How to design a large resistorMeg Ohm in IC chip?

[leonken]

for examle, I need a serveral Meg Ohm resistor. If we do not consider the accurate do you have some good approaches to make it in IC chp?

 

[Hughes]

If you do not consider the accuracy, you may use active resistor, a mosfet biased in linear reagion.

 

[selvaraja]

Hi

can use diffusion resistor if resistor value big but accuracy is trade offf

regards

selvaraja

 

[sunking]

Hi poly resistor or well resistor

 

[tlihu]

There is HR (High Resistance) poly, usually 1k/sq.

 

[IanP]

Better then MOSFET will work FET transistor as from its nature it is electrically controlled resistor. And for you it probably does not matter if it is in linear or non-linear area as long as you can achieve required resistance... 8)

 

[Hughes]

Poly and diffusion resistors are process related. In mixed-signal CMOS process, poly resistors with 1kOhm/sq. sheet resistance may be available. But in submicron standard logic CMOS process with salicide, the sheet resistance of poly and diffusion may be a few ohms/sq.

 

[xinyin]

It depends on your process. Check with you process manual to find proper resistor, or active resistors can be used.

 

[rambus_ddr]

This resistor is too large to use a real reistor, which have large capacitor and area, i think it should be an active reistor, such as a mos transistor.

 

[sunking]

An other method is R=T/C, using switch cap.

 

[cetc1525]

"Meg Ohm" is very large in IC design.we don't suggest designers use a real resistor.did you think of using a current source ?

 

[extraord]

current source has many limits, I don't think it what the poster really want.

 

[tedchen]

Usually, the N-Well got the highest resistance per square but worst accuracy so not many people wanna take that risk. The alternative would be poly resistors. According to TSMC documents, the Rsh of P- poly w/o silicide can goes up to around 1k ohms/sq while P+ poly w/o silicide is only slightly over 300 ohms/sq, for 0.18um. BTW, N+ poly w/o silicide got 200 something, nearly 300 ohms/sq.

About how to layout it out, there are couple books talking about resistor layout, capacitor layout, etc. You can probably check the book "The Art of Analog Layout" first.

About the size, if you are going to use poly resistor, I don't think the size will be too big compare to capacitors or inductors. I personally laid out couple 5x MOhms resistors but the capacitor wes the biggest area consumer.

 

[atmaca]

In my opinion the best method is to use switched capacitor circuit. R= delta t / C. As in the formula you can change the resistor value by changing switching period.

Thanks

 

[devrimaksin]

N-well is the layer to go

 

[terryssw]

Depends on applications, I think switched-cap ones can only used when discretized mode of resistance is required (which means we must have some clock)

 

[bastos4321]

I would recommend a switched-capacitor solution. Is the best in term of area for a Meg resistor.

Bastos

 

[Fom]

N-well is not the best option because pitch (min width + min space) rule too large. Usually I use the following figure of merit: min one square area divided by sheet resistivity. After that you can easily estimate resistor area. Just multiply this coefficient by resistor value.

 

[cetc1525]

I just want to say that this resistor must occupy a large area of the layout.so, I think your structure must has some problems.

 

[leonken]

1.switched-capacitor need another clock circuit
2. and if I need a big resistor, the frequency should be smal. So, how to generate a large period signal ? a big resistor or capacitor maybe needed.
3. If I need several resistors in series. The switched-capacitor approach also works?
thank you!