Kiwi1 suddenly goes deaf (and slow) [fixed!]
Hello,
I've had more Kiwi problems during the past 3 weeks than the past 10 years combined. Here is the latest.
My Kiwi1 suddenly went deaf. It still works and receives everywhere but with around 60 or 70 dB attenuation. This happened suddenly, earlier it was just fine. It's sunny outside (no thunderstorms) and no transmissions happened around the antenna. (In case you might wonder I have other antennas and other radios and can confirm with absolute certainty that this attenuation isn't related to the antenna, the cable or the connectors. This is caused by the Kiwi itself).
When the attenuation first appeared I also noticed another oddity: the waterfall was extremely slow (like 20 times slower than normal). Upon inspection I noticed that kiwid taking more thaIn 95% CPU.
In an attempt to remediate this mess I initiated a firmware update from 1.810 to 1.815 (because... why not?). Both problems persisted after the update was successfully completed.
Then, out of desperation, I switched the received off and separated the boards. When I powered back on, the slow waterfall was fixed (!!), but the attenuation stayed.
This has never happened before and I have a feeling my receiver might be at the end of its rope. I hope I'm wrong because the Kiwi2 is pretty expensive. Is it really over?
Thank you for your help!
Alain
Comments
So a few things. First, kiwid always runs at full speed. To achieve the required realtime response on the relatively slow Beagle it uses a programming technique called demand polling. And that means it runs flat-out all the time. This also means that if any other Linux program(s) of any consequence run simultaneously the realtime performance suffers. Since the waterfall is the lowest priority task in the system it will always slow down first when there is a lack of Beagle CPU cycles.
You say "separated the boards" but then don't say what you did after. Did you run the Beagle alone for a time? Did you simply reconnect the boards together again and power up?
After disconnecting the board I just reconnected them, nothing else. To be more precise, I disassembled the Kiwi after I slightly touched the top board to see if anything changed, and it did: the moment I pressed the top board the waterfall went from super slow to normal. Unfortunately it did not fix the attenuation problem.
My initial estimation on the amount of attenuation was exaggerated, by the way. The actual value is between -35 and -45 dB at 21 MHz (this is estimated by comparing a separate receiver using the same antenna).
For whatever reason, higher bands are suffering more than lower ones. The receiver is particularly affected above 3 MHz.
The following might or might not be related but it is exceptional so I'll mention it anyway.
The Kiwi sits right below a ham radio map on the wall. The map is made of paper and is held straight by two plastic sticks. Today the wind blew the map off the wall and made it land right on top of the Kiwi. Could the plastic stick have damaged the boards? It's a small map, everything is very light plus the Kiwi was it its plastic enclosure, still the timing matches so it makes me wonder.
The part about pushing on the board affecting the waterfall speed is completely bizarre. The only thing I can imagine is if there was some corrosion on the inter-board header pins causing the SPI clock to be intermittent. But that seems unlikely. Although a board separation/reconnect could have wiped corrosion off the contacts I suppose.
Do you live in a high salt-spray environment? I live next to the ocean and stuff here is constantly corroding.
I live half a mile away from the sea but corrosion isn't strong at all. I don't think I ever experienced corrosion on anything indoor before (or even outdoor, for that matter).
I wonder if anything can be done about the attenuation issue. I basically have 3 options: repair, replace, QRT.
I think I found the problem: somehow the map falling on the kiwi managed to rip off L407 from the board (see attached picture).
I even tried to short it with the tip of a small screw driver and, surprise surprise, signals went back to normal while doing that.
That is good news because it might mean the pad is still there, making the repair possible. All need to find more details about L407 and someone with a microscope/hot air rework station (this is my vacation place!).
I believe I could try dropping that map 1000 times on this board without any problem. How could a single component be removed like that will remain a mystery.
L407 is a ferrite bead on the 3V3 going to the ADC preamp. You could (carefully) bridge those pads over with a solder blob using a small soldering iron. It isn't critical to replace the bead.
I fixed it! Thanks
Doesn't your Kiwisdr live in a box? Even a cardboard box is better than nothing.