Hello everyone,
                      I trying to find a relation between DNL and elemental mismatch for a resistive ladder dac. I know that the worst case DNL is when the ladder position is at the end. In my case there are 2 ladders..
                      I have got a relation between INL and mismatch but dont know how to go about finding a relation between DNL and elemental mismatch..
                      Any help or reference material which I can use..
Thanks a lot in advance
BR
Vabzter

panditabupesh wrote on Feb 16^th, 2007, 7:36am:

Hi Bhupesh
                Yup I have gone through them. What I am looking for is relation between σ and DNL. In those relations he has derived that the max DNL is at the 2 ends and max INL is in the middle position of the ladder. I want the relation to size the resistive ladder in dac
Thanks
BR
Vabzter

panditabupesh wrote on Feb 16^th, 2007, 11:40am:


Hi Bupesh,
                  Yup the above equation is right. I have also got the value of r_alpha from the reference manual from the foundry. The question is I am designing a 10 bit ladder DAC and I want to derive a relation between the DNL and mismatch.Then by using the above equation I can size the resistance values. Like for INL I found this relation from Berkeley site for the leactures on Data converters(references are not mentioned on the leacture slides...)..
B = 2 + 2log2(σ_inl / σ_mis)  .......(log2 = log to base 2, B=no. of bits)

So in the similar manner I want to find a relation between DNL and mismatch..

Thanks for your help
BR
Vabzter

Hello
Though the paper uses terms like  precision and linearity for N, but I think it means same as accuracy. For example let us consider an 8-bit  two-step  (or subranging)  ADC, and we  are extracting 4-bit MSBs and 4-LSBs in two steps. Without digital-error correction, even though we are extracting only 4-bit MSBs (resolution n= 4)   the DAC ( a resistance ladder etc )  involved will need to be 8-bit accurate (linearity N = 8). But, for a single-step 8-bit ADC, e.g flash, the resolution and accuracy will be same and equal to 8-bits.


Johns & Martin spend a couple of pages, section 13.5, on this topic.

bye
Bupesh