start-up circuit for widlar bandgap

[chu]

widlar bandgap

hi all,

i want to implement a start-up circuit for widlar bandgap.
i know that V2 must be at least one Vbe volt so that Vref will be stable to 1.4V. Two condition when V2=Vbe.

1) when i ramp Vcc from 0V, there is a state where QG16 still in cutoff and V2 follows Vref and V2 reached Vbe.

2) when Vcc continue ramp up, QG16 and QG15 fwd active, thus V2=Vbe, and Vref stablize.

how do i implement start-up to by-pass 1st condition ??

thanks in advance.

rgds,
chu

 

[biff44]

widlar bandgap circuit

Try connecting RG5 5KΩ resistor to Vcc instead of Vref. You might have to make the resistor value a little larger to maintain the same quiescent bias point.

 

[10kangstroms]

bandgap initial condition start up

Isn't LP07 supposed to be an NPN?

 

[chu]

start up circuit widlar

hi 10kangstroms,

ya, u r right, QG21 suppose to be NPN, but i also tried PNP. From simulation, PNP is giving slightly more current compare NPN.

can u tell me what is the effect on these difference ?
u hav any idea on the start-up ? thx again.



hi biff44,

thx. i don't think i can connect vref to vcc because my vref is an output of that widlar circuitry. And vref is connected to a diff-amp (with -ve feedback) & emitter-follower.

do u have other method ? thx a lot.


rgds,
chuliang

 

[10kangstroms]

bandgap widlar

hi 10kangstroms,

ya, u r right, QG21 suppose to be NPN, but i also tried PNP. From simulation, PNP is giving slightly more current compare NPN.

can u tell me what is the effect on these difference ?
u hav any idea on the start-up ? thx again.
rgds,
chuliang

With QG21 as NPN, you have a negative feedback loop with QG10 (correct). With QG21 as PNP, you have a positive feedback loop (incorrect). I simulated it both ways. I could not get it to work with any startup circuit or initial condition with the PNP (this was before I discovered the "error"). With an NPN, I ramped the supply from zero to 5 volts, and it worked correctly. I don't see why you think you need a startup circuit.

 

[chu]

hi 10kangstroms,

u r right. i had also simulated. It's look fine now without startup. thanks to u.


rgds,
chu

 

[liuyonggen_1]

hi 10kangstroms,

ya, u r right, QG21 suppose to be NPN, but i also tried PNP. From simulation, PNP is giving slightly more current compare NPN.

can u tell me what is the effect on these difference ?
u hav any idea on the start-up ? thx again.
rgds,
chuliang

With QG21 as NPN, you have a negative feedback loop with QG10 (correct). With QG21 as PNP, you have a positive feedback loop (incorrect). I simulated it both ways. I could not get it to work with any startup circuit or initial condition with the PNP (this was before I discovered the "error"). With an NPN, I ramped the supply from zero to 5 volts, and it worked correctly. I don't see why you think you need a startup circuit.

can you tell me why the positive feedback is formed with PNP of QG21 ?
thank you!

Added after 8 minutes:

hi 10kangstroms,

u r right. i had also simulated. It's look fine now without startup. thanks to u.


rgds,
chu

In my opinion ,you should design startup circuit to make sure that your reference can work in any condition even though you havn't met any problem when you simulated it

Added after 8 minutes:

hi all,

i want to implement a start-up circuit for widlar bandgap.
i know that V2 must be at least one Vbe volt so that Vref will be stable to 1.4V. Two condition when V2=Vbe.

1) when i ramp Vcc from 0V, there is a state where QG16 still in cutoff and V2 follows Vref and V2 reached Vbe.

2) when Vcc continue ramp up, QG16 and QG15 fwd active, thus V2=Vbe, and Vref stablize.

how do i implement start-up to by-pass 1st condition ??

thanks in advance.


can you tell me the role of the QG20 ?for biasing?

rgds,
chu

 

[chu]

hi 10kangstroms,

ya, u r right, QG21 suppose to be NPN, but i also tried PNP. From simulation, PNP is giving slightly more current compare NPN.

can u tell me what is the effect on these difference ?
u hav any idea on the start-up ? thx again.
rgds,
chuliang

With QG21 as NPN, you have a negative feedback loop with QG10 (correct). With QG21 as PNP, you have a positive feedback loop (incorrect). I simulated it both ways. I could not get it to work with any startup circuit or initial condition with the PNP (this was before I discovered the "error"). With an NPN, I ramped the supply from zero to 5 volts, and it worked correctly. I don't see why you think you need a startup circuit.

can you tell me why the positive feedback is formed with PNP of QG21 ?
thank you!

Added after 8 minutes:

hi liuyonggen_1,

example: if voltage at base of QG10 increase, at collector QG10 (npn) decrease, this also mean base of QG21 (pnp) decrease, thus collector QG21 increase, this also mean increased the base of QG10 some more, so this loop is a +ve feedback.

if QG21 replace with npn, then it stablize with -ve feedback and i will not get vref stuck to zero-bias point.

thanks,

rgds,
chu

 

[10kangstroms]

In my opinion ,you should design startup circuit to make sure that your reference can work in any condition even though you havn't met any problem when you simulated it

Why? Do you think op amps have a startup circuit? You don't need one unless the circuit has two stable operating points, and one of them is undesirable. I can't see two stable operating points in the (corrected) Widlar bandgap circuit.