LM3886 grounding - Am I doing it right?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
.The second is still yours, but I've moved the connecting point for the capacitors and other connections. No other changes.
Not, it's not anymore because the audio current loop and the charging current loop share a track again.

Right, with the connecting point moved to Mr. Self's "short spur" point you've simply redrawn my schematic.
Nah. I attach below exactly what Self means in his book (removing all unnecessary things for now).

edit: in the second attachment, I highlighted in red the piece of wire that is the problem.
 

Attachments

  • audio-amplifier-dual-power-supply-ground-001-c-shield.jpg
    audio-amplifier-dual-power-supply-ground-001-c-shield.jpg
    58 KB · Views: 228
  • aaa.jpg
    aaa.jpg
    53.8 KB · Views: 217
Last edited:
.
Now wanting to make sure I understand what you're saying, which at the moment I don't think I do.

Your second attachment above I yet again edit, just to make some room.

Is my arrow with the note, "This is a short jumper wire or pcb trace" in accordance with what you're saying?

(Or do we agree that this is, in fact, a short jumper wire or pcb trace?)
.
 

Attachments

  • aaa_1.jpg
    aaa_1.jpg
    74.4 KB · Views: 210
Last edited:
What I'm saying is...

the track you're pointing must find its origin at the caps terminals (it's the red track in my first pic). Currently, it is connected to the wire coming from the CT and going to the cap . See again my sims in the other thread to see why it is a bad idea.

edit : basic fact: the most "stable" points wrt voltage in your circuit are the terminals of the reservoir caps. The audio returns must go to that point. Allright that far ? Let's go back at the circuit just above. Every 1/100 of a sec, the point where all the wires come together (at the "bottom" of the short wire) is raised by the impedance of the red track multiplied by the charging current spikes. It isn't thus a stable point and a fortiori not the place to put the common ground.
 
Last edited:
.
Seems original poster tbj is long gone, and who can blame him. Still I'm finishing the current thought with an updated schematic.

Many thanks, 00940, for your help. I was looking right at the solution without seeing it, but no longer.
.
 

Attachments

  • audio-amplifier-dual-power-supply-ground-001-b-noshield-c.jpg
    audio-amplifier-dual-power-supply-ground-001-b-noshield-c.jpg
    117.6 KB · Views: 322
Thanks for the update bentsnake

Please pardon my ignorance with the following questions:

What do you mean by "decoupling caps" which you have going to 'A' ? I have seen decoupling caps in series with the signal, but not grounded. Do you mean additional storage caps near the load?

Ca - A - Cd "conductivity." My meter doesn't do siemens. :eek: For practical terms is this typically measured in ohms? Is it measured from the -Ca lead to +Cd lead?

Rod Elliot in his Project 72 notes (LM1875 chip amp) says:
"Note that the speaker must return to the central 'star' earth (ground) point. If connected to the amplifier's earth bus, you will get oscillation and/or poor distortion performance."

On your diagram his advice can be seen by the connection of speaker return (and audio commons) to the star, via jumper. Is that right?

Can you explain why the A-B jumper is needed?

Thanks a lot!
Rich
 
What do you mean by "decoupling caps" which you have going to 'A' ? I have seen decoupling caps in series with the signal, but not grounded. Do you mean additional storage caps near the load?

I do mean the additional storage caps near the load (the op amp itself), commonly specified as 100nF, by me as 0.1uF.

In my experience capacitors in series with the signal are usually called just blocking capacitors. Not to imply that "decoupling" is not a correct term.

Note also the bypass caps I call out in the power supply. Very unfortunately, the terms decoupling and bypass are used interchangeably by many, so nobody knows what anybody is talking about. I googled both terms, and used the first definitions that came up in Wikipedia.

Ca - A - Cd "conductivity." My meter doesn't do siemens. :eek: For practical terms is this typically measured in ohms? Is it measured from the -Ca lead to +Cd lead?

Don't measure anything, it just means use the same size (gauge) wire, or the same width trace on the printed circuit board, throughout. Which you'd probably do anyway, but don't use heavy gauge conductors throughout the power supply, and then go down to #24 for jumper A - B.

You can come into point B with #24, or whatever size wire or trace you're using for the audio circuit. Which with me is usually just the component leads.

Rod Elliot in his Project 72 notes (LM1875 chip amp) says:
"Note that the speaker must return to the central 'star' earth (ground) point. If connected to the amplifier's earth bus, you will get oscillation and/or poor distortion performance."

On your diagram his advice can be seen by the connection of speaker return (and audio commons) to the star, via jumper. Is that right?

I looked at the project ( Single Chip 25W Amplifier (Project 72) ), but I'm not certain of Mr. Elliot's meaning. As far as I do understand the answer to your question would be yes.

Can you explain why the A-B jumper is needed?

Ask me anything, but don't ask me that.

On its face the whole thing looks silly to me. What difference can a quarter- or half-inch length of conductor make?

However, more than one guru (someone published, or else widely recognized as knowledgeable) insists that such a jumper is needed. Supposedly they have the expertise, and hopefully the measuring equipment, needed to make such a claim, so I simply accept it as fact.

In such a case as guru Texas Instruments they certainly have the equipment. Whether they use it I don't know.

Thanks very much for asking. Clarity is what I'm working toward, so finding out that something is not 100% clear is valuable to me. For instance, I wasn't crazy about the term "conductivity" in the first place but it's all I could come up with. I hope something else will occur.

In a sidebar I guess you're thinking about building an LM1875 amp? Good move, they're very widely built and liked.
.
 
Last edited:
Ca - A - Cd "conductivity." My meter doesn't do siemens. :eek: For practical terms is this typically measured in ohms? Is it measured from the -Ca lead to +Cd lead?
This just means that A is roughly in between the - terminal of Cc and the + terminal of Cd.

Can you explain why the A-B jumper is needed?
"Jumper" is a bit misleading. It must be a thick piece of wire (or a large track or a plane on a pcb).

If all wires had zero resistances, it wouldn't be needed. The problem, illustrated in this post, is that you must avoid placing the audio "0" reference in the path of the charging currents going from the transformers to the caps Cc and Cd. In practice, not anywhere in the wire going from the transformer's center tap to the point A.

By creating this "tee" A-B (to use Self's term), you ensure that you're not doing that mistake. Another solution is to make the point A a very low impedance point (a plane on pcb, a small copper plate) and then everything can be connected to A.


As an aside, while 100nF caps are kind of a standard in decoupling opamps and the like (and I'm guilty of using them quite often), it's not always ideal. With modern low esr electrolytics in //, the combination can lead to resonant peaks. These days, it might be advantageous as a rule to go to 1uF or 2.2uF x7r caps; they're pretty cheap today. Decoupling is another wide topic....

As another aside, I'm not certain Ca and Cb are that useful to be true.
 
.
Thanks from me also, 00940, for your further comments. Just to mention it, along with "conductivity" I'm also not crazy about the term "jumper." But I haven't thought of anything better yet considering the space restraints. Also considering the intended audience, I'm obviously not doing this for engineers.

<< ...while 100nF caps are kind of a standard in decoupling opamps and the like (and I'm guilty of using them quite often), it's not always ideal...I'm not certain Ca and Cb are that useful... >>

Nothing is standard, nothing is ideal, nothing is always effective. This stuff drives me absolutely crazy. When is somebody going to come up with something better than electricity? Tesla had his suspicions, but nobody has picked up the torch yet (except maybe Bedini?).

Obviously one yields to the schematic if there is one, but sometimes information is lacking--or there's entirely too much information. In either case my purpose is that a guy with a brand-new soldering iron will have a successful build without doing 6 months of research. Well, we'll see.
.
 
.....................

Ask me anything, but don't ask me that.

On its face the whole thing looks silly to me. What difference can a quarter- or half-inch length of conductor make?................
Every Member that understands charging of the smoothing caps and the pulses that flow from the rectifier knows why this short stub is required.
It is written about by many Desingners and ignored by many Builders.

I posted many years ago about a subtle trick used by J.E.Sugden in a Thread on the P128 amplifier that surprised me and yet is so obvious after I had seen it.

They use a double sided PCB with the PSU and amplifier on a common board.
The charging traces are on one area and on one side of the PCB. Other area share both sides. At the Zero Volts junction between the smoothing caps is a bolt passing through the PCB. One side carries the "charging" pulses. The other side is the Main Audio Ground (MAG) and also has the lug that brings the Speaker Return back to the SOURCE of it's Flow Current.
The MAG side is connected to the the CHARGING side by the length of the bolt shank. It is only 1.6mm long between the traces, yet isolates the charge pulses from the Audio circuits.
Costs nothing and performs perfectly.
 
Well, I chose a layout. Sounds just fine to my ear.

XIRya8M.jpg


YhxHqZ7.jpg


It's actually based on another design I found, here: http://www.diyaudio.com/forums/chip-amps/126904-lm3886-noise-issue-2.html#post1570826

Which in turn appears to be based off Rod Elliot's design here: Single Chip 50W Stereo Amplifier

I went for this because it seemed to make sense grounding-wise - the output compensation and feedback/input grounds return through different paths, as does of course the load ground.

Thanks for all the help.

Just constructing a PSU for it now. Need to make the decision whether running off 24 volts will give me enough power - I briefly tested using a +/-35v power supply which I am using for another purpose and it sounded loud enough even turned down low, but I don't know if +/-24 will put enough oomph into my 8 ohm speakers :D
 
.
Looking good, very good. However...ummm...

<< I don't know if +/-24 will put enough oomph into my 8 ohm speakers >>

I guess it depends on what you call oomph. Look on page 12 of the data sheet, which is here: http://www.ti.com/lit/gpn/lm3886

Look at the graph labeled "Output Power vs Supply Voltage." 24v/30watts at full blast in an average living room will cause hearing damage. With 35v/60watts you can communicate with aliens.

However, bear in mind that doubling output watts does not double output sound levels. The progression is logarithmic, not linear.

The rule is that amplifiers don't produce sound, speakers produce sound. The question is how big an amp will drive my speakers to full (factory recommended) output.

But you're already past the hard part, which is making a decision. The rule there is that any decision will do, just make one. You did, and wishing you smooth sailing for the rest.
.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.