Dummy capacitor for MiMCap

[hrkhari]

Hi Guys:

I find that placing dummy capacitor surrounding the MiMCap is quite area consuming, is it possible to surround the MiMCap with dummies significantly much more smaller (W X L) than the MiMCap itself.

Rgds

 

[yibinhsieh]

mim dummy

Yes, I usually do it. It will save much area. And the matching is acceptable.

Yibin.

 

[layes2]

mim capacitor dummy

not use the same cap
just use part of it is enough

 

[hrkhari]

capacitors for dummies

Hi Guys:

I'm quite confused, shall I use the same value of dummy capacitor as suggested by layes2, or smaller values as suggested by yibinhsieh. By the was layes2, What do you mean by just using part of it?, pls clarify.

Rgds

 

[showtime]

edge matching capacitor

Of course you can use smaller one, it doesnt matter

 

[nafaiz]

dummy structure capacitance

usually i used dummies with smallest DR width but the length follows ur MIM structure.

 

[uncle_urfi]

dummy capacitance

All you need to do is to match the boundary conditions on all sides of the 'actual' capacitor. It doesn't have to be surrounded by caps of the same size. A small strip would do just fine.

 

[alanbrooke]

mim cap dummy

Normally, If I use 10*10um MIM, I will place some 2*10um dummy MIM capcitor by the side of the capcacitor .

 

[xinsu]

dummy of a capacitor

Sometime i will use the same cap,and sometime i will use small one , this is determined by my moule's floorplan.

 

[drasta]

drumy cap

I think that you can for sure use half of W around with very good characteristics.

 

[Btrend]

dummy capacitor

I think that you can for sure use half of W around with very good characteristics.

Is there any theorem about ur suggest?
if i want to do a more accurate matching, said 20bit , how should I do with the dummy cap ?
there should exist some basic rules about this matching issue in quantitative view, not just qualitative view , right?
thanks

 

[blueant]

Capacitor Dummy

I used same high mim-width

 

[hrkhari]

Re: Capacitor Dummy

Hai Guys:

Thanks for the beneficial input. Really appreciate it.

Rgds

 

[JPR]

Re: Capacitor Dummy

The sizing and need for dummy devices depends heavily upon the requirements, and on the array edge effects that your process causes. In some cases, no dummy may be needed to provide sufficient matching (dominated by device matching over array edge effects)

In slightly more edge sensitive cases, a ring around the array may be all that is needed.

In even more sensitive cases, partial dummy devices may be needed to provide exactly the same adjacent topology as the main array.

In extreme cases, multiple dummys may be needed around the array in order to suppress the effects fully. I would guess that if you need 20 bit matching, you would be in this extreme case. Your dummy array may even end up being larger in area than your device array, and that still might not be sufficient.

There are alternatives to dummy devices that you may wish to consider. If you have some capacitors that do not need to match as well (due to being averaged with many other devices), you might be able to place an edge device in this portion of the array, for example.

You could also rotate how each device is used in the array, thus using average capacitor values, rather than specific values. Even this might not make the 20 bit accuracy.

 

[hrkhari]

Re: Capacitor Dummy

Hi JPR:

Do you mean we can use for example an ac ground capacitor, which does not require much accuracy as a dummy for other on chip coupling capacitor. Kindly confirm

Rgds

 

[layes2]

Re: Capacitor Dummy

you can use thin and small cap

 

[JPR]

Re: Capacitor Dummy

What I was trying to say is that the requirements for dummy capacitors are determined by the accuracy needed on the primary capacitors.

I would typically connect both plates of the dummy capacitor to ground, but this will ultimately be determined by what is happening on the other capacitors in the array.

The purpose for the dummy capacitors is to create "equal" surroundings for your primary capacitors. It is simply a question of how equal things need to be to obtain the accuracy that you desire.

I was trying to point out that I would not use dummy devices if I need to match capacitors to 1% of the total capacitance (7 bits or so...). 8 or 9 bits could be done without dummys, but will be easier with dummys. 10 bits or more will most likely need dummys. By 13 or 14 bits, additional dummys, and special care must be taken in layout to keep consistent surroundings for each capacitor. In order to obtain additional accuracy, additional steps must be taken.

I would use the same structure for the dummy capacitors that I am using for the array capacitors. If I only need a little help (8 or 9 bits), I would use smaller capacitors for dummys. If I need to make 13 or 14 bits, I would likely surround each unit device with a shield. This will increase the unit capacitance and will produce equal surrounding for each unit capacitor. I would probably also have a row of dummys outside of the array.

Dummys alone do not provide all the accuracy. You will still need common centroid layout, for example, or all of the effort to add dummys will be lost by gradients.

You, as a designer, must decide whether dummys will help or hurt. Use the matching per total area as a measure of how good various approaches are. There should be matching data for your fabrication. If you are lucky, there will be some data that will demonstrate the additional effects, such as edge of array effects or gradients.

In one process, the gradients and edge effects might be small, but matching not so good. In this case, you want to use your area to make large capacitors, and skip the dummys.

In another process, the gradients and edge effects might be larger, but with good matching on adjacent capacitors. In this case, you would want small capacitors to keep them close together, with dummy capacitors and shielding rings around the array.