Is this Embedded System design of a Data Logger Sufficient?

This is a design i have composed for tracking the acceleration and direction of travel of a moving car.
I was hoping to get some input for improvements or possible faults.


A 3.7V 18350 or 18650 provides the power to the lm2623 switching boost regulator to provide 5V for the micro the gps and the accelerometer.
A low drop out regulator takes this 5V and drops it to 3.3V for the SD card.
The sd card has a voltage divider on it to interface it with 3.3V from the 5V output of the micro.

The gps module and accelerometer are use for tracking acceleration and direction of travel.

I did consider a magnetometer but i felt that GPS was better, i would certainly be keen to hear peoples opinions on alternative methods of determining heading.

I am hoping to use the RTC from the GPS to timestamp the collected data and store it in the SD card.

The accelerometer has a buffered output to enable its interfacing with the ADC of the PIC.
The GPS module uses UART to interface the PIC while the SD card operates in SPI mode.

I will do some sums to determine the battery life from a decent cell, but i need to ensure more than ten hours operation with a recording frequency of 10s.

I guess i could get better power efficiency with a magnetometer over the gps as i wouldn't need to keep the module active constantly to maintain a frequency of 0.1Hz for heading acquisition.

Any thoughts are greatly appreciated!

- - - Updated - - -

I am now thinking that i will need a magnetometer given that i have just realised heading and course are fundamentally different things to track.

You should inform what are the main requirements and constraints of this project, namely cost, features, etc...

Anyway, I think that you should consider to use a dedicated RTC instead of occasionally synchronize an internal timer of the uC. Let suppose a condition on what the vehicle is parked for long inside a place where GPS signal is not strong enough.

Another cheap device that for some particulars occasions could give you a useful information would be a I2C/SPI thermometer.

Finally, just to aid some field servicing or runtime debug could utilize the remaining USART, and for onboard programming, an ICSP connector could be added.

I have very few requirement specifications.
The device must be portable, battery powered and last for 10 + hours taking a record of acceleration and heading every 10 seconds and storing this in an appropriately sized memory location.
I have added the magnetometer and may add an RTC as you suggested, for extra redundancy.

The GPS is becoming less and less useful than i first believed haha.

I may keep it, but just to add additional functionality to the device.

I understand you want to use a PIC processor. Then save yourself a lot of troubles interfacing to the SD card by using a PIC that's running at the same 3V voltage. The many voltage divider schematics you find on the internet don't work for the PIC. The PIC has a schmitt trigger SPI input and requires 0.8 * Vdd for input high voltage.
0.8 * 5V = 4V
The 3.3V from the SD card isn't high enough.
These schematics are for other processors like the Atmel's which have a TTL level SPI input.

GPS is nice but draws a lot of energy, for 10+ hours and a sample every 10seconds you'll need larger batteries.

Buy mikroElektronika's Clicker2 for PIC18FJ, RTC2 Click, Li-po battery, GPS2 Click, mikroBus Shield, Accelero Click, mikroSD Click and make your project.