Hi Team, in relation to my previous question, this...
# analog-design
j
Hi Team, in relation to my previous question, this is the schematic of the 2 stage op amp that I'm doing. Do you have any suggestions to better improve this design?
l
Use a current source instead of resostor for the biasing circuit. In a real circuit, we have biasing currents from other circuits.
Use multipliers for your transistors. Try to avoid different lengths and widths for current mirrors. Multipliers also help with layout, since you will need to use techniques such as common centroid and interdigitation.
The output current of the second stage is too low, as one can infer from current mirrors. Miller compensation only works properly if the second stage gm is much higher than the first stage. If you want to keep low currents at the second stage, you will need to use at least a resistor in series with the miller cap.
j
Hi Luis, thanks a lot for the feedback. In your first suggestion, how do I calculate what current source I should create to mimic the same resistance?
l
You don't need to. It was already measured as 22 uA.
j
What I meant was instead of a resistor, you suggested a current source right? How do you suggest I should do that? I assume its a self-biased transistor?
l
There is a current source symbol such as the current source symbol
https://github.com/lhrodovalho/sky130_rodovalho_ip__amp/ Look for ampc_tb.sch. It is a small-signal testbench for a two stage amplifier.
You should make a subcircuit for your amplifier and use pins. Your biasing current should be external and use an dedicated pin.
j
Yes I'm aware of the symbol. I guess the concern is that I will also layout this schem on Magic, and for practical uses I used a resistor so I can lay this out on magic as if it was for fabrication.
l
Then you should design a constant-gm biasing circuit.
message has been deleted
j
Thanks for sharing this Luis
Just curious, what's the disadvantage of using a resistor instead of a current mirror? Aside from temp dependency.
l
I don't know what this current mirror biasing you're mentioning. The resistor only connected between the positive supply and current mirror input is a bad idea because any error in the typical supply voltage will change the biasing current and the amplifier bandwidth. Its power supply rejection will be decreased too. At least it's a simple solution and doesn't have start up issues.
j
Thanks for the inputs Luis. The current mirrors I'm mentioning is the self biased current source that you've shared.
So if i understand it correctly, this picture is the constant gm biasing circuit that I can replace with my resistor?
l
You can replace your resistor with that