The topic has been discussed many times I'm sure, and I've done a fairly extensive search both on this board and the rest of the web. I want to summarize my thoughts here, though, and seek feedback.
Description:
My project is a glorified GPS tracker - it reads GPS, processes the NMEA feed, extracts some information and logs it to an SD card. From there, it will periodically dump it via bluetooth to a client application.
It needs to be fairly weather resistant, so it will be as sealed as possible, hence the wireless data interface. The only port it's going to have is for power. The power to the unit is provided from an external 12v battery, which also powers some other things around and will be installed for the day, then taken out and recharged, etc.
I only need for the device to function when this external battery is connected; however, I cannot afford corrupting the state of the SD card when the battery is disconnected. I need to be able to gracefully "shutdown" on the external power loss. It does not matter whether I persist the last bit of data after the power is disconnected - I just care about not corrupting what I already have there.
I'm thinking to try the super capacitor route.
For the backup power, I'll have a bank of super capacitors feeding a boost regulator, feeding the system (e.g. Pololu 5V Step-Up Voltage Regulator U1V11F5). The capacitors need to be charged from the 12v source, so I was thinking of using a buck regulator from 12v to the super capacitors (e.g. Pololu 1.8V, 500mA Step-Down Voltage Regulator D24V5F1).
Effectively: 12v -> buck -> supercapacitors -> boost -> system
The question is, how do I provide the main power to the system - feed it in parallel via two diodes?
Alternatively, feed the system through the backup power circuit constantly?
To detect power loss, I could hook up the output of the buck to an analog input on the board (and isolate it from the capacitors with a diode). All the logic I really care for with regards to the power loss is not to initiate any new writes to the SD card if it's not there.
Check the working voltage of the super cap, 12V is high for one of those.
It is vital you know the current you will draw, while super caps are indeed super they can not provide much current for long. I would use a rechargeable battery for this project rather than a cap.
Have an ON-OFF switch button, that using programming, gracefully shuts the controller down.
Then when an indicator light goes out, replace the battery.
If using supercap, then sense loss of battery and use limited power in cap to gracefully shut controller down.
3)If SD card has room, save two files one a couple of seconds after the other, so any corruption due to switch off during saving a file will only occur to one.
I have used this strategy, No 3, when a toshiba 1600 portable was pressed into service as a data logger and needed the 3 1/2 floppy as the storage device. (Remember 3 1/2 floppy! !)
weedpharma:
A capacitor stores electrons on foils separated by an insulator. A battery stores power by changing the chemistry.
Symantics.
Did you know, the difference between a biscuit and a cake was set by legal precedent when McVities invented the Jaffa Cake. At the time, biscuits attracted a higher rate of tax than biscuits. HM Inland Revenue prosecuted McVities, claiming tax evasion as the Jaffa Cake is, in reality, a biscuit. McVities claimed that the difference is, when left in the open a biscuit gets softer, whereas a cake gets harder.
The court ruled in McVities favour. I'm sure if the tax incentive was there, we'd suddenly find car batteries being marketed as super electrolytic capacitors.
Grumpy_Mike:
Check the working voltage of the super cap, 12V is high for one of those.
It is vital you know the current you will draw, while super caps are indeed super they can not provide much current for long. I would use a rechargeable battery for this project rather than a cap.
One concern that I have is limiting current to the caps, not sure if relying on the current limiting feature of the regulator is enough.
As for the battery, I'd rather avoid the more complex charge circuitry needed for the battery. After all, my objective is not to keep the device running all the time, so keeping the battery maintained for those 500ms (I hope) or less to finalize whatever the last write was happening seems like an overkill.
I expect the current draw to be fairly small. I'm certainly going to build the core system before tackling the backup power, so I'll know what I'm dealing with.
Have an ON-OFF switch button, that using programming, gracefully shuts the controller down.
Then when an indicator light goes out, replace the battery.
If using supercap, then sense loss of battery and use limited power in cap to gracefully shut controller down.
3)If SD card has room, save two files one a couple of seconds after the other, so any corruption due to switch off during saving a file will only occur to one.
I have used this strategy, No 3, when a toshiba 1600 portable was pressed into service as a data logger and needed the 3 1/2 floppy as the storage device. (Remember 3 1/2 floppy! !)
Tom......
Option 1 not gonna happen - it's going to be used by many different people and most of them won't even know of its existence - they're going to simply plug the battery in like they used to, and unplug it at the end.
Option 2 is what I'm after.
Option 3 is interesting, but I don't think it guarantees non-corruption - controller can decide to fiddle with the partitions, etc.
Use a boost regulator on the Cap to hold the 5V up as long as possible to the micro & SD buffer/card (assuming a 5V system).
Parallel diodes let the regular power drive the uC/SD, when it dies the super cap/boost drive them.
What am I missing?