Current consumption with HSEBYP on F103T8

Weller.Bernhard · ‎2012-04-17

Posted on April 17, 2012 at 09:45

I'm currently evaluating the current consumption of the F103T8.

The current measured is considerably higher than the one given in the datasheet, but this has been discussed before here (seems to be dependent on the code executed). The problem I'm facing right now is the use of the HSEBYP. I expected the current to be considerably lower using the HSEBYP mode compared to the HSI or an external oscillator with HSE.

I'm using an external squarewave with 2MHz and divide it by 2, to get a system clock of 1MHz. The current I'm measuring is 2,3mA. Using a 4MHz external oscillator, divided by 4 I measure only 1,95mA. If I use a similar configuration with the HSI (divided by 8 then) I get a current consumption of 1,46mA.

These tests were performed at 3V, the debug interface was disconnected.

I guess my configuration contains some error, as I'm very new to the whole ARM Cortex-M3 architecture and the the F103T8 as well.

The code I used is attached (I'm using IAR EWARM). I appreciate any comment on this.

Weller.Bernhard · ‎2012-04-19

Posted on April 19, 2012 at 16:36

I've done some further measurements and the results are very counterintuitive.

I am feeding a rectangular signal from a signal generator into the OSC_IN pin of the device, which is configured to run from HSE in bypass mode. My target frequency is 1MHz and my results are:

1MHz / 1 -> 4,5mA

2MHz / 2 -> 2,65mA

4MHz / 4 -> 1,76mA

8MHz / 8 -> 1,48mA

This result alone is strange, but I also measured that the calculation time (for some measurement calculation which I cannot share) needs 0.2ms less (1.45ms instead of 1.67ms) if I use 1MHz directly.

I guess the Cortex-M3 is not exactly tailored for low power, low frequency applications - but my results are exactly the opposite of what I expected (and yes I made my notes in the right order).