Identifying what 6502s from Aliexpress really are

[Martin A]

Looking at the underside of the chip will also give you clues as to what they really are. Quite often there will be a manufacture location, or printed text that can be matched to known devices.

[Atlantis]

cjs wrote:

Atlantis wrote:

Out of a dozen only one works perfectly fine with my computer.

You don't mention anything about the board you're using. Is it designed to work with NMOS CPUs? Given that many failures, and that one of your parts kinda worked for a bit, I'd start to suspect that there may be something about your board incompatible with them.

It is my own design. And yes - it was designed for classic NMOS version of MOS6502. I uploaded some photos and schematics here.
I've checked some (two or three) in my other project, which normally uses WDC65C02 (also clocked at 1MHz). Results were also negative.
I will try to check other chips on breadboard, as you suggested. All I need is time.

Quote:

I've got six used "R65C02" CPUs from various vendors (including at least four from AliExpress), and of those fully half of them are actually relabeled 1 MHz NMOS parts.

Yes, I had the same situation. I bought chips that supposed to be R65C02, but they were plain NMOS CPUs. I believe there was no single CMOS among them...

Martin A wrote:

Looking at the underside of the chip will also give you clues as to what they really are. Quite often there will be a manufacture location, or printed text that can be matched to known devices.

Can you telle me something about them? I was unable to Google anything. And yes, I am aware of broken pin in one of them.

[Chromatix]

The two marked MEXICO could be Rockwell parts. This doesn't necessarily mean they're vanilla 6502s, or even any kind of 6502 at all, but Rockwell parts have a good reputation.

Curious that two others actually have "6502" printed on the bottom, with another row of text crudely scraped off. These two also have consistent package moulding with each other, so that's probably from the original factory.

Another matching pair are the ones marked "5L09" and "5G04", which would be empty-stock movement headcodes on the British railway.

That leaves "LD04", "TW03", and the one with the long string of text as oddballs within this collection. The latter is the most likely to be traceable with some precision, given the right reference.

[Martin A]

I've seen the U3020 marking on re-marked chips before. It's apparently signifies a Synertek SYU6502, which is 1mhz NMOS I believe.

[Atlantis]

Chromatix wrote:

The two marked MEXICO could be Rockwell parts. This doesn't necessarily mean they're vanilla 6502s, or even any kind of 6502 at all, but Rockwell parts have a good reputation.

I tested the rest of the chips in my WDC65C02 board. They also didn't work.
Did Rockwell produced some other 6502 CPUs, with slightly different pinouts/application schematic. If I remember correctly R65C02 supposed to be a drop-in replacement for standard for NMOS 6502.
BTW the single working chip also has "MEXICO" designation. It is "R6502F MEXICO 1738".

Martin A wrote:

I've seen the U3020 marking on re-marked chips before. It's apparently signifies a Synertek SYU6502, which is 1mhz NMOS I believe.

In that case, shouldn't it be 100% compatible with standard MOS6502? If yes, it should have run in my computer...
Was I simply unlucky and bought batch of mostly damaged parts? Or is there a glimmer of hope that they are still useful, but slightly different versions?

Anyway... Can you recommend me a trusted source of MOS6502 or R65C02 on eBay or Aliexpreess?

[Chromatix]

R6502F is not a real Rockwell part number: http://www.cpu-world.com/info/id/Rockwe ... ation.html You could run my little software routine, over in the Programming section, to find out whether it's really an NMOS or CMOS part.

I do suspect you've ended up with damaged parts that were, once upon a time, working 6502s. A non-6502 with the right number of pins would probably exhibit a short-circuit through an ESD protection diode structure or a CMOS latch-up path, and thus get very hot very quickly when powered up as one. But there's also an outside possibility you've got an NMOS 6512 or two in there, requiring a Phi1 input and a BE pin pull-up which you aren't providing. Or a 65C102 which has different clock characteristics, so won't interface correctly with a system that assumes a standard 6502 clock arrangement.

[BillO]

Atlantis wrote:

Anyway... Can you recommend me a trusted source of MOS6502 or R65C02 on eBay or Aliexpreess?

https://www.ebay.ca/itm/10PCS-Microproc ... 2749.l2649

I bought a lot of 10 of these. $15 shipped for the lot. That's only $1.50 each.

I do not know if they are genuine Rockwell parts, or re-labeled. However, they are all new, they are all 65C02s, they all work and they all run at 4MHz.

I have bought plenty of stuff off polida2008 over the years without a problem.

[cjs]

BillO wrote:

I do not know if they are genuine Rockwell parts, or re-labeled. However, they are all new, they are all 65C02s, they all work and they all run at 4MHz.

Or both genuine and relabeled! I'd be curious to know what the date codes are on your parts.

According to BDD:

BigDumbDinosaur wrote:

In 1999, Rockwell Semiconductor was spun off into Conexant and MPUs with the Rockwell mark were no longer produced. Ergo any Rockwell 65C02 with a date code later than 1999 IS A FAKE. Conexant itself was bought out in 2017—65C02 production had ceased well before then. I've not seen a Conexant 65C02 in the flesh.

I had wondered if Conexant had perhaps continued manufacturing parts with the Rockwell markings, but a image search immediately turned up an R65C02P3 with Conexant logo and 0026 date code, so I'm inclined to believe that seeing a Rockwell name/logo with a later date code is indeed a reliable way of identifying a remarked chip.

But remember, remarked does not mean the chip isn't the exact model it says on the label, or possibly even a better one. As you can see from the testing I mentioned above, I have three "R65C02P" or better CPUs (marked "T" in the image below), that all have CMOS-level power draws and run a NOP test at 4 MHz with reasonable waveforms; it looks like two are probably the original marking and the other has been remarked with the exact same markings as the NMOS part immediately below it. I expect they just throw everything into a bin and remark it all, regardless of whether that's helpful or harmful.

I should mention that the "4" and "1" markings on my labels just indicate whether my NOP tests saw large or little waveform degradation as I changed the clock through 1, 2 and 4 MHz; that should be taken only as the "1"s will almost certainly not run reliably at 4 MHz, not that the "4"s will.

[BillO]

cjs wrote:

BillO wrote:

I do not know if they are genuine Rockwell parts, or re-labeled. However, they are all new, they are all 65C02s, they all work and they all run at 4MHz.

Or both genuine and relabeled! I'd be curious to know what the date codes are on your parts.

9326 - all 10 of them.

However, back a few years ago I bought 2 R65C02P4s that have 0815 date codes on them that were 65C02s, they work and run at 4MHz.

Not to dis BDD, but Conexant sold it's Rockwell IP and the Mexico fabrication plant to NXP in 2008. However, the deal did not go through until August of the year. This is all very confusing. We have no proof that 65C02s were not manufactured under the Rockwell brand, either by Conexant or by NXP, after the Rockwell name and logo went up for grabs. It's all just speculation. I know it's easy to brand "those guy's" as criminals, but we just don't know enough to paint them all with the same brush. JMO.

[Chromatix]

Okay, on the MEXICO chips, the code 11450 seems to be an internal Rockwell designator for the 65C02. The corresponding 65C102 code is 11451. An NMOS part would carry R6502 here. So those appear to be genuine, just non-functional.

The suffix may indicate a die revision or a speed bin; H1 without a dash doesn't match any top-printed examples I can find, while pictures of the underside are rarer. Earlier Rockwell parts don't have the internal code printed on top, so yours may be examples from that period.

[cjs]

BillO wrote:

Not to dis BDD, but Conexant sold it's Rockwell IP and the Mexico fabrication plant to NXP in 2008. However, the deal did not go through until August of the year. This is all very confusing. We have no proof that 65C02s were not manufactured under the Rockwell brand, either by Conexant or by NXP, after the Rockwell name and logo went up for grabs. It's all just speculation.

Right. I point out the sale to Conexant and my discovery of a Conexant-marked R65C02P3 not to say that BDD is absolutely correct there, but to say I've found a bit of independent evidence that adds some plausibility to his claim.

And I can also see why vendors might remark parts just to change the manufacturer name; of three major forms of the 6502 (NMOS, R65C02 and W65C02) Rockwell seems the best-known brand name for R65C02s (I'd never even heard of Conexant before these threads in this forum) and it could well be that many beginners have a strong preference to buy "Rockwell" parts over others.

Sure, those of us with the knowledge and the time are not going to trust any of the labels now and use technical means to qualify the individual chips we're using when we're not buying new stock WDC parts, but that's a fair amount of work, and it is nice to have some heuristics to (we hope) increase our chances of getting the thing we want when ordering from random cheap sources on the 'net.

But yeah, it would be nice to have some more definitive information on what real markings actually existed and got shipped.

Quote:

I know it's easy to brand "those guy's" as criminals, but we just don't know enough to paint them all with the same brush.

Well, and as usual it just takes one bad guy in the supply chain to mess things up. For a retail vendor, doing full qualification of every chip you receive is expensive and probably puts you into a different price bracket and, thus, market. I don't immediately blame the guy who sold me a chip when it turns out to be mislabeled or otherwise broken.

[BillO]

Just to add another data point to this,

The 9326 dated parts have the code 11450-14, while the 0815 dated parts have the code 11450-13 like the Conexant part CJS found.

On the bottom the 9326 dated parts have:
11450H1
3434
MEXICO

The 0815 dated parts have:
11450E
MEXICO
3204

[BillO]

Just to add a little more fuel to the fire, I just contacted NXP. They were kind of guarded regarding their reply. They said that R65C02P4 is not one of their part numbers, neither have they marketed any component with the brand Rockwell on it. However, they did also say they manufacture all sorts of components for third parties with all sorts of brand names on them to which they added that because these are third party contracts they could not share any information about them.

Now here is the weird thing. They said they would trust parts from 2008 on. When I pushed them on that they said I should contact Rochester Electronics who are their only distributor authorized to offer obsolete parts, but again stressed that if they were a part manufactured for a 3rd party, no detail will be forthcoming.

I put in an information request to Rochester Electronics. I'll post their reply.


Edit: Update: No reply from Rochester, but they have started sending me SPAM ...

My guess is that there is no one there that knows or cares.

[Atlantis]

Ok, I tested my supply of 6502 bought from China during last few years. Unfortunately it turned out that there is more of faulty ones that doesn't work inside my computer.
I performed "freerun" test, forcing NOP instruction by pulling data pins up/down with resistors.
As for now CPUs were wires as standard NMOS wersion. Only six of them generated proper signals on A0-A15 pins. Rest of them either did nothing at all or generated faulty signals. Another one was close to being fine but two highest lines were scrambled.

Is there a possibility that those six chips are fine, but there is something in my computer design that makes them not compatible? I don't know... Voltage levels? Timings? EPROM/RAM I've used?
Or is it simply possible that they are damaged in different way, that allows them to perform freerun/NOP test?

Does it make a sense to test rest of them (those who generated nothing on address lines) using different application schematic?

Also: Can I easily modify my breadboard test circuit to test WDC65C816? I bought few of them some time ago and now I wander if they are fine.

[Dr Jefyll]

Atlantis wrote:

Is there a possibility that those six chips are fine, but there is something in my computer design that makes them not compatible?

Yes. Even a chip that's perfectly OK will act flaky or fail entirely if the surrounding circuit isn't right. For example, flaky operation is something that can arise (among other causes) from a poor power supply or improper application of bypass capacitors.

I don't recall seeing the computer you mention. Has it demonstrated its reliability, or is it still something of a question mark? Can you supply photos and other info? Or supply a link, please, if this info is posted elsewhere.

cheers
Jeff