Two deaths in the family

I love my Kiwis, but for whatever reason, I can't seem to keep them alive.

Power supply is a 3 output Lambda desktop linear supply . I see 4.95 V at the Kiwi board as measured at the 5VE pin on both Kiwis. Antenna is an attic mounted dipole for the past 6 months. There's an overload clamp also in line with the coax feed. There's nobody near the equipment but me.

One died, about a week ago, I disconnected it , measured supply voltage from the Lambda and verified that it was sourcing 5v at its characteristic .75A and today this one too is dead. The 5V at the board level is rock solid stable.

When I attempt a power on, I see a brief blip of a blue LED then zip. No 5v or 3.3v on the Kiwi pins. Both of them.

I had replaced the fans with ball bearing versions this summer and all was well (other than a bit noisy).

While it doesn't explain why they are dying, is there a diagnostic I can run to see if the beagle is alive (perhaps while disconnected from the Kiwi)

As you can likely tell, I'm not a linux expert.

Thanks
John

PS: RSP1A has been running off of same multicoupler antenna feed for months and all is ok. It's using a cheap Chinese switcher to feed it's micro USB

Comments

  • Remove the SDR Cape from the BB and treat the BB standalone and see if it boots
  • edited November 2020
    @john I assume you have read the PSU threads about tweaking the supply voltage up to around 5.3V?
    The Beaglebone will fail to boot if it dips during boot kiwisdr.com/quickstart/index.html#id-power and if you disconnect the power and just do the same again it (erroneously) seems to be dead. I saw your previous posts mentioning the voltage drop on the leads but I didn't spot what the supply was set to?
    Also (in my experience) leave the linear PSU on and disconnect from the Kiwi, don't power down from the AC side, as the voltage ramp up may cause issues.

    Also I'd be temped to try to cut the DC path between the Kiwi and other SMA items (like the RSP), I have managed to cause issues previously with the Beaglebone where it was grounded at the antenna when used with marginal voltage linear supplies.
    Stu
  • I'll try powering the BB separately and see what happens. I'm assuming it'll have some sort of LED heartbeat indicator if it's alive....or at least a steady LED.

    Just to be clear, the two Kiwis were on their own Lambda supply, so there was nothing else powering it. I now have them on the bench (also a Lambda variable/current limiting linear supply. I'm a fan! ) . Should I get these running again, I'll use the Vsense connections to overcome the lead drop issue.

    Let me try the BB standalone and report results.
  • edited November 2020
    Both BB's boot without the Kiwis being connected. "Boot" means LED's flicker and the heartbeat LED seems to be doing its thing.

    This is not good.
  • Why is it not good?

    I know feeding 5.3V to the Kiwi sounds a bit agricultural but I have unintentionally fed much more than that into the Kiwi (not Beaglebone direct) over an extended period without issue.
    Unless you are going to use sense wires in production I'd tweak the supply up to that. A good few of those who post on the forum set the supply above ~5.2V to reduce the chances of issues caused by brown out as described in the quick start. There is DC filtering and traces before it gets passed to the BeagleBone.

    Power the lambda on, tweak to 5.2-5.3V, connect only one Kiwi, report back.
    k5mo
  • edited November 2020
    OK. I set the bench Lambda to 5.25v. Inserted barrel connector. Measure exactly 5.25 v at the 5VE pin on the kiwi. 0.65 volts at the 5V pin directly adjacent it on the single kiwi I am testing in this configuration.

    One blip from the BB LED, then zip.

    "not good" comment was because BB's are much cheaper than Kiwis!
  • OK so let me get this right (from other posts) both the new Kiwi and the secondhand one act the same?

    Have you tried both capes on either BB?

    I think there are other posts where the John (jks) describes fault finding this sort of issue if I find I'll post a link otherwise I'll go quiet to avoid filling the thread with noise.
    I do wonder if you still have a PSU problem, or at least we haven't ruled it out, can you find a switcher capable of 5V/2A to try, not to use just to check the Beaglebone is not doing what is described in the docs - paraphrased "Shutting down due to momentary dip".
    Also was it just the Kiwi+BB or are there still antennas + ethernet connected?
  • edited November 2020
    You're not filling the thread with noise! I appreciate the input. I know from earlier struggles that the power quality is key (I've redone my cabling as a result and power is the first thing I've checked.

    Summary again as I know it's easy for me to overlook details.

    Two Kiwis, two BB's . Both were fed from same linear supply. They stopped (LEDs out) about a week apart. Power was checked and was found to be OK at the Kiwi end (I have screw terminal barrel connectors to enable bigger wire to be used AND to allow 5V to be checked while plugged in.

    Moved both of them to my workbench with another I and V regulated linear supply.

    Trying them on the bench one at a time, I see the blue LED on the BB come on briefly as 5V is applied then nothing. This is the same with both Kiwi "systems" (SDR+BB). I see no current being drawn at the supply metering (usually this seemed to be about .75 A)

    Measuring DC resistance at the Kiwi barrel connector with the two boards connected , shows about 500K (this is with a fluke 87 DMM).

    With the Kiwis together with the BB as a system, I see only the supply voltage in at the header (5VE) , nothing at the 3.3 and a .6v diode drop at the 5.0 pin.

    When I power the BB from USB without connection to the SDR, I see the power supplies (5 and 3.3) on the BB as OK.

    There's nothing (Ethernet, Antenna) attached at the present time.

    I have tried 3 different linear bench supplies (2 Lambda, one "power mate corp" . All are variable V and variable current limiting.

    WIth either of the BB's disconnected, the status lights blink merrily away when powered by USB.

    To ME, it looks like there might have been some sort of catastrophic power input issue, but then why are did the BB's apparently survive?
  • I have 4x BBAI+Kiwi connected to single 10A Acopian Linear supply set to 5.16VDC. They've been running for 1+ year, never an issue.
  • Yeah, these have been running for about that time, one of them probably closer to 2 yrs.
  • I've been reading Olaf's thread.

    In my case, when the BB is disconnected from the SDR I can see Chip Enable on pin 1, system 5V on pin two and 3.4 ish on pin3 on the TL5209. So, I think the BB's are fine .

    When I connect the SDR, Chip Enable and system v are both just a diode drop and nothing on pin 3 (3.3 output).

    ....sigh.....
  • I have several of these around from Odroid/XU4 work. For a switcher, they are quirt but they also provide a good Kiwi Power source for troubleshooting or workbench config etc.
    https://www.amazon.com/Adapter-Switching-100V-240V-Charger-WeBoost/dp/B081RGZH37/ref=sr_1_6?dchild=1&keywords=odroid+xu4+power+supply&qid=1604862035&sr=8-6
  • Supplies are all still good. If anything happened there, it was a transient issue. I put a 6 ohm power resistor in the 5 volt line and let it run for an hour on all 3 supplies in turn. No issues (other than a warm wirewound)
  • Quote "and variable current limiting" - what did you set the current limit to? Did you scope the voltage during boot?
    There is an inrush peak which can hit the current limit (dropping the voltage - result no BB won't boot), I assume you watched for that?
    I know I keep saying the same thing about boot voltage solidity but I'm confused from your post history like "it's dead, it was just PSU, it's dead", in my experience the Kiwi cape is hard to kill with just DC power at the power connector (within reason).

    To test I think it is best to have a fast response PSU that does not current limit, some decent barel connectors and conductors (if nothing is getting warm and the BB are still good).
  • Did you try measuring DC resistance (with no power applied) on the kiwi's main voltage rails? 5V, 3.3[this comes from the BB], 3.3G, 3.3A, 1.8V, 1V? Measure these with the kiwi board disconnected from the BB.
    There are some pins on P9, filter caps and some test points that make these measurements pretty easy:

    P9-1: GND
    P9-4: 3.3 (from beagle)
    P9-6: 5VE
    P9-8: 5V
    C5: 1.8V
    C20/C703: 1.0V
    TP near GPS can: 3.3G
    R406 near RF can: 3.3A

    See if you have relatively high-impedance on all of these rails. If you have a short (< 100 ohms or so), then we need to track that down. If all looks good, try applying 5V (with a supply capable of 2A or more) to the barrel connector and measure DC voltage at all of the above points. Only P9-4 should show no voltage. Conversely you could try a current-limited supply and look for hot spots, but I would expect your DC resistance measurements to indicate an area of concern. If the BB boots fine on its own but dies when the kiwi is attached, either your supply is incapable of sourcing the current or something on the kiwi is pulling down the rails on the BB. You need to find it.

    The schematic is available here: http://www.kiwisdr.com/docs/KiwiSDR/kiwi.schematic.pdf
    A high-res picture of the board here: http://kiwisdr.com/ks/Seeed.sample.1.jpg

    Good luck,
    Nick W1NJC
    johnk5mo
  • OK ... life gets in the way and the Kiwis sat for a couple weeks, and time to bare my soul again and share that indeed it was power supply issues again. This has been a strange journey made stranger by several problems that were all of the same type (power integrity) and happening at the same time.

    I am pretty sure that the PS issue(S) have now been solved and I appreciate your patience and advice.

    When I believed I had damaged Kiwis, I put them on the bench supply (current limiting Lambda) and did NOT have the current limit set high enough for it to sustain the voltage. I did this knowingly thinking that I had a low impedance short somewhere so I had the available current dialed back. I could see none of the rails with a low impedance short , and nothing on the board felt hot (like it should it if was sinking a couple watts) . I then dialed up the current and inserted the barrel connector and was met with the happy blue lights.

    The software updated and we're off to the races once again.

    Thanks to you all. I had several other SDR's in line in the meantime and I did not get anything near the same performance and easy to use UI as I do with the kiwis.

    Powernumpty
  • Always good to hear it worked in the end!

    That Beagelbone voltage dip thing has caught a lot of people but the general power management and control has probably saved more hardware and installs than we will ever know.

    I can still vividly remember the shockof discovering I had not managed to kill my Kiwi with serious over voltage. I corrected the voltage it just went on without holding a grudge. It's that sort of event that reminds me there are still some things available that have been engineered not just built.

    johnk5mo
  • Both issues, "cable brownout" and lab-grade power supply voltage drop due to current limiting, have happened to me personally and were totally unexpected at the time. That's why they are documented: http://kiwisdr.com/quickstart/index.html#id-power

    WA2ZKDjohnk5mo
  • Yup...Occam and his razor are still a "thing".


    I found another industrial power supply in my stash (open frame linear) that's 6A at 5 that I let cook for a day loaded at 2A and it's been fine. I'll swap it out and move on to my next debacle.

Sign In or Register to comment.