Skip to main content

ESPCLOCK4 - Reduce operating current by removing UART module

The ESP32 Mini uses a CP2104 UART module to drive the USB port.  The CP2104 is configured in self-powered mode, which means that even if nothing is plugged into the USB port and the board is powered by a 3.3V source, it is still drawing ~100uA of current. I wanted to see the effect of removing this module on the power consumption of my test circuit.

Removing the UART module means losing access to the USB port. However, I should still be able to program it using a standard FTDI programmer, which I wanted to verify. So I took out the same FTDI programmer that was used for programming the ESP8266, and built a custom holder for the ESP32 Mini.

I started with the same connections as used with the ESP8266:
  • ESP32 VIN <-> FTDI 3.3V
  • ESP32 GND <-> FTDI GND
  • ESP32 TX <-> FTDI RX
  • ESP32 RX <-> FTDI TX
However, I could not get it to work. After some fiddling around, I discovered I could only get it to work by connecting RX - RX and TX - TX. Not sure why.

But now I am confident I can still program the board without its USB, so I proceed to remove the CP2104 module (marked in red) with a hot air gun.

After removing the UART module,  I verified that the board was still working by uploading the code here, and plugging it into the test circuit. Sure enough, the clock started ticking. It worked!

I then tried measuring the current consumption using the LTC4150. Unfortunately, the current seems too low to be measured. After doing some research, I decided to swap out the 0.05ohm sense resistor with a 1ohm resistor. Because I didn't have a 1% part, I simply used a normal 5% metal film resistor. However, I did calibrate it against a multimeter with a 10kohm resistor over a 3.3V power source. I found the measurement to be within ~20uA tolerance.

Note: with such small current, the power used by the LTC4150 itself becomes very important. From the datasheet, at 2.7V, its typical power usage is 80uA, with a max of 100uA. With a 10Kohm passive load, I got a reading of 430uA. Substracting 90uA from the reading gives me 340uA, which compares well with the multimeter reading of 326uA.

Using a 1ohm sense resistor means I can no longer use WiFi in my test code. The 500mA current drawn will cause the board to brown out and reset.

So finally, with the UART module removed, the current draw is 650uA - 90uA = 560uA, or ~0.6mA. This is a hugely improvement over the 1.13mA that I measured previously with the UART module intact.

So it looks like I am going to go ahead and create the full ESPCLOCK code with this setup.


Popular posts from this blog

Update: Line adapter for Ozito Blade Trimmer

Update (Dec 2021): If you access to a 3D printer, I would now recommend this solution , which makes it super easy to replace the trimmer line. I have been using it for a few months now with zero issue.

Filament Joiner Part 2 (With Display and Knob)

Thanks to the current corona-virus crisis, the parts I ordered for the filament joiner project were taking forever to arrive. But now that they have finally arrived, I can put them to good use. These were the parts ordered: 0.96" OLED display SSD1306 Rotary switch encoder KY-040 Here is the final circuit diagram: The OLED display is connected to the SDA and SCL pins of the Nano (A4 and A5 respectively), and powered by 5V and GND. The rotary switch encoder is connected as follows: VCC => 5V GND = > GND CLK => D9 DT => D8 SW => D2 My prototype board now looks like this: The updated code for driving the knob and display is available in  heater-with-display.ino in the Github repository . We now have a fairly compact (about 7cm x 5cm) and independent filament joiner (no need to connect to PC) that is driven solely by a 12V power supply. Here's how to use it to join printer filaments. More usage details in my previous post .

Adding "Stereo Mixer" to Windows 7 with Conexant sound card

This procedure worked for my laptop (Thinkpad E530) with a Conexant 20671 sound card, but I suspect it will work for other sound cards in the Conexant family. I was playing with CamStudio to do a video capture of a Flash-based cartoon so that I can put it on the WDTV media player and play it on the big screen in the living room for my kids. The video capture worked brilliantly, but to do a sound capture, I needed to do some hacking. Apparently, there was this recording device called "Stereo Mixer" that was pretty standard in the Windows XP days. This allowed you to capture whatever was played to the speaker in all its digital glory. Then under pressure from various organizations on the dark side of the force, Microsoft and soundcard makers starting disabling this wonderful feature from Windows Vista onwards. So after much Googling around, I found out that for most sound cards, the hardware feature is still there, just not enabled on the software side. Unfortunately, to