MOS resistor for tunable CMOS_C filter design

[Junus2012]

Dear friends,

For sure no linearity is comparable to the real resistor like poly resistor,

However, authors are justifying the use of non-linear MOS resistor in fully differential circuit where second order distortion can be canceled.

In the same time a lot of work have been presented to linearize the MOS resistor, as an example the one below

A 1-V 450-nW fully integrated biomedical sensor interface system

This paper presents a 1 V 450 nW fully integrated bio-signal acquisition IC in 0.35 μm CMOS technology which includes a tunable band-pass filter, a variable gain amplifier, and a 12-bit ADC. The ultra-low power is achieved by using an energy-efficient system architecture and a novel tunable...

ieeexplore.ieee.org

or the one below

A 1-V 450-nW fully integrated biomedical sensor interface system

This paper presents a 1 V 450 nW fully integrated bio-signal acquisition IC in 0.35 μm CMOS technology which includes a tunable band-pass filter, a variable gain amplifier, and a 12-bit ADC. The ultra-low power is achieved by using an energy-efficient system architecture and a novel tunable...

ieeexplore.ieee.org

in the same time, technology also offer what so called High Resistance Poly Resistor (HRES),

I need your help to discuss about the advantages and disadvantages

Thank you very much

 

[arthorios]

Hi Junus2012,

It seems that you want to design a very low cut-off frequency filter. How low must this frequency be? Do you need HPF or LPF? Specify this.
For the second reference, the feedback resistor controls only the location of the first pole of the HPF transfer-function. You can see that from Fig. 5(a).
As for the distortion, remember that second-order distortion will always be suppressed by differential operation.

The high-value MOS-based resistors you are showing are usually referred to as pseudo-resistors or tunable pseudo-resistors. Some people like to call them "back-to-back" pseudo-resistors since they form two diodes with their anode connected together.

The main advantage is clear: you can get very large resistance (~TOhm) with relatively small transistors. Meaning that pseudo-resistors are very area efficient. So you can see that for your LPF (?) you would need a very small capacitor to get a cut-off frequency of a few Hz.

As for disadvantages we can cite:
- Non-linearity: since their resistance is based on the leakage current of a reverse-biased diode, the resistance will change accordingly. So you can only consider them "quite" linear for very small swings.
- Bulk-leakage current: very small leakage currents will result in large DC offsets.
- PVT: the resistance value can change as much as 2 orders of magnitude with process and temperature.

Hope this helps.

Regards

 

[Junus2012]

Dear arth

Thank you very much for your reply,
The thing is that I need to design LPF with tunable frequency from few hertz to 5 MHz, this would not be possible with normal resistors

Coming to the MOS resistor, yes you are right absolutely MOS resistor is not that good in linearity, but if I would implement a filter to cover the same range by using a gm-c filter then I will also end with limited linearity

I have tested the filter with the MOS linearized resistor presented in figure one and it shows good linearity and good tunability