Jump to content

ANALOG GROUND <> DIGITAL GROUND


techristian

Recommended Posts

  • Members

Hey Craig

 

I have never really understood GROUND LOOPS, but that is only part of the topic today. I'm designing a circuit with MIDI, digital components and analog audio output. I need some ideas on isolating digital ground from analog ground. I have heard well designed systems (where no digital noise could be heard from the outputs) and poor designs (where much digital noise could be heard from the outputs). For an example, when my wife plugs her cell phone into the AUX plug on her car stereo, she must turn her phone volume way up or the digital noise will be much louder than the music.

 

When I design my circuit I don't want such noise. I was also thinking about filtering out the noise as well.

 

Dan

Link to comment
Share on other sites

  • Members

Your example of the cellphone doesn't apply.

 

Midi contains no audio data. Its simply list of commands telling a program what it should do.

 

Midi sounds are created by Tone generating software on the computer itself, or stored on memory chips inside a unit.

 

The point of conversion from digital to analog, (the converter chip) may be a weak area. Win sound cards are notorious for passing zipper noise from a mouse. I'd suspect its the 3.5v pulse from the decoder led's in a mouse which emit stray EMF. The sound does not have grounded chassis so any spikes occurring in close proximity can be amplified. The pulse may be spiking the power supply too and making it back to the sound cards supply that way.

 

Some computers don't exhibit the zipper noise. I suspect they have better power supplies, better components or better placement/board architecture that keeps them away from the power supply or lines that carry such interference.

 

A ground is the same for analog and digital, both use the same power supply. Some components I believe are supplied a +12v and -12Vv instead of say +24 and 0 ground. The difference in voltage is the same but the components don't use a ground as you would normally think of it. You may also have different outputs to different sections of a computer. Drive voltages are usually isolated from the other boards for example and may be on a different transformer tap.

 

A normal ground loop is caused by the length of conductors to ground.

If one conductor is longer, it has more resistance. That resistances sets up a difference in potential between the two devices and grounds.

 

If one has nearly zero ohms to ground and another has 100 ohms, you have a voltage drop over that resistance just like a resistor. Since Current takes the path of least resistance, if your body touches both grounds, your body acts as a conductor to provide a second path for the ground that has more resistance.

 

Whether those two grounds, having two different potentials, will produce any noise in a circuit is questionable. You'd need some real specific examples and testing to even pinpoint it down to that level.

 

Components often have a 1~10% tolerance and can run completely noise free so thinking a ground issue will be the cause of noise wouldn't be my first guess. Ground provides a negative supply voltage and if the ground is poor you'd have bigger issues with components being supplied voltage.

 

Most use a DC grounds are very short and unlikely to have many issues with loops so long as you have a good design. AC grounds can travel hundreds of feet back to a breaker box ground buss so they often have much higher resistances, and therefore much larger differences in ground potentials.

 

In something like radio where you have tuned oscillators that produce stray EMF you often have to tune component placement, bending or twisting components so their emf emission is out of phase with another and creates a null to a squeal. With a scope you can see the noise and you basically tune the circuit for minimal noise, moving wires against the chassis and that kind of stuff.

 

If anything I'd say you may want to focus on shielding your audio section so digital EMF does not get amplified. The rest would be sure your midi cables are high quality and your power supply caps filer well. The rest comes down to the quality of the design, components, and the placement of those components. After that, if you do have noise you troubleshoot that away using test tools designed to identify the source of the problem and then you modify things to minimize the issue the best you can.

Link to comment
Share on other sites

  • CMS Author
Hey Craig

 

I'm designing a circuit with MIDI, digital components and analog audio output. I need some ideas on isolating digital ground from analog ground. I have heard well designed systems (where no digital noise could be heard from the outputs) and poor designs (where much digital noise could be heard from the outputs).

 

MIDI is (officially, anyway) opto-isolated, so that should keep noise from a MIDI input from getting into your analog circuitry via the ground. For a good hybrid design, the analog and digital circuits should have separate power supplies with their grounds tied together at one, and only one point. Economical designs that use a single power transformer with separate winding for analog and digital ciruit power feeding separate regulator circuits can work pretty well. This isn't impractical seeing as how if you're using conventional op amps, you'll want bipolar 15v for the analog side and 5v for the digital side.

 

There's an article about grounding on my web site.

Link to comment
Share on other sites

  • Members

All the above advice is spot-on, so I don't have much to add except maybe a little theory.

 

The idea of separate analog and digital grounds is because every current that flows, digital or analog, will have a return current flowing through ground back to the signal source. By keeping all the digital grounds separated, the digital return currents do not add to the analog return currents. If they could, they would show up in the analog signal as noise. Digital signals, in general do not suffer much from interference, as they are designed to have 'noise margins' built in.

 

Also, the loop area of a circuit (from the signal source, to whatever it drives, and back to the source through whatever ground path is available, is important. The inductance (tendency to act as a coil or transformer) rises with the square of the loop area; keeping this inductance at a minimum decreases how much the signal 'couples' (leaks) into other circuits in the area. This is why quality PC board designs use ground planes directly beneath the top layer, so there is always a path directly back to the source. When analog and digital are present on the same board, the ground plane is often split into analog and digital sections. These are unconnected to each other except, as noted above, at one and only one point.

 

Link to comment
Share on other sites

  • Moderators
Your example of the cellphone doesn't apply.

 

Since Current takes the path of least resistance

 

You perhaps meant to say, "Current takes all paths in proportion to their resistance," but otherwise your post was excellent.

 

Terry D.

 

Link to comment
Share on other sites

  • Members

 

You perhaps meant to say, "Current takes all paths in proportion to their resistance," but otherwise your post was excellent. Terry D.

 

That's another way of putting it if you view the circuit as a whole. "Current takes the path of least resistance" is one of those old clichés or golden rules you are taught in electronics. Path of least resistance is an experienced based term from folk physics that can describe very simple situations of what occurs.

 

Other similar examples like "what goes up must come down" aren't laws of physics as much as they are common sense terms. In a more complicated situation these kinds of rules fails to give even approximate results. In electrical circuits, current always follows "all" available paths, and in some simple cases the "path of least resistance" will take up most of the current if there are only two choices.

 

If you have three paths of nearly the same resistance, the "path of least resistance" will only take up about a third of the current.

 

If however you have three paths and one "has" zero resistance and only one source potential the other paths are mute situations that have no bearing on the current conductance. This is exactly why ground makes a chassis safe to touch and why it must be as conductive as possible. The conductor will conduct all the current removing any difference in potential from the other parallel paths of resistance, so the cliché can be true if the path of least resistance (like a ground) is a pure conductor.

 

In the case of an AC ground loop, you are dealing with two different paths in parallel. by placing yourself between the two ground loops you are adding a series resistance which conducts electricity between the two loops so the overall picture gets more complex, but you can focus on just that series resistance of your body acting as a conductor and it will apply to the cliché.

 

Of course absolutes aren't possible in the physical world. Even wire has some resistance, but that's how you use the theory as you work with it. Everything else is dealing with actual component tolerances which is that "other" reality many overlook when trying to explain it how something invisable works.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...