This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

OPA857EVM: EVM stopped working

Part Number: OPA857EVM
Other Parts Discussed in Thread: OPA857,

Hi Experts,

My customer had a question about OPA857-evm product.
He purchased 2 of them and it worked. However, there was a power glitch in their building and the evm stopped working.
Just curious if you could provide sometime to help debug the OPA857-evm board

The board stopped working after the power glitch. He had two board. One board: He connect JP1 to Vcc to enable TEST mode
He input 2.1VDC + 100mV AC @500KHz square wave from TEST IN
He had 3.3V DC power supply
From OUTPUTP, he didn't get any signal. The DC of OUTPUTP is 1.8V
For another board, everything is the same, but the DC of OUTPUTP is 1.14V.
Both of the board gave amplified signal from OUTPUTP before. However, they stopped working and gave no signal after the power glitch.
The power glitch might influence my DC power supply but customer is quite sure and think he used a pretty good DC power supply Agilent E3640A.

Please advise. Thank you.

Kind regards,
Gerald

  • Hi Gerald,

    sounds weird. By glitch you mean a sudden decrease of mains voltage like when a fuse is blown? Do you know the nature of this dip? Sometimes a dip is not just a simple dip can be accompanied by very insane overvoltages immediately before the dip. Even more complex overvoltage and undervoltage patterns can occur which even a good DC power supply might not be able to fully suppress at its output. The "good old" linear voltage regulators powered from mains by the help of a bridge rectifier and a big stortage cap were free from this trouble. Even in the event of mains voltage fluctuations the output voltage was stable and only changing the output voltage slowly. But with the modern switch-mode power supplies an insane output voltage condition cannot be fully excluded. So, unfortunately, there's a certain risk that the DC power supply has produced an insane output voltage during the glitch. I have learened that it can be wise to have all cables disconnected from the DC power supply when turing it on and turing it off.

    I would simulate all kinds of glitches at the input of naked DC power supply (with the EVM being disconnected from the output) and check with the scope whether insane output voltages are generated. I would also add a dummy load to the output simulating the current consumption of EVM.

    And I would check the test signal generator in the same way.

    Another nasty cause of failure can be compensating currents flowing across signal ground or protection earth connections from one mains power supplied device to another. Think of your DC power supply and the oscilloscope, for instance. These currents can produce voltage drops which become superimposed to the signal voltages and supply voltages and can destroy a sensitive input or output or supply voltage pin. This is the nature of surge and burst, by the way.

    I don't want to say that TI has made a mistake here, but because of being a rather paranoid circuit designer myself (Relaxed), me thinks that the EVM would profit from more protection measures at the supply voltge pins. I miss a bigger decoupling cap here, and I miss especially a TVS across the supply voltage. My boards have alway installed low leakage TVS.

    There's a nice anecdote when I studied physics. We worked on GaAs microstrip detectors for the ATLAS experiment at CERN and needed a bias voltage of about 120V DC to deplete our microstrip detectors. Everytime we used a certain 1000V DC power supply our equipement becomes damaged from time to time. But only when we took this certain power supply. It turned out that everytime the output voltage was altered to change the detector's bias voltage, a very narrow and short lasting 1000V voltage pulse appeared at the output. That wasn't funny of course. But funny was, that the department director who was a very arrogant and unfriendly person had designed this 1000VDC power supply himself...

    Kai

  • Hello Gerald,

    I am looking into your described behavior of the customer's OPA857EVMs.  Have you began to investigate the two boards for shorts and/or damaged components?  The power glitch could have manifested in a number of ways, and any additional information about the current state of the boards (which are no longer working, as you stated above) would beneficial.  

    To clarify:

    Both boards have a 2.1VDC + 100mV AC @500KHz square wave at TEST IN.  Power supply is 3.3V DC for each board. JP1 is shorted to VCC to enable TEST mode.

    Board #1 has no AC signal at the output (OUTPUTP), with a DC of 1.8V.

    Board #2 also has no AC signal at the output (OUTPUTP), with a DC of 1.14V.

    To confirm:

    Are there any differences in passive components, signals applied, power supplied, etc between the two boards?  

    If the loss/glitch of building power is the root cause for this behavior (which is fairly certain), Kai's response captures the essence of how many things could have affected the boards.

    Please allow me a bit of time to investigate.

    Best,
    Alec

  • Hi,

    Thank you so much for all the responses and inputs. 

    Are there any differences in passive components, signals applied, power supplied, etc between the two boards?  

    No, both of the boards are the same. Customer compared the good boards and the damaged 2 boards current consumption at Test Mode (2.1V DC+ 160mV AC 500kHz square). Here is a table:

    Board

    Mode

    Input

    Power supply voltage (V)

    Power supply current (A)

    Good functional EVM board

    Test Mode

    2.1V DC + 160mV AC

    3.3V

    0.031

    Good functional EVM board

    Test Mode

    None

    3.3V

    0.027

    Broken EVM board #1

    Test Mode

    2.1V DC + 160mV AC

    3.3V

    0.02

    Broken EVM board #1

    Test Mode

    None

    3.3V

    0.022

    Broken EVM board #2

    Test Mode

    2.1V DC + 160mV AC

    3.3V

    0.017

    Broken EVM board #2

    Test Mode

    None

    3.3V

    0.019

    Broken EVM board #1

    TIA Mode

    PD

    3.3V

    0.075

    Good functional EVM board

    TIA Mode

    PD

    3.3V

    0.089

    Customer was using E3640A power supply to power two EVM boards. The EVM boards was working in TIA mode with Test IN to 3.3V, JP1 shorts GND. The total current consumption is 0.178A (0.089A each board).

    Kind regards,
    Gerald

  • Gerald,

    Thank you for providing details regarding the performance of the EVM boards.  I have taken a look at the data you have sent; unfortunately, it is likely the power glitch has permanently affected the EVMs.  As the nature of the 'power glitch' is unknown, it is difficult to recommend debugging tactics or advice on potentially rescuing the EVM boards.  The current measurements provided indicate the affected boards are drawing smaller supply currents in TEST mode than the good EVM.

    Considering the Test IN pin was connected to the power supply during the power glitch, it is possible the OPA857 (and passive components along the path the 'power glitch' travelled) was damaged or had the absolute maximum range exceeded by the glitch.  For the OPA857, the input current at VIN is limited to 10mA absolute maximum, and the input voltage limited to +4V maximum.  Depending on the glitch, these levels could have been exceeded.  

    If the passive components around the OPA857 on the EVM are inspected and found to be in good condition, the power glitch may have caused damage to the OPA857 IC.  Unless your customer finds a shorted or damaged component on the board, it will be difficult to know if a fix or change will resolve the observed issue.  

    I would suggest proceeding with the remaining good boards for any remaining test or characterization; there is not a recommendation I have for attempting to repair the OPA857EVM at this time.  If power supply transients and other glitches are of (possibly new) concern for the customer, they might explore adding a larger decoupling capacitor at the supply pin of the OPA857, as Kai mentioned above.  

    It does appear the customer was operating the EVMs correctly during their testing of the OPA857.

    Best,

    Alec

  • Hi Alec,

    Thank you so much for all the support.  Customer do have a one quick question, not related with this topic. He want to apply reverse bias voltage (20V) to the photodiode. He curious if the EVM configuration will allow him to do that. He understand the V_Input bias voltage is 1.83V now and He connect my PD cathode to the V_Input and his PD anode to ground via SMA connector. What can he do if he want to connect external bias voltage?

    Kind regards,
    Gerald

  • Hello Gerald,

    I will look into this shortly.

    Best,

    Alec

  • Hi Alec,

    Thank you for your support on this. I wiill wait for your update.

    Kind regards,
    Gerald

  • Hello Gerald,

    Could you work with the customer and add the voltages you are considering to the attached image?

    Thank you; I will be better able to assist if I have clarity on exactly what your customer intends to do with the EVM.

    Best,

    Alec