I’m really sorry, but I do not fully understand, what exactly you mean with this. The voltage at the PPK2 was set to 5 V, as I was powering the XIAO at the USB port.
The way shorter pulse duration I was mentioning earlier was “measured” with the uCurrent and an oscilloscope. At that time, I didn’t have a PPK2 yet.
Or were you referring to a different response?
I am applying voltage from PPK2 to the 5V pin of XIAO and observing the current. when 5V is applied, the period of the pulsed current that appears during the sleep period is 20mS. when the voltage is reduced to near the lower limit of the ETA4054’s operating range, the pulsed current period is 1.1sec. It appears that the operating mode is changing, but I do not know what is happening.
This is just information; try it when you get PPK2.
Hi there,
So looking at the S3 spec’s on sleep I see this foot note?
GL 
PJ
The waveforms also show that the average power consumption during deep-sleep mode is 26.85 μW, and the average power consumption during active mode is 78.32 mW .
I have just run into this issue when powering the C3 from the batteries of an external board.
It’s a pity that the XIAO boards don’t have a way to completely disable the charger when using the 5V pins.
I observe an average deep-sleep consumption of ~270uA (when using the 5v pin with 4 AAs in series totally 6 volts):
I believe the power consumption peaks come from the regulator and the led, which, as others have observed, blinks (permanently) with an odd frequency.
I wonder how much would we save by removing the charging LED (or the full charging circuit for that matter)
OK, I bit the bullet and desoldered the charging chip. Consumption goes down to ~150uA in deep-sleep. This, the charging chip + LED were consuming around 120uA. I still don’t understand why Seeed didn’t incorporate a way to disable them …
There are no spikes anymore
Here are some pics of the process. This was my first real experience with a hot air station. I protected the USB port and buttons with aluminum foil tape but I clearly should had added more layers of foil to the buttons since they melted anyways (which is fine for the most part since I don’t really use them for flashing):
Next, I think I will try to desolder the R9 pulldown resistor (if I manage to identify it), since AFAIU it’s not needed anymore. That should bring the deep-sleep power down to ~100uA.
Thanks to Seeed sharing the Schematic/PCB I have managed to identify R9 and removed it:
After that, the power consumption goes down to ~90uA
It would be good to know where the extra 45 uA (compared to the spec’ed 44uA deep-sleep) are going to. Probably it’s just that I am using 6V to power the board and it’s being dissipated by the regulator.
If you put ppk2 in source meter mode and reduce the voltage applied to the 5V pin to about 3.8V, you will see the current decrease.
Good point @msfujino !
Interestingly, when using source meter mode and separating the XIAO from the main board, the power consumption is higher: ~100uA with 3.7v and ~120uA with 5v.
I can’t explain why it’s higher (the main board is supplying 6v) but something other than the charger and R9 is consuming ~50uA over the spec’ed 44uA deep-sleep consumption.
If there are no connections on any pins or pads other than the 5V and GND pins connected to ppk2, it is strange that the sleep current is over 100uA.
It could be leakage current from C10 and C39, but I don’t know if 50uA is a reasonable value.
For reference, the sleep current is less than 50uA when powered from the battery pad as shown in the link below.
Yep, that was measured without any other connections. Maybe there is a software issue. However, I am using a barebones esp-idf example which just goes to sleep …
I will see if removing C39 makes a difference. AFAIU I shouldn’t be removing C10 permanently.
I found the culprit. I had the USB port connected (using a cable with the power removed though, so I thought it wouldn’t make a difference). After detaching it the XIAO consumes just 40uA
As a side note, I had also removed removed C39 but it didn’t make a difference
