Subthreshold Two Stage Op-Amp Vbias (V_REF1 & V_REF2)

Hi,

The text says:

Once appropriate aspect ratios were
determined for all the MOSFETs, the DC bias voltage VREF1
was set based on the gain considerations after performing a DC
analysis on the first stage of the op-amp. VREF2 was
subsequently set to meet the desired gain of the overall
amplifier.

Click to expand...

I guess you or better say your application decides the gain requirement.

Klaus

KlausST said:

Hi,

The text says:

I guess you or better say your application decides the gain requirement.

Klaus

Click to expand...

Hello KlausST, I appreciate your answer. Where should I connect the V_Ref2 and V_Ref1? Should I bias both of the terminals? Or should I connect it to Vdd and Gnd respectively?

Hi,

The text says
* you need to consider the (required) gain. (We can not know this)
* you need to do a DC analysis.(it makes no sense that anybody else than you does the DC analysis)
Then you will find out Ref voltage.

I guess it is somewhere inbetween GND and VCC.

An experienced IC designer may give you more detailed informations.

Klaus

Those FETs are used as current sources as I see (M5: diff pair. tail current source, M6: active load of 2nd stage).

So don't connect VREF1,2 to supply rails, M5,6 won't saturate, they going to operate as small resistors.
Connect them to diode connected FET-s rather.

The diode connected devices will generate bias voltages for the M5,M6 current sources (=Vgs,Vsg) if you connect a constant current to the diode connected device.

And if you want to save some time and pain connect M7 gate to VREF1 and M6 gate to output of the 1st stage.

To keep M7 in saturation a relatively low potential should be on its gate, M6 need a potential closer to VDD for the same reason.

About biasing, CMOS OPAMPs you can learn a lot from Behzad Razavi's book, Design of CMOS analog integrated circuits. Good luck!

The circuit you attached has nothing directly to do with it being biased in subthreshold. It is probably the most common type of two stage amplifier out there with Miller compensation.

If being used in subthreshold the size and bias of the transistors is what matters and that isn't given.

I suggest looking, as frankrose suggested, in Behzad Razavis book for reference. And if you search some more you will find papers where more detail is given for the circuit which might give you some more insight.

Also think about why you want it in subthreshold. Do you need lower power? Lower supply voltage? A lot of performance parameters gets a hit when an amplifier is biased in subthreshold. Some might not be obvious at a first glance such as matching.

Since you say it will be used in a bandgap and you might not have a current reference, read up on constant-gm circuits.