H guys

I've created a prototype of my design and I've got a few issues. One of the biggest is extensive ringing on the whole circuit. I think this is down to me using veroboard and that I've use the outer tracks as the power rails. Now, I've seen Garth and others refer to a star-type ground. Unfortunately, I can't seem to find a good example of this.
Does it basically just mean taking leads from each ICs VSS/GND pin to a central 'area' of copper rather than relying on tracks? I.e. a glue on a central square bit of copper and use this as ground and all wires go to it? If any one has a picture of an example (not a printed circuit board version though) I'd be very grateful.
Sorry if this seems like a very basic question.

Hi banedon. The whole circuit? This makes me wonder whether your oscilloscope is being used properly. It's necessary that the business end of the 'scope probe have a good, short connection to circuit ground. If you omit that connection and rely instead on a separate ground lead, or on the ground included in the AC mains power cord, then the instrument will show ringing -- but it's an artifact of the hookup. IOW a good circuit viewed with poor oscilloscope technique will erroneously appear to suffer severe ringing throughout.

-- Jeff

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

Hi Jeff

Thanks for advising. You're quite right - I'm using a link from the ground further in to the circuit.
I'll try it near the ground 'input' pin from the battery and see how it goes.

I would still like to know about star grounds though

Usually the scope probe comes with a short ground lead. One end has an alligator clip and the other attaches on the business end of the probe. But these ground leads have a habit of getting lost or overlooked. That's OK for some work but not for what you're doing now. Think of that alligator clip and the probe tip as you would think of the probes on a multimeter. IOW, be conscious that "here" and "here" are the two points we're measuring between.

For your project, work toward achieving a ground plane rather than a star ground. That means the ground of any IC can be directly connected to the ground of any other IC -- connecting two grounds together can't be wrong. Not so with a star connection. In that case there's a star pattern to preserve -- useful, for example, when analog & digital circuits must share only a single, well-defined point of reference. But the pattern you want is a plane.

You can only approximate a plane, but an approximation will suffice. If your IC's are laid out in rows & columns (X Y), then you could run a ground lead from chip to chip horizontally for each row, and likewise another ground vertically chip to chip per column. (You could also do an X-Y grid for the 5 volt connections, but that shouldn't be necessary as long as every IC has its own bypass cap intimately connected.)


-- Jeff

ps- re ringing, I notice my own scope probes sometimes get a little funky where they connect to the instrument, and then I'll get a noisy display. Things clean up after I give that BNC connector a twist. It pays to be on guard. Even on a poorly-built circuit you should be able to find at least a few clean signals, as long as the scope's connected properly (probe tip grounded to the IC whose output you're observing). If you can't find any good signals then the instrument is suspect.

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

Hi jeff

You were correct. I'd attached the probe's ground connector (crocodile clip) to a ground pin off of one of the ICs. Now it's connected to the ground pin near the rocker on/off switch the "main" part of the probe reports far less ringing. I get about the same as I did when I breadboarded the circuit.

I've also found why the thing wasn't working - bad solder joint on the NAND chip dealing with the chip decoding. Just got to get the wretched VIA to work now.

Cheers for your advice

That's still a worrisome observation -- it says the connection between those two ground points needs improvement (perhaps with an X-Y grid as I suggested). Don't be shy -- with the crocodile on one ground point, try scoping the other ground points. That may seem a pointless exercise but it's NOT!

Good spotting re the NAND connection. Sounds like your attention will be shifting to the VIA now; good luck with that. Just remember that that, too, could be a ground issue.

-- Jeff

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

I'll certainly do as you suggest. To be honest my soldering is not too good on this proto type so I'm currently going through and redoing it all.
I've uploaded a picture of the rear of my project. The red wires/links are for the +5V rail and the black for the ground. The 5v regulator is in the top left.
As you can see, it uses the outer "tracks" for these rails.

The grounds of the main chipset ICs are all (except for one) soldered to the main ground rail on the edge of the board without long traces. However, the VCC/VDD pins do have to run via links.
Should I solder these together in the middle perhaps?

That soldering looks pretty good to me. But I have strong misgivings about the wire you're using. It's rather thick, which means things will get terribly crowded before you're done. Also you have issues with the insulation stripped back too far, with a lot of bare wire exposed -- and thus a risk of shorts. Both of these problems are reduced if you use wire-wrap wire (even though you're not wire wrapping). The photo here shows an example of point-to-point wiring using wire-wrap wire -- you see that connections can be very closely spaced (and use both sides of the board). Yes it's necessary to acquire some skill (and a fine-tip soldering iron if you don't already have one) but honestly I don't think the wire you're using will allow you to complete that project -- not if you're doing all the address & data lines with that same thick wire. Of course I've been wrong in the past.

Sounds good. Remember you're trying to approximate a plane.

cheers
Jeff

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

Hmmm I hadn't considered that the wire is too thick. The insulation being too far back is down to me heating then reheating the wire to reflow the connection - tends to melt the insulation. Originally, it was all nice and tight. I did make sure that with ones which had melted that the wire is held higher, away from the board where possible.

BTW here's the project which shows the data and address buses on the front side of the board: viewtopic.php?f=4&t=2660&start=45#p32720

Thanks very much for you advice and help.

Okay -- I'm certainly relieved to learn you weren't planning to add all that stuff on the bottom. I hadn't really been following the other thread. Now that you've provided the link I see you're doing alright. Keep us posted!

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

Do you use 30awg?

I probably need to re-work the part of the primer than mentions the star distribution of power and ground, and add some material that helps understand the matter of inductance (which is generally bad), mutual inductance (which is generally good), ground planes, and transmission lines, without heavy math that makes people run away fast.

The first thing I would point out with the power and ground going down the edges and across the end is this: Suppose you have an IC in the middle of the right side putting out a bit signal to one in the middle of the left side, on a board that 4x6". (In this case the two ICs may even be so close together that there's almost no room between them.) An example of the most troublesome signal is the clock signal. Its return current now has to go all the way around the edge of the board, essentially making a square coil that's 3x4" in size. That's a lot of inductance for this application. There is unavoidable capacitance too, and now you have a resonant circuit that will ring like a bell.

As the ground returns get closer and closer together, the size of that "coil" diminishes, and so does the inductance. When the return runs right next to signal line, two good things happen. The width of the coil goes to nearly zero; but also, the inductance of even a straight piece of wire now is mutual, meaning that the signal line and the return line have the opposite effect on the magnetic field surrounding them, so you don't build up a magnetic field, meaning you've lost your inductor, since the current in the two adjacent lines is traveling in opposite directions. Without the "coil," you also don't radiate to, or pick up noise from, neighboring lines. (For these reasons, you should keep the wires to your separate single-cycler clock board twisted too, at least loosely.)

This is the beauty of a ground plane too. Since mutual inductance cuts out the unwanted coil, that's where it's easiest for the high-frequency signal to travel, so the return current in the ground plane does not take the shortest possible path, but rather takes the shape of the trace directly over it, since that becomes the "path of least resistance." (Ideally, every other pin of an IC would be a ground pin so each signal's return is super close, but that's not going to happen, at least not in the ICs we're using. Differential signaling was made for this purpose though-- basically the same idea.)

You don't have to rebuild your circuit (again) to improve the behavior. Even if you didn't leave traces down the middle you could use for power and ground, you can bridge across from IC to IC in many places, to reduce the size and length of the return-current paths. This way, adding just a few wires goes a long way in improving AC performance.

_________________
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?

Thanks for the explanation, Garth. I've added some more lines as both yourself and Jeff have recommended and the ringing has reduced even further. I've attached some photos.

Cheers!