CPU Current In-rush

RE: CPU Current In-rush

Dave_couling — Thu, 04 Feb 2016 23:06:01 GMT

I used a 10 ohms resistor and an oscilloscope to measure. The board is connected to a power supply.

I had removed most of the large decoupling capacitors to verify my Tx and Rx peaks and durations. The only caps on board are the small value caps (0.1uF x3, 1nF). That's when I noticed this very high spike. It wasn't seen when I had my normal power caps placed on the board. Since the decoupling capacitance is so low I was not sure it would cause such high current spikes to charge.

~0.3uF charged via 10 ohms time constant = 3usec

The spikes I am see ramp up to 26mA in 0.4usec, and remains at this peak current for 1.2usec

The current then ramps down a reach CPU level. Total Spike duration 12usec or 4RC. So I supposed the math does match up.

I was just surprised to see such high charging current on such low capacitance when connected to a power supply, which should keep the caps charged.

-KL

RE: CPU Current In-rush

Dave_couling — Tue, 02 Feb 2016 20:06:54 GMT

Hi Einar,

Yes I've considered that scenario. However I'm using SDK 6.1, S110 7.1. So there is no concurrency with CPU and radio. Furthermore the pre-processing region of a connection event should have no Radio consumption as the HFCLK has not started yet.

RE: CPU Current In-rush

Einar Thorsrud — Tue, 02 Feb 2016 11:35:57 GMT

This is expected in some cases. If you sum radio peak current with no DCDC (13 mA), CPU current (4.1 mA) and current consumption of other resources used simultaneously (HF clock and maybe some other peripherals) you can see a current consumption peak in that range.