Need suggestion on a discreet ldo design.

[opa627bm]

Hello everyone,
                     I got assigened to design a voltage source that is capable supplying +-100ma at -+10V. After I build the circuit, I notice that there is some fluctuation at the output side. I probed with scope and there is no oscillation at the gate or opamp. The Vref into the opamp is extremely stable 0.70160X volts. but my output fluctuates around 1mv. The mosfet is FR5305. I need some advise on getting this thing more stable....

Thank you!

[opa627bm]

It actually drifts around 1mv per 20 seconds 
halp....

[KJDS]

What are the three op-amp pins doing as the output varies?

Just hook up four meters on those points and plot a graph with time. The answer to what is happening will then be obvious.

[ohmineer]

You could add a capacitor  to the output so as to stabilize the loop. Values of 4.7u to 10u should be fine.
For everything else, a stability analysis would be required. Have a look at this application note (http://www.ti.com/lit/an/slyt194/slyt194.pdf) of TI just in case you need some help.

[David Hess]

You have a feedback loop stabilization problems here.  The common source MOSFET adds voltage gain inside the feedback loop and presents a capacitive load to the operational amplifier.  Operating with gain helps but at the very least, an output capacitor is needed but this is one of those cases where oscillation is likely if the ESR is too low.  Similar integrated voltage regulators designs often place a resistor in series with the output capacitor to control this.  Pole-zero compensation with a series capacitor and resistor from the operational amplifier output to the inverting input will help but you will have to modify your reference filtering network.  The circuit would be easier to compensate if the gain was fixed and the reference voltage was adjusted.

Compensation can be checked by doing a transient response test.

Since this is a low dropout design without a bias supply for the MOSFET gate, I assume you drew the arrow for the p-channel MOSFET backwards.

From a noise and drift perspective, the OPA227 is the right amplifier for the impedances shown but its high bandwidth exasperates the feedback stability problem.  An OPA177 which is slower might be better and would allow easier reference filtering.  I like using LT1012, LT1097, and OP97 style operational amplifiers because they allow easy adjustable overcompensation but I am not sure if TI has an equivalent.  Their even lower input current would allow for easier reference filtering.

If precision is important, then leakage in the reference filtering capacitors may be an issue but the film or ceramic capacitors you show should not be a problem.  Their their small size means they are not doing anything for low frequency noise though.

I would probably put the variable resistor in the low end of the voltage divider so if it fails open, the output voltage drops to the minimum.  Grounding one side of a potentiometer often has noise advantages although that probably does not apply here.

If the output current is not high, I would use an NPN output transistor with a unity gain level shifting network.

As shown your circuit can only source current.  If you want +/- 100 milliamps, then you need a different design.

[opa627bm]

You have a feedback loop stabilization problems here.  The common source MOSFET adds voltage gain inside the feedback loop and presents a capacitive load to the operational amplifier.  Operating with gain helps but at the very least, an output capacitor is needed but this is one of those cases where oscillation is likely if the ESR is too low.  Similar integrated voltage regulators designs often place a resistor in series with the output capacitor to control this.  Pole-zero compensation with a series capacitor and resistor from the operational amplifier output to the inverting input will help but you will have to modify your reference filtering network.  The circuit would be easier to compensate if the gain was fixed and the reference voltage was adjusted.

Compensation can be checked by doing a transient response test.

Since this is a low dropout design without a bias supply for the MOSFET gate, I assume you drew the arrow for the p-channel MOSFET backwards.

From a noise and drift perspective, the OPA227 is the right amplifier for the impedances shown but its high bandwidth exasperates the feedback stability problem.  An OPA177 which is slower might be better and would allow easier reference filtering.  I like using LT1012, LT1097, and OP97 style operational amplifiers because they allow easy adjustable overcompensation but I am not sure if TI has an equivalent.  Their even lower input current would allow for easier reference filtering.

If precision is important, then leakage in the reference filtering capacitors may be an issue but the film or ceramic capacitors you show should not be a problem.  Their their small size means they are not doing anything for low frequency noise though.

I would probably put the variable resistor in the low end of the voltage divider so if it fails open, the output voltage drops to the minimum.  Grounding one side of a potentiometer often has noise advantages although that probably does not apply here.

If the output current is not high, I would use an NPN output transistor with a unity gain level shifting network.

As shown your circuit can only source current.  If you want +/- 100 milliamps, then you need a different design.

Thanks for the response David. Yes I did draw the arrow backwards lol. It can source or sink because the ground point is in the middle of the two batteries so when the output is below 12V, it sinks current to the other battery ... I have tested the circuit with dummy load and it does output +- 100ma... Thanks for all the suggestion!

[opa627bm]

So I did a long term stability test and it drifts alot..... nooo

[free_electron]

heating of the resistors.
why do you pull the resistors to -12 volts ? (pot and the 270 ohm…)
if you turn the pot to 0 ohms that 270 ohm resistor will get VERY hot !

any current flowing throught e feedback loop will cause the resistors to warm up. they thermally expand : the wiper moves a bit and the thing drifts.

make the feedback a static divider and change the set voltage going in to the opamp.

[opa627bm]

Strange....... I replaced the opamp with OPA606 and 602 and the circuit doesn't work at all .....

[opa627bm]

heating of the resistors.
why do you pull the resistors to -12 volts ? (pot and the 270 ohm…)
if you turn the pot to 0 ohms that 270 ohm resistor will get VERY hot !

any current flowing throught e feedback loop will cause the resistors to warm up. they thermally expand : the wiper moves a bit and the thing drifts.

make the feedback a static divider and change the set voltage going in to the opamp.

you are right... middle is opamp, far right is 399 and left is the 270