Replacing a 6502 with 65816

Post by **Mike Naberezny** » Tue Jan 06, 2004 7:48 pm

Daryl,

8BIT wrote:

Sorry about the previous link. I was in a hurry to get home from work. Here is an HTML doc with the picture included. It should work.

http://65c02.tripod.com/conv.html

Would you mind if I placed a version of this in the Hardware Mini-Projects section on 6502.org? I think it would fit in very well there and people ask about this from time to time.

http://www.6502.org/mini-projects.php

Regards,
Mike

8BIT · Post by **8BIT** » Tue Jan 06, 2004 11:02 pm

Mike,

I don't mind but this has not been tested on hardware yet. Upon a suggestion from Garth Wilson, I have tried to come of with a trial schematic.

Before publishing in the miniprojects, someone ought to test it, or at least validate it.

Thanks!

Daryl

Post by **Mike Naberezny** » Wed Jan 07, 2004 1:24 pm

8BIT wrote:

Before publishing in the miniprojects, someone ought to test it, or at least validate it.

Absolutely. All of the projects in the mini-projects library get built and tested before they are posted, so I will probably build one of the adapters myself also once the design is hammered out.

Regards,
Mike

8BIT · Post by **8BIT** » Thu Jan 08, 2004 4:45 pm

I'd like to ask for input on the actual adapter's physical design.

There are many ways to go about it. Ideally, we would want to keep it the size of a 40 pin DIP so as to facilitate installation into an existing layout. However, we are talking about putting 4 IC's in the space of one. There are SMT varieties of the 74 series IC's as well as SMT PLCC sockets for an 65816 which can help to reduce size. I have not used any SMT devices but have been told by some that they are easier to solder than it seems.

Here are some detail that I've been pondering over. We'll need a PCB thats the size of a 40 pin DIP to use as the base. We'll need to attach pins to this pointing down to be used to connect to the host's 6502 socket.

My question is about sourcing these pins. They have to be able to fit into a standard DIP socket. The IDC header pins cannot be used because they are square and would either not fit or damage the host's socket. Does anyone have any suggested sources for these pins?

Now, we can place the three SMT 74 series IC's on the top side of the board down the center of the DIP40's pins. Finally, we stagger another 40 pin DIP slightly up and left and place a socket on top of the SMT components. This will be for the 65816. The PCB ends up being slightly longer and wider than an 40 pin DIP socket. See the layout image here:
http://sbc.rictor.org/support/conv.html

What do you all think? Will it work? Will there be enough roome to place such a module into an existing 6502 socket?

What do we use for the pins???

Daryl

GARTHWILSON · Post by **GARTHWILSON** » Thu Jan 08, 2004 7:48 pm

I don't have an answer on the pins yet, but it might work well to put SMT parts on both sides. Instead of going for the '816 in a PLCC, I would go for the PQFP which is smaller and probably much easier to solder by hand since it has the pins sticking out the sides instead of curled under.

Avoiding the extra 40-pin DIP socket with the '816 sitting up there high might be what someone in Fabrice's position needs, since he apparently has a pretty limited amount of vertical room too.

8BIT · Post by **8BIT** » Thu Jan 08, 2004 8:11 pm

GARTHWILSON wrote:

I don't have an answer on the pins yet, but it might work well to put SMT parts on both sides. Instead of going for the '816 in a PLCC, I would go for the PQFP which is smaller and probably much easier to solder by hand since it has the pins sticking out the sides instead of curled under.

Avoiding the extra 40-pin DIP socket with the '816 sitting up there high might be what someone in Fabrice's position needs, since he apparently has a pretty limited amount of vertical room too.

I'll look into the PQFP package and see what I can do with that. The PLCC sockets, either in PCB mount or SMT mount styles, would not fit between the pins of the DIP, which would have required a second board and much more complexity.

Thanks!

Daryl

Euphoric · Post by **Euphoric** » Tue Jan 13, 2004 11:14 am

Quote:

The PCB ends up being slightly longer and wider than an 40 pin DIP socket. (...) Will there be enough roome to place such a module into an existing 6502 socket?

Well, I don't have access to all the 6502 systems, but I have never seen a system having the 6502 socket so close to another chip that you cannot overflow all around the chip...
I would say (with no guarantee though) that you could use 1/10 of inch all around the socket without any problem...

Cheers,

Fabrice

PS: I have worked on the software part for that 65816 module... I can announce a full rewrite of my 6502 C code generator that gives outstanding 65802 code !

8BIT · Post by **8BIT** » Thu Jan 29, 2004 11:09 pm

I'm still working on the board design. I'm having a hard time trying to place the parts in a way to allow all the traces to fit in the limited space a DIP 40 provides.

If anyone else would like to try, you can download the ExpressPCB schematic file from here:
http://sbc.rictor.org/support/conv.html
Daryl

Euphoric · Post by **Euphoric** » Fri Jan 30, 2004 11:04 am

Quote:

If anyone else would like to try, you can download the ExpressPCB schematic file from here:
http://65c02.tripod.com/conv.html
Daryl

I'm going to try your Phi2 generation...
On my prototype board I used two inverters to derive Phi2 from Phi0, and I don't if this is related or not, but I have problems when using 70 and 80 ns DRAMs, whilst it works ok with 150 ns ones. So, I'm going to try adding a 08 chip (room is sparse...) and see if it works better with it...

Fabrice

PS: do you want me to start regrouping french orders for your module ?

8BIT · Post by **8BIT** » Fri Jan 30, 2004 1:54 pm

Quote:

PS: do you want me to start regrouping french orders for your module ?

I would wait until we have 1) tested the design and 2) have a board layout complete.

Let me know how your experiment comes out.

Daryl

Euphoric · Post by **Euphoric** » Wed Feb 04, 2004 9:56 am

Quote:

Let me know how your experiment comes out.

Daryl

Ok, I have added a HCT08 to my module in order to generate Phi2 like you. It works "better"...
I mean, on my system, it works when I have 150ns DRAMs.
It still does not work completely when I use 70 or 80ns DRAMs, but it is better than the two-inverters-only Phi2 generation because now I can see that the system boots and only crash when the "memory mapper" starts doing some job (the system didn't start with the two-inverters).

This lets me think that we still need to adjust the schematics if we want to be really compatible with any 6502 system... IMHO, the timing is not exactly the same, and that could affect some systems...

So, what is the intended use of this module ? most of the 6502s we want to upgrade are used in 1 MHz or 2 MHz systems, am I right ?
I think we should try having the same Phi2 than the one generated by a Nmos 6502A, don't you agree ? Also, I'm not sure that all the different systems will have a Phi0 at CMOS levels... I would suggest using an HCT04 and HCT08 instead of their HC brothers...

I have borrowed a scope and will try measuring the Phi2-Phi0 time difference, but I'm not a pro, so it would be nice someone else does experiments too...

Best regards,

Fabrice

Euphoric · Post by **Euphoric** » Fri Feb 06, 2004 7:18 pm

Quote:

I have borrowed a scope and will try measuring the Phi2-Phi0 time difference, but I'm not a pro, so it would be nice someone else does experiments too...

Here it is, sorry I had to copy by hand what I was seeing on the screen...
I would say that we have a very good replacement for a 65C02

But, oh my god, what's that crappy Phi2 positive edge with a 6502A ???
(I've verified it on two different systems...)

http://oric.free.fr/HARDWARE/phi2vsphi0.jpg

Best regards,

Fabrice

8BIT · Post by **8BIT** » Fri Feb 06, 2004 8:52 pm

Fabrice,

I'll bet a 74HC04 will bring those waveforms even closer. Good job!

Daryl

vbriel · Post by **vbriel** » Sat Feb 07, 2004 12:53 am

Hey Daryl, anything I can do to help? I had to do a piggy-back board for serial I/O on my replica I. I needed to intercept the I/O lines of the 6821. I didn't try to cram everything within the 40 pin space, I used duel sockets on the bottom of the board to elevate it, and then laid the comonents on the side of the socket adapter. This is a down and dirty solution, but you could do 2 small boards, 1 adapter into the 6502 socket and a sipp board that goes vertical with the components. Just my $.02 worth. Here's a picture but it's a little blurry. The original socket is on the right of the piggy-back board.

http://www.applefritter.com/apple1/memb ... serial.jpg

8BIT · Post by **8BIT** » Sat Feb 07, 2004 6:23 am

If we were just doing the 65816 with the clock modification then it would be easy. But adding the 74hc245 is causing a real road block. Also, since some designs cannot spare a large space, the goal was to keep the board just slightly larger .1" wider and longer, and approximately the same height as a DIP. Using the tqfp 65816 and soic 74HC series keeps the height within specs, but the real estate on the board is very tight.

I've also been working on other projects and have not had much "extra" time to work on the board. I have not given up, just moving forward very slow.

Daryl

Replacing a 6502 with 65816

6502 > 65816

6502 > 65816

6502 > 65816