Tested a bit more. The backup battery needs to be fully charged; newly maxed out from the charger, for the Aviogard to "accept" it, and not go into some kind of indefinit "break down mode" or whatever by making ticking noises and heating up. Even when fully charged, it would still enter in this indefinite loop at every other try. I switched batteries. Took the half way charged smaller battery as the input, and the larger one as backup and turned on the power. The idea was to see what happens when the "main" battery slowly discharges, a typical situation when the alternator stop working.
That seemed to work. The Aviogard switches off and the backup takes over. It did not switch cleanly off though, it happened with lots av ticking and heating up, the same way as if a not fully charged backup battery was installed. I could see on the EFIS the voltage varied also.
After that the Aviogard was dead. It does not turn on again. It smells burned also.
Either I have a faulty device (it obviously is now), or the Aviogard simply does not work properly with a backup battery. I'm just glad I was curios enough to test it on the table before mounting it in the plane.
Because of the smell, I had to take off the cover and look inside. Not much in there actually. It didn't smell good, but looked OK (to me). It looks like this:
And there are a few capacitors and resistors mounted on inside of the cover. The main unit is just an off the shelves isolated DC-DC converter. It's a Murata Power Solution UQQ series unit. More precisely is is a
UQQ-12/8-Q12NB-C. It's sold by
RS for instance, and I bet other component warehouses. This is useful, because full documentation is readily available.
The schematics of the UQQ looks like this:
The pinout and some specs:
The interesting part is the note that say Important!
Always connect the sense pins;
see Application Notes. With emphasis on
always. These sense pins are not connected on my device! I don't know what that means in detail though, but the Aviogard is certainly not made according to the specs of the UQQ.
Further down the following can be read about the sense pins:
Note: The sense and VOUT lines are internally connected through low-value
resistors. Nevertheless, if sense is not used for remote regulation, the user
must [emphasis in specs] connect + sense to + VOUT and -sense to -VOUT at the converter pins.
Sense is intended to correct small output accuracy errors caused by the
resistive ohmic drop in output wiring as output current increases. This output
drop (the difference between Sense and VOUT when measured at the converter) should not be allowed to exceed 0.5V. Consider using heavier wire if this drop
is excessive.
Sense is connected at the load and corrects for resistive errors only. Be careful
where it is connected. Any long, distributed wiring and/or signifi cant inductance
introduced into the Sense control loop can adversely affect overall system stability.
If in doubt, test the application, and observe the DC-DC’s output transient
response during step loads. There should be no appreciable ringing or oscillation.
You may also adjust the output trim slightly to compensate for voltage loss
in any external fi lter elements. Do not exceed maximum power ratings.
Then the trim. The UQQ is nominally 12V out, but can be trimmed up to +10% and down to -20%. This means MAX 13.2V and minimum 9.6V. The trimming is done as explained below in the specs:
Now, the trimming on my unit was done between the "trim" and Vout+, but should be done between "trim" and "sense+". Of course, if sense+ and Vout+ were connected (as they should, but isn't), this wouldn't matter.
I think this is the root to all problems.The Aviogard is trimmed to 13.8V. This is +15% and thus above the recommended +10% by an amount of 50%. The sense+ and Vout+ was not connected. According to the red square in the specs above:
Trim adjustments greater than the specified +10%/–20% can have an adverse
affect on the converter’s performance and are not recommended. Excessive
voltage differences between VOUT and Sense, in conjunction with trim adjustment
of the output voltage, can cause the overvoltage protection circuitry to
activate (see Performance Specifi cations for overvoltage limits).
This "hiccup" mode looks very much like what happened to my unit.
This all seems strange to me. It's not obvious how the unit is expected to maintain 13.8V while charging a battery that has dropped to 11-12V without entering into a never ending "hiccup mode", also due another reason:
The UQQ is used in the Aviogard without adhering to the specs of the UQQ. The UQQ was obviously never ment to charge batteries, as it will enters a "hiccup mode" that it won't get out of without removing the load, ie. removing the backup battery.
A closer look at the inside of the cover, and I could see the fault. It was the
output line on the PCB that had burned off. Probably due to hiccup mode?
