OPA548: output latch and triangle output

Nate,

I am adding this information to that you previously provided.

Regards, Thomas
Precision Amplifiers Applicaitons Engineering

Hello Thomas,

Please see my comments and questions.

- Parts are heat sunk onto a large piece of Aluminum and can set finger on after a long period of time (once thermal equilibrium is achieved the part temperature is between 50 - 60C. Doesn't explain the latch state that the part goes into?

The OPA548 junction temperature operating range in normal usage should extend up to 125°C. Datasheet sections 11.3 through 11.5 cover the thermal requirements quite extensively. If the OPA548 die temperature reaches approximately 160 °C, the output current shuts down until the die cools sufficiently. The triangulation of the output waveform is more indicative of the edge rates being limited by reduced output current and not a full shutdown.

- As far as transient causing the part to go into this state.... We have injected high levels of ESD using a ESD gun into the output of the device loaded and unloaded as well as both polarities upto 15kV. No problems.

We haven't subjected the OPA548 to in-circuit HV ESD.

- We have tried to create a lot of conditions to try and force this device to go into this latch state and can not do it. We have set inputs to have voltages when no power supply voltage is on ....doesn't cause the issue.

Piezoelectric loads present a huge capacitive load at the amplifier output. There may be an inherent stability issue with the circuit design, which is something we are going to assess. Even if the phase margin isn’t sufficient to assure stability that likely isn’t the cause of the output triangulation. A piezoelectric load can produce a significant back EMF if they are subjected to mechanical shock. The lack of output protection diodes in the output circuit as shown on datasheet Figure 38 can result in charges accumulating in the OPA548 die substrate. If that happens, the normal internal circuit operations can be disrupted. It is important to know if adding the diodes corrects the behavior. Note there also needs to be an appropriate transient voltage suppressor connected from the supply pin, to ground, for the protection diodes to do their job.

Originally the piezoelectric load was described as a 16 nF load. However, there is a TINA schematic showing the capacitance is 36 nF, shunted by a series L, C and R. Which one is the correct load? Also, is the OPA548 output being slammed against one, or both, of the output swing rails? 

- Both modifications as you mentioned above (changing to max current limit and to add resistance across inputs help but do not fix the problem on some parts.

There may be a thermal issue at work here.

-One other note is that the power has to be applied to the device and then inputs pin 1 and 2 need to be shorted to get the part out of the latch mode.

Originally, there was a mention that momentarily shorting the non-inverting input to V+ temporarily brought about the expected output. Now it appears that shorting the inputs together does that too. There is not enough known about this issue to speculate why that happens.

I suggest that you request a Failure Analysis be started on some of the devices that exhibit the triangulated output condition. The request is made through the Authorized TI distributor from whom the OPA548 devices were obtained. They will provide you with a Failure Analysis Request form that provides complete instructions.

Regards, Thomas

Precision Amplifiers Applications Engineering

Hi Thomas,

Sure, let's take this off the E2E so we can discuss the issue in depth. However, please allow us to complete the stability analysis of the circuit first before making the call. We hope to have that completed by tomorrow. Then we can then rule that in, or out, as a possible factor.

To answer your question, we have not seen or had any other reports of this type of behavior with the OPA548. It may be something unique associated with driving the high capacitance piezoelectric load. 

My number is (520) 750-2169.

Regards, Thomas

Precision Amplifier Applications Engineering

Hello Thomas,

I have another appointment later in the afternoon. Please try to call me by 2130 UTC (2:30 pm MST).

Regards, Thomas

Precision Amplifiers Applications Engineering

Hi Thomas,

Any chance we can meet 1/2 hr later than you suggested. 22:00 UTC, 3:00pm MST, 5:00 EST

Thanks,

Tom 

Hi Yang,

It sounds like it may be a stability issue. My recommendation is to post this in a new thread here since this is a separate question. Can you also do a load step test and see what the overshoot is? (use a small signal step, not full scale).

Thanks,

Nate

Thank you for your answer!

• Now,I describe my circuit. In the circuit that the power of OPA548  is ±30 voltage , And when the load is 1Ohm  the OPA548 output  5 voltage .First,When the OPA548 output  positive 2 Amp(And the load is 1Ohm),the output waveform is like the picture 1. (Notice : my input signal is   DC signal all the time ) In the picture , the waveform should be straight line ,because of the input signal is DC .But the  wave is square wave.(I think that is very ***)    Second, when the OPA548 output negative 2 Amp ,the output waveform is  picture 2. the wave of the picture 2 is oscillation ?  I  add that when the OPA548 output  either positive 2 Amp or negative 2  Amp ,the first  3  seconds is fine sometimes .There are two ideals that the first is stability and the second is overheating protection. For stability , I add the RC snubber  at the  exit of OPA548 ,but the way is useless. But  I think  overheating protection is more possibility . So far my problem is still unresolved.Can you help me??   What happened ?        
• 

                           picture 1

   picture  2  

Hi Yang,

Thank you for the quick reply. Please post this information in a new thread so that we can properly track and close your question. When new questions are added to existing threads tracking becomes difficult if there is already a suggested answer.

Best regards,

Nate