About CMIM mismatch????????

[ira2008]

Hi everybody,

Can anybody tell me what tool can verify accurately layout mismatch .
For example, my design engineer told me there are two mim caps . Their mismatch need to reach one thousandth . I have drawn this cmims. but i don't know if it satisfied his demands. What tool can verify it?

thanks for your reply .
Amy

 

[steve_guo]

u can refer to design rule wihch may include some of the key parameters related to mismatching.
Depending on these, the size of MIM cap can be determined.

 

[ira2008]

u can refer to design rule wihch may include some of the key parameters related to mismatching.
Depending on these, the size of MIM cap can be determined.

First of all, I appreciated your quick reply. According to your prompt, i looked design rule which is not any about some of the key parameters to mismatching.
Thanks
Amy

 

[ajay181]

Hope you have used unit caps of same size with common centroid arrangement ?
the ratio match between cap can be calculated by checking the extracted cap values between them.

 

[ira2008]

Hope you have used unit caps of same size with common centroid arrangement ?
the ratio match between cap can be calculated by checking the extracted cap values between them.

Very very thank you. by the way, the ratio match between cap how to calculate. What tool can verify it? or it can be calculated internally. yes or no?

thanks in advance.

Amy

 

[erikl]

... the ratio match between cap how to calculate. What tool can verify it? or it can be calculated

The mismatch ratio can be estimated with the Pelgrom equation and a mismatch coefficient to be provided by the foundry. For the calculation, see the following (resistor) mismatch examples.

Extraction tools (Diva, Assura, Calibre) can only find additional associated (parasitic) values due to routing and possible environment differences.

 

[ira2008]

The mismatch ratio can be estimated with the Pelgrom equation and a mismatch coefficient to be provided by the foundry. For the calculation, see the following (resistor) mismatch examples.

Extraction tools (Diva, Assura, Calibre) can only find additional associated (parasitic) values due to routing and possible environment differences.

Hi erikl,
I really appreciated you. Maybe occupying much of your spare time. on the whole, i understood your ideal. thanks again for your help. Hope you happy everyday.
Best wishes
Amy

 

[ira2008]

The mismatch ratio can be estimated with the Pelgrom equation and a mismatch coefficient to be provided by the foundry. For the calculation, see the following (resistor) mismatch examples.

Extraction tools (Diva, Assura, Calibre) can only find additional associated (parasitic) values due to routing and possible environment differences.

Hi erikl,
I have another question. there are two 2pF cap, How to make A cap and B cap to get good matching? Besides common centroid arrangement and add dummy cell . Actually I want to ask you how to choose unit cap value? Please give some advices.
thanks very very much!
Amy

 

[erikl]

How to make A cap and B cap to get good matching? Besides common centroid arrangement and add dummy cell.

Hi Amy,
seems you already know all the necessary tricks ;-) . Try to get a square block, see the PDF below, p. 19 : View attachment Analog_Layout_Techniques_in_VLSI.pdf

Actually I want to ask you how to choose unit cap value? Please give some advices.

Unit caps usually are provided with the foundry's PDK - often several units depending on the cap type: MOSCAP, PP, MOM, MIM, fringe, ... - always square.
Mine were in the order of 10×10(µm²), around 100..350 fF each (0.18µm process).
Best regards, erikl.

 

[ira2008]

[/COLOR]

Hi Amy,
seems you already know all the necessary tricks ;-) . Try to get a square block, see the PDF below, p. 19 : View attachment 49938


Unit caps usually are provided with the foundry's PDK - often several units depending on the cap type: MOSCAP, PP, MOM, MIM, fringe, ... - always square.
Mine were in the order of 10×10(µm²), around 100..350 fF each (0.18µm process).
Best regards, erikl.

First, Very thanks. and I am sorry about that I am forgetting to tell you that unit cap can not be designed for square. Because my room is rectangle (width=40um length=500um) not square. so unit cap is not fit for square. Do you think so? I have selected unit cap equal 0.18um each. A and B MIMCAPs value is separate (0.18umx12)=2.16Pf. (0.18 UMC process). Do you think it is a good choice? , By the way ,hope you understand
my expression.

Best wishes.
Amy

 

[erikl]

... unit cap can not be designed for square.

Sure it can. See this PDF: View attachment MIM-cap_mismatch.pdf

Because my room is rectangle (width=40um length=500um) not square. so unit cap is not fit for square. Do you think so?

Sorry:no. The unit cap layout - with which you build the caps you need - has nothing to do with your available real estate on the chip.

I have selected unit cap equal 0.18um each. A and B MIMCAPs value is separate (0.18umx12)=2.16Pf. (0.18 UMC process). Do you think it is a good choice? , By the way ,hope you understand my expression.

Unfortunately not: how can you get picoFarads from a multiplied length? On the other hand: 12 MIM caps with an edge length of 0.18µm impossibly can create a capacitance of ≥ 2pF !

 

[ira2008]

Sure it can. See this PDF: View attachment 49956
.

Thanks for your professional suggestions.
So for the MIM-CAP, the shape of the unit cap should be either square or rectangular. But the size should be kept less than 30um per side. Yes?

For the width of metal lines which are used to connect top-plates and bottom-plates, I think minimum width(such as 0.3um) is enough, because wider width will cause more parasitic capacitor. Yes?

I found the file you provided to me is very useful. Could you do me a favor sent the whole pdf file to me for my reference. I appreciate your help very much.

Unfortunately not: how can you get picoFarads from a multiplied length? On the other hand: 12 MIM caps with an edge length of 0.18µm impossibly can create a capacitance of ≥ 2pF !

OK 0.18 process :MIM capacitance =0.98fF/(um)*2. Length was selected randomly Or by my feeling to choose. It is not any basis. By the way, total MIMCAP value= (width x length )x0.98fF/(um)*2 x 12=(9um x 20 um )x0.98fF/(um)*2 x 12= 2.16Pf.

Thanks again for you. Hope you happy every day.
Amy.

 

[erikl]

... for the MIM-CAP, the shape of the unit cap should be either square or rectangular. But the size should be kept less than 30um per side. Yes?

Better square than rectangular! Max. edge length depends on the process.
For my former 0.18µm process, a max. MIM cap edge length of 150µm was allowed, s. the above PDF!

For the width of metal lines which are used to connect top-plates and bottom-plates, I think minimum width(such as 0.3um) is enough, because wider width will cause more parasitic capacitor. Yes?

Right!

I found the file you provided to me is very useful. Could you do me a favor sent the whole pdf file to me for my reference.

Unfortunately this isn't possible: The above PDF page has been extracted from our foundry's process analog characterization docu, which of course is confidential info (that's why I've made it irrecognizable by the yellow top & foot notes ;-) ). Hope for your comprehension!

OK 0.18 process :MIM capacitance =0.98fF/(um)*2. Length was selected randomly Or by my feeling to choose. It is not any basis. By the way, total MIMCAP value= (width x length )x0.98fF/(um)*2 x 12=(9um x 20 um )x0.98fF/(um)*2 x 12= 2.16Pf.

By the way: For me, (9um x 20 um )x0.98fF/(um)*2 x 12= 2 * 2.12pF

Hope you happy every day.
Amy.

Right so, thank you, same to you! - erikl

 

[ira2008]

Unfortunately this isn't possible: The above PDF page has been extracted from our foundry's process analog characterization docu, which of course is confidential info (that's why I've made it irrecognizable by the yellow top & foot notes ;-) ). Hope for your comprehension!

Hi erikl,
I think you are a very nice . and I can understand you. By the way, If you came to China in the future and gave me a opportunity, I would invite you to eat Chinese fine food.:grin:
Thanks again.

Best wishes
Amy!