Is my KiwiSDR about to fail?
Hello,
I have been running my Kiwi for 10 years or more and I recently noticed that reboots sometime fail.
Upon closer inspection I see that the Kiwi doesn't respond to pings at all (it is connected via wired ethernet). Some other times it repeatedly responds to a handful of pings and then stops responding for a minute or so. Either way, the issue has been solved by attempting the reboot numerous times with no other changes.
I am starting to wonder if the Beagle (or the Kiwi?) board is about to fail and if anyone else experienced the same issue. In case it fails, what kind of Beagle board should I get as a replacement?
Thanks!
Comments
I'd suggest checking your power supply, as an initial part of any investigation..
I've had similar problems, and it's nearly always turned out to be power supply related.
On particular problem is poor quality electrolytic capacitors that "dry" out over time, and the power supply smoothing and regulation suffer as a result. This is not always immediately obvious, but sometimes becomes more apparent when the KiWi and Beagle draw more current during certain operations.
Regards,
Martin
Yes, it is possible the Ethernet on the Beagle is beginning to fail. Specifically the Ethernet PHY chip. There was trouble with this some years ago that seemed to be a bad batch of PHY chips, although this was never conclusively proven.
You could consider replacing the Beagle with a new Seeed BeagleBone Green. Roughly $50 USD + VAT. Be careful not to use any other Beagle type. Specifically not the Seeed BeagleBone Green Wireless. It is not physically compatible with the Kiwi board. If you can find the older BeagleBone Black that will work also. But make sure it is a "rev C" board that has 4 GB eMMC (not 2 GB).
After another long session of failed boot attempts I have some updates on this issue in the hope they might be relevant to other KiwiSDR 1 users out there.
The Ethernet issue does not happen when the Beagle board is powered up by itself while separated from the Kiwi board (this seems to be ruling out the failed PHY chip hypothesis).
Moreover, separating the two boards and then putting them together also solves the problem immediately.
Rebooting the two boards without touching the hardware is the least effective option and could take forever to be eventually successful.
I wonder if the problem I have extends beyond the Beagle and eventually into the Kiwi.
Thanks for your help
So this sounds like our old nemesis: a marginal power supply (or voltage drop on a marginal power cable) that is causing trouble when there is higher combined current of the two boards together.
For whomever might be reading this in the future, here is the solution to my issue.
Long story short, turns out Martin was right. The delay relay I have been using for years caused the issue I described above. By using a straight cable the problem went away.
I still can't understand how this is possible (it also happen with a never before used used spare relay) but it is what it is...
Regards
Alain
Just like a real computer the Beagle has strict requirements on the applied +5V voltage waveform. But because the Kiwi are being fed by user-provided quiet power supplies (hopefully quiet) there are no standards enforced. As you'd see with a computer switching supply that has gone through a strict qualification process.
The waveform "ramp up" time, 10% to 90% voltage, has to be less than 50 milliseconds. Otherwise the Beagle's power management chip (PMIC) will refuse to power up. So a slow relay (or one that has become slower over time) could definitely be an issue.
This is the same reason linear supplies with large output caps that take time to charge can also be an issue.
I see what you mean and it happens when power goes out and then comes back: the ramp up of my power supply isn't steep enough and the beagle stays off (this is the reason why I introduced the delay relay in the first place).
What happens now, however, seems different: after the relay successfully powers up the beagle, the board will respond to pings for about 10 seconds, then it stops and reboots. This endless cycle can be broken either 1) removing the relay with a regular cable or 2) temporarily separating the two boards.
I noticed the relay introduces a 0.02 V dropout in tension. Could that have developed over time, to a point that it is now somehow relevant? That is my best guess right now.
As I commented in my original post regarding the power supply.
"One particular problem is poor quality electrolytic capacitors that "dry" out over time, and the power supply smoothing and regulation suffer as a result"
I would suggest that it is also worth checking the DC power supply output for ripple using an oscilloscope, as this can also indicate degradation of the smoothing capacitors.
I have set my power supplies to deliver 5v on the internal KiWi / Beagle header connector when the KiWi is running.
There is a slight voltage drop through the KiWi V1 DC input filter, and this usually means that 5.4v is required at the DC input connector to provide 5V internally.
Regards,
Martin
I've been through a few iterations of Power Supplies, including what I thought was an adequate linear one, that ended up not firing up the Kiwi first time round (using the Kiwi supplied cable too!), and running hot, to settling on this:
https://www.aliexpress.com/item/1005006697287204.html
It runs cool, and seems to meet the demands of the Kiwi with no start up issues at all. I've had it open and its well-made and very clean and uses high quality components.
Very interesting. Thank you!
Did you get the bingzi or the talema version? Whatever that means the cost differs significantly
I settled on the cheaper one. the difference is the manufacturer of the transformer. In this case I dont think it matters too much.