lm386 Headbanger shielding?
 

experts please help!

i've got half of the Headbanger, one chennel, mono, working pretty well on breadboard, and have a shielding question now, when planning about putting it all in a little box.

http://www.minidisc.org/schem.gif

the author said he added some shielding to the bottom of the case, in order to avoid lawn mower engine noise. the shielding, case, and case of the pot were all connected to the INPUT ground, not output ground, nor system ground.

my question is, why does which ground matter?

scenario 1: connecting to input ground.
sure if noise is picked up, it would be the input net, but this seems irrelevant, because we are trying to stop the noise from reaching the input net. wouldn't connecting the shielding to the input ground introduce noise to the input ground, and therefore noise to the input signal?

scenario 2: connecting to system ground.
noise finds its way to system ground, the input net wouldn't see any noise, no ripples introduced to the input ground, mission accomplished, yeah? what did i get wrong here?

what's the advantage of connecting the shielding to the input ground and not any other grounds?

all helps/explanations/advices/comments appreciated!

If you grounded it somewhere else down the line, you would still want to connect the input to shield it. This may or may not produce a ground loop.


Generally when you shield an input cable, you want to ground the "upstream" end of the shield, same with shielded bi-polar capacitors, these are marked with a line, not to be confused with the negative symbol "-". You also want this isolated from the amplifiers ground.

The short answer - I don't know.

You might be able to get more info from NwAvGuy's O2 blog; the O2 case is also grounded at the input.
My understanding is that this grounds the noise at the input therefore it is attenuated (Or dissipated. I'm not sure which is more correct.) there and doesn't make it to the gain stages. It doesn't create a common-mode situation on the + signal input.
Entire books are written on grounding, RFI, and EMI. It's a complex topic.
I lined my plastic Pac-Tec Headbanger case with foil tape grounded as per the docs and have never had a problem with noise from appliances or transmitters.

thank you Alvis! all information is useful and helpful to me, i am a newbie:)

upsteam huh, alright it makes sense, so that noise is eliminated right from the beginning to the end of the whole signal path. got you!

but then, what about introducing noise to the input ground? signal is relative to its ground, correct? wouldn't the signal get noisy because its ground has become noisy? so this way, noise is introduced right from the beginning of the whole signal path and gets amplified? :confused:

That shielded input effectively does more to block interference then create it at audio frequencies.

Small signal ground fluctuations on high power stage are better to have then high power ground fluctuations on a small signal ground.

The idea is to catch the radio waves and short circuit them to ground, before they have a chance to enter the system. (They turn into miniscule amounts of heat.)

So the idea is to "shield" or put a "cage" around all sensitive amplifying components, and this cage needs to go directly to ground. Short to ground, the earlier the better.

This "shield" in effect is an antenna, that blankets around sensitive devices.

Kind of like how a lightning rod works, but for RF signals. Divert the energy around the structure/device rather then through it.

I think of an amplifier as a machine that measures the voltage difference at the two inputs and sends a multiplied replica of that voltage difference to the output (with some added noise, some added distortion and some added interference).

If you accept that the machine measures the input signal diff, then all interference signals must not add to that diff signal. The screens and grounds and etc must be connected such that they do not change the diff signal.

That requires the enclosing screen to be connected to the Signal Ground in unbalanced systems.

BUT.
the gain in a headphone amp is usually so low that interference is rarely a significant problem.

Do you need any screening?
Simply inserting an RF filter on the amp input may be sufficient.

thank you sofaspud! man, that's a very thorough treatment of a subject! let me try read it and see what i can understand, i'm a newbie:)

here's my way of thinking, and i couldn't turn around, please help me.

noise is in the environment, not in the input net yet. with shielding, we are trying to prevent it from reaching the input, so that the input signal as well as the input ground could remain pristine. now the noise is caught by the shielding, my intuition is the pristine input ground is the last spot for grounding the dirty shielding. grounding to the system ground, say, is dirty to dirty, the pristine input side is left alone and doesn't come into the picture at all, wouldn't that be the best?

if the noise is in the input net already, no grounding will help any. if the input side is pristine, grounding to input ground is introducing noise into the input ground, which will have the same effect as introducing noise to the input net.

now i know that many well designed example all ground shielding to input ground, so my "logic" must be wrong, but i couldn't figure out where i got it wrong.

i didn't know grounding is such a complex issue, thank you for pointing that out to me!

how did you solder a wire to the aluminum foil? or it is mechanically connected?

sure do accept Andrew.

that's why i tend to think screens-to-input-ground connection will add to the diff signal, because the input ground is disturbed now.

could you please explain "unbalanced system", and why screens-to-input-ground connection will satisfy the diff signal requirement? sorry i'm a newbie. if that's too much trouble, could you please point me to some books on this specific topic of "unbalance system" where screens-to-input-ground is requried for the diff signal to remain unchanged? thank you!

Think of the shield as being used to capture that noise in the environment, before it can make its way into sensitive circuitry. Once captured within that shield, we want to "turn it loose" into a harmless portion of the circuit. By connecting it to the input ground, it dissipates into the lower impedance ground portion of the circuit rather than the higher impedance input portion of the circuit.
Now AndrewT has me wondering if, despite my previous statement to the contrary, it actually does create a common-mode situation on the + input. That does make sense.
No solder. I didn't use a pigtail plug. I used a panel jack, and made sure it made good electrical contact with the foil tape.

alright, impedence makes sense to me. a lower input impedence will get more energy transferred, right?

damn, an answer gives rise to more questions! i always thought impedence involves two terminals, a signal, and a reference/ground. however, we are only connecting the shield to a ground, a single terminal, not two. can the ground terminal alone, by itself, have impedence?

or, is it that a circuit can be taken in a way such that the normal net names of signal and ground can be switched around? taking the ground as signal and the input as reference, the shield is fed to the switched-signal terminal then, and from the view point of the input net, it's got a lower input impedence input where the noise gets transferred more effectively into the switched-signal net, which is a ground in fact?

what is a "common-mode situation"?

see. foil tape is a good idea, the lining must be very neat, thank you for the tip!