With the launch of ESP32, the big question was - is the bulky QFN package really suitable for wearable tech? That has now been answered with the launch of the ESP32-PICO-D4 as a little System-in-Package (SiP).
You no longer have to worry about routing decoupling caps close to the ESP32 while leaving space for the RF trace and all of that. It is all inside the ESP32-PICO-D4 package.
The most important stuff to know is that the PICO-D4 integrates:
- Flash memory
- Crystal oscillator
- Most of the decoupling caps
In practical design, this makes life a lot easier and reduces via count as well as component count - speeding up your manufacturing process and cost per unit (assuming that the PICO-D4 does not cost a whole lot more than the regular ESP32 itself).
A look inside the ESP32-PICO-D4
This is what you have inside the PICO-D4. The flash memory is mounted on and bonded with the ESP32 die with bond wires directly. That is how Espressif achieves the extremely small form factor. Direct bonds off of the die also means that the production costs will be (hopefully) very low. The base board seems to have a rather usual trace resolution observed in LGA packages, which is also important for keeping costs down and yields high.
Something to worry about
We have not seen the under side of the ESP32-PICO-D4 package, but if the package does not dissipate heat well enough through the base material, your design could be in for some trouble.
The crystal mounted right next to the ESP32 die is exposed to a lot of potential heating and cooling. When the ESP32 is clocked at 240 Mhz, it generates a fair amount of heat. You may have observed this if you have worked on the NANO32 development board, which has some heating issues. The temperature inside the SiP can reach to over 70 degC rather easily and also come right down to ambient really fast as well (say if you stopped using Wi-Fi or put the chip to sleep).
Based on the datasheets of some crystal oscillators such as the one used in the ESP32-PICO-D4, you can calculate that such a swing in temperature may cause drifts of almost 1 kHz in the crystal frequency. If your application is affected by this, the PICO-D4 is not the best for you.
For example, here is a snippet from folks who know the frequency business really well, Connor Winfield "Aging Performance in Crystals":
So with the PICO-D4, you might effectively end up doing some aging tests on the integrated crystal if you stress the ESP32 to its maximum capabilities often in your design.
That is something to test properly and watch out for! The best you can do is provide proper pathways for the heat to dissipate and consider this factor in R&D if you aim to design critical systems that won't look pretty if they failed or drifted off a little bit.
We would definitely like to test the temperatures and drifts to share the results when we have PICO-D4 engineering samples with us!
Find the article interesting? Scroll right down to the bottom of the page and SUBSCRIBE to us to have full articles and tutorials on ESP32 delivered right into your inbox.
Need help with embedded system design? We work a lot with the ESPs!