Assembled boards are back from retirement

6/16/2023 - Improved schematic added, as well as top and bottom renderings of both boards.

We have long offered high quality ESP32 and ESP8266 project prototyping boards; these have a combination of surface mount and through-hole prototyping area (in the "solderable breadboard" style), as well as pads for an ESP8266 or ESP32 module and the required components for operation. These can accelerate the process of building a prototype of your ESP8266 or ESP32 project, and the soldered connections provide greater reliability and a more compact package than breadboard or tangles of jumper wires.

Thanks to improved equipment and assembly methods we are now able to assemble the boards with modules like the ESP32 and ESP8266 mounted without prohibitive expense.

We have revised the design of the ESP32 prototyping board, adding pads to make better use of otherwise "dead" space near the surface mount footprints - this can now be used for placement of surface mount passives to support those parts (such as decoupling capacitors, which can now be that much closer to the IC they're associated with) and general prototyping use. This new version features ENIG gold surface treatment, with variants for WROOM and WROVER modules, and we use ROHS 158C solder on the ESP32 boards to eliminate risk of damage from excessive temperatures and silkscreen discoloration and ensure a lead-free product

We realize that soldering the ESP modules can be difficult for some hobbyists, so we are now happy to offer a version with the ESP8266 (ESP12F module) or ESP32 (WROOM32 or WROVER32) soldered down now that improvements in our processes have enabled us to do so without unreasonable expense. We always attempt to get modules with the largest memory available, and of the latest silicon. We are not always successful. We are confident in the ones ending in E being just that. The others are older silicon revisions (see the official ESP32 errata, which details the slight differences - unlike certain other semiconductor companies who shamelessly release new products containing the same silicon bugs they admitted to years ago in other product lines, when Espressif ships hardware that doesn't work 100%, they rev the design and make corrected silicon available in a somewhat timely manner.

These do not include a serial adapter on the board or the RTS/DTR-based autoreset to bootloader mode, you use an external serial adapter (like an Arduino Pro Mini, as opposed to an Arduino Nano (those are likely the two most used microcontroller boards (largely because they're cheap as dirt from shoddy manufacturers in China). As far as serial adapters, while I've always been a big fan of CH340's, if I was buying just for this setup, I would skip the CH340 - maybe a CH342 or CH343, even in a pinch the HT42B534, in preference to a typical CH340. The $2 CP2102, and to an even greater extent, the (likely counterfeit) $3 FT232RLs good for this, as both of these are guaranteed to use 3.3v logic levels, with no way to get damaging voltages on the serial pins (many CH340's, even if they let you switch the Vcc output, don't actually change the chips operating voltage. So you get Vcc = 3.3V, but the logic levels are still 5v, and only the combined action of the 1k-2k series resistor usually in series with Tx, and the protection diodes of the target, standing between an out of spec voltage and unplanned smoke release. I am working to bring superior boards to market based on the CH342 (dual) and CH343 (single) serial adapter ICs, which have one key advantage over their predecessors: They have a VIO rail dedicated to the I/O logic supply, so the ugly voodoo needed to make CH340s's run at even just two voltages is replaced by the ability to run the I/O voltage levels at any voltage! But these are not ready yet. (Regarding the counterfeit FTDI adapters, there was a time a few years back ago, the scandal known as FTDIgate, where FTDI took bold steps to try to stomp out the counterfeiting, but instead got so much backlash that they were forced to stand down. They were using windows update to distribute a driver "update" that would corrupt the VID/PIDs of counterfeit boards. Microsoft was reportedly furious as it erodes trust in the update system, as were many end users, who saw equipment built and sold in good faith turn out to apparently have fake FT232s' in them. I spent an evening working by flashlight to replace the serial adapter buried in an assembled device after it got bricked during the FTDI-gate scandal - the flashlight being needed because the device that was bricked controlled the lights in my room). And it wasn't any shady cheapo one either, it was a purple-solder-mask adapter with a convincing Adafruit logo and a $10 price tag and which looked totally legit... Anyway, the backlash forced them to roll back the bricking-the-fakes drivers, so the fakes are an option once more; The fakes generally have acceptable performance, and in fact have the second best throughput I know of so far (the Holtek garbage, when it works, does win by a non-negligible margin in raw speed, if you neglect to consider the times during which it's hung and the time you spend restarting the serial using app to unfreeze it, and rebooting the computer more often to get rid of phantom HT42's that are still listed in device manager but not connected - among other bugs). Since the FT232 and CP2102 never use 5V I/O levels, (though the pins are 5v tolerant), they can be used to power these provided their 5V rail is tied to Vin (feeding the regulator).Hold Flash and press and release reset in order to boot into serial upload mode. The absence of an on-board serial adapter, while yes, it makes uploading less seamless, also means more prototyping space available on a board of the same size, lower cost, and that you do not have a hard-wired serial adapter blocking up a serial port that you might want to connect to an additional device. As I said, WCH semi is really coming up in the world, and I hope to bring to market some of the most affordable parts with support for Vusb (5V Nom, which could power Vin) 3v3 nom (from on-board regulator) or use neither of those, and instead expect to see a voltage on one of the devices they're connected to (note that both devices VIO is connected to the same rail, so you cant do everything) - so these will be best for things like combined serial-UPDI adapters where the same part will be connected to Vdd, or where two serial connections are required to the same device or devices powered from the same supply. But this is getting off topic.

These are some of the best boards around for making modesst ESP32 amd ESP8266-based devices. In case you need a review, the ESP8266 has less ram, is available with only smaller flash, and cheaper by a large margin. It is well suited to IoT devices that will perform limited network connectivity related actions. The ESP32, on the other hand is considerably beefier. The peripherals blow the doors off the ESP82566 (a low bar, the doors were held on with tape and chewing gum - it's no secret that the peripherals on the 8266 were lackluster; these are considerably better), an also adds support for bluetooth in addition to generally better peripherals and higher specs (and price to match). In all cases we use the largest (16MB) flash and (for the WROVER) the largest PSRAM (respectively these are designated XXH28 (129 with the 1 truncated to fit, and XXN16R9 (16MB flash 8MB PSRAM)