Right, but how does one determine what any of that good design is, if discussion of it is using non-standard meanings of common words, without clearly defining their meaning for a specific scenario, additional ambiguous terminology, like, "relatively large power supply voltages and currents," (which, for many cases, including most cases for power amps, are going through the audio circuit, after the first fraction of a second from power-on), and then ambiguous diagrams (fig 3-5 could be describing a star, ground bus, or just screwing into the chassis wherever)?
Then if you look at #38 and #41, in one the resistance of the chassis is ignored (all electrically the same point), but then in the next, it is considered, for the same scenario, as if in #38 a loop is meaningless, but then by #41, you want to watch out for those loops. Yet, the diagram in #38, if you flesh it out, makes a loop, though it is claimed to be very good practice for grounding in #4. This sort of thing only continues the confusion around grounding.
One way is to simulate the circuit/layout, add as many parasitics as you can, add the ground loops.
That only leaves you with the problem of having to explain it to others in a way that is both correct and understood by all.
That only leaves you with the problem of having to explain it to others in a way that is both correct and understood by all.
The conductors, as seen by us coarse humans, must get within millimeters, sometimes a fraction of a millimeter, of each other. Must, not might or maybe. You have to bow to physics at every step, and at point can the circuit "grounds" not be connected. Connecting them all back to neutral just adds to that. It's how convenient the path is from one point to another is, as far as return v. signal paths (power may as well be considered another signal), and how likely unwanted induced currents may be, for how to place them if in parallel. Every power output rail connects to the same reference common as the signal, unless balanced from end to end (even then, that often happens, just with the shield not pulling double duty), and the currents have to flow through it.
I'm saying I've heard and read that for years, but have yet to see or hear it explained in a way that makes any sense, because all the ways to say it, with all the synonyms, do not make sense in regards to a fleshed out physical diagram, circuit diagram (no, "and a miracle happens here," white space, except for the insides of ICs), nor any basic electrical theory application I know of. But that, if it exists, I'm open to knowing it, as I lack formal schooling on the matter. If there is an accepted set of definitions, they should be easy to find, and/or clearly stated. Because of that, and other over-simplifying, the confusion gets continued by advice of the matter. By trying to be so basic, different rules come to contradict one another, or end up missing the goal entirely.
Only if far over spec (I'm not saying I'm not lazy enough to do just that, but it's not a requirement). With a little management of the reservoir capacitor charging, there's no reason the transformer cannot be merely a bit over that of the powered circuit's needs. That is commonly the situation with commercial products, since transformers are pretty expensive in volume.
The layout of any precision circuit can make or break the circuit. I saw an order of magnitude (10x) difference in THD between a poor layout and a decent one. The data is in this post: http://www.diyaudio.com/forums/chip-amps/252436-lm3886-pcb-vs-point-point-data-3.html#post3846783
I suggest reading the thread in its entirety. I actually beat the data sheet performance of the LM3886 by a bit.
You'll get the best performance with a low-impedance ground. This means a ground plane. If you have quiet circuitry, I suggest creating a quiet ground plane and referencing it to the power ground plane using a low-inductance connection (think really wide trace or a copper pour) at the output connector's ground connection.
~Tom
In the past, in most US cities, homes Grounding Electrode Conductor (GEC) were only connected to all metal water pipe systems. That made a much better connection to both the power company's Neutral and to the soil. But now with plastic water pipes the common connection to the soil is two ground rods (as this is less expensive than measuring one ground rod). The 2 ground rods are a poor connection to the power company Neutral and only a fair connection to the soil.
But the quality of the connection to the soil has nothing to do with the quality of the AC power for our audio systems.
But the quality of the connection to the soil has nothing to do with the quality of the AC power for our audio systems.
This is an extensive subject area, not just a few rules of thumb. Read several of Ralph Morrison's books to get some background.
Otherwise, it's best to try to duplicate an existing, successful effort if you just want a working amplifier.
Last edited:
Yes, the one rule to follow in grounding is:
"Don't follow rules!"
Instead, understand how to minimise inductive coupling (reduce loop areas, and increase distance) and capacitive coupling (reduce surface area, and increase distance) and how to avoid injecting hum/buzz voltages into signal circuits (hint: every conductor, however short and thick, has non-negligible resistance and inductance) by thinking about exactly where all the currents flow (hint: currents always flow in loops).
"Don't follow rules!"
Instead, understand how to minimise inductive coupling (reduce loop areas, and increase distance) and capacitive coupling (reduce surface area, and increase distance) and how to avoid injecting hum/buzz voltages into signal circuits (hint: every conductor, however short and thick, has non-negligible resistance and inductance) by thinking about exactly where all the currents flow (hint: currents always flow in loops).
Especially not rules of thumb siphoned off the internet... 🙂
Yep. Series inductance matters too as any AC current through it will create a frequency-dependent error voltage across the conductor, leading to higher THD at higher frequencies.
Minimize the impedance.
~Tom
As a specific, and this time non-rhetorical, question: without a special-purpose tool for audio range measurements, what is a good way to check this out? For example, is there a good way to tell relative distortion with an analog scope (and if so, what specifically would I be looking for)? Or, is a reasonably good single-ended sound card good enough, with software like RMAA or ARTA?
Yes, I have 3 of Ralph Morrison's one dozen books. But we should note:
a] The books are written in field theory rather than circuit theory. The can be challenging reading for those who are not engineers or scientists.
b] Large sections of the books are devoted to US NEC rules from 20 to 30 years ago.
c] These books can be purchased used for $10 to $15 dollars.
Last edited:
I think you'll find this one very useful and up to date.
Grounding and Shielding - Circuits and Interference
Fifth edition, 2007 Ralph Morrison
I very much like this one. I underlined what is the priority.
It rules out the common, blindly followed, grounding "rules".
- Status
- Not open for further replies.