6502.org

I've replaced 10k resistors with 1.5K ones (I really don't have 3.3k ones)
I have 24Mhz and 32Mhz oscillators. Divided these by a counter to test 6, 8 and 12 Mhz. 6 and 8 ok, 12 Mhz fails. But my 24mhz logic analyzer can't properly display 12mhz clock too at that frequency.

If this is a counterfeit product by the way, it's not that strange they chose the name Winbond, in the below link they are wholesaling a part with Winsome brand

http://www.alibaba.com/product-detail/- ... 76269.html

That's just a typo by the Alibaba web site. The markings on the chip say Winbond, right?

Not sure what to say about your 12 MHz test. There are several things that could cause a failure at high frequency. Can you post a photo of the test rig, especially showing the bypass caps and the counter etc used to generate the clock? What type of IC are you using for the counter?
Hmm. If you go with a dummy opcode of $A9 then the address bus should increment every cycle, as Mike said. So, how about this. You can fool the logic analyzer by giving it A0 and "pretending" that's the clock. A0 will have one-half the frequency of PHI2, so now the "clock" is lower. Then connect the other logic analyzer lines to observe A1, A2, A3 and so on -- they should show the incrementing behavior.

(BTW even 1.5K may be too high for those resistors. I won't try to predict what the actual limit is -- that would require actual, as opposed to published, timing specifications; also actual input capacitance and actual input-voltage thresholds.)

Actually the company's name in that link is Winsome. There is no picture of the IC they sell but the part number is W65C816.
A quick hack on the breadboard. I use a 74HC4040 12 bit counter IC. 220nf caps are there for all the supply pins to the parts. Logic analyzer's cables disrupt the view. I removed the reset part and feeded counter's most significant output to the reset pin so cpu would reset every 2^11 cycles.
High res image : http://www.tepetaklak.com/data/20160122_225553.jpg
I used counter's Q1 output to get a 6Mhz signal to feed to logic analyzer since chip has failed with both $A9 and $EA opcode.

Here is a photo of the chips.. When I closely examined them on the photo I decided that they are indeed counterfeit products. Small cavities (dots) under the chip's notches are non standard among the chips. They are all different sizes.
High res image : http://www.tepetaklak.com/data/20160122_225617.jpg

Well, you know what they say, right? "You Winsome, you lose some" ... I'll show myself out, thanks.

Mike B.

Sorry for the delayed response—it's been a long and busy day.

I've been swapping E-mail with David Cramer at WDC, who in turn, has been swapping E-mail with Bill Mensch. They have never heard of Winbond making discrete devices—and I would think they'd know. I have not pulled any of this out of my hat.

<Groan>

He's here all week, ladies 'n germen.

Could they be older 8mhz pulled parts that have been "tidied up".

The top chip is a 20 year old pulled W65C816SP-8, it's not too clear in the photo, but the word TAIWAN and what looks like a batch code (93161X) are engraved into the bottom of the chip roughly in the middle between the 2 grey lines. One of the dots is blank, the other has B4 moulded in it

The newer chip pictured at the bottom is a much more recent W65C816S6PG-14, and in this case TAIWAN is moulded into one of the dots and 6J4 in the other

@Martin A : Mine look a lot similar to the below picture.
One of them has Taiwan G2, and an empty dot.
One of them has Taiwan D3, and an empty dot.
Two of them have PHILIPPINES and BR
One has TAIWAN (very hard to read) and A2

All of them have non uniform bottom paint giving the impression that they got some processing. (sanding and or painting)

By the way, where can I get information about these 8mhz parts?

I'm disappointed that these chips are not what they appear to be. But at least you got something functional (apparently, so far). Of course the red flags are up. No quality control assurance on these babies!
I haven't checked what's in the 6502.org document archive but here's an old GTE datasheet for 8 MHz '816 and '802.

• GTW_G65SC802_G65SC816 (20 pages).pdf

  (2.12 MiB) Downloaded 91 times

Nice trick with the Reset! But, from what I can see, connections to the bypass caps have a lot of room for improvement. Admittedly it's a contradiction that you would be using a breadboard to create a circuit whose maximum clock speed you wish to determine. But if you MUST go that route then I suggest the bypass cap for the CPU be attached to the chip itself -- soldered on top, straddling the 40-pin package.
Thanks for the clarification. Busy day and delayed response for me, too! It seems to me WDC themselves should already be on record with a first-person statement of who the alternative sources are, but it's not your fault that isn't the case. What's the accepted practice in the industry? It seems a sad state of affairs when a fake Winbond chip and a legit GTE chip could both get tarnished with suspicion of being counterfeit.

[Edit: maybe I'm expecting too much. For every alternative source agreement there was probably a press release at the time. But now the info isn't easily accessed.]

-- Jeff

There's a copy of an early (c 1986?) WDC datasheet for the W65C816 which covers 2,4,6 and 8mhz parts, within the Apple2 GS hardware reference found here http://apple2.gs/library.php.

Its a low resolution scan so not the most readable document.

Yes, in Chapter 10 -- pdf page 115; or page 205 according to the original numbering. Thanks, Martin!
I think the main problem has to do with washed-out gray scale (before-&-after tweak below). Looks like interesting material in this manual, though.

Bummer...

Didn't think of the possibility of counterfeit 65816 chips.
Sorry to hear they are not what they suppose to be.

Hopefully I should be safe if I stick with Mouser and DigiKey...

Major electronics suppliers such as Mouser and Digi-Key have programs in place to try to catch counterfeits. In the case of the 65C816, WDC has publicly disclosed the foundries that produce their devices, which tends to make it easier to trace the parts back to their progenitors.

I commented on this issue in my previous post, then tempered my remarks with an edit on Saturday to insert a post-script. Apologies if this was too easily overlooked -- I maybe should've made a new post.

Certainly the info about alternative sources isn't secret, but neither is it on record if the best we have to go on is, "BDD said that Cramer said that Mensch said... " Also, Winbond is just one part of the puzzle. The inquiry revealed nothing about other potential sources.
FWIW I have '816 datasheets from VLSI and from GTE (later acquired by CMD, I believe) -- and no Winbond datasheet. But the latter's absence didn't alarm me, given that old datasheets can be hard to find. And regarding the actual chips, GTE, CMD and VLSI markings can be faked. WDC, too.

I guess the takeaway is: either deal with major suppliers (for the reason BDD noted) or accept a degree of risk and be on the lookout for anomalies in the chips themselves. i_r_on, I hope you'll keep us informed regarding your further success (or otherwise) with your purchase.

-- Jeff

I think you should re-think your strategy and ask WDC for a meeting if you need volume. They will certainly give you much lower price on volume than what some dodgy parts from "winbon" will be. They are nice folks to talk to.

A usual strategy from such companies is to give you 10 working parts that are top notch. If you order 1000pcs you will be lucky to have 10 more working. This is no joke.