My idea was to use a coin cell (watch battery type) and put a diode inline on the positive side. Then attach this without removing the normal vbat connection.
Some people have reported that their GPS signal quality drops when they use a battery. This could be from any number of reasons. A diode on the battery would essentially prevent the GPS from seeing the battery, prevent the APM from charging the battery, prevent the battery from being drained while the APM is powered, yet still provide the minimum voltage on vbat so that the data can be saved. That should give you a lot faster TTFF.
I haven't tried this yet, but it should work and nobody here has seen any problems with the idea. It's on my todo list, but not a priority at the moment.
Having a battery prevents the GPS from needing the empheris data at every start and keeps the real time clock running. Having this will make the GPS lock a lot faster.
Thank you Jake, this may work. But what do you think of the idea to put a switch between the coin cell and the GPS and switch off the connection immediatelly after powering the APM? I think Steven Harsanyi did it this way, posted in this thread.
I think a diode would be easier and work better.
I posted on this and came up with the diode idea here...
The data sheet shows 2.0V-4.3V for Vbat on the GPS. The APM should be providing 3.3v.
If you start with a 3v (3-3.4v fresh) coin battery and have a 0.6-0.7v drop across the diode, you'll be supplying about 2.3-2.7v to the GPS. That should hold the data and run the RTC.
There's such a tiny current draw that the battery should last months or years. The diode is mainly there to prevent the 3.3v from the APM from charging the battery, which would eventually destroy it. But it should also prevent the circuit from draining or seeing the battery while it's running.
You should be able to pull pretty much any diode out of any piece of old electronics and it should work fine. There's really no current to speak of and the reverse voltage is low, so any diode should work.
The switch you see in the picture is there so I can isolate the battery to charge it. On my other gps units I installed replaceable batteries with no switch. Once you connect your battery you do not want to disconnect it til you have to replace it. You will know when to change/charge the battery when you start having gps problems! I think Jake's method will work as well. As far as the batteries causing interference I don't find that to be true for me. I have only experienced positive results. Here is a better shot showing the switch.
Thank you very much Steven. This looks quite easy and may help a lot of people, having trouble with GPS lock.
I will try it with a non rechargeable battery, without switch but a diode on the vbat line. Will report as soon as it is working.
I have installed the backup battery with a diode on +.
And IT IS WORKING.
Thank you Jake and Steven.
When disconnecting Lipo for seconds or minutes and than repowering, GPS-Lock is there in 10-20 sec.
But I´m still not so satisfied with it, because I only have 4-6 sats and the accuracy is very poor, about 10-20m. Maybe this is because of the interferences, which prevent the GPS to get lock without the battery. I think I have fought only the symptoms and not the cause.
But I have no idea which part or configuration is emitting the interferences. I have nothing special installed: UBEC (5,22V) on the IN-rail, 6x Hobbywing Skywalker 20A (each min. 2" away from the APM) and one of them connected with +/- on the OUT-rail, all others S and -. No telemetry, and for testing no RX.
Maybe it helps when installing the Diydrones BECs, or/and the original power distrtibution board? Any suggestions?
Great job! All you need to do is some testing. Telemetry would be pretty helpful.
I walked around a bit with just my battery, BEC, APM, and telemetry radio. Then checked in my plane. I'm actually lucky and get great performance. Usually it's dead on and occasionally jumps to 1-2m off when near trees, buildings, etc..
In my other testing I have found GPS units to be very susceptible to power quality issues. Trying different power configurations may help.
Unfortunately, it is not working anymore. The battery is empty now. It was a brand new one with 3,3V, and now it has only 1,4V. I tested the diode, it is working and connected in the right direction. Any idea?
I also tested GPS accuracy with APM off the copter, connected to PC via USB. Position in MP is walking around in a circle of about 10m while APM is laying on the table with 4-6 sats connected. Is this normal behavior?
4-6 is not a very good lock. 4 is the bare minimum required to get a 3D fix.
Maybe the battery is feeding power back into the APM circuit. If you've got a meter, check the current draw. Might have to cut the power trace and throw a diode in that side also.
The 4-6 sats were outside, blue sky, without backup battery connected, only the APM off the copter, powered by USB. This is really not very good.
This evening I will check if the backup battery is feeding power back into the APM. Maybe the ground pin 3 or 8 must be used for ground. But Steven also used the daughter boards ground pin.
I found something intersting in the data sheet of the MEDIATEK-3329:
The vbackup (Pin4) must be connected for normal operation.
The backup battery is feeding 1,4V back into the APM. Maybe a bad solder joint?
Probably just the way the circuit was designed. I guess we'll have to cut a trace and install a diode on the APM supply of vbat or use the switch method.