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.

Comments

Popular posts from this blog

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

Hacking a USB-C to slim tip adapter cable to charge the Thinkpad T450s

This hack is inspired by this post . A year ago, I bought an adapter cable for my wife's Thinkpad X1 Carbon (2nd Gen) that allows her to power her laptop with a 60W-capable portable battery (20V x 3A). A USB-C cable goes from the battery into the adapter, which converts it to the slim tip output required by the laptop. Everything works out of the box, so I didn't give much thought about it. Recently, I decided to buy a similar cable for my Thinkpad T450s. I know technically it should work because the T450s can go as low as 45W (20V x 2.25A) in terms of charging (though I have the 65W charger - 20V x 3.25A).  I went with another adapter cable because it was cheaper and also I prefer the single cable design. So imagine my surprise when the cable came and I plugged it into my laptop and it didn't work! The power manager just cycle in and out of charging mode before giving up with an error message saying there is not enough power. After much research and reading the Thinkwiki

Using Google Dashboard to manage your Android device backup

I used to use AppBrain/Fast Web Install to keep track of which apps I have installed on my phone, and to make it easier to reinstall those apps when the phone gets wiped or replaced. But AppBrain had been going down the tubes, and Fast Web Install had always been a hit-and-miss affair. Android's own "backup to the cloud" system had previously been even more unusable. There isn't a place where you can see what has been backed up. And when you setup a new phone with your Google account, you just have to wait and pray that your favorite apps will be restored to the phone. Typically all the stars have to be aligned just right for this to happen. More often than not, after waiting for an hour or so and nothing happens, you just curse under your breath and proceed to install your favorites apps manually via the Play Store. But I just looked again recently and was pleasantly surprised that things are much more civilized now. Firstly there is a place now where you can loo