Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Matching property of nwell resistor?

Status
Not open for further replies.

hebu

Full Member level 4
Joined
Nov 15, 2004
Messages
194
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,296
Activity points
1,816
I would like to design a low frequency circuit, for saving the area, I'll use
the nwell resistor. but how about the matching propery of this kind of resistor
and any technique to improve matching?
 

Is the matching accuracy high? The resistance of nwell resistor is dependent on the width and the voltage on it.
 

it is not very good for n well res.
you can use other kind res while the size small(L samll)
 

As far as I know, there is no special requirement for matching for NWELL RES. Maybe you can consult EDR for min. Ls.
 

For nwell resistor design nature, the well resistor is not good.

As we know, the well resistor is formed by a indenpent well and two contact to form the metal connect.

Because the well implant is more deep, so there is lateral diffusion which is a main factor to do effect the accuracy of well resistor.

In the mean while, the different drop voltage between terminals of resistor results in a different resistor value . this is called voltage coefficient of resistor.

Due to the two disadvantge, i think if you have a matching requriement , other type resistor can be selected for improving performance of circuit.
 

I don't really worry about the voltage coefficient, because the voltage difference
between the two matching resistor is not big. And the voltage coefficient is less
than 8e-3 by foundry document. My matching requirement is the error should be
smaller than 3%. error=2*(R1-R2)/(R1+R2)
I know it's easy to achieve this by poly resistor, but how about n-well resistor?
I need your opinion.
Thanks,
 

I think the 3% matching can be acheieved by the nwell resistor. Make sure your square number is bigger than 5 and width is bigger than 4um. That's my experience.

Yibin.
 

Any skill to improve the matching of nwell resistor?
Such as poly resistor matching, we often add dummy to prevent the mismatching introduced by etching.

But how about nwell resistor because it's a process of diffusion rather than etching.
 

Due to the electronic nature, i think you can design two types nwell resistor such as R1, R2 . From your error=2*(R1-R2)/(R1+R2), i think which one is larger than other one for R1 and R2.

In order to reduce the erro (R1-R2)/(R1+R2), the matching of R1, R2 is very important. There are some ways to improve the matching for resistor.

1. as possible as you can to use a kind of value resistor , and you can receive the value you want by connect multiple. This is mean you have to find the common divisior among many resistor to be designed.

2. enlarge the width to make the lateral diffusion neglectable.

3. add some guarding rings to make the codition the same.

4. add the taps which connect to GND between the nwell resistances.

5. add some nwell dummy as option for debugging.

6. the most one is using the same width of resistance.
 

    hebu

    Points: 2
    Helpful Answer Positive Rating
Thanks a lot, and what strategy is best for fractional ratio between R1 and R2?
For instance, R1=4k, R2=6.3k.
Assuming the small unit resistance like 0.1 is not feasible.
By the way, what do you mean "square number is larger than 5"?
What is the square number?
 

HI, hebu,

If the accuracy is the most important parameter for you , i think the 300Ohm/squre and 1kOhm/squre can be design for receiving the value you want.

Such 4k=1k*4 +(300) % 300 use as a dummy
6.3k=1k*6+300

(Note the all nwell is drawn the same length but the metal contact is different for 300Ohm and 1k Ohm as below:
_______________________
*|______________________|*

%%%%%%%%%%%%%%%
________________________
*|_______*______________|

* represent the metal contact to well diffusion
% represent the taps which connect to GND between the nwell resistances

)

Maybe , there are some erro due to the lateral diffusion.

what we do is to get the most result as possible as we can!

and many techniques for matching is a depth consideration. you can learn and

use them.

May you secuss for layout and silicon chip.
 

i think it is not good to use nwell resistor, because it has bad process acuracy and bad temperature coefficient and voltage coefficient.
for 0.35um CMOS process, the accuracy of it isd about ±30%~50%.If you do use Nwell resistor, you can do careful layout, such as maching using dammy cells, cennecting to stable Nwell power, etc.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top