EEVblog Electronics Community Forum

Electronics => Projects, Designs, and Technical Stuff => Topic started by: Starlord on August 13, 2015, 10:21:27 pm

Title: Which of these regulators has less noise?
Post by: Starlord on August 13, 2015, 10:21:27 pm
I have a buck converter on a board which can be powered by up to 16V, and I need to drive this DAC:
http://www.ti.com/lit/ds/symlink/dac101s101.pdf (http://www.ti.com/lit/ds/symlink/dac101s101.pdf)

I have already prototyped my board, and the noise floor is, well... not great.  You can definitely hear hiss when no other audio is playing and you don't have to listen carefully.  It's adequate, but barely.

Now, I'm not sure if this hiss is due to the DAC being only 10-bit (doubtful?) using only two 100uF capacitors for the amplifier, or due to not specifying that they be low ESR, or due to the switching regulator noise, but I know I'm going to have other sources of noise in my final circuit, so in addition to replacing those 100uF caps with a couple low ESR 470uF caps, I'd also like to use a separate linear supply for my DAC, just so I have the greatest chance of getting results that I'm satisfied with.

The TI documentation suggests this regulator for use with the DAC:
http://www.ti.com/lit/ds/symlink/lp2980-n.pdf (http://www.ti.com/lit/ds/symlink/lp2980-n.pdf)

It alternatively suggests the LP3895 but that does not work at up to 16V.

The LP2985 however does:
http://www.ti.com/lit/ds/symlink/lp2985-18.pdf (http://www.ti.com/lit/ds/symlink/lp2985-18.pdf)

Now here's the thing.  The 2980 says it has a PSRR of 63dB.  While the 2985 says it has a PSRR of 45dB.  So the 2980 is better right? 

But wait... the 2980 also says it has an output noise voltage of 160uV.  While the LP2985 says it has a noise voltage of 30uV.

So which parameter should I pay attention to here? 

It should be noted that the board will be powered by a 12V battery most of the time.  And even when powered from a wall I would be using a switching power brick rather than some old hunk of junk, so there shouldn't be AC noise.  I assume that's what's meant by "ripple".  So what I'm mostly concerned with here is getting rid of any noise induced in the supply by the board itself, or by the switching regulator powering the board if any.  I guess there may be some "ripple" in the supply voltage as the current demand changes, but I don't know if that matters with a 12V supply when I'm dropping a whole 7V.

Btw, the DAC only requires 20mA, so I don't anticipate that dropping 7V here is going to be an issue as far as heat goes.
Title: Re: Which of these regulators has less noise?
Post by: Paul Price on August 14, 2015, 03:42:27 pm
You are not posting a schematic, so you are making this a guessing game.

That being said, it is likely that your amplifier connected to the DAC output might be generating the noise you hear or else the digital signal driving the DAC.  It is likely that it is neither the DAC nor its power supplies are creating the noise problem.
Title: Re: Which of these regulators has less noise?
Post by: dom0 on August 15, 2015, 08:42:08 am
10 bits only gives a SNR of ~60 dB - that's your hiss right there.
Title: Re: Which of these regulators has less noise?
Post by: Starlord on August 15, 2015, 09:44:09 am
10 bits only gives a SNR of ~60 dB - that's your hiss right there.

I don't think so.  I've converted audio files to 8 bit before and though you get some hiss, I don't think it was nearly as extreme as this, and 10 bit is 4x as good.

Also on my last board I used a 12-bit DAC, which is only 4x as good as 10-bit and I don't think this hiss is only 4x as loud as that.  But I don't have them both to compare side by side.

I suppose I could order a 12-bit version of the DAC from Digikey and do some rework on my test board. 

But I also recall that in my tests I tried adding and removing bulk capacitors from the board and adding more capacitance seemed to have helped somewhat.  Of course I don't know if I merely reduced it o the lowest possible point.  But I didn't use low ESR caps either and I suspect that could affect it.