nRF52 Quiescent Current (EN = 0)
Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.

nRF52 Quiescent Current (EN = 0)

by magoldsm on Fri Sep 07, 2018 8:09 pm

I haven't measured this, but from the schematic it would appear that powering down the board via the EN pin should result in a parasitic load of around 27 - 42 uA (depending of SOC of LiPo, and ignoring the system voltage sag when the LiPo falls below the 3.3V regulator output). This would be the current through R5, the 100k pullup on EN. The standby current of the AP2112 is another 1 uA (according to the data sheet).

According to the AP2112 data sheet, EN is internally pulled down through a 3M resistor and the enable threshold voltage is 1.5V. Thus, R5 could easily be replaced by a 1M resistor, thereby reducing the parasitic load by an order of magnitude to 2.7 - 4.2 uA. Given that the 1.5V is listed as "typical", this still gives you plenty of headroom to avoid a premature shutdown of the regulator should the battery voltage fall enough for EN to approach, but not reach, the particular chip's actual threshold.

Have you given any thoughts to this? Or is my analysis wrong?

Michael Goldsmith
WinStock Engineering

PS Can you please update the .sch and .brd files? I spent far too long reconciling observations against the way the CP2104 on-board regulator is used (or not). I have the newest boards (ordered in April) as confirmed by visual inspection and comparison of the .brd file to the actual board. Confusingly, the DATE field on the schematic is listed as 7/25/2018, so I at first thought it was the latest, forgetting that the file date stamp is the date of the download.

Posts: 7
Joined: Sun Feb 21, 2016 3:20 am

Re: nRF52 Quiescent Current (EN = 0)

by adafruit_support_mike on Fri Sep 07, 2018 10:47 pm

The quiescent current of a Feather is dominated by parasitic leakage through the semiconductors. The low-Vf Schottky diode between BAT and VBUS has a specified reverse leakage of 200uA, for instance.

In practice, you can expect the quiescent current to drop to the 80uA to 100uA range when you pull En low. You won't get it into the single-digit-microamp range without spinning a whole new board with components specifically chosen for low parasitic leakage.

Frankly, trying to reduce quiescent current is a losing strategy if you're looking to maximize quiescent battery life. You're better off reducing the amount of hardware connected to power with a low-current timer like the TPL5110:


That controls all current to the load, and the leakage through the P-mosfet is in the picoamp range. The default quiescent current through that is around 20uA, and that's almost entirely leakage through the power-on LED. If you cut that trace, the quiescent current drops below 100nA for supply voltages down to 2.5V. That's a couple orders of magnitude better than you can do with the Feather on its own.

In my own tests of the TPL5100, the load was 75nA to 80nA in the 2.5V to 5V range.

Posts: 58856
Joined: Thu Feb 11, 2010 2:51 pm

Re: nRF52 Quiescent Current (EN = 0)

by jps2000 on Mon Dec 24, 2018 4:03 pm

If you change the EN pullup resistor R5 from 100k to 1M you save about 40uA. The other improvement is the replacement of D5 MBR120 by a better one. For example a PMEG3005AESF.
With both adaptions you can bring the off current ( EN = 0) below 10 uA. which is about 1/10 of the original design.

1 year is 8760 hours so 10uA eats about 90mAh per year. Not much for a 500mAh cell.

The advantage is that you can still use the battery connector and you keep the charging function.
When you put a switch in the battery line as an alternative you always have to make sure the switch is on for charging.

A nice design is to have a reed contact from EN to GND. You can mount a small magnet in a cylinder that can be rotated for example with a coin. When the magnet is in line with the contact = on when it is perpendicular the contact is off.

Posts: 449
Joined: Fri Jun 02, 2017 4:12 pm

Please be positive and constructive with your questions and comments.