Bandgap with opamp give a wrong voltage after using resistor from TSMC40

[Maoxuewang1]

Hi

I have designed a bandgap reference circuit with an opamp on Virtuoso cadence. The bandgap reference works well when I use ideal resistor. However, after I replaced the resistor with rnpoly from TSMC40, the reference voltage went wrong (like from 700mV to 900mV). This error occurs when I used 'vdc' in analogy lib as my VDD. But if I used step voltage (from pulse) as my VDD, the reference works well.

Do you have some ideas about it?

 

[KlausST]

Hi,

Opamp common mode input voltage problem?

Klaus

 

[dick_freebird]

Working with a transient-ramped supply, but failing at
DC, often indicates that you have either not built in an
explicit startup circuit, or it needs work.

Real resistors have capacitances and maybe this has to
do with startup "catching" or not.

A well designed bandgap reference will start every time
with a DC sweep, a DC .OP or a transient ramp on the
supply.

 

[Maoxuewang1]

Yes, the dc current of two path is not the same in opamp

 

[Maoxuewang1]

Thanks for your reply.

I have a startup circuit in the Bandgap Reference.
I think if it is startup problem, then the current flow into the triode should be zero. Inversely, in this case the current of bandgap is high. Is it also a start up problem?

Actually, the problem occurs even I just put a floating real resistor into the schematic . That is quite strange. I find this problem when I connected the reference to my main circuit. Then, I find if there is any real resistor in the circuit, the circuit goes wrong(by the way real cap is totally OK). But if I connect it to my main circuit with ideal resistor, it works well.

 

[Maoxuewang1]

The transient simulation failed when there is a floating real resistor. I guess it is a problem of resistor model. To solve the Kirchhoff function for a floating resistor, the simulator just give some default value of the resistor's node, and this value makes a wrong transient response. So in this case ramp VDD supply is better for simulation than DC?