How to initiate deep sleep of T114?

#1

Hi, T114 powered from the battery seems to drain quite a lot of power (cca. 8 mA) in the sleep mode. According to the datasheet it should consume cca. 11 uA in the deep sleep. There is meshcore project upstream issue:


According to the comments it maybe the LoRa radio which is not sleeping. The board has attached GPS (am I right that for the GPS it is enough to put the Vext pin to LOW?) and a display. How to properly put the board into the deep sleep and achieve 11 uA power consumption? With the exact steps to do and the recommended GPIO settings known I will do the meshcore project PR fixing this problem.

#2

My observation so far:
According to the datasheets it IMHO requires:

GPIO21: LOW - power off GPS
GPIO15: HIGH - power off LCD LED

RGBs LEDs (2 x SK6812) are IMHO not used by the current Meshcore firmware, so no need to touch them, otherwise they would require one wire programming to the OFF mode.

IMHO the sd_power_system_off() call for the nRF52840 chip doesn’t handle these…

The LoRa radio chip SX1262 can be put into the standby mode by setting the NRESET pin (GPIO25) LOW, the LoRa radio will then consume cca. 0.6 mA, it’s still quite a lot for using the board without the HW battery switch.

The LoRa radio can be put into the OFF mode when it consumes cca. 160 nA by sending the SetSleep command (opcode 0x84) followed by 0x00 sleep command parameters. Then it should enter the sleep mode by the rising edge of the NSS pin. Unfortunately, it seems the NSS pin is hardwired to the logic 1 on the T114 board, so I am afraid it would be impossible to enter this sleep mode without the HW mod. Am I right? Or is there a way how to enter this LoRa sleep mode on the T114?

#3

IMHO, you are wrong. Between the NSS pin on the SX1262 and the pull up resistor, there is a label - LOAR_NSS - which connects to P0.24 on the MCU.

#4

Thanks, I also discovered it shortly after I asked the question :slight_smile: It isn’t clear in the schematic (there should be a node) and also wrongly marked as LOAR on the nRF side.

#5

It’s a signal label, not a sheet to sheet node.

Both versions show LOAR_NSS on the P0.24 …

#6

It’s a signal label, not a sheet to sheet node.

I disagree here, check e.g. the Vext_Ctrl near the U5 (https://resource.heltec.cn/download/Mesh_Node_T114/MeshNode-T114_V2.0_SCH(2).pdf) which is quite common way how to mark it. The LOAR signals are marked pretty inconsistently then and it’s quite confusing.

Both versions show LOAR_NSS on the P0.24 …

I am pretty sure it shouldn’t be named LOAR, but LORA.

Nevertheless thanks for quick assistance, I did meshcore project PR: https://github.com/meshcore-dev/MeshCore/pull/1138 and I am curious whether it improves the situation.