OPA2607: shorting V and V-

Hello Roy,

Could you share a diagram or circuit schematic of both the system and the test you are describing?

Thank you!

Best,

Alec

Hi Alec,

Sorry for late reply, below is our test diagram and the circuit we used with OPA2607.

Test diagram and schematic:
Short V+ and V- then turn on the power.

Below is the circuit we use with the OPA2607, and the output waveform of the good and bad units. The input signal is 42kHz sine wave.

Schematic:

Regards,

Roy

Hello Roy,

I will take a look a these details and get back to you early next week.

Best,

Alec

The test schematic is not very useful. The only thing that test can determine whether the inpu offset voltage is positive or negative.

I do not understand the circuit schematic. VG1 is shown as a floating voltage source, but I doubt that this is actually the case. Please describe exactly where the two input signals of U1 come from, and if and how they are referenced to GND.

Hi Ladisch,

VG1 is a microphone so it didn’t reference to GND. Below are the good and bad unit without any input source, which mean V+ and V- is open. But its output voltage is still different.

Regards,

Roy

What kind of microphone? Can you link to its datasheet?

Can you show an oscilloscope trace with the signals at the input and output pins of the opamp?

Hi Ladisch,

The attachment is the datasheet of the microphone, and the is as shown below.

There is another project in GRMN also found the same issue and the problem is oriented to OPA2607 after implementing ABA swap.

We want to clarify this issue whether this is caused by our design or just a corner case of IC since this IC only used on GRMN two projects. Do you think FA or any other test can check it?

 apc_10-1101_Rx.pdf

Regards,

Roy

I am not a TI employee.

So VG1 actually is a capacitor. I do not know your input voltages; if the problem occurs only with the OPA2607, then it might be related to its common-mode range.

Ladisch, we observed that it only occurred in some specific socket and ABA can't be fixed. That's why we need to know what different parameter cause the issue.

Alec Saebeler How do you think?

Regards,

Roy

Hi Roy,

The issue being seen when the device has no input and is left floating is most likely due to the device being in an unstable configuration. The device is decompensated, meaning it requires a minimum gain to be stable. I was wondering if you could maybe elaborate on what is being seen, is the issue with the shorting of the inputs different than what is being seen when the transducer is used? Could you also explain in which situation the unit is performing bad. You mentioned it occurs in a specific socket.

Best Regards,

Ignacio

Hi Ignacio,

The main issue is the different output signal. Shorting the input is just one way of trying to know the difference between a good unit and a bad unit.

And below is our finding. Yellow waveform is VF1.

Regards,

Roy

Hi Roy,

There could be a chance the device is marginally stable and due to process variation; some devices do not show this instability. Just to confirm, is this coming from a separate unit, or do you see this issue from one channel and on the same unit the other channel is normal? If this is from separate units, is the same configuration done on both channels or is the unused channel appropriately configured? I was able to run a stability simulation on the configuration shared and it does appear to be marginally stable. It is hard to simulate the real-life circuit with parasitics of the board and transducer, but one suggestion would be to increase the feedback capacitance of the circuit to say 5pF to 10pF just to see if this makes any difference in the response.

Best Regards,

Ignacio

Hi Ignacio,

This issue is from separate units, we use the second channel as a different amplifier.

After changing the feedback capacitance to 8pF, this issue was resolved. Can you share more details about how you choose the capacitance value and why lower capacitance would be a problem? Thanks!

Regards,

Roy

Hi Roy,

I am glad it helped the stability problem. When we have a configuration like the one this customer implements, stability is affected by the noise gain at the higher frequencies. At these higher frequencies it is determined by the capacitance of the transducer and the feedback capacitor. Adjusting the capacitance at the feedback can be a quick way to try and improve stability. The value is figured out when testing the system as different factors such as board parasitics will change the value you choose. It is very common to take a guess and check approach as simulation does not account for parasitics and one value in simulation can often result in little change when actually testing it.

Best Regards,

Ignacio

Ignacio, 

Thanks for the information!

If the simulation was run through Tina, can you also share the file with us?

Roy