Prof Ali ...Thank you very much for uploading your lecture to help guys like me to learn from one of the best analogue/RF designer. If I may to draw your attention that, this the last lecture I can watch on RUclips? Thank you for your help, May Allah give you all the best!
in 4:32 if the horizontal axes represents digital numbers, then half LSB does not exists. in 13:38 because we are considering the analog signal, I think that It has to settle under delta/2 = Vref / (2^ (N+1)) , and not LSB/2
With Delta R, the equation assumes that each resistor in the ladder deviates by the exact same amount (of Delta R) ... the error of each resistor's resistance are "matched". In *actual* voltage, shouldn't Delta R be different for every resistor?
Prof. Hajimiri assumes that every deviation is fixed. This is for our own ease of understanding and to illustrate the calculation of INL/DNL for this specific case. In reality, you are correct and all resistors may not have the exact same deviation (it really depends largely on the layout of the resistors) and delta R is more a statistical quantity (having a normal distribution) with zero mean (ideally). Thus delta R can have different signs along with different values. If you plot delta R for the 65,000 resistors on a distribution plot, you would see a plot similar to a normal distribution. With good matching on-chip, you will see a tighter distribution of delta R with a lower sigma value. Although, a point to note here is that you CAN have something close to a fixed delta R for all resistors if they are laid out in a chain-like fashion. This arises because of the gradient in the diffusion of the polysilicon layer on the chip and thus Prof. Hajimiri's case is not far away from actual real world scenario. In fact, this is how it generally turns out on silicon.
I enjoy these lectures, dear Prof., but you didn't seem to be so sharp today. I think you confused yourself in several places. But we all have days like that ;-)
Hi sir its very nice class...Sir can you please do some problems related to this topic ie,how to find out DNL and INL.I have some problems but felt difficulties to solve that ..so please help me out sir
Your lectures are invaluable, thank you so much.
Prof Ali ...Thank you very much for uploading your lecture to help guys like me to learn from one of the best analogue/RF designer.
If I may to draw your attention that, this the last lecture I can watch on RUclips?
Thank you for your help, May Allah give you all the best!
the death by 1000(1024) cuts lol😂😂
in 4:32 if the horizontal axes represents digital numbers, then half LSB does not exists.
in 13:38 because we are considering the analog signal, I think that It has to settle under delta/2 = Vref / (2^ (N+1)) , and not LSB/2
With Delta R, the equation assumes that each resistor in the ladder deviates by the exact same amount (of Delta R) ... the error of each resistor's resistance are "matched". In *actual* voltage, shouldn't Delta R be different for every resistor?
Prof. Hajimiri assumes that every deviation is fixed. This is for our own ease of understanding and to illustrate the calculation of INL/DNL for this specific case. In reality, you are correct and all resistors may not have the exact same deviation (it really depends largely on the layout of the resistors) and delta R is more a statistical quantity (having a normal distribution) with zero mean (ideally). Thus delta R can have different signs along with different values. If you plot delta R for the 65,000 resistors on a distribution plot, you would see a plot similar to a normal distribution. With good matching on-chip, you will see a tighter distribution of delta R with a lower sigma value. Although, a point to note here is that you CAN have something close to a fixed delta R for all resistors if they are laid out in a chain-like fashion. This arises because of the gradient in the diffusion of the polysilicon layer on the chip and thus Prof. Hajimiri's case is not far away from actual real world scenario. In fact, this is how it generally turns out on silicon.
I enjoy these lectures, dear Prof., but you didn't seem to be so sharp today. I think you confused yourself in several places. But we all have days like that ;-)
Hi sir its very nice class...Sir can you please do some problems related to this topic ie,how to find out DNL and INL.I have some problems but felt difficulties to solve that ..so please help me out sir