Odd grounding problem

[leftshoe]

I've run into an interesting problem with my LM3886 amplifier. I get quite serious amounts of hum whenever the volume control on the amplifier is at anything other than 0 or full volume, when plugged into my diy DAC. This doesn't occur when plugged directly into a iPod, and interestingly it occurs even when the DAC is not plugged in! If the signal cable is not connected to anything it's fine, but just plugging it into the DAC without any power cable going into the DAC triggers it. Plugging in the DAC and turning it on doesn't affect the level of the hum.

I've also noticed that when turning the amp off, the hum stops immediately, while the music keeps playing for a few seconds. This seems weird to me, I would expect both to continue while the caps deplete.

I've tried floating the amp's ground by just not connecting it to main ground, but it didn't help. I can upload some pictures of its internals if need be.

[random007]

Try attaching the DAC to your amp by just the signal wire alone (disconnect the signal ground wire on either the amp or the DAC). Avoid using the amp with no safety earth! hehe

[Omega_Void]

I'm a relative novice, but here's a few observations:

Originally Posted by leftshoe

I've also noticed that when turning the amp off, the hum stops immediately, while the music keeps playing for a few seconds. This seems weird to me, I would expect both to continue while the caps deplete.

That implies that the source of the hum is the transformer/rectifier in the LM3886 PSU.

Originally Posted by leftshoe

...interestingly it occurs even when the DAC is not plugged in!

Hypothesis: a loop or something in the DAC output is picking this hum up. This antenna effect, being passive, would still happen with the DAC off.

Does the DAC work humlessly with other amplifiers when the LM3886 PSU is switched on in the vicinity?

Does the hum reduce when the LM3886 PSU is moved further away from the DAC?

Originally Posted by leftshoe

...serious amounts of hum whenever the volume control on the amplifier is at anything other than 0 or full volume

When the volume is full off the 3886 input is shorted to input ground reference, so no hum.

Why the hum goes away when the volume is full up is less obvious to me. I'm gonna lurk around this thread 'til one of the smart guys explains it.

Good luck sorting it out.

[abraxalito]

Originally Posted by leftshoe

I've run into an interesting problem with my LM3886 amplifier. I get quite serious amounts of hum whenever the volume control on the amplifier is at anything other than 0 or full volume, when plugged into my diy DAC. This doesn't occur when plugged directly into a iPod, and interestingly it occurs even when the DAC is not plugged in! If the signal cable is not connected to anything it's fine, but just plugging it into the DAC without any power cable going into the DAC triggers it. Plugging in the DAC and turning it on doesn't affect the level of the hum.

Presumably this is a stereo set up, so does the hum disappear or change noticeably when only one input is connected to the amp? It could be your DAC, by joining together two signal grounds, is causing some kind of loop. Another thing - what's the DAC connected to via its digital in? That could be the culprit - not all DACs have isolated digital inputs and many CD/DVD players don't either.

I've also noticed that when turning the amp off, the hum stops immediately, while the music keeps playing for a few seconds. This seems weird to me, I would expect both to continue while the caps deplete.

Its quite normal The source of the hum is going to be either the transformer's magnetic field (solve it by putting some more physical space between the amp and the trafo) or its the result of the high current charging pulses running into the main res caps cross-coupling into some sensitive place.

I've tried floating the amp's ground by just not connecting it to main ground, but it didn't help. I can upload some pictures of its internals if need be.

Certainly would be helpful to see a wiring diagram (with grounding clearly shown) and a schematic too.

[leftshoe]

I did some more experiments:

- Disconnecting signal ground resulted in no music.
- Hum occurs when only 1 channel is plugged in, only comming out of that channel.
- Not connecting a digital source doesn't help, niether does using a computer as source.
- Also, the hum is significantly louder than the music except at near full volume. It ramps up linearly at low volumes then stays fairly constant volume until near 100% volume.
- If I move the DAC about 6 feet away, the hum is the same as before with the amp off or on and the power cord plugged into the DAC, but when I unplug the power cord, the hum goes away. This sounds more like a normal grounding problem.

The DAC is a kit, the one in the recent Silicon Chip, metal cased, with a star grounding scheme. I think the DAC is properly grounded, so it must be something I did wrong with the amp. Any pointers on how the amp should be grounded? Its plastic cased.

[leftshoe]

I've attached photo. The photo doesn't show the volume control, which was a latter addition.

[gootee]

Look up "Faraday's Law".

Your main problem is that one or both pairs of your internal input signal and input signal ground wiring is forming open "loop area".

A time-varying (e.g. AC) magnetic or electromagnetic field induces a time-varying current in any conductive loop.

When an input is connected, the loop is closed, enabling induced AC current to induce voltages across any impedances in the loop, which would include the source impedance, and any resistance to ground at the amp input pin (as well as the resistance and inductance of the wires themselves).

The solution (mitigation, actually) is simple: Your input jacks should be isolated from chassis ground and each of your internal input signal and signal ground wire pairs should be inside of a shielded cable, the shield of which should be connected to chassis ground at the input end only, with the other end of the shield connected to nothing. If you can't use shielded cable, then tightly twist together the wires in each pair, for as MUCH of their length as possible. Only the shortest bare-minimum lengths of the wires should be unshielded (or un-twisted). And keep the wires as close to each other as possible, for those short unshielded or untwisted paths, even following against the amp input resistor to its ground end, with the ground wire, from where the signal wire connects to the resistor's other end.

Also make sure that you tightly twist together all OTHER conductor pairs, such as the AC input pair, transformer secondary pair(s), AC wire pairs to/from switch, rectifier output pair(s), DC power pairs, audio output pairs, etc, noting that Faraday's Law also works in reverse, i.e. a time-varying current in a loop will cause a time-varying field to be radiated. So tightly twist together ALL pairs, in order to minimize the effects for both "transmitters" and "receivers".

And ALSO, keep all small-signal and DC wiring as far away as possible from all AC and large-signal wiring and components. If such wires must approach each other, they should be perpendicular to each other.

Edit: Your amp is PLASTIC-cased?? Bummer. Then you might also have to worry about the interconnect cables from the DAC. But if you twist all of your internal AC wire pairs, you might not have to worry much about the interconnects.


Cheers,

Tom Gootee

[leftshoe]

Thanks for the input. Would RG-59 wiring be suitable? I can easily replace the signal wiring. Also, should I place a sheet of metal between the transformer and the main board? Some sort of shielding has to help.

[gootee]

You would not want to use coaxial cable that has one center conductor. You want a cable with two insulated conductors inside, plus an outer shield. If the shield is foil then there should also be an uninsulated "drain wire" for connecting the shield. (Tightly twisting should be "almost as good" as shielded, especially if you also twist all of the AC and other pairs. It's a bit of a pain but is something good to try if the shielded cable isn't immediately available.)

I would also twist all of the AC wire pairs before worrying about a metal shield. It might still be necessary but it's more difficult than you might imagine to shield magnetic fields. Better to first try to eliminate the fields radiating from the wiring, then see what's left. Toroids usually radiate mostly axially, anyway.

(Sorry, have to go to sleep now.)

[AndrewT]

I agree with Gootee.
Minimise the loop area of every wire pair/triplet.

The +ve and -ve and Zero Volts wires from PSU to Amplifier must be twisted as a triplet for the return current field to cancel the flow current field.