does anybody understand this POR circuit?

[incol]

por circuit

see the figure. this POR circuit looks like a bandgap, and it also has very small temperature effect. but I don't understand why it has very small temperature effect, is it relate to bandgap circuit theory? but there isn't a feedback to keep Vx=Vy, why can this circuit have the character of the Bandgap?

 

[DReynolds]

por hysteresis

incol, what controls the pmos in your circuit? It looks like the circuit is using deltavbe like a bandgap, but as you point out it does not drive Vx and Vy to be equal... in fact it relys on the fact that they are not equal to work. If I run too little current in the diodes then Vx-Vy will have one sign and if I run too much current Vx-Vy will have the opposite sign and I can tell if the power supply is up or not.


Hope this helps

David Reynolds

 

[AndrzejM]

The temperature effect is small because all diodes Q1 = 8 in parallel and Q1 have exactly the same temperature coefficients (say 2 mV/deg).

The current through diodes is small when it flows through R4 only and higher when the current is supplied also through MP1 , R5 .

Let us assume that R1=R2 ( I do not know if this assumption is right but there is no value on the drawing)
For low current Vx is < Vy (the current through single diode of Q2 is 8 times less than through Q1 so voltage must be lower) an additional voltage on R3 (I assume that its value is very low) is also not significant.

When MP1 change its state to a conductive one, the currents will increase, the voltage DIFFERENCE between Q1 and Q2 remains the same but the voltage drop on R3 will became significant hence Vx > Vy and this will change the comparator output.

 

[incol]

I am sorry that I didn't describe the circuit clearly. PMOS MP1 is used to generate hysteresis, and it didn't relate to temp effect. When we consider the reason of the small temperature effect, we can delete MP1, just consider R1, R2, R3, R4, R5, Q1 and Q2.
I think the reason of small temperature effect is that: small temperature effect means that the POR threshold voltage has a small temperature effect. When does POR threshold voltage generate? Only during Vx and Vy changed from Vx<Vy to Vx≥Vy. so what we care is when Vx=Vy, there is a small temperature effect. And when Vx=Vy, the circuit is a bandgap, so we get what we want. Is it right?

The temperature effect is small because all diodes Q1 = 8 in parallel and Q1 have exactly the same temperature coefficients (say 2 mV/deg).

The current through diodes is small when it flows through R4 only and higher when the current is supplied also through MP1 , R5 .

Let us assume that R1=R2 ( I do not know if this assumption is right but there is no value on the drawing)
For low current Vx is < Vy (the current through single diode of Q2 is 8 times less than through Q1 so voltage must be lower) an additional voltage on R3 (I assume that its value is very low) is also not significant.

When MP1 change its state to a conductive one, the currents will increase, the voltage DIFFERENCE between Q1 and Q2 remains the same but the voltage drop on R3 will became significant hence Vx > Vy and this will change the comparator output.

 

[jordan76]

Hi incol,

I agree with your analysis about small temp effect, but I doubt about the statement
"PMOS MP1 is used to generate hysteresis". Instead, I think MP1 is used to switch between two states(Vx<Vy and Vx>Vy) which is controlled by VRC.

regards,
jordan76

 

[incol]

Hi jordan76,

Yes, MP1 is a switch, at the power-on mode, MP1 is off, the effective resistance is only R4. When MP1 is turned on at the power-off mode, it causes R4 and R5 to be in parallel, the effective resistance decreases. This generates two different POR threshold voltage, so provides the hysteresis.

Hi incol,

I agree with your analysis about small temp effect, but I doubt about the statement
"PMOS MP1 is used to generate hysteresis". Instead, I think MP1 is used to switch between two states(Vx<Vy and Vx>Vy) which is controlled by VRC.

regards,
jordan76

 

[dreamteam]

That's right.
In fact the temperature behavior is due to the structure. The crossing point of this POR is very stable in temperature. Unfortunatly, the crossing point is for vdd=Vbg=1.2V. If you try to fix another value, the crossing point will have a lot of spread.

 

[dipswitch]

Hi Incol;

Your analysis is correct; think of it as a bandgap, but instead of applying feedback at the bandgap voltage to keep Vx=Vy, it does a comparison AT THE POINT of Vx=Vy. That comparison latches in the state when VDD is sufficiently high, giving you the "power-on okay" signal.

Note that this is a POR, so you never have the case of Vx=Vy as a steady state condition. It only switches at the point where Vx=Vy and effectively latches that state until the next power down or the next reset.

Also note that because the point of Vx=Vy is based on the Delta-Vbe (bandgap voltage), it is as stable across temperature as is the typical PTAT based bandgap (i.e., tempco of 0.08 mV/C).

Also note that by tuning R2 and R1, the circuit can trip at any voltage, not just VDD=1.2 as dreamteam suggested.

-Bill