_________________
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 thought I had something functional but reading your AC performance page makes it seem that I am deluding myself, as I have built the prototype on solderless breadboards and used long wires that ought to be a disaster area. I'll have to test at 1Mhz and see it fail. I'm not even sure how fast I've run the circuit so far - probably just 1Khz or 2Khz, so it "working" may be quite the overstatement.I also haven't really paid attention to what parts I'm using, it's a mishmash of what I had available and what I needed to order. There are very few parts though, so I could always replace them.The code I saw seemed to be intended for an LCD that is connected to the bus. Mine is on the far side of a VIA, and is already working; I wasn't planning to do much with it so I considered that module finished.

P.S. Sorry for the slow response it took me a while to realize my post had a page 2 <blush>

You can get to 2Mhz on breadboards without issues.

This works just fine:



-Gordon

_________________
--
Gordon Henderson.
See my Ruby 6502 and 65816 SBC projects here: https://projects.drogon.net/ruby/

Or even faster, with good construction. Blue Board runs at at least 5MHz (which is a slight overclocking of its vintage CMD parts.) I haven't tried to go faster.

Edit: That bit has to do with what Garth mentioned above. Those CMD parts are a lot more forgiving when it comes to edge rates than new WDC parts are!

_________________
"The key is not to let the hardware sense any fear." - Radical Brad

On top of what Garth already pointed out, you are looking at half that estimate as you are dealing with accesses on both phases of the clock, so your RAM is effectively seeing a 2MHz system.

Then you have the 25nS delay for the address and data to be valid on the bus after BE goes high again. Playing with BE is an expensive transaction (See P25 of the datasheet). Meaning that you need to (or should) wait until you are half way through high phase of PHI2 (at 1MHz) to ensure you aren’t peppering the RAM with corrupted data.

Taking all of that in to account, your 55nS RAM starts getting tighter all of a sudden.

I only know all of this because I’m presently doing the same thing in my present project and have had to resort to some trickery to get my own 6502 homebrew running stable, which I only managed today in fact.

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

Unfortunately I’m on a hiatus from this project for a while so I haven’t had a chance to see if my breadboard can go fast or not, but I am using a static core part and so am expecting problems. Once I have those problems, it’d be nice to be able to solve them without understanding analog electronics … as I’m about to demonstrate my lack of knowledge in the futile quest for a solution:

I must not really understand the issue, because it seems to to me that putting a small capacitor to ground ought to enable me to soften any edge that’s too fast? Please bear in mind that I realize that what I am saying must be nonsensical but hopefully explaining to me why it won’t work will enlighten me (and perhaps others).

Let’s say we want to have edges that are 10 ns long, and we’ll use 500 ohm pull down resistors on all the address and data lines so that they’re never floating. Then we want T = 2 nanoseconds (because we need 5T to fully charge the capacitor) and T = RC so our capacitor needs to be 2e-9/500 = 1/250 e-9 = 4e-12, so we put 4 picofarad capacitors in series with the pull down resistors and now our rising edges are about 10-15 nanoseconds long, and similarly for our trailing edges. Or maybe 2k pull downs work making the capacitors 1pF instead, I’m not really clear on the right way to size pull down resistors either. Aside from the exact values that are desirable, there must be some other flaw in this approach that I don’t understand?

What’s the flaw in this approach to avoiding hard edges?

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

I wouldn’t let posts deter you. I’d say give it a go.

I’ve personally had a stable-ish 38 MHz on a W65C02S on cheap and nasty breadboards alone. If you are looking at a few MHz you’ll have to do something really silly to stop it from working. And we are all here to bounce ideas and failures off anyway. We’ve all been there.

Give it a go. What’s the worst that can happen?

Experiment is good, but it can also be helpful to do some study - in particular, one can look at simple designs which already exist and which work.

I would note that one rarely sees pull-down resistors. Not never, but rarely. If you don't have a good reason to think you need them, based on some understanding or example, don't add them in. Similarly for other embellishments!

Good luck on your journey.

I agree on all fronts here. Pull-ups, in some cases like reset, RW, BE, etc. But pull-downs, not so much. Unless there is a specific goal in mind. Series resistors, sure, to take the edge off some signals. Not necessarily as requirement, but can be great for that fine tuning stage of the project.

Start with the KISS method and build from there, I say.

Garth’s primer, while awesome, can be an intimidating read. Making you feel that if you don’t have a dedicated ground plane or don’t use wire-wrap, that you are destined for failure. It’s not like that at all, I find the W65C02 to be quite forgiving. I’ve had masses of long wires all over the place on breadboards and still managed clock the chip easily beyond double or triple its specification.

Take the plunge, you won’t regret it.

In my case, the pull downs are on the monitor module which contains the LEDs for all the address and data lines, and there's a pull up on RWB. The effect of this ought to be that in the transition from the CPU as the bus master to the video card as the bus master - which I intend to do on the rising edge of Φ1, which ought to be slightly after the falling edge of the clock - all that might happen is a spurious read of address 0. Now, the modules are meant to be removable, and of course the debug monitor is a prime example of a module I'd expect to be able to take out of the stack, so if it turns out these pull downs are actually necessary I will discover it only when running tests of the "production" system.

All the posts here have done so far is made me think I should write a simple ram test that I can run at speed before I bother adding more to the system, and that rather than adding the mono display next that I should add either the keyboard interface or a serial interface next, both of which are recognizably smaller increments than the mono video card (if less exciting). Everything seems to be working fine at the moment but I haven't tried running it at 1Mhz yet.

I do appreciate the encouragment

This. I have read a lot of it, but couldn't describe it as a 'primer'. It is more of a collection of pointers to things I don't understand and it's hard to suss out the priority of understanding those things as they are currently not getting in my way. Off the top of my head, here are a few things I don't know that the primer seems to assume I should know or will need to know:

- open drain vs open collector meaning and purpose
- familiarity with 74-family ICs
- what the function of a ground plane is and why twisted pairs couldn't serve the same purpose

and basically my experience of reading the primer is that I have to go do more research to understand each sentence. At the end of the day it's probably less productive but significantly more fun to actually wire up a working circuit, which is what I have, while worrying that it won't -stay- working without fixing the multitude of sins I have committed in the process.

100% this. Get in there and start wiring things up.

He cops a bad rap around these parts for whatever reason, but I recommend checking out Ben Eater's videos. He does explain things simply and to the point and you'll get something that works if you follow along, barring things out of his control like dodgy connections etc.

Sure, there are things you'd improve upon and change over time, but when is a "Hello world" example ever intended to be a production run?

_________________
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?