Looking for bias circuit for micropower design

[puyeng]

about micro power design

i would like to ask if anybody know the suitable bias circuit (with startup) for micropower design. the standard topology (for strong inversiong) used now can give current about 45 nAmp, but there's a problem with the speed and stability. the SR+ and SR- are very far from being symmetric, the worse is that the settling time is very bad (about 2 ms).
is there any way to solve this? thanks!

 

[paulux]

Re: about micro power design

Since the schematics of your circuit is not available, my comment is just for reference. But 45nA charging current in 2ms seems to be reasonable because the startup current is too small. The startup time could be reduced by increasing startup current.

 

[ambreesh]

Re: about micro power design

Hi,
I have been trying to use a contantgm for generatin of bias current as low as 100nA. And I am using the formula given by Vittoz in his paper.

I=Thermal voltage * Ln(K) / R

The problem I encountered was when we designed for 100n it was giving 150n.
Are you using gm/Id methedology to design. We are also going through it.
Are you using EKV models for simulation, what simulator are u using.

 

[paulux]

Re: about micro power design

Hi ambreesh
If you want to get back 100nA, you have to fine-tune the circuit.
I heard gm/Id method and tried before but I just use square-law or simple weak inversion approximation.
Since I don't have EKV model, I have not used it. Hspice is my favourite simulator.

 

[ambreesh]

Re: about micro power design

Hi paulux,

Yes we had to fine tune it. But 50% increase in current was quite a setback.
I tried it in 2 different foundry and different techonology. And observed the same in both case. We are still trying to debug the issue. One was simulated in Spectre and other in HSPICE.
The papers we have read, do mention that weak inversion requires a good reference current.
Do you know of any material that talks on opamp design in weak inversion and also telling how input swing, output swing and other parameters could be predicted.

 

[paulux]

Re: about micro power design

May I know which papers do you refer in your design? Or do you mind to provide your schematics? The problem may not be due to opamp.

I forget the exact title of the paper at this moment .....its title is .....
"A Fully Differential 1.5v Low-power CMOS Operational Amplifier ....."from ieee
If u can find the exact title, please also let us know.

But I am not sure whether this amp suits your micro power design requirement.

 

[puyeng]

Re: about micro power design

Hi ambreesh,

we actually don't really use gm/Id method in calculating the aspect ratio, we just use it for estimation on how the parameters can be increased/decreased to get the spec wanted.
we use HSPICE. it is very difficult to match the calculation result to the simulation result. so after calculation, usually we always estimate ourselves and simulate repeatedly. the result of device characterization seems very different from that of simulation.