Is the common centroid method obsolete ?

[muldersun]

Common centroid

I have seen a layout , it's a full different opamp,the input stage are not matching,and also are not close,they have a long distance.
I'm ver..........y confused.

 

[jianbinma]

Common centroid

For M1 and M2 matching,I don't think using common centroid is a good way,generally common centroid is suitable for small W value transistor pairs. In this case ,I think it's better for you just use integradient fingers to make M1 and M2 as compact as possible.In addition ,the resistor is a single big one,why you don't make it by 10X4k ,may be it's a little easy for you layout.It's just my own opinion.

 

[duron999]

Re: Common centroid

For M1 and M2 matching,I don't think using common centroid is a good way,generally common centroid is suitable for small W value transistor pairs. In this case ,I think it's better for you just use integradient fingers to make M1 and M2 as compact as possible.In addition ,the resistor is a single big one,why you don't make it by 10X4k ,may be it's a little easy for you layout.It's just my own opinion.

If this is an analog layout, will any problems come out when
using 10x4K resistor ???

 

[raymond33]

Re: Common centroid

Well I give you my feeling on this topic
M1-M2 common centroid , minimun lengh +0.01u or +0.02u (depending if is odd/even pitch) but big W because area play (mitsmatch in Vt) a big role and of course because the bandwith depend of the gm of these transistors too.
M3-M4 Lenght never minimun and W.L big and you need to think than the gm3,4 needs to be below gm1,2
M8-M5-M7 lengh play a big role for both raisons but only L nothing else and integer ratio is mandatory between these transistors.The lenght our current value but never it's possible to use L min for them.
M6 is your common source transistor , don't forget than this transistor give a big part of the gain of the second stage (Miller efffect is better in this case) so be carefull with them.
I guess if you write equations ,Gain Bandwith and offset it's easy to found values for these transistors.Just onether stuff , when you reduce the offset by the same way the noise decrease too.I hope I helped you.

 

[lakeoffire]

Re: Common centroid

Why bother trying to match the PMOS with NMOS. It looks that this is a twin-well process (because all NMOS bulks are tied to sources), ans as a result, PMOS and NMOS fets are sitting in different wells and they even cannot be placed close enough to each other.

You have a design problem. The ratio between M5 and M12 should be an interger number : 7 or 8, not 7.5!
No way to draw a good layout. I suggest you to increase a little your ref resistor to keep the same current in your diff pair and go for a ratio of 8.
Then, your layout becomes : M12,M12,M12,M12,M5,M12,M12,M12,M12 :idea:

Hello, i don't understand what mean "the ratio in interger number:7 or 8, not 7.5!"

and below is the overall circuit, which transistor need to do matching ???
How to see which transistor need to match ??? any rules ???

If PMOS can not match with NMOS, any method to match them?
Can i change the W/L ratio for matching so that the size can match?
For example: orginial is 18u/0.8u = 22.5, can i change to 4.5u/0.2u ?

 

[yan2004]

Re: Common centroid

Hi all, I'm aware of common centroid shall not be used for high speed circuits. Is there a suitable way to layout a moderate large width (say a W/L ratio of approximately 40) transistors for high speed circuits?

 

[cretu]

Re: Common centroid

Hi all, I'm aware of common centroid shall not be used for high speed circuits. Is there a suitable way to layout a moderate large width (say a W/L ratio of approximately 40) transistors for high speed circuits?

It depends on what you call 'high-speed'. What's the frequency your are talikng about?

Added after 8 minutes:

I am working as an IC designer for the last 10 years. I have seen at least 10 chips where the common centroid technique was used with very good results(very low offset and minimal mismatches). Like I mentioned in a previous message, youmight also want to consider 2nd order centroid if matching of that diff pair is very important

Added after 1 minutes:

100's of MHz should be no problem with the common centroid

 

[yan2004]

Re: Common centroid

Hi cretu, thank you for the reply. For high-speed circuits, I mean circuits that operate at GHz frequency of range. It seems at frequencies above/over 1 GHz, the parasitic capacitances resulted from using common centroid become obvious and deteriorate the frequency response.

 

[cretu]

Re: Common centroid

By the way..if you are interested in the IN-Out capacitances..I remember reading some articles about using some IN-Out capacitance for a diff pair which has as a result an increase in speed. That was also for signals with 3-400 mv coming into the differential pair

Hope this helps..I will try to point you to the papers I mentioned

 

[yan2004]

Re: Common centroid

Hi cretu, for the IN-OUT capacitances, do you mean Miller capacitances? If you can get me the references that will be helpful. Thank you.

Regards.

 

[cretu]

Re: Common centroid

Hi cretu, for the IN-OUT capacitances, do you mean Miller capacitances? If you can get me the references that will be helpful. Thank you.

Regards.

hi again, not the Miller caps I am talking about. These articles where actually having capacitances between gate and drain(for the NMOS) added extra by the designer to increase the speed of the diff pair. That means that you end up with something that will swing faster those 3-400mv. I don't remember though the price that you pay..but I suspect power consumption

 

[radram]

Common centroid

generally we do matching for current mirrors and diff pairs.

Added after 42 seconds:

but we dont match pmos and nmos transistors.

 

[cretu]

Re: Common centroid

generally we do matching for current mirrors and diff pairs.

Added after 42 seconds:

but we dont match pmos and nmos transistors.

Correct. That will be impossible almost to match they vary in a different way

 

[layes2]

Common centroid

COmmon is not needed in most design

 

[kumarans]

Re: Common centroid

Hai Guys,
I have one dout
which below mentioed pattern is good and avoid the gradient and offset voltage??????????

ABAB ABBA
BABA vs. BAAB
ABAB BAAB
BABA ABBA

Rgds
Kumaran.S

Added after 10 minutes:

Hai Guys ,
Here I have attached the basic PDF for the DIFF amp layout it would be useful for the Starters
Rgds
Kumaran.S