High-Speed Digital Design Online Newsletter: Vol. 2 Issue 12
I am teaching a class out of "High Speed Digital Design" to first year grad students. We recently discussed grounding issues and there was a lot of confusion about "digital" ground vs. "analog" ground vs. chassis/earth ground. Grounding in a cellular system, etc.
If there are clever ways to explain these concepts intuitively, I (and my students) would appreciate the help. How do you view these issues from a practical, everyday point of view?
Thanks for your interest in High-Speed Digital Design.
First, thanks for using my book as the basis of your course. Few university professors realize the importance of teaching high-speed signal concepts to digital engineers. I'm glad you're doing it (regardless of whose book you use).
About teaching the subject of grounding...
In my opinion the most important point to make with regard to grounding is that the input to every digital logic gate is a DIFFERENTIAL amplifier. That's right--a differential amplifier. This differential amplifier compares the digital input signal to some local reference (often generated inside the chip), and decides which is bigger (more positive).
It couldn't be any other way. It's not like we could make a chip that had a private wire leading out of the package to the center of the earth, to pick up some "ideal" ground voltage. That would be absurd. In the real world, every chip MUST evaluate its inputs only with respect to its own local reference. I don't move on to even thinking about fancy grounding structures until I'm sure my students get this point, and get it so well they will never forget it.
Another point to make is this: every logic signal on a chip's data sheet actually has TWO pins that can activate it: the input pin, and the ground pin (or, for ECL families, the more positive power rail).
From this vantage point, we can then go on to explain that if MY local ground reference is different from the one YOU used to generate your signals, we can have problems communicating. This is the basis of understanding the problem of signaling across different ground regimes. I show the example of ground bounce caused internal to a chip, and another example of the same sort of ground bounce caused by a connector.
Once we've got the ground bounce thing down pretty solid, I can show the example of an A/D converter with separate ground pins for the analog and digital sections. In many systems, these are for the purpose of preventing digital output switching currents from affecting the A/D inputs through the mechanism of ground bounce internally generated inside the chip package. External to the part, in many cases, these two ground systems should be tied together directly to the PCB board ground.
Another example I often relate is of a daughter-card plugged in to a large motherboard (in a 19" rack-mount configuration). When high-speed logic on the daughter card drives current into the motherboard, the returning signal currents, as they flow through the ground pins on the connector, create a ground shift between the motherboard ground and the local ground on the daughter-card. Hopefully we have bought a good enough connector to limit the magnitude of this effect.
Now let's assume that we must place a very sensitive analog component on that same daughter-card, and that analog component needs to receive signals from the outside world that are referenced to the chassis ground of the product (for example, a video feed). I will sometimes place the sensitive analog stuff on a ground "island" on the daughter-card and connect it to chassis ground with ground pins on the motherboard connector that are separate from the ground pins used to carry high-speed returning signal currents to the main digital ground.
In effect, I let the analog circuit "reach through" the connector to touch real chassis ground. As long as we can limit the currents flowing on these analog ground pins, the analog ground patch on the daughter-card will be at the same potential as the chassis ground on the motherboard. In other words, we have made an island of "quiet" analog ground.
Once this has been done, you can see that we had better not allow the flow of any high-frequency currents between the digital and analog sections of the daughter-card. If we do that, the return path for those currents will circulate out through the ground pins on the analog section to the motherboard chassis ground, and then back in through ground pins of the digital section, defeating our entire purpose. We were supposed to keep high-frequency current OFF of the analog ground pins, not force it ONTO those pins. The only correct way to connect the digital and analog sections, once their grounds have been split asunder, is to use isolating components (optical or magnetic) or extremely well balanced differential coupling. These methods avoid the circulation of high-frequency signals on the analog ground pins.
That's kind of a long answer, but it's the way I like to teach the subject in my classes.
Dr. Howard Johnson