So after much reading and research into the whole "backup" BEC saga. I think I've found a solution!
My first idea was just to run a 5.5v backup lipo in parallel with the 5v UBEC output.
This idea has been vitoed by most. Too many unknowns apparently.
The other was to parallel all the BECs up, but this would cause harmonics due to all the square wave interaction.
So I figured we cant mix power supplies. Other than using a microproccessor controlled mosfet to switch
power supplies when and if the bec dies.
We use a relay :) Bear with me.
The main power supply powers the relay. With the relay closed (NO) power is routed via the bec to the apm.
If the bec fails, the power to the relay fails and is switched to the auxillary power supply (battery or another bec)
This completely seperates the two power sources.
Will the relay be able to switch fast enough to the auxillary power source without the APM losing power?
If not, im sure a power capacitor will keep the APM alive long enough for the bi-pass surgery to take effect?
Above is a recording of the ciruit just in case my explaining skills are dodgy :)
You need to be a member of diydrones to add comments!
Comments
Sorry, the filename is correct but I copied the link from the previous post.
Version 2 UAVPWR file
I don't know what the APM 2.5 power system looks like, we use a custom autopilot.
Design complete and files uploaded to http://71.6.151.114/pub/UAVPWR2.zip
Includes LDO for 5V regulated output, LC filter in input to LDO, mounting holes and BOM file with exact recommended part numbers.
Notes:
I don't think it will do much good to put LC filtering on the output to the servos to clean up the servo power. Putting LC filters at the servos themselves (or at the end of their leads) would be more effective, but I suspect the real benefit comes from filtering the ESC and servo switching noise out of the 5V power supply that ultimately supplies the various sensors. The Exar LDO has a fairly good power supply noise rejection ratio, and the LC filter on the inputs helps further reduce the noise. On the APM board, there will be another LC filter and LDO (pretty much all sensors are 3.3V) further reducing the power supply noise.
Be careful substituting the 10uf ceramic caps on the LDO. Ceramic capacitors have significantly reduced actual capacitance with operated with a bias voltage (6.5V ish on the input and 5V on the output in this case). The X7R caps are better, but you will want a 16V cap (or more) to get most of the 10uf we are hoping for.
The mounting holes are sized for #3 screws.
The board should run fairly cool in free air, don't thermally insulate it in any way. If you are paranoid, eliminate the solder mask on most of the top and bottom land areas (keep some around the surface mount pads and both sides of the through hole pads to ease assembly).
Jeff, sorry, I forgot about that. I will post another board in the morning with an additional regulated 5v output using the OKI switching regulator and an additional LC filter on the output to limit noise into the digital/sensor section.
Dean
I humbly suggest again to look at just using diodes to OR your supplies/batteries ?
Get practical, take a diode like 1N5819, run 500mA through it, measure the drop.
And it can easily handle peak currents up to 10 A ! Looking at Dean's circuit, with those output capacitors, the diodes will present an almost flat supply voltage to the load.
No complicated electronic devices to go belly-up, and you can build the thing on Vero Board......
Design complete and files uploaded to http://71.6.151.114/pub/UAVPWR.zip
Contains PCB Artist (free download) CAD files and PDFs of schematic and board layers.
Dean
It would not be wise to use diode drop to generate a regulated 5V anyway, as the drop depends on the current flow across the diode and can be much smaller than the rated 500 or 700mv value at low currents.
It would seem to be a good idea to put a 5V LDO on the board with an LC filter on the input. The problem here is that you will end up needing
It looks like the best parts to get the job done are the LTC4412 ideal diode, using STAT to drive a second mosfet, the Si7615ADN pfet (dual pfet, 0.006ohm rdson @ 5V Vgs, >10A continuous current limit), and an Exar SPX3940M3-L-5-0 LDO (1.5A peak output, 280mv drop@1A, low noise (high PSRR)).
The board will have two outputs, one for the servos and other loads and one for the APM. The servo output will be at the same voltage as the inputs, the APM output will be 5V +/- 1.5%.
Mouser does not stock Linear, so a selected parts from Digikey so anyone could get them in a single order. Looks like $7 quantity 1, plus shipping and the board. None of the parts are very fine pitch, so the board should easily assembled using a hot air gun or toaster reflow and without a stencil. I will look to see if there is an SO8 pfet we can use, as the PowerSO package is leadless which makes it difficult to see if everything worked out.
I have a tube of LTC1473's which we no longer use in product which looks like a good candidate for the board. It does require a small inductor to generate the gate drive for the NFETS, but that is a cheap part. I will post a schematic and board layout on Sunday, and put the layout on a board already being ordered next week. I should have boards back a week later and will test the board and report results.
You can use a pair of Schottky diodes, but even 500mv is a pretty large drop when you are dealing with a (nominal) 5V system (10%). If your UBECs are running over 5.5V, then the diodes might be a better idea (simple equals reliable).
I will take a look at the available chips and design something over the weekend. I don't think the 4415 is suitable, as turning off inductive loads (like servos) will produce spikes that exceed the 5.5V limit on the 4415, probably resulting in destruction of the device and an unpredictable outcome.
I took a look at the LTC4370, and I am no longer sure you could use it as an over voltage protection device as well as an ideal diode. Doing over voltage protection would require back-to-back fets rather than a single fet, and it looks like the device depends on the intrinsic diode in the fet, which would be backwards in the second fet, preventing the circuit from working.
I will look for a device available in a leaded (TSSOP/SOP/etc) package and use an SO-8 dual mosfet if it is an external fet device (which I believe it will). This will result in a pretty small board, and a low RDS-on FET running at a max of 3A will not require any significant land area on the board for heat-sinking.
I will draw it up in PCB artist as I already have to order a prototype board next week from Advanced Circuits, you can redraft it in DipTrace if desired. I will probably have a few boards left over for the first few people who want one for the cost of postage.
Gareth.,
You don't want to use relays.
Switching them introduces spikes, brownouts, etc, etc.
Much better to OR all your supplies using Schottky diodes, with a fairly large capacitor on the final output.
In fact, on most of my stuff I use plain old 1N4007 diodes to OR the different supplies.
At low current the diode drops much less than the "textbook" 700 mV, much closer to 250 mV.
Most ESC'c use linear regulators, and you can quite safely run them in parallel.
That's what I'm doing in a H9 Twin Otter, and now flying in it's 4th year.....
For that 'plane I'm not using a capacitor or diodes, the BEC outputs run straight into a NiCad battery...
In Diptrace, the easiest way to export to PDF is to install a PDF printer driver (CutePDF, etc) and just print the schematic and both sides of the board.
When you say the other model of chip, which chip are you referring to? Many of the linear chips that have internal FETs or use N channel external FETs don't operate below 9v.
If you are going to design something, I would recommend the LTC4370. It uses external fets, but you can use physically small fets at BEC currents. You can also drive the EN inputs with a voltage divider from the BEC input. This will form a moderately accurate over voltage protection function that will disconnect a BEC from the output if the output voltage of the BEC goes over some preset limit. This would prevent both brown outs and destruction of your servos and other electronics due to a runaway BEC.
Post another note if you don't think you can design the board, I can put one together (in Diptrace or PCB artist) in a few minutes.
-
1
-
2
-
3
-
4
-
5
of 10 Next