm

I think I got it working. The links to "artfromny" don't work, as he seems
to have changed his business. His pages do seem to be in the archive.
And another site: https://flippers.com/fluke.html

Looks like there is also a Tektronix device - 8001 / 8002A
Found a picture of a later unit.
https://archive.org/details/bitsavers_tektronix8_7189364
Hard to tell what board is being debugged, but look at tke keypad and
displays tacked on to the side of the main design.







On Sun, Aug 9, 2020 at 12:35 PM Lee Hart <leeahart@...> wrote:

Martin wrote:
Thanks, Lee - this is helpful. My evidence is that these are likely all
1802s (because of the back side markings matching those on my Intersil
CPU), but for all we know they've been recycled in a fairly brutal
fashion with no handling precautions, then quite effectively beautified
as long as they have all pins. I doubt that they are tested, because
the fallout rate is low enough (20%) that they can outsource the testing
to, well, you...
Exactly. People like an honest bargain. But it's dishonest to hide or
lie about the provenence of these parts (where they came from, what's
been done to them).

The old "Poly Paks" surplus company (remember them?) was famous for
buying "floor sweepings" scrap parts, bagging them up, and selling them
in bulk for absurdly low prices. Their advertising was (nominally)
honest: Things like...

- We sell good junk
- No chance to test
- Buy 'em by the barrel
- Only a little out of spec
- Memory chips, 99% good bits
- Blemished 7-seg display
- How good? Who knows!
- Unmarked resistors, capacitors, diodes, etc, 100 for $1.98
- Used parts; their loss is your gain

I don't think the Fluke system that I use is particularly
sophisticated, it is all TTL inside. So the test part just doesn't work
if it fails, it isn't some corner condition being exposed.
I had a Fluke 9000, though with the Z80 probe. For those who don't know
it, it's a clever piece of equipment. The pod has a set of analog
switches and a ROM. You unplug the CPU, plug the pod in instead, and
plug the CPU into a socket on the pod. A (serial?) cable connects the
pod to the main unit, which is like a big desk calculator.

Testing a CPU chip is hard. Testing some random circuit connected to it
is even harder. So the Fluke uses a technique called "signature
analysis". In effect, it's like connecting headphones successively to
each pin, and listening for the "noise". You compare the noise to a
working systems "by ear", and decide whether it sounds normal or not.

The Fluke uses the ROM in the pod to run a known program, and captures
the bitstream "signature" on each pin with a shift register. It displays
the result as a 4-digit hex number, like a CRCC check. Any incorrect or
misplaced bits will change the signature.

Then, I am sure that these parts are vulnerable to handling
electrostatic discharge, which to your point makes them unreliable in
the long term.
I test my 1802's in a Membership Card, hooked up to a bench supply that
meters the voltage and current. An 1802 with a blown input generally
draws excessive supply current. I also run a simple program that tests
the memory and I/O.

Lee

--
A designer knows he has achieved perfection not when there is
nothing left to add, but when there is nothing left to take away.
-- Antoine de Saint Exupery
--
Lee Hart, 814 8th Ave N, Sartell MN 56377, www.sunrise-ev.com




Doug Jackson
 

Sigh.

Yep - Negative feedback is the only thing these sellers listen to.

Kindest regards,

Doug Jackson

ph: 0414 986878

Check out my awesome clocks at www.dougswordclocks.com
Follow my amateur radio adventures at vk1zdj.net

-----------------------------------------------------------

Just like an old fashioned letter, this email and any files transmitted with it should probably be treated as confidential and intended solely for your own use. 

Please note that any interesting spelling is usually my own and may have been caused by fat thumbs on a tiny tiny keyboard.

Should any part of this message prove to be useful in the event of the imminent Zombie Apocalypse then the sender bears no personal, legal, or moral responsibility for any outcome resulting from its usage unless the result of said usage is the unlikely defeat of the Zombie Hordes in which case the sender takes full credit without any theoretical or actual legal liability. :-)

Be nice to your parents.

Go outside and do something awesome - Draw, paint, walk, setup a radio station, go fishing or sailing - just do something that makes you happy.

^G ^G ^G ^G ^G ^G ^G ^G- In more laid back days this line would literally sing ^G ^G ^G ^G ^G ^G ^G ^G 





On Mon, Aug 10, 2020 at 5:09 AM coolcash3030 via groups.io <coolcash3030=yahoo.com@groups.io> wrote:
You also have to be wary of ordering cheap or even not so cheap chips on Aliexpress....many are pulls and refurbished....most sellers are honest about that but I did hear of some outright fakes being sold by some AE sellers....of course these pulls are a gamble ... I recently got some 2764s 10 for $10 .... 9 tested good and 1 doa. My main beef with the AE chip sellers is the packaging....Saran wrap! I gave some negative feedback on the packing....I know these pulls were probably abused but let's give these chips a fighting chance.


ian may
 

On Sun, Aug 9, 2020 at 10:22 AM, Martin wrote:
OK I dragged out the tester and went through all the parts, notes attached to the pictures. 
I've got a close match to your printed on 2 lines 6FVLA devices - 6FVKJ an RCA device "449" so 49th week of 1984. I have an RCA "318" which has FVXMH. The next one by date is "336" which has 6FVVY and is the first one I have that starts 6F.  The last one I have has 6FVMA date code "706" so 6th week of 1987. Just to confuse things I have a "610" 6FVBO which suggests that the letters after the V were not a linear progression. So two line starting 6F parts are RCA made perhaps some time between week 36 of 1983 (but not before week 18 1983) and at least the 6th week of 1987. I need to look into the single line 6FVT662281. I think I have some single line 6F devices somewhere in my collection. Anyway I'm reasonably confident that none of the two line 6F devices were made by Harris.
Cheers, Ian.


Martin
 

Thanks, Ian.

There's an inconsistency, though. The Intersil processor, which has a laser etched logo and markings, also has a 6FV.. code on the back. I didn't see that until I took the picture, as it is very faint. 

So the Intersil processor:
   - could be a laser remark
   - could be from a shared production line
   - could be a packaging house mark

I guess we need some more codes. 


On Mon, Aug 10, 2020 at 9:55 AM ian may via groups.io <fps16xn3=yahoo.com@groups.io> wrote:
On Sun, Aug 9, 2020 at 10:22 AM, Martin wrote:
OK I dragged out the tester and went through all the parts, notes attached to the pictures. 
I've got a close match to your printed on 2 lines 6FVLA devices - 6FVKJ an RCA device "449" so 49th week of 1984. I have an RCA "318" which has FVXMH. The next one by date is "336" which has 6FVVY and is the first one I have that starts 6F.  The last one I have has 6FVMA date code "706" so 6th week of 1987. Just to confuse things I have a "610" 6FVBO which suggests that the letters after the V were not a linear progression. So two line starting 6F parts are RCA made perhaps some time between week 36 of 1983 (but not before week 18 1983) and at least the 6th week of 1987. I need to look into the single line 6FVT662281. I think I have some single line 6F devices somewhere in my collection. Anyway I'm reasonably confident that none of the two line 6F devices were made by Harris.
Cheers, Ian.


ian may
 

Martin, is that 6FV code on one line like 6FVZOX6455 or two lines like
6FVEO
Z6270
?
That single line code is from a group of 4 ceramic package Harris CDP1802D "Z 949" (1989 week 49) chips that I have. That is why I thought Harris used a single line code. Further digging through my chip box produced another group of 4 chips with that two line code but they are plastic CDP1802AE made by Harris "Z 036" (1990 week 36). So the "Anyway I'm reasonably confident that none of the two line 6F devices were made by Harris"  I wrote in my last post is WRONG.
At some stage Harris changed to a 4 digit date code, the earliest one of those that I have found in my collection so far is a CDP1802ACE marked "J 9447" on top and on the bottom:
XX75J
MALAY
I have found 3 other Harris plastic chips marked (top side date - bottom top line/mid bottom line (if any)/bottom bottom line) 9531 - XO43E/MALAY, 9540 XH55G/MALAY and 9549 K6127/VT03C/SINGAPORE. I think the change over from Harris to Intersil was in late 1999 so "6FV" on an Intersil device looks suspicious.
Cheers, Ian.


Martin
 

The Intersil device (laser etched logo and numbers) is labeled:
6FVZA 
V6918

In the Harris "poor topside remarks" I have of note:
6FVT6228IU (single line, nicely printed)

and

6EVZA
V6513
(which looks a lot like the code on the back of my Intersil part).



The underside markings are usually pretty rough - inconsistent placing, random angles. 

There are still some mysteries to be unraveled, by the looks of it. I wonder how many fabs were actually building 1802s. By any standards, it was an outdated process and design. so it would have to run in a legacy fab. 


On Tue, Aug 11, 2020 at 11:31 AM ian may via groups.io <fps16xn3=yahoo.com@groups.io> wrote:
Martin, is that 6FV code on one line like 6FVZOX6455 or two lines like
6FVEO
Z6270
?
That single line code is from a group of 4 ceramic package Harris CDP1802D "Z 949" (1989 week 49) chips that I have. That is why I thought Harris used a single line code. Further digging through my chip box produced another group of 4 chips with that two line code but they are plastic CDP1802AE made by Harris "Z 036" (1990 week 36). So the "Anyway I'm reasonably confident that none of the two line 6F devices were made by Harris"  I wrote in my last post is WRONG.
At some stage Harris changed to a 4 digit date code, the earliest one of those that I have found in my collection so far is a CDP1802ACE marked "J 9447" on top and on the bottom:
XX75J
MALAY
I have found 3 other Harris plastic chips marked (top side date - bottom top line/mid bottom line (if any)/bottom bottom line) 9531 - XO43E/MALAY, 9540 XH55G/MALAY and 9549 K6127/VT03C/SINGAPORE. I think the change over from Harris to Intersil was in late 1999 so "6FV" on an Intersil device looks suspicious.
Cheers, Ian.


thinkpast
 

http://www.retrotechnology.com/restore/chip_re-labeling.html

In support of the discussion, here's what I've accumulated on the matter. The Web page describes itself of course; a series of people give their experiences and considerations they are aware of. The problems and issues go up the entire IC food chain. One can read electronic trade "magazine" articles (Web blogs and sites) about its impact in industrial and military work, where one really needs full-spec parts.

In earlier cosmacelf discussion, Ian May described 1802 IC brand and year identification by the imprint marks on the plastic packaging. I summarize that information too:

http://www.retrotechnology.com/memship/cosmac_1802_production.html

"Legacy fabs" have been mentioned in this thread. That's actually an industrial niche. companies produce small runs of legacy devices. In some cases they retest and re-certify. Some have dies (unpackaged IC's) which they package. Rochester Electronics is a reputable legacy fabrication company; they source some (Intel) 8085's for instance.

Regards, Herb Johnson


Martin
 

Thanks, Herb!

Looks like we need to start collecting more of those back side codes. I'll take a look at the units that passed my tester. 

 BTW, many years ago I had a conversation with a  lower-tier chip supplier that went something like this:

"what do you do about defective parts"
   "what do you mean"?
"customer returns - parts that fail in test"
   "oh those - we mix them in with the next batch, and as long as we are under 2 percent bad - problem solved".
(2 percent was the contractually agreed error ceiling)

Of course, the parts were way cheaper.


And many years ago, Chinese recyclers were heating scrap boards over braziers until the parts came loose. Not so much any more. 



On Tue, Aug 11, 2020 at 8:03 PM thinkpast <hjohnson@...> wrote:
http://www.retrotechnology.com/restore/chip_re-labeling.html

In support of the discussion, here's what I've accumulated on the matter. The Web page describes itself of course; a series of people give their experiences and considerations they are aware of. The problems and issues go up the entire IC food chain. One can read electronic trade "magazine" articles (Web blogs and sites) about its impact in industrial and military work, where one really needs full-spec parts.

In earlier cosmacelf discussion, Ian May described 1802 IC brand and year identification by the imprint marks on the plastic packaging. I summarize that information too:

http://www.retrotechnology.com/memship/cosmac_1802_production.html

"Legacy fabs" have been mentioned in this thread. That's actually an industrial niche. companies produce small runs of legacy devices. In some cases they retest and re-certify. Some have dies (unpackaged IC's) which they package. Rochester Electronics is a reputable legacy fabrication company; they source some (Intel) 8085's for instance.

Regards, Herb Johnson