Post Go back to editing

AD1938 output coding

Category: Datasheet/Specs
Product Number: AD1938


I will use AD1938 as 24bit ADC and could I ask how to calculate input voltage ->LSB ?

( which the number of full-scale  or VREF in the datasheet should I use ?)

Which equations are correct ?

Differential input voltage / ( VREF (1.5V) *2)  = outpu code ( LSB) or  differentail input voltage / (Full-scale(1.9Vrms) *root(2)) =output code ?

Please let me know your advice.



  • Hello Teli,

    I will try to get back to you with more detailed information. I do not want to delay some of the answers for you. This thread went into a lot of this detail.

    (+) AD1938 CODEC Full-Scale Input Voltage (Differential) - Q&A - Audio - EngineerZone (

    For the LSB part the positive voltage is split up by 23 bits,from the common mode voltage (which is zero for the 2's compliment number) to the full scale peak DC value. So that is 1(2^23) times the voltage difference between CM and peak positive. 

    On the negative side there is one more bit but with the bit size being so small that it would be difficult to measure this. 

    Dave T

  • Hello Mr. Dave

    Thank you for your advice.

    I understood 1.9Vrms *2root2 = about 5.37Vp-p and full scale is the dirrefence between positive and negative pin (+- 2.69V => 5.37Vpp) from your adviced thread.

    Following equation is output coding equation for bipolar operation in common ADC with gain ( offset binary).

    I can't understand whether I can apply below equation to AD1938 and please let me know your advice how to calculate the output code in the case of AD1938.

    (for example, below AIN to differential input voltage( Positive input  - Negative input ) , below VREF to 2.69V ( 1.95Vrmms /2 *root 2) )



  • Hello Teli,

    I do want to ask why do you need this? Most people who use this part for audio do not need to calculate the exact code. They just need to know what the full scale peak to peak voltage will be. If you are planning to use this as a data converter keep in mind that it is designed for audio and not data so it will have a lot more drift and gain error than a data converter will have. Its DC performance will not be great.

    This will be a little difficult to follow. Here is the formula simplified:

    code=(4*(Ain-CM)* 2^23-1)/Vpp

    Ain is the voltage on the pin relative to ground and assuming a nominal CM voltage.

    CM= nominal Common Mode voltage which in this case is 1.5V.

    Vpp is full-scale input voltage as a differential peak to peak number. 

    The result will be the 2's compliment code number. Keep in mind that this is one pin and the assumption is that the other pin is sitting at the complimentary voltage. If you just use this single ended with the other pin sitting at the CM voltage, then the result will need to be divided by 2. You need to use both pins to reach digital full scale input. 

    I got this by doing the following calculations then simplifying it afterwards. 

    Ain-CM gives the full scale positive voltage swing. 

    You then divide that by the voltage per bit which is obtained by taking the Vpp and dividing it by 4. Each pin takes half the peak to peak voltage and then each pin further divides it by two around the CM voltage. So you get Vpp/4

    Then take the Vpp/4 and divide that by the number of codes and that is 2^23-1

    It is a 2's compliment result so the 24 bit number has a sign bit therefore only 23 bits are available for a magnitude. 

    I worked this all out on a spreadsheet if you care to look at it. I did it in pieces then combined it over on the right where I tested the simplified formula.

    Dave T


  • Helo Dave

    Thank you for your suuport .

    We aret going to use AD1938 for instruments ( not audio products) therefore we need exact code.

    As your advice, we understand that  some performance  of AD1938 are different from the those of standard data converter.

    I will test the performance of AD1938 before I use 1938 for our products.