Thanks to all for your thoughts. It's a confusing subject, but the material Toshi posted is quite revealing. Allow me to flesh in the details:

• due to the possibility of a page crossing, the NMOS 6502 always uses an extra cycle for indexed writes. The extra cycle is optional for reads, but for writes it is mandatory. That's because an invalid read typically is tolerable, but an invalid write is not. Hence, for writes, the NMOS chip must always wait the extra cycle, during which time the ADH may be corrected to reflect the page crossing.
  
• OK, let's say the original '816 -- the one that never got produced -- had logic capable of computing the entire address immediately (as today's 816's do in Native Mode for branches-taken with a page crossing). That would allow the extra cycle we are discussing to be eliminated in all cases -- reads or writes, page crossing or not. A valuable optimization!
  
• Unfortunately, the Apple disk controller, conceived for the NMOS 6502, "needs" the extra cycle. Therefore WDC was forced to redesign the '816 so as to reinstate it. Bye-bye, optimization!
  
• VDA and VPA outputs, used as intended, cause I/O devices not to be active during the extra cycle -- hence the recommendation that, "VDA and VPA should not be used to qualify addresses during disk operation on Apple systems."

It's easy to imagine WDC's frustration if, for the sake of one customer, they were forced to undo an optimization intended to benefit everyone. Here's a larger excerpt from the link Garth supplied... http://web.archive.org/web/201003310757 ... /65xxx.txt

cheers,
Jeff

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

It looks like you and Toshi solved the riddle. FWIW, I thought that the '802 was a neat little unit.
It was even mentioned in the WDC manual that the '802 was a more suitable 'plug-in' replacement
than the 'c02 for existing NMOS designs, due to its better cycle count mimicry ... as long as no
undocumented op-codes were used, which would have caused problems with either choice. It's
too bad that not enough people bought it to justify keeping it going.

Mike

Daryl has a design to supply it in a hybrid, but IIRC, the problem was that just as he finished it, the '816 in PQFP became unavailable. I don't think the board desgn has been thoroughly tested. At the time, there was quite a bit of interest.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

The issue has already been identified and explained in previous messages.

Toshi

It could be done with the PLCC-44 version of the '816 soldered to the PCB instead of being in a socket. In terms of total foot print, the PLCC-44 package is not a whole lot bigger that the PQFP.

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

The problem with the PLCC is that it doesn't fit between the rows of pins that are spaced to drop into an existing 6502 socket. There are ways to make the pins not come up to the top surface of the board, so the usable area on the top of the board can be wider, but I don't know how that's done. It might have to be like a multilayer board that gets one part drilled before being laminated to the other. That kind of thing is done for blind and buried vias, although these wouldn't be for vias but rather for the pins that stick out the bottom.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

I've not seen Daryl's design, so I'm sort of thinking "blind." It was just a thought.

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

It's in my link above. Maybe colorblindness obscured it?

Yeah, I think "blind" means they only come out to one side, and "buried" means they're only between inner layers so you can't see them from either side. That's something I've never done in a PCB design since I didn't want my employers to freak out about the cost.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

You guys are just through-hole dinosaurs, and one of you actually claims to be one.

I think someone like Ironwood makes the kind of adapter that you are after. If you want to make one, the bottom side, with the pins to stick into the DIP socket or even a single row receptacle for 0.025" square pins, would be soldered to a surface mount array of alternating pads. These single row headers have alternating pins at right angles to the header pins so that when placed on the surface mount pads it will stand up straight. I think that Samtec, Sullins, or there's a German company that makes these type of headers.

On the top side, a surface mount pattern for the PLCC finishes off the adapter. Only through vias are required to construct the PCB for this kind of adapter.

I think that I saw someone's project on this site constructed in this manner. IIRC it may have been a CPLD adapter described by Windfall (John Kortnik). It may have been on his site.

I've attached a picture of the SMT Sullins S1043E-20 header of the style I described above. I think that it will do exactly what you guys would like to do for a PLCC44-DIP40 adapter.

_________________
Michael A.

If we're still talking about Daryl's substitute hybrid for the now-unavailable 65802, it has four ICs on it, on a board made to take the space of a 40-pin DIP and plug into a 6502 socket in existing boards. Here's the layout, from his website:



IOW, it's not just putting a single PLCC into a DIP socket. Also, .025" square posts are too thick for IC sockets.

I use SMT for my PCB layouts for work, but I never do connectors in SMT, since I've seen them rip the foils away from the board when any force was put on them. Connectors basically need to be thru-hole because of how that has the effect of riveting them to the board. Thru-hole connectors also take less board space, because they don't have to have wings sticking out the sides.

I have not checked prices recently, but traditionally Ironwood has been three times the price of Aries which gives high quality. One of my early efforts for a multi-megabyte SRAM module was to stack these SRAM ICs on an Aries adapter like this:



Besides not looking nice, this was way too labor-intensive. The PCB I laid out for the moduel I sell stands vertically and takes less area on the board it plugs into:



It has four ICs on each side.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

Well I recollected incorrectly the details of the SMT CPLD to DIP adapter. It was Windfall's ReTuLa. Instead of the SMT header I described above, he used single row through hole headers, but the middle pins don't come through the PCB. Instead it appears that he manually trimmed them and soldered them in some manner. Here is a photo of the component side of the ReTuLa, and here is a view of the circuit side.

_________________
Michael A.

I am blue-green color blind. Links embedded in text often escape my view because the link color blends in with the text color and background. I went back and looked at your message and had to resort to moving the mouse over the text to find the link. I always boldface links and make them a contrasting color, mostly for my own benefit.


Daryl's layout is ExpressPCB, who don't provide blind via, even if the via only attaches to an inner layer.


As have I. Mechanical strength is only as good as the bond between the copper and laminate.

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

I'd noticed you edit your posts to make your links a different colour. Of course it doesn't affect any links in anyone else's posts... perhaps a bookmarklet would help? The following text can be dragged to the bookmarks bar of Firefox or Chrome, and then if you like given a helpful name such as "GreenLinks":


There are surely other tools you could apply to make global changes to the CSS applied to sites by your browser, but they would be (more) browser-specific. Making changes to your browser's presentation should be more effective for you, compared to changing the properties of your own posts.

Cheers
Ed

The problem, of course, is that each site differs in various ways, making a universal solution elusive. As I've always had vision problems, which have only worsened with age and infirmity, I've tended to use colors and contrasts that I see well, and would be seen well by others with vision maladies. My POC site's makeup reflects that.

My recent surgery has helped in some ways (no more cataracts and blood clots on the left retina) but not helped in other ways (right eye's acuity is markedly poorer than the left, especially in dim light—there's a complication that may require more surgery). I'm able to read now without having to enlarge fonts, but still am grappling with the color perception problems that can't be fixed by any means.

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

Probably your best bet is a user stylesheet, whereby you could set all colours styles and fonts as you choose.