[Solved] Power consumption deep sleep Pysense/Pytrack

[Emmanuel Goudot]

According to specs of pytrack peripherals,

GPS (Quectel L76-L):
acquisition/tracking : 22/29mA
standby: 0.5mA (after sending "$PMTK161,028\r\n")
backup: 7µA

Accelero (LIS2HH12):
Normal: 50µA (ODR 10Hz)
Power down: 5µA

After setup_sleep(30) and go_to_sleep(), are GPS & LIS still powered ?
What is the PIC doing concerning them with CMD_SETUP_SLEEP & CMD_GO_SLEEP ?

[Pwntus]

Thanks for the replies @jubbs @this-wiederkehr @jmarcelino
It has been suggested that the peripherals are not put into "standby" while in deep sleep.
Shouldn't this be done by the Pysense/Pytrack by default? In addition, we do not initialize/use any peripherals in our test program.

We did some more testing with the Pytrack shield.
LoPy firmware: 1.7.6.b1
Pytrack firmware: 0.0.3

Our test program disabled WiFi and Bluetooth in boot.py and ran a 30s normal sleep (time.sleep()) and a 30s deep sleep (py.setup_sleep(30) + py.go_to_sleep()).
Here's our results:

USB power
Normal sleep: ~75-80mA
Deep sleep: ~5.3mA

3.7V LiPo battery
Normal sleep: ~73mA
Deep sleep: 18μA (!)

It seems like the Pytrack gives acceptable readings while using a 3.7V LiPo battery while the Pysense still has some issues.

[dbrgn]

I can confirm that the pysense does not seem to properly go into deep sleep.

On a 3.7V LiPo, it still pulls 2.7mA in deep sleep mode. (Note, the USB cable in the photo below is not connected to any power source.)

Pysense firmware is at 0.0.3.

[bucknall]

@Pwntus (+ others) Thanks for raising this! We're now investigating this, it appears that the issue only pertains to the the Pysense. Thanks for your patience!

[jcaron]

@bucknall Hi,

Have you made any progress on this? Any idea if this is hardware-related or just a firmware issue yet?

Thanks!

[Pwntus]

@jcaron issues with the Pysense should be fixed in the pysense_0.0.4.dfu firmware release, but we have yet to test it. I'll post the results tomorrow.

[jcaron]

@Pwntus Ah, wondered about that, but didn't see anything in the chagelog that seemed to be related. Thanks for the (very) quick reply!

[bucknall]

@jcaron @Pwntus I'll double check with Daniel regarding the power concerns. I'll confirm if this is correct (I'm London based - Daniel works from the Netherlands).

[Pwntus]

@jcaron @bucknall (+ others) here's our new (unofficial) Pysense test.

LoPy firmware: 1.7.6.b1
Pysense firmware: 0.0.4

Our test program disabled WiFi and Bluetooth in boot.py and ran a 30s normal sleep (time.sleep()) and a 30s deep sleep (py.setup_sleep(30) + py.go_to_sleep()).
Here's our results:

USB power
Normal sleep: ~57.2mA
Deep sleep: ~7.35mA

3.7V LiPo battery
Normal sleep: ~49.1mA
Deep sleep: 2.2mA

It still appears that the Pysense has issues with deep sleep.

[bucknall]

@Pwntus Ok I'll feed this back to Daniel again - apologies for getting your hopes up - I'll keep you in the loop with how this progresses.

Thanks for your patience,

Alex

[jmarcelino]

I saw the same results for the Pysense as @Pwntus

Also noticed that if you used the MPL3115A2 pressure sensor (not sure about the others) it goes into deep sleep with it in active mode increasing consumption by about 1mA. But I think this one can be solved in the Python library by setting MPL3115_CTRL_REG1 to 0x0 (to clear SBYB bit in CTRL1 register)

mp = MPL3115A2(py)

mp.i2c.writeto_mem(MPL3115A2.MPL3115_I2CADDR, MPL3115A2.MPL3115_CTRL_REG1, bytes([0x0]))

before entering deep sleep.

[bucknall]

@jmarcelino Ah good catch! I'll test this out - Thanks!

[jmarcelino]

@bucknall
Just to be clear that 1mA is on top of the lingering 2mA... I'm not sure what's causing that one yet :)

[Emmanuel Goudot]

@jmarcelino Are you sure sensors on the PySense are powered during deepSleep ?
I have done tests on pytrack, and it looks like 3.3V is 'cut' for LoPy, LIS(accel) & GPS, explaining the ~17µA of the pytrack during deepsleep.

[jmarcelino]

@Emmanuel-Goudot
I see 2.6V on the MPL3115A2 Vdd pin during deep sleep so I guess it is active, for example on the Si7006-A20 (temp/humidity) I don't.

My best guess is certain sensors can act as wake-up sources (e.g. pressure going above/below threshold) so they have separate power nets.

[dbrgn]

Is there any ETA on the deep sleep fix? This is delaying the shipping of one of my projects. We simply can't send it to the customer if the battery only lasts a few days or weeks.

[dbrgn]

Sorry for bumping this again, but now that a new LoPy firmware has been released, any news about the low power fix?

The specifications of the PySense on the website still says:

Ultra low power operation (~1uA in deep sleep)

...since that promise is still off by a factor of about 2500, at least an ETA would make me happy :)

[Pwntus]

@dbrgn Not really any changes with 1.7.8.b1.

LoPy firmware: 1.7.8.b1
Pysense firmware: 0.0.4

USB power
Normal sleep: 48.8mA
Deep sleep: 7.4mA

3.7V LiPo battery
Normal sleep: 40.2mA
Deep sleep: 2.2mA

[dbrgn]

Cool, this is now fixed: 10.9uA :) https://forum.pycom.io/topic/1022/root-causes-of-high-deep-sleep-current/73

[Pwntus]

@dbrgn Thanks for the notification, I can confirm that it works!