Help me size input transistors for a differential amplifier

[suhas_shiv]

Hi all.

I am designing a differential amp(both Pmos input and nmos input transistors with nmos diode connected loads and pmos diode connected loads respectively, like an OTA). How do i get to know the W/L of the pmos diode connected loads(for Nmos input diff amp) and nmos diode connected loads(for pmos input diff amp)? The expected gain is around 5-10. Power dissipation is around 3-5 mW).
0.25um TSMC
Supply 3.3V
ICMR 0-2.4V
Minimum input differential level 100mV

It would also help if you could explain how you would size the Input transistors(I have done that part, and now trying to optimize the W/L for diode connected loads).

Appreciate the help.

 

[osha]

Re: Differential Amp

you can go site
www.analog.com

 

[gszczesz]

Re: Differential Amp

There are several equations you need to balance that most books (like Grey + Meyer) will give details on. Being an IC designer, I can tell you how I would get to the first-cut rough sizing.

Firstly, the W/L ratio of the diode NMOS should be such that the net current density is approx. 1uA/square. So if your width is 2, and your length is 1, this diode is perfect for 2uA of current. If you use less current, you start increasing the offset. Using more current makes you loose voltage headroom. 1uA is a very approx. rule of thumb for a first-cut that balances offsets and voltage headroom, and yes, the current density does affect how much a "weighing" an offset voltage gives to the final output current.

Given a chosen current for bandwidth + slew rate considerations, I would next choose the combined W*L size of the PMOS and NMOS such that each contribute roughly equally to the offset. The target offset will dictate what that size is. For example, suppose you are shooting for 2mV 1-sigma offset. I would therefore size the PMOS and NMOS to each give me 1.414mV of offset.

The input diff pair now needs to be determined the W and L proportions for the given area. Generally, the smaller the L the better the gain and headroom (given that you have a diode connected pair on top), so I would size it for the target gain, assuming headroom is not an issue. Keep away from very small current densities. Typically one goes below the 1uA/square density for the input diff pair, but don't go to extreme values (like 0.1).

Hope this helps.

Greg

 

[hi_hedonist]

Re: Differential Amp

you may cosult anybook of analog electronics aur opamp by gaykward

 

[beckchm]

Re: Differential Amp

you must know the direction to do the work

you can refer some books to get it