Draw them in yourself, then you'll be sure your meaning is as you intend. Your input will be appreciated not only by me, but by many I'm sure.
.
I thought you were trying to explain it to us?
I can put a simple diagram together soon enough.
jn
Bent,
take as an example the two 9V batteries and the 4 capacitors labeled Ca to Cd.
Looking at the leftmost battery, trace out the current loop from +ve, to Red, to the wire to top of Cb, through Cb, to bottom of Cb along the wire to Black to -ve, through the battery.
That is a loop1.
There is a similar loop around the second battery into and through Ca. That is loop2.
Then from Cb to Cd there is loop3.
From Ca to Cc there is loop4.
From left most +ve across red, Black, through right most, through Red, through Ca then through Cb and finally back to battery via left Black. That is loop5.
And finally from top of Ca to top of Cb through Cb, through Cd, across to bottom of Cb, through Cb then through Ca to end at top of Ca. That is loop6.
Those two batteries and four capacitors have six loops.
The wiring interconnecting them should be arranged to minimise the loop areas of all six loops.
The wiring at the terminals of all six components should be arranged to minimise the loop areas.
This is most easily done by using twisted triplets for the battery to first caps. then another twisted triplet from first caps to second caps.
That makes a big difference to radiated fields. It is worth doing and more so when all the components share a small volume with other sensitive circuits.
That type of LOOP AREA analysis should be applied to EVERY wire pair connection between modules and between components.
As a final hint:
The output from the second pair of smoothing caps should be a twisted triplet to the load circuit. Those are the separated thick black lines labeled "Power Supply Circuit" AND the thinner grey line that SHOULD go to the client circuit where it's Power Ground will be located.
take as an example the two 9V batteries and the 4 capacitors labeled Ca to Cd.
Looking at the leftmost battery, trace out the current loop from +ve, to Red, to the wire to top of Cb, through Cb, to bottom of Cb along the wire to Black to -ve, through the battery.
That is a loop1.
There is a similar loop around the second battery into and through Ca. That is loop2.
Then from Cb to Cd there is loop3.
From Ca to Cc there is loop4.
From left most +ve across red, Black, through right most, through Red, through Ca then through Cb and finally back to battery via left Black. That is loop5.
And finally from top of Ca to top of Cb through Cb, through Cd, across to bottom of Cb, through Cb then through Ca to end at top of Ca. That is loop6.
Those two batteries and four capacitors have six loops.
The wiring interconnecting them should be arranged to minimise the loop areas of all six loops.
The wiring at the terminals of all six components should be arranged to minimise the loop areas.
This is most easily done by using twisted triplets for the battery to first caps. then another twisted triplet from first caps to second caps.
That makes a big difference to radiated fields. It is worth doing and more so when all the components share a small volume with other sensitive circuits.
That type of LOOP AREA analysis should be applied to EVERY wire pair connection between modules and between components.
As a final hint:
The output from the second pair of smoothing caps should be a twisted triplet to the load circuit. Those are the separated thick black lines labeled "Power Supply Circuit" AND the thinner grey line that SHOULD go to the client circuit where it's Power Ground will be located.
Last edited:
Also remember that opamps typically consist of a differential input stage, an integrator (Which we call the open loop gain) and a power stage, that integrator capacitor connects to one or the other of the opamp supply pins (Which one depends on the particular chip), this pin in particular is a signal node right alongside the obvious ones, it also needs consideration.
The art to this is to consider current loops and where the common impedances are, forget the gross simplifications 'star earthing' and think loops and common impedances.
Regards, Dan.
The art to this is to consider current loops and where the common impedances are, forget the gross simplifications 'star earthing' and think loops and common impedances.
Regards, Dan.
So say some. While others, especially the point-to-point crowd, have less flattering opinions. Some even say that such things belong to the high speed digital world, not necessary to audio at all.
But that's the nature of these forums, is it not? 50 different posters have 50 different opinions, all of them pronounced as unassailable scientific fact, none of them with anything but "hey, listen to me" to back them up. Not greatly unlike any Arabian bazaar.
Of your charity, spend a tear for this unfortunate. Unknowing of the cost to myself I took on the job of separating the gold from the ain't-I-smart. Hercules and his stables, bentsnake and his posts, to which falls the greater task?
I continue to solicit input.
.
Last edited:
@jneutron
Is it most important that you have a twisted triplet for the rails and the bypass cap's ground because they are the highest current... and then just close couple the other grounds zip tied to them?
Because the triplet wires are going to radiate out the most of any other wires and the tighter you twist them the more it cancels that out? Am I wrong in thinking this?
Is it most important that you have a twisted triplet for the rails and the bypass cap's ground because they are the highest current... and then just close couple the other grounds zip tied to them?
Because the triplet wires are going to radiate out the most of any other wires and the tighter you twist them the more it cancels that out? Am I wrong in thinking this?
I include the short excerpt only to say that I'm pretty sure I've never mentioned ground loops at all.
Which actually leads into my main point, which is that in a rather long post #136 you make no reference at all to the posted schematic. You know, the one in post #119, the one that's supposed to be the point in the first place.
Not that you're the only one to do this. Cohorts come sweeping down in a thunder of hoofbeats, the din of their war drums, the snap of their battle flags. Closer and closer still, now bare steel flashes in the uncaring sunlight, now are the screams of men in mortal combat, and now at the high moment, the instant to thrust home with a great cry of, "This is how all this relates to your circuit..."
Nothing. No hint of what it's all for. Whatta letdown.
Well, if people want to wax poetic about this or that principle--albeit sometimes a shaky one--then wax away, sez I. But while all of that goes on I continue to solicit input regarding...pssst...wait for it...the posted circuit. The one away back there in the aforesaid post #119. Anybody got any thoughts, love to hear 'em.
.
Yep, just did this with some ideas from Cordell's Designing Audio Power Amplifiers.
Mr. Cordell is very nice and has a thread about the book. He'll probably say 'fine, no prob'.
I love this whole internet community thing ...
Yep, just did this with some ideas from Cordell's Designing Audio Power Amplifiers.
Mr. Cordell is very nice and has a thread about the book. He'll probably say 'fine, no prob'.
I love this whole internet community thing ...
Yep, Mr. Cordell said no prob just reference the full name of the book.
.
.
.
.
.
.
.
.
.
.
.
Oh and bentsnake, now I have an 'opening' with Mr. Cordell and may get some advice and help with the circuit from the man himself.
You could try this with Mr. Self, you might be surprised ...
Last edited:
Yes, twisted triplet is important when you worry about the output current into the load. If you only twist pos and neg, then positive current into the load will not go back to the supply via a twisted pair. Same with negative current into the load. If you look at bentsnake's schematic in 119, he twists the supply rails, but the load return is somewhere else. No magnetic cancellation by common centroid.
Bringing the three supply current wires together and twisting or braiding reduces the radiated field by common centroids (on average). Triaxial supply runs actually cancel radiated field entirely.
EDIT: I added a picture of a triax I made back in 2006 for exactly this. I took a bog standard 14awg orange insulated wire, pulled some mike cable braid over it, blue heatshrink over that, another braid layer, then a yellow heatshrink over that. I used the outer braid as ground, inner braid as negative rail, and core wire as positive rail. Everything from parts express, they had a good deal on 1000 feet (maybe 500, been so long) of mike cable, so I picked it up. Plenty left..
Doesn't matter. Your schematic in 119 is horrible, everybody is explaining why, you ignore.
I left the trashing of your drawing on the cutting room floor (which is pretty much how I'd explain it.) I didn't see the need to point out how badly it considers EMC concerns within the chassis, violating simple understandings.
When you learn the concepts, you will be able to identify the errors in your schematic. Merely repeating what another says isn't getting you anywhere.
Waxing poetically using warcraft scenario's is not intellectual at all from an engineering perspective. I recommend you lose that attitude, and ask salient questions.
edit: btw, you've no idea what a long post is. Ask around..
It's garbage, drawn by somebody who has no clue as to electromagnetic theory, electromagnetic compatibility, faraday's law of induction, ampere's law...the list goes on.
Is that how you'd like it? I've been very gentle in trying to get you to step back and learn something, you've not been nice in return.
1. You've twisted the rails only, this does not cancel rail to ground currents.
2. You've shown load ground starred to signal ground. It is not possible to create orthogonality between multiple grounds connecting to the same point, so load currents will couple to the ground reference used in the rest of the circuit.
3. BY DESIGN, you are also creating external chassis to chassis ground loops. Horrid design techniques.
I've watched others spell out reality to you, but you still argue.
Don't.
Learn instead.
jn
Attachments
Last edited:
I think you are exaggerating the case here. Both rails are at AC ground. The impedance between either rail and ground is tiny. 3 wires are awkward to wind regularly. Not that I intend to add weight to bentsnake's case.
Both rails are tied to ac ground via caps. Time varying currents will cause external magnetic fields that are time varying as well. That allows coupling.
The DC resistance is tiny. At frequency, it is no longer tiny.
Go to the diagram in post 119, and explain why that will not oscillate at high frequency. That is how badly the inter-chassis or board layout is.
Not really. Nor is it difficult to braid. Nor is it difficult to make a triax.
But that's in the hands of the beholder, some are better at it than others.
edit: 4, 5, and 6 wires, now that's difficult..7 wires is easy.
jn
Last edited:
It would be far more instructive if you'd explain why it will. It would also be to the point of this thread.
Equally to the point, the layout is not necessarily inter- chassis or board. It could as well be a single board, as is stated on the schematic. In the case of a single board, is it still your view that the circuit will oscillate?
.
Last edited:
You only need a few tens or hundreds of mV induced in a ground loop and a low loop impedance to get 10s or 100's of mA of current to flow . . . The result is hum and noise. So, reducing the loop area by creating twisted triplets is a very important technique. Ditto the prevention of loops between equipment which gives the same bad result.
Dude, you named this thread "comment on grounding scheme" and that's exactly what he was doing.
What... in the hell... is with this weird a$$ dungeons and dragons talk?! Stop insulting the smart people!
jneutron's post #144... B, he's spelling it out for you. What more do you want?? No one is going to learn grounding fundamentals for you. Step up and cut your teeth on it boy. That means: put it together, test it, take pictures of it and rip it apart and start over when you are told what is wrong with it.
Either way, adjust your attitude to one of humility instead of proud blatant ignorance and follow this project to it's end in the right way and you'll be able to help some others in this forum following along learn with you!
Returns, Ground, and Chassis Bond - Update
.
This has been covered before, but it seems an update is needed. Not an information update, just bringing the same information forward to this location in the thread.
===
Texas International data sheet for the LM1875 (and others), section headed "Stability."
"...it is important to return the load ground, the output compensation ground, and the low level (feedback and input) grounds to the
circuit board ground point through separate paths...."
The full section goes into more detail and explanation, it's freely available here: http://www.ti.com/lit/gpn/lm1875
===
Douglas Self, "Audio Power Amplifier Design Handbook," section titled "Power amplifier PCB layout details," page 400.
Mr. Self goes into some detail on how to run returns to both circuit ground and star ground (my audio common). More detail than can be lawfully quoted, I think, so I can only hope that readers of this thread will refer to his book.
Regarding what I call the chassis bond he says:
"Mains/chassis ground will need to be connected to the power amplifier at some point. Do not do this at the transformer centre-tap as this is
spaced away from the input ground voltage by the return charging pulses, and will create severe groundloop hum when the input ground is connected to mains ground through another piece of equipment.
"Connecting mains ground to starpoint is better, as the charging pulses are excluded, but the track resistance between input ground and star will carry any ground-loop currents and induce a buzz.
"Connecting mains ground to the input ground gives maximal immunity against groundloops."
===
If somebody wants to take on the combined forces of Mr. Self, and the engineering staff of Texas International, then I, for one, am eager to witness the match. Taking all bets.
Or, as before, if I've misunderstood or misinterpreted something, then I'd like very much to hear about it. Necessarily in the context of the posted circuit, since if you don't refer to that then how do you expect your comments to be understood? Unless, of course, you post a schematic of your own.
.
.
This has been covered before, but it seems an update is needed. Not an information update, just bringing the same information forward to this location in the thread.
===
Texas International data sheet for the LM1875 (and others), section headed "Stability."
"...it is important to return the load ground, the output compensation ground, and the low level (feedback and input) grounds to the
circuit board ground point through separate paths...."
The full section goes into more detail and explanation, it's freely available here: http://www.ti.com/lit/gpn/lm1875
===
Douglas Self, "Audio Power Amplifier Design Handbook," section titled "Power amplifier PCB layout details," page 400.
Mr. Self goes into some detail on how to run returns to both circuit ground and star ground (my audio common). More detail than can be lawfully quoted, I think, so I can only hope that readers of this thread will refer to his book.
Regarding what I call the chassis bond he says:
"Mains/chassis ground will need to be connected to the power amplifier at some point. Do not do this at the transformer centre-tap as this is
spaced away from the input ground voltage by the return charging pulses, and will create severe groundloop hum when the input ground is connected to mains ground through another piece of equipment.
"Connecting mains ground to starpoint is better, as the charging pulses are excluded, but the track resistance between input ground and star will carry any ground-loop currents and induce a buzz.
"Connecting mains ground to the input ground gives maximal immunity against groundloops."
===
If somebody wants to take on the combined forces of Mr. Self, and the engineering staff of Texas International, then I, for one, am eager to witness the match. Taking all bets.
Or, as before, if I've misunderstood or misinterpreted something, then I'd like very much to hear about it. Necessarily in the context of the posted circuit, since if you don't refer to that then how do you expect your comments to be understood? Unless, of course, you post a schematic of your own.
.
- Status
- Not open for further replies.