Pro Audio Design Forum

Today I was having a look at this TI article on single supplies and op-amps.

http://www.eng.yale.edu/ee-labs/morse/compo/sloa058.pdf

What picked my attention was figure 3 in page 5 for a noninverting circuit. The VCC/2 reference is connected to the gain chain, which is different to what Walt Jung used on his book.

Can anyone explain which is the right one and why?

There's something fishy about that second figure: There's no bias resistor for the non-inverting input.

Well, that's the official TI report on that matter. Have a look at the original file.

I did. It's still fishy.

The TI version is unclear, but if you follow the thrust if the incomplete text they say the input impedance should equal R1 in parallel with R2. What they mean is the + input termination should equal R1 in parallel with R2 to cancel input bias current effects.

It is poorly executed and incomplete as drawn.

JR

{edit there is no one right way to do this, while the TI example is incomplete and will not work without a dc path for + input. /edit]

I went into the TI Forum to find out how to adapt the LME49600, and I think I understand their paper better now.

What they propose is to float the system completely from chassis ground and only use Vcc/2 as ground for everything. Maybe that's the fishy thing you talk about.

But I wonder how does that work for isolating RFI, hum and other interferences.

Have a look at their last comment to my question:
>
To run off of a single battery, the ground must be VCC/2. This is the virtual ground that I discussed earlier.

It is simply creating a voltage tap in the center of your supply so that the op amp V- rail is below the virtual ground and V+ is above the virtual ground. This makes the output of the op amp centered around half of the supply voltage, or 6V if you are using a 12V supply. This is why your output must be referenced to VCC/2

It is exactly the same principle if you had +10 and -10 bipolar supplies, the op amp is centered around 0V (the center point between V- and V+) thus you can reverence everything to 0V or true Earth ground potential.

You are correct, you are floating the output so to speak 6V above true 0V earth ground. It should not be an issue for most things because ground is a relative term.

I can reference a speaker to a 1KV ground and apply 1.001KV on the other terminal. The entire speaker would see it 1V and it would work just fine. It doesn’t care how far it is above earth ground. However if I did earth ground it, then we would have a big problem because we would now have a new reference voltage!

Therefore you cannot tie your virtual ground to the chassis of your enclosure and then tie the chassis to earth ground.

For headphones, you may want the check with the manufacturer if they are active and noise canceling. These may expect to be reference to true 0V earth ground but maybe not.

>

I feel like I am repeating myself, maybe we don't need so many threads for one project.

If you use a single supply with v/2 instead of split supply with real ground, that V/2 needs to be "ground like". It would take too many worlds to completely describe what ground is "like" but for RF and such you want the chassis and audio shields bonded to ground.

You can probably get away with a hybrid ground where V/2 is bonded to ground at RF with HF capacitors.

JR

I think our major disagreement is that for your suggested way to do things you need a DC-DC supply that generates something like +/-15v and drains batteries very fast.

Sound Devices gear is designed like that, and the major complaint everybody has about them is how fast batteries go.

You may like what TI proposes then, which you can see on this topic from their forum:

http://e2e.ti.com/support/amplifiers/au ... pi258929=2

They suggest to make a split point with resistors and capacitors, and make that the only ground to be used.

We don't have a disagreement, and I do not advise using 2x the supply voltage you actually need. I make a battery powered product and i use a class D chip to drive a speaker very efficiently, with good battery life.

IIRC I was suggesting a split supply with ground in the center.

This is your design so do whatever makes you happy. I am not arguing with you but trying to share some of my experience designing products. If you read my last post i suggested that you use HF caps to make the V/2 appear more ground like at RF.

JR