I'd add my vote to this video, I had seen it before and for me it helped solidify my notion that the right model here is the setup time of the magnetic field as the wire (whether open or closed) is charged to the applied potential, with a very concrete experimental basis.

Hmm, if you said electric field I'd be with you. There's no current in one branch - what's stabilising is the voltage.

I definitely agree electric field would be clearer because the electric field survives the transient state.

But I believe that when the voltage is first applied, an electric and magnetic field runs down the length of the open wire as the voltage charges the skin of the conductor. There ought to be a transient current flow during this setup time, and then the current stops; therefore there is a small transient magnetic field. The electromagnetic field setup time is probably on the order of 0.6c * length of wire. In steady state I'd expect the surface of the conductor to carry a positive or negative charge along its entire length, and the transient current will have stopped; an electric field will remain.

The small current bouncing off of the open circuit should be possible to time and it is presumably on this principle that the equipment that measures the length of open cables from one end works.

Please, if what I have written above is incorrect, correct me. This material is new to me so it's more than possible I am describing it incorrectly.

It may be because of my history - I came into EE for the purposes of designing chips, and digital chips at that. So, for me electrostatics, and electric fields, do almost everything I need of my mental model. On-chip, at least, crosstalk effects were always adequately modelled by thinking of capacitance and voltages. It was rare indeed that I needed to think about currents, and rarer still that I needed to think about inductance.

And, naturally enough, I find my mental model to be good enough for me!

(I continue to think that you are digging a bit too deep for your own good - if you were doing mechanical engineering, or civil engineering, and you expressed concern about, say, the gravitational influence of Mercury or Mars, I'd be similarly concerned. Physics will tell us that everything influences everything else, but engineering is all about taking suitable approximations to reality.)

(In our hobby digital electronics world, I'd say signal is the voltage we care about, and noise is any deviation from the ideal. For most purposes, most of the time, it doesn't matter where that noise came from. What matters is to keep it low enough, at the critical times, to keep the digital signal above or below the threshold of each input. Usually, we do that by careful wiring, adequate decoupling, and an adequate power supply. We don't do it by solving 3 dimensional field equations.)

These points resonate with me! The reason I am down this rabbit hole at all is because we said (much higher up in the thread) that my breadboard computer was unlikely to work for reasons I don't understand, and I am probably floundering a bit trying to figure out which of the many words I don't understand actually will affect my circuit in a meaningful way. I neither want nor intend to learn about a ton of material that isn't in my way, but my approach appears flawed to everyone: I think the consensus of the board is that I should just "follow good engineering practice", without understanding why it is necessary, rather than build on breadboards and expect to be able to figure out and understand the details at an ever deeper level in order to solve specific problems.

As a side point, since I have to leave the project aside until September and do very little on it, I have rather more time than is useful to read books and watch YouTube and rather less time than I'd like to actually work on the circuit. So that's probably not helping either

I certainly don't expect to need to solve things in terms of electric fields in order to build my circuit. But I expect to be able to understand when and why I will need to switch to a PCB at some point during this project and to be able to point at specific measurements that show when my breadboard failed, it failed for a specific reason that moving to a PCB or to wire wrap will solve so that I can continue building. This expectation may be unrealistic.

_________________
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 would advise the usual caution about information you find on the internet: it's not all equally valuable. That applies here as much as anywhere. And I'd certainly advise against thinking in terms of a consensus - there isn't one. A few people post very often, and often post very early in a thread. There are many other people, who post less often and may post later in a thread, who might - or might not - have better advice. Don't judge anything just by the number of times it's repeated. There are people who haven't posted who have even better advice, I'm sure.

Of course, you have to find your own balance, between following advice you get, and doing experiments. Between reading here, and reading elsewhere.

I think, though, that if you can build a simple working system, you can learn a lot from that starting point.

Let's see if I've managed to learn anything by trying to understand ground bounce / Vcc droop on a trivial clock circuit. The oscillator I have is a MXO45HS (datasheet attached). The test circuit shown in the datasheet is:
As ever, my understanding is weak but I am interpreting the mA and VM symbols to be "meters" - that is, not really in the circuit, but labels for the operating current and operating voltage. So the circuit I built started with just the oscillator, and not even the 0.01μF capacitor that is called for on the input side, and I didn't verify that my oscilloscope probes were under the capacitance required for the output but just assumed that was the case (if I am reading the datasheet correctly, for my 1Mhz oscillator the probe capacitance has to be below 50 picofarads). With the circuit in this state and channel 1 probing the voltage from the ground pin of the oscillator to its Vcc pin, the scope shows this trace:With the 0.01μF capacitor across the power rails of the breadboard, the scope shows:With an additional 1000μF capacitor from the ground rail of the breadboard to the power pin of the oscillator, the scope shows:If I move the ground probe of the scope to be across the legs of the 1000μF capacitor, we get:So, what does all this add up to in my updated understanding?

Initially, I was expecting the capacitors to make a difference to the voltage between the ground pin and the output pin of the oscillator. But they don't. The final image shows that the 1000μF capacitor is working, so the issue must be the reactance in the wiring and the breadboard between the ground rail of the breadboard and the ground pin of the oscillator - that is, even though the capacitors are trying to keep the ground at level, the reactance in the wiring is actually measurable on my scope, so that when the leading (edited) edge of the clock arrives the "inertia" of electrons arriving at the ground pin means they aren't coming in fast enough, and so the voltage between the ground pin and Vcc falls. Presumably the closer I can get the bypass capacitor to the ground pin of the oscillator, the more I can reduce the effect.

Does that explanation sound correct, or am I misunderstanding or misusing my oscilloscope?

Even if I am understanding correctly, the amount of bounce I am measuring across the device doesn't look significant. But, I could put the scope across other chips in my real computer circuit looking for traces where I am playing with the tolerances on the datasheets, and presumably capture a trace of ground bounce if it was in fact significant enough to flip a bit. Such a trace should be possible to capture when running the circuit at 4Mhz, which is the speed at which the computer currently fails ...

As a bonus distraction, I tried putting the clock signal through a 74HC04 inverter gate, and the result was:which seemed to me to suggest that the HC chip both took out the overshoot on the clock signal and very slightly softened the edge. Whether that's useful or not I don't know.

Off topic is where I shine

However that leads to a question: should I be making new threads for each topic in the hope that the threads will someday be findable by someone searching for specific discussions, or is it fine to just have one giant thread here? This particular thread started as a reasonably focused question on RDY / BE pins for the purpose of building DMA for video, but has long ago ceased to be about that topic; I personally don't care at all, but I don't know what the general consensus is on making things findable (e.g. the PDFs of the two-generation-old work by Kuphaldt).

_________________
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 guess I don't understand then, because if there's a capacitor already across the power pins of the oscillator I wouldn't expect to be able to see the voltage oscillate on every clock edge - what are the internals actually like?

_________________
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?

(For me, if a topic has drifted from the original subject, it often makes sense for the original poster to edit the subject of the head post accordingly. Splitting a topic is a good idea, but I've very rarely seen it happen here (another way of saying this is that Garth very rarely splits threads.) The best practice is for the person who has the new idea to start a new thread with a new appropriate subject. It's very easy to post a link back to the original, and even to post a link forwards to the new thread. Once a diversion has had one or two replies, it's really hard to recover.)

Thanks for the picture. Assuming they're all much the same, the internal bypass capacitor should make the effect I showed in the above screenshots almost invisible/impossible. So, perhaps unsurprisingly, it turns out my measurement was flawed.

Specifically, I had a short length of wire coming out of my breadboard to clip the ground leads of my probes to. Even that short piece of wire had enough reactance to show up on the scope. Doing it again with more care, probing closer to the breadboard, shows a near perfect situation around the oscillator:And then, just to show how bad it can be when one isn't careful, I put a very long wire between the ground pin and the ground clips of my probes to get this "measurement":So indeed, as Garth said, the oscillator can is already designed with the bypass capacitor in place, and measuring these effects requires some precision as to how one probes the circuit.

So then I decided to try running the clock through the output enable of a 74HCT541N 8-input buffer, with one of the outputs hooked up to a resistor to sink the current to ground. This seems like a bad idea for reasons I don't yet understand, as the output signal is less than ideal looking:but as I don't really care about that part of the problem at the moment I just ignored it. With one or two outputs hooked up the whole circuit draws about 100mA, but with all 8 lines hooked up it draws about 300mA, and the effect across the power pins is more pronounced:Even this though, looks like not really a problem.

For now, as near as I can tell, I can ignore ground bounce until it actually negatively affects things I want to accomplish with my breadboard computer, and try to measure it when I start to see failures/flakiness. When I can get back to the project for real, I think I'll focus on building my next "working" module rather than testing for stability, as the circuit seems stable enough to be working on at the moment.