Increased current drain of system while powered with switchr

[msyrjala]

current drain

I have a following kind of setup:
1. Battery (LiFePO4, 3.2V) driving a switcher (ADP2504), components according to datasheet (10uF input, 2*10uF output, 1uH coil)
2. Load is Freescale MCF51QE128, Atmel flash. Totaling about 20uF caps, two LC-filters to smooth Vcc for ADC section.

Here's what happens:
When I power up the load from DC source, the idle current is 30uA. Very good.

When the load is powered from the ADP2504, the voltage has 100mV ripple due to PSM (power saving mode). The DC current drain is now 400uA (with ripple according to voltage).

I tried powering the load with function generator using a similar waveform (DC+ripple), but the current is now again 30uA+associated ripple.

Any guesses where to continue hunting ?

 

[mghoffman78]

current drain forum

This sounds very normal to me. You have to remember that you are now using a switching power supply. the ADP2504 requires power too. Now you are supplying 30uA to your load as well as current to drive your IC. You also have switching losses and copper losses associated with the converter.

400uA sounds about right to me. I think you are trying to hunt for something that is not there.

 

[msyrjala]

sistem while

This sounds very normal to me. You have to remember that you are now using a switching power supply. the ADP2504 requires power too. Now you are supplying 30uA to your load as well as current to drive your IC. You also have switching losses and copper losses associated with the converter.

I forgot to mention that the load current was measured between the switched and MCU. I also tested with dummy load of 10-900uA and the switched internal current usage was really low.

I finally found out the reason. Sometimes getting out of the lab and sleeping over a problem really helps.

Here's what I did: I supplied MCU and switched from separate supplies. Whenever one side only (switched or MCU) was on, the current draw was 30-50uA. I have four connections between these sides: GND, +3V3 (supply current), a control pin to Switcher and a battery voltage measurement (Vbatt divided by 2). Only the +3V3 line was supplying the current.

I guessed that the reason is probably 1 single bit error. I was right.. The ADC input pin on the MCU that was used to measure the battery voltage was not set as analog input. When a CMOS input floats close to VCC/2, what happens ? Current drain. In the prototype board the pin was set up correctly, but to make PCB layout cleaner I moved this measurement to an another pin and had forgotten this bit.

Now I get 47uA total current drain when the MCU is sleeping and the device shelf life goes back to the planned 3 years.