Whats using current during deep sleep?

[schuh8]

I have a TTGO T-Display board configured as a internet radio. If I upload a simple sketch which does nothing for 10 seconds and then goes to sleep for 10 seconds, the current draw is 42 ma while idleing and 0.2 ma when in deep sleep. So far so good.

But if I upload the radio sketch it draws 130 ma while streaming audio and 9 ma while in deep sleep mode ... far too high.Since the hardware hasn't changed why am I drawing 45 times as much current during deep sleep in the radio sketch ??? (Yes it really is entering deep sleep mode !). I had thought deep sleep shut down pretty much everything except the Real time clock and the RTC memory.

So the simple question is: Is there anything you can code in a sketch that will increase the deep sleep current draw??

[EDsteve]

I had similar experience with another dev boards. I found out that the UART chip is involved in that issue. Not sure if that also is the case with TTGO boards. But worth to look into it.
Here a quick video of my strange UART power consumption behaviour:
https://youtu.be/AiU6VKYXP0k

The only solution that worked for me was to cut off the traces from the UART chip. Ugly. But I didn't find another way.

[schuh8]

Thanks for the suggestion EDsteve. Problem is I can get low deep sleep current simply by uploading a short sketch with just a deep sleep command ... only about 0.2 ma. That would seem to indicate that the TTGO development board is smart enough to turn off the UART and any other peripherals that use juice. Its only when I run the Internet radio sketch on the same board that I get 9 ma current draw in deep sleep. Therefore it must be something in the sketch thats causing the high current draw in deep sleep. Can't begin to imagine what though!

[chegewara]

There are many questions like that already answered on this forum, but:
- devkits are not prepared to "work in deep sleep mode", because 0.2mA is still high
- you have explicitly to turn off wifi and ble, if you are using them
- you should to turn off power domains for all peripherals which you dont use in deep sleep mode

I think most of this is described here:
https://docs.espressif.com/projects/esp ... modes.html
https://docs.espressif.com/projects/esp ... d_option_t
viewtopic.php?f=13&t=18529
viewtopic.php?t=19325#p89590

[schuh8]

Thank you very much for answering my question .. it helped me solve the problem.
I would like to point out that the documentation you sent me to, is confusing to anyone but an expert in the field. The following two sentences in the wifi/BT section appear to conflict.

-In Deep-sleep and Light-sleep modes, the wireless peripherals are powered down.
-Before entering Deep-sleep or Light-sleep modes, the application must disable Wi-Fi and Bluetooth using the appropriate calls

Perhaps I don't understand what powered down means. But if so, most of the other enthusiasts on various forms also think going into deep sleep totally powers down wifi/BT.

All that said simply entering:
WiFi.mode(WIFI_OFF);
just before entering
esp_deep_sleep_start();

reduced the deep sleep current on the TTGO board from 9 ma. to 0.16 ma. That may still be high compared to a bare esp32 chip but it's pretty good for a development board and more than enough reduction to satisfy my requirements for the project !!!

Again, thanks for your council.