What purpouse does 100K res serve on audio shield

On the Audio sheild, there is a 100K resistor on the Vrefa pin. There does not appear to be one in the example circuit in the DAC’s datasheet:

And a google search for Arduino audio shield 100K turned up this page unrelated to it:
http://docs.macetech.com/doku.php/shifty_vu_shield

But it did have this paragraph which was was caught my eye in the search:

A potentiometer should be chosen with a high enough resistance that it doesn’t draw too much current; but small enough that the voltage reference is not affected by external factors. 10K to 100K would be within an acceptable range.

And it got me thinking that I didn’t want my analog voltage reference to be affected by external factors and maybe I have enough noise in my circuit that a lower value resistor would be prudent.

So I stuck a 10K on there instead of the 100K, and while I can’t be sure it’s less noisy, my 4000hz test tone seems noticably louder and more peircing, so I’m thinking maybe the square wave the piezo is getting is less noisy.

What do you think? Using the 10K won’t harm my dac, right?

Using the 10K won't harm my dac

It won't but it will actually be a more noisy solution. The input impedance to the Vref is 165K so by using 10K you are sending more current down it but not more than twice. However, the external capacitor forming an LC circuit will have a higher cut off frequency with 10K and so for any given frequency of noise will let 6dBs more noise in.

100K it is then.

But why is it so much louder? Either way, isn't the voltage reference still 5v? And doesn't my DAC therefore swing from 0 to 5v when I send it a square wave?

How is this affecting the loudness of my piezo?

sn't the voltage reference still 5v

No with a 100K you only have about 3V reference because it acts like a potential divider with the input impedance.

So that 165K of resistance... is that coming from those resistors going to ground in the block diagram? If so I can see what you're saying about a voltage divider.

And I just did the math... 3.11v out apparently. And with a 10K I get 4.71v out.

Perhaps I should go with the 10K after all. Now that I think about it, I'm more concerned about volume than noise. These props will be used in loud bars and at conventions.

So that 165K of resistance... is that coming from those resistors Going to ground in the block diagram?

yes the value of 165K is given in the data sheet in the characteristics table.

I'm more concerned about volume than noise

That's your choice, connect it up to +5V if you want and squeeze that bit more out of it.