Opamp with out an external bias current source!

[findsriharsha]

I am designing an opamp for super low power application. I am not getting a fixed bias current (generated using bandgap) to bias up the amplifier. I am trying to use pmos mosfet stack in triode region (acts as resistor) to generate bias current internally. The problem with this current is it varies ~100% across PVT and so does my gain! Does anyone, who has come across similar situation, knows a workaround/better solution, please let me know..Thanks in advance!!

 

[dick_freebird]

I think the simplest you can do is a delta-VT PTAT loop,
4 transistors and a resistor. But if your supply span is
large relative to VT then a simple large value resistor
could be "good enough". A 4 transistor + 2 diode +
resistor loop will make you a "low quality" bandgap PTAT,
one more resistor and you can have a voltage reference
of "housekeeping grade".

MOSFETs in triode make lousy resistors by any measure.
Their sole virtue is "cheap".

 

[electronrancher]

You don't really want a fixed current over temp, you want a current that keeps your transconductance fixed over temp. That should fix your gain problem. Vt (or d-Vbe) is the right choice for BJT input stage, dVgs is the right choice for MOS input stage. Dick's advice is a good one, although there are many methods to creating a PTAT current with just a few transistors. If you have good resistors in your process it should be easy. If not, you can build a bit of a custom tempco source using what you have available. Often, oscillators need a weird tempco to make them flat over temp so it's not unusual for me to generate one custom tempco current just for that block.