Your Kiwi must be running the v1.690 (or later) release to use the proxy service and be in the public Kiwi list (rx.kiwisdr.com, map.kiwisdr.com)

Please visit kiwisdr.com (documentation) and kiwisdr.nz (online store)

Troubleshooting after lightning strike

2»

Comments

  • Excellent news :-)

  • Was it only the ADC chip and nothing else? Not the ADC preamp (inside the RF shield) or any other other smaller parts? Does the GPS work?

    Great job getting it fixed!

  • Yes, it was just the LTC22481 and the GPS is working.

    The preamp is ok. I didn't had a differential probe to properly measure the gain, but I measured about 10 dB gain to ground, so I thought it was good enough, and it turned out, it is ok.

    The antenna was disconnected during the lightning strike, so I don't know what exactly caused the ADC to fail and why only on one of the kiwis, but I'm glad that I got it working again.

    I've been playing with it the entire afternoon and don't see any anomaly.

  • Something seems to be off though. During the evening hours in the presence of strong signals I noticed a higher noise floor.


    Noise floor without cable attached: Good Kiwi: -119 dBm, fixed Kiwi: -124 dBm

    Noise floor with -20 dBm test signal on 28 MHz: Good Kiwi: -119 dBm, fixed kiwi: -110 dBm


    So there is still a problem somewhere. I think I will do more measurements on the frontend, but it is not a severe impact in my usecase.

  • jksjks
    edited August 2023

    In hindsight, it probably would have been worth the minimal cost of replacing the ADC preamp chip at the same time. It seems odd that the signal/no signal behavior is different. Or maybe not -- my knowledge of RF gets sketchy real fast here.

  • jksjks
    edited August 2023

    Some more thoughts: It's possible the TVS diode(s) have degraded under such a harsh event. Although it seems unlikely that would manifest as only moderate attenuation. Usually they just go full short when they fail.

    Knocking them off the board with a small soldering iron is not too hard. Replacing them is another story (I would need my microscope). But you can add back the TVS protection externally. Or fit TVS/diode parts across the SMA solder pads.

    Try and measure the special low-resistance values of R402 (66R5) and R403 (28R7) at the preamp inputs. They are critical. And check that none of the caps there or in the preceeding 30 MHz LPF have shorted. Schematic

  • The TVS diode shows >200 MOhm, I think it should be ok. Resistors are good and caps are not shorted.

    I probed the ADC output again and i noticed that two pins are not switching, D3 and D4 are both always low. Does this explain the behavior?

    (Note on the measurement: the noise floor shows a bit variation, so the values below -110 are an estimate)


  • I'm coming to this late so maybe I'm missing something you've already tried but it would seem that D3&D4 should definitely be moving around. Maybe they aren't stuck by the ADC (now replaced) output but by the input they are driving?

    This would cause a problem similar to ADC nonlinearity which smears the error (now large) across the entire spectrum when higher/MSB bits are exercised.

  • edited August 2023

    It's the resistor array. The resistors for D3/D4 are open. I don't know how i missed that before. And I don't know how that could happen.

    Considering the circumstrances, the FPGA input might be damaged as well. Its input resistance is not suspicious though. If there is a simple way to read out the bits, i could apply an external voltage and see if they toggle.

    Edit 17:38: I tried to solder a 100 Ohm resistor across, but it's too small. BUT: the array resistor for D4 suddenly works again and the noise floor improved. So perhaps the array got damaged during the ADC-resoldering? And the heat from my soldering attempt partially "fixed" it again.

  • Hot air rework effecting surrounding parts is a real problem. And that crappy lead-free solder doesn't help matters either.

  • I got the resistor arrays replaced and everything is now as it should be. Also the noise floor without antenna connected, which was too low with the dysfunctional D3/D4.

  • That's great!

Sign In or Register to comment.