Theory of operation for ADAU1962

Hello Moffle,

Sorry for the delay but there is a lot to unwrap in your questions so I needed a decent block of time to respond. 

I certainly understand your confusion and it is confusing until you study this the correct way. The specifications for the RMS output levels are measured between two pins. To calculate the actual voltages on one pin you have to look at just one pin at a time then look at both of them together. 

Question 1:

First I need to correct something. If you want to send a full scale positive value to the DAC it would be a 0x7F FF FF not an 0xFF FF FF. The DAC works with a 2's compliment number input so the MSB is the sign bit. An 0xFF FF FF is one bit negative below what we would call zero. (in this case 1 bit below the 2.25V common mode voltage) A full scale negative number is 0x80 00 00. 

One little detail is that this means that there are only 23 bits to divide up the positive voltages and 23 bits to divide up the negative voltages. Well, technically the positive side has one bit less because zero uses one of the 2^23 available numbers to represent a voltage on the positive side but the negative side uses all of the 2^23 different possible numbers. These voltages are so small and the noise floor and error of the common mode will swamp this technical detail so I will simply use 2^23 steps for both positive and negative. You would never be able to measure the difference. 

For the rest of question 1. Using the values in Table 1 of the datasheet the typical differential output voltage is 3.00Vrms. By the way, this was my first datasheet I ever edited at ADI. 3V RMS converted to peak to peak values is: 3 * 2.828= 8.484 but 2.828 is actually truncated so the result I put in the datasheet is a little more accurate. 8.49Vpp. This is where it gets confusing because we are talking about two pins. So divide that in half to get the voltage on one pin. Then divide that in half again to get the voltage swing from the common mode voltage of 2.25V. So 8.49V/4=2.1225V . You had these calculation correct to thumbs up to you! 

You are correct that one pin will go from roughly 4.37V for full positive and 0.13V for full negative. This is a swing of 4.24V 

Each pin will do this but in opposite direction. So now let's look at this from the perspective of an OpAmp differential input. It will take one input level and subtract it from the other. So when the sine wave is at full scale positive the opamp math would be 

4.37V - 0.13 = 4.50V positive output. (assuming it is setup for unity gain of course)

When the sine wave is at full scale negative this is the equation:

0.13 - 4.37 = -4.24

The full voltage swing from positive 4.5V to negative 4.24 = 8.74V. This is different from the datasheet volts peak to peak because we were using 0V as out reference. Both of these pins will not go down to 0V and the opamp should be using the CM voltage of 2.25V as its reference. Subtracting the 0.26 gives me 8.48Vpp which is what is in the datasheet. 

This part is designed for audio so the specs are all centered on RMS values of a sine wave and peak to peak values of a sine wave not for DC levels. You can use the peak to peak values to get an idea of what it would do with just DC or close to DC. 

Question 2:

We could go through the math of dividing the 2.12V positive swing by 2^-23 or written in another way 1/2^23 and coming up with a voltage per step. Then doing more math to convert from 5V to 3.3V AVDD but I will instead give you some info that is not in the datasheet. 

The history here is that we released this part as being able to operate from 3.3V to 5V for AVDD. The specifications were not as good at 3.3V of course. Then we came out with the "A" versions of this part. The ADAU1966A which is only a 3.3V part. I would recommend you use that part if you are planning on using 3.3V for AVDD. the specifications will be a little better. Also, that part you can use single ended. This ADAU1966 part you must never use single ended. There is a lot of noise on the pins. We fixed it with the "A" version parts. 

Since I wrote these datasheets I still have the original version that shows it operating at 3.3V. I cannot put up the entire datasheet but let me share two relevant tables. This first one is the specs for AVDD=3.3V at ambient temperatures. I need to do it in two screenshots because it wraps to the next page. 

You can see the common mode voltage is different. There is a register that controls it and I removed it from the datasheet since we did not plan on this part running at 3.3V after we came out with the "A" versions. Here is the register:

Question 3:

SigmaDelta converters are not great for DC applications. They can be used and you may be able to work around the errors you will see. 

First, look at the Gain Error specs and the gain drift and you will see it is not great. These specs will come into play when using these for DC. 

There are a couple of sources for these errors. One of them is the voltage reference itself. It does drift and looking at the min/max you see a spread. If you were to shut off the internal reference and drive the CM pin with an external precision voltage source you could improve the gain error and gain drift specifications. You also could trim the output voltage if the precision reference is fancy enough to be adjustable. I suppose you could put in the trim pot and have a factory calibration routine. 

What you want to check is the gain error at the temperatures your product will be exposed to and then monitor it over a long time period and with different signal levels to see how much it drifts. So this will take some work on your part to see if it is acceptable for your application. I cannot say that it will be since there are many other factors involved. 

My suggestions is to drive it with an external reference and would look into using the ADAU1966A version if you want to run at 3.3V. I gave you the info to be able to use the non-A version in case you purchased a few trays already. 

I hope this helps 

Dave T