OPA548: a.Is the part sensitive to power supply sequencing?

Hi Robert,

In general, op amp designs be it the OPA548 or another, are not intended to experience power supply sequencing when dual supplies are being used. When we are using dual supplies in the lab, we use ones that come up to both supply levels very close in time to each other. The reason that we want this is the op amp internal circuits are not designed to have one supply turned on before the other. If just one supply is on the internal circuits are not correctly biased. When that occurs, the electrical isolation between components on the op amp die may not occur and undesirable parasitic latch structures can be developed.

These parasitic structures may be of the form of a 4-layer semiconductor device. You can think of these as SCRs. If they become present, the SCR structure may not be eliminated as the other supply comes up to level. Very high current can flow in some cases and the op amp doesn't attain normal operation.  Some op amps are fine with supply sequencing but much depends on the circuit design and their semiconductor processes.

How exactly an op amp behaves when the supplies come up at different times, or different sequences, usually isn't characterized because the assumption is the dual supplies will come up very close in time with each other. And it can even be different from one production run of a particular op amp to another due to minor process variations.

During an abort power down, we do see some oscillations on the output while +V supply is off and the -V supply is still discharging.  From what we have captured on a “good” part (one that’s still functioning) is about a 2V-3V (with peaks to 6V-7V) while V- only is applied (~7V-8V).  Is this still considered within the SOA condition?  The primary windings of our transformer is only 200 mΩ of resistance.

Because things internal to the OPA548 are in power down normal current paths begin to shut off. The normal internal schematic of the OPA548 ceases to exist and some other form of circuit momentarily exists. I expect the oscillation is occurring in a temporary local loop, and it is not related to phase margin issues that can occur in a normally operating op amp circuit.

The SOA question comes down to how much current is flowing through the OPA548 output transistors during the oscillation and shutdown period. You mention 6 to 7 volt peaks, but it sounds like the oscillation is a very momentary condition. Likely because the OPA548 is shutting down the power dissipation isn't anywhere near what the power op amp delivers when it is delivering 5 Amperes into a load where half the supply voltage is dropped across the load, and the other half dropped across the active output transistor. That is a high power, high dissipation condition.

We suggest that when there is an unavoidable power supply sequencing issue that transient voltage suppressor (TVS) diodes be added to the supply lines. If one supply lags the other the TVS will provide a return current path for the supply that comes up first. Then, as the other power supply comes up to full value the TVS diodes become reverse biased. You can see this concept in the image shown below.

We have found that adding the TVS diodes to the supply lines really helps in thwarting unexpected lath conditions.

Regards, Thomas

Precision Amplifiers Applications Engineering

How long has the addition of TVS diodes Zs1/2 been known about and was there failures in the field like what we are having that drove that design suggestion? 

Thank you. The TVS diodes were not implemented in the design. We are looking at an anomaly during power supply shutdown of the +/- 25V rails that causes an oscillation on the output. The opamp is enabled via the -25v. We are concerned if this is causing the bond wires of out put to -25v to fuse open .  It was mentioned that the PS should shut down power rails at near the same time. Our ramp rate for the +25on a scope plot  starts at -.11s +25V down to near zero at -.01sand negative rail starts at -.078sat -25V to near zero at +.018s. So when the +25V rail is near zero there is still -10V on the Negative rail. Between the time that positive rail reaches zero and the negative rail reaches zero there is an oscillation occurring on the output between time -.003s -6V and .008s-4V with an amplitude of up to +/8v duration 11ms.Do you think this is significant enough to fus open bond wires? 

Again the schematic is very similar to your evaluation board SBOU132.

Hi Robert,

It's not easy to say something useful without seeing the circuit. I would take a fresh OPA548 and would make some measurements of the output current and the current through the protection diodes.

It's not unusual that an amplifier oscillates at the output when the supply voltage goes down. We see this in audio circuits all the time and the best remedy is to increase the supply voltage decoupling capacitances.

But in your case I think there's something different going on. If you abruptly shut down the output of OPA548 while a big current is flowing through an inductive load, a heavy inductive kick can occur which can generate a hugh voltage at the output of OPA548. Having freewheeling diodes from the output of OPA548 to the supply rails is a good idea, but if the currents are not shunted to signal ground via a TVS or at least into a big supply voltage decoupling capacitance, the OPA548 might be destroyed.

Unbalanced supply voltages during power-up and power-down should not be an issue, provided that the input voltages do not exceed the supply voltages by more than 0.5V. If this is cannot be excluded, you should think about the use of an input protection scheme.

In any case, I would mount the TVS at the supply voltage pins of OPA548, as already suggested by Thomas and I would increase the supply voltage decoupling capacitances of OPA548.

Can you show a detailed schematic of your application?

Kai

Hello we have some new scope shots which we can't show you but may answer the question. Here is the description:

1. the +25V supply is shut off and -25V is still on. The voltage at the output of the opamp is zero (going through coil of transformer). Op Amp is enabled. Primary side of transformer connected. Current increses to about 1.3 amps that the ILIM is set to.Then the output voltage goes above the +25V rail which was turned off and at zero. All of a sudden the Output voltage drops to a negative voltage then the sinking current  to the opAmp goes to 5 amps as if the ILIM stopped working because the output went above the +25V rail?