Sorry if this is asked before, I've searched the forum but couldn't find a direct topic related to this.

I just bought a bunch of cheap 65C816 processors from China. I want to test them before I confirm the transaction if they are dodgy/broken or not. (They are Winbond made labeled W65C816S8P-14)

The question is : Is there anything else I need to care to do the NOP test on the 65C816 different from the same test that can be done on 6502?

I'm a schematic-disabled guy so I'll just describe the connections I made,
- 1Mhz Oscillator Can connected to PHI0 IN
- RDY, IRQ, NMI, ABORT pulled high by 10k resistors.
- VDD to 5V, VSS to GND
- RESET connected to a tactile button, open position is pulled high with 10k Resistor, when shorted connects it to the GND.
- D0..D7 pulled high or low with 10K resistors to make the NOP opcode (EA). 11101010 (1=pull up, 0=pull down)
- A0..A7 directly connected to logic analyzer.

The first chip I tested put continuous DA on A7..A0. The others made some other weird things

Now I'm unsure whether my test fixture is bad or actually the chips are bad. Any idea?

Winbond-made? I wasn't aware that they were ever licensed to produce the 65C816 as a discrete device. If they were, the part number definitely would not be W65C816S anything, as that is the genuine WDC part number (the 8P part refers to the geometry and packaging, and the -14 is the speed rating). I'm suspecting that you may have counterfeits.


Short of running these devices in a 65C816 system that is known to correctly operate, I doubt that you can devise a test regiment that will 100 percent prove that you have fully functional hardware.


There is no PHI0 IN input on a 65C816. Also, you need to be running a much higher Ø2 frequency to fully qualify the part.


Make that 3.3K resistors and don't forget BE, which must be high in order for the MPU to be able to drive the buses.


A push button alone will not generate a clean reset, due to contact bounce. You may be able to get away with a simple R-C delay circuit that dampens the bounce. An alternative is to use a Maxim DS1813 reset generator.


Again, make that 3.3K resistors.


Presumably you are just monitoring the address lines. Make sure that they are all clearly in logic zero or logic one voltage ranges at all times.


Make sure BE is pulled high and then test again. As I said above, I think you may have been sold counterfeits.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

In addition to a hard-wired $EA, you could also try a hard-wired $A9, to see if the address bus counts twice as quickly. And prior experience tells me that you should heed BDD's advice about the reset pin ... a suitable RC circuit would suffice, but a simple switch and pull-up resistor will likely cause inconsistent (or even consistently bad) behavior.

You would probably need to do something like borrow a working IIgs as a test-mule to be a bit more confident, but even that wouldn't answer the "-14" part of the performance question.

Mike B.

I looked into this some more. Although Winbond has an IP license with WDC, they have not made discrete 65C816s for general sale. It definitely appears that you have counterfeits. Could you please advise where you got them? You can PM me if you don't want to go public with that information.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Well, one thing I forgot was indeed BE line. I pulled it up to VDD. The other problem was with me actually not in the setup. The circuit was expected to free run actually. My badness I thought I should have seen EA on the A7..A0

Now I tested one of the chips today and it indeed free runs... I just see address lines incrementing

It's no secret, I bought these chips from AliExpress, they have genuine like winbond markings on them.

Here is where I bought them (not affiliated with them) : http://www.aliexpress.com/item/5PCS-W65 ... 65980.html

It seems they are out of stock now Even though only I purchased from them... Seems dodgy Maybe they are just 6502 chips
Second test, pull BE low... Indeed address lines read 0xFF... not so dodgy

ps: Tested all five chips and they all look good so far.

Genuine W65C02S and W65C816S MPUs are only sold through distributors authorized by Western Design Center. Going to anyone else almost guarantees that you will be sold "dodgy" parts. Incidentally, you paid over twice as much for those devices as they cost at authorized distributors.

As I said, Winbond did not make these devices.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Well, my "genuine" remark was for the "Winbond" marking on the chip. Maybe these are produced by Winbond or not I don't know. They are not in the wild in Aliexpress so I just can't be sure that some company just stole the IP and used Winbond's name on them.

Actually it's cheap because most of those authorized distributors charge at least the price I paid for five of them for shipping only from US / UK or so on. It's ~$4 a piece shipping included.

On mouser, 5 65c816s chips cost 60€, makes ~$13 each one. 3X the cost.

I can't delve into the IP rights/ethics issue because I didn't know if Winbond actually produced these chips or not when I ordered.

I would gladly pay an extra $10 for peace-of-mind, but that's just how I like to roll. Now, if I was buying in bunches, I might be tempted to re-think that ...

Mike B.

According to Wikipedia, Winbond's W55V9x series ASIC's use the WDC '816 core, so clearly a business relationship exists between the two companies. I find it plausible Winbond also made some '816 CPU's. The only thing remotely strange (to me) is the "W" prefix on the part number, suggesting WDC. Or Winbond, I suppose.

This search suggests 816's bearing the Winbond name are not entirely uncommon. And the prices are attractive. But AFAIK Winbond is not a well-recogized source for this product, so wouldn't the hypothetical counterfeiters fake a source that is well recognized?

_________________
In 1988 my 65C02 got six new registers and 44 new full-speed instructions!
https://laughtonelectronics.com/Arcana/ ... mmary.html

...which I did note in an earlier post when I said that Winbond has an IP license with WDC.


WDC has seen this in the past. I was told Winbond did not make 65C816s in discrete devices.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Oops, I overlooked your mention of that point. Anyway we're in agreement, for whatever an IP license proves -- and I'm not saying it's conclusive, just suggestive.

I know VLSI and GTE made 816's. Why is another source implausible? Maybe i_r_on's initial failure planted doubt in our minds, but I think that's straightened out now (thanks to you; good catch on the BE thing). Now he has an apparently working, apparently Winbond chip. I'm not clear what you're telling us. Someone at WDC told you that Winbond never made 816 CPU's? (If we believe Winbond 816's never existed at all then we have to believe the counterfeiters REALLY pulled a boner in choosing that brand name! But it's a strange world we live in.)

_________________
In 1988 my 65C02 got six new registers and 44 new full-speed instructions!
https://laughtonelectronics.com/Arcana/ ... mmary.html

Do the chips LOOK like they've been re-marked or altered

http://www.aeri.com/counterfeit-electronic-component-detection/

I was given some suspect Z80s to play with, it's pretty obvious in the photo which one is the Mouser supplied genuine part, and which 3 came from China.

In this instance the suspect chips are NMOS Z80's (probably 4mhz Z80A) rebadged as 20mhz CMOS parts. They work fine, until you try to clock them faster than their original rating

Did you try higher frequencies? 10MHz or more?

At higher frequencies i_r_on's NOP setup might indicate a fail even with a fully functional part. That's because the value of the resistors used to apply the $EA (or $A9) becomes critical. 10K is too high (although it'd be fine at low frequency), and even 3.3K may be marginal. During an opcode fetch the resistors have less than one-half clock cycle to replace the bank address with $EA.

In the event of a failed high-frequency test I would reduce the resistors to 2.2K or 1K just to see what happens. Better yet, ditch the resistors and apply the $EA by means of some sort of tri-state buffer (eg: 74hc245 or '244 or similar), enabled only during PHI2 high.

Proper supply-bypass capacitors also become more important at higher frequencies -- another potential cause of an erroneous "failed" test result.

_________________
In 1988 my 65C02 got six new registers and 44 new full-speed instructions!
https://laughtonelectronics.com/Arcana/ ... mmary.html

(Added a few sentences to my post above, BTW )


There's a big difference between a counterfeit part and a legit part not intended for general sale! One we can expect to operate flawlessly; the other not.


BDD, earlier you said they have not made discrete 65C816s for general sale. Fair enough, maybe something new came to your attention. But this is a key point, because a legitimate part not intended for general sale is hardly the same as a counterfeit. It will be helpful if you can share not only your conclusions but also the details of the supporting information.

_________________
In 1988 my 65C02 got six new registers and 44 new full-speed instructions!
https://laughtonelectronics.com/Arcana/ ... mmary.html