Professional docs and toolchain for ESP32 in commercial product???.....

[Smokey]

I'm starting a new project that needs Bluetooth support on the microcontroller, so it's new micro selection time.  Yay.

I was surprised to see Espressif stuff (ESP32) is actually cheap and in stock at Digikey with lead times around 8 weeks (not 52w like some of the usual suspects).  I remember the days when these guys were much more "Padauk" style with no official documentation, no dev tools or SDK, and the only information anyone had came from a bunch of crazy people on electronics forms that reverse engineered the hardware (btw, thanks and I love you guys).  But then again, Digikey has a "marketplace" now so not sure that's exactly a good measure of confidence in a part.

For someone coming from the Microchip, TI, and STM worlds of officially supported dev tools, documentation, and example code, is ESP32 and the dev toolchain still a commercial liability in late 2021 that will make me crazy to work with it?  Arduino will not fly here, so it will have to be the "official" dev tools, which appear to be this github: https://github.com/espressif/esp-idf

Can I treat working with an ESP32 like any other major manufacturer of micros yet, or is the documentation and dev tool stability still not up to quality standards for real commercial products?

::And, yes, I know there are "real" products that use ESP32 out there. 
My question put another way would be: "If I worked in a real engineering department and I suggested we use an ESP32 in a new project would I get approved, laughed at, or fired?"::

[SiliconWizard]

The answer to your question would all depend on WHERE you work. What the company's policies are. Etc.

ESP32 SoCs are proven tech, but there are a couple things to keep in mind: first, the technical points. Read datasheets carefully and make sure they fit your requirements. In particular, AFAIR, two things I'd consider (not exhaustive!): time to start will usually be significantly longer than with your average MCU with "integrated" Flash. Time to get out of a low-power mode is also to be checked carefully. Which gets me to the second point, power consumption is likely not to be anywhere as low as you can get with a typical STM32 or TI MCU. And, also check the embedded peripherals can also fit your needs.

Then there's the relationship with Espressif itself. While you can have both a technical and commercial relationship with the usual TI, STM, On Semi.... I have no clue how good or bad it is with Espressif, but I would expect this not to go anywhere as smoothly. This may matter in the long run, and if you work in a large organization, it's likely that your managers will raise an eyebrow.

[Smokey]

That's an interesting question.  Has anyone every got any support from Espressif direct?
It sorta looks like the esp-idf is community developed.  Which is sort of crazy for that to be the "official" toolchain.

[nuclearcat]

The answer to your question would all depend on WHERE you work. What the company's policies are. Etc.

ESP32 SoCs are proven tech, but there are a couple things to keep in mind: first, the technical points. Read datasheets carefully and make sure they fit your requirements. In particular, AFAIR, two things I'd consider (not exhaustive!): time to start will usually be significantly longer than with your average MCU with "integrated" Flash. Time to get out of a low-power mode is also to be checked carefully. Which gets me to the second point, power consumption is likely not to be anywhere as low as you can get with a typical STM32 or TI MCU. And, also check the embedded peripherals can also fit your needs.

Then there's the relationship with Espressif itself. While you can have both a technical and commercial relationship with the usual TI, STM, On Semi.... I have no clue how good or bad it is with Espressif, but I would expect this not to go anywhere as smoothly. This may matter in the long run, and if you work in a large organization, it's likely that your managers will raise an eyebrow.

On relationship, to me, it looks like Chinese and Western IC manufacturers have swapped places in terms of quality and support in last few years.
I had exactly opposite experience, when ST didn't bother to fix known bugs in their devkit for years, released under ST branding shields with missing documentation and questionable design quality, and on top of cake is those recent shortages that hit them really bad. Perhaps, on some large orders, everything is different, but so far ST is disappointing. Hopefully they will take the necessary steps to improve their products/support.
And at the same time, I got free and useful support in very short time , on forum, private email and github, from Espressif.
P.S. For BLE NRF52 are the best, and ST are the worst in this matters. Nordic allow more low level access to RF hardware and have even completely opensource BLE stacks.

[floobydust]

I would be leery about using Espressif in a commercial product. Are they not a chinese company bound by their constitution to support the State's activities?
I went through this designing in huawei cell modems and despite their too good price, the USA and other nations refused to buy any product using them. It was a shitstorm.

The ESP8266/ESP32 price too low, the hardware too powerful. The honey is too sweet. And unaffected by the semiconductor shortages?

[fenugrec]

Read datasheets carefully and make sure they fit your requirements.

I would add, make sure to read erratas before selecting a specific part... you may be surprised at the bugs that have "no workaround available"...

[Smokey]

More community reverse engineering and documenting of ESP32.

"Remoticon 2021 // Uri Shaked Reverse Engineers ESP32 WiFi" =

[PKTKS]

(..)
The ESP8266/ESP32 price too low, the hardware too powerful. The honey is too sweet. And unaffected by the semiconductor shortages?

humm...  cause CHINA committed to highly competitive industrial electronics a couple decades ago investing material and human resources like nuts to achieve that...

And cf. latest declarations from their industrial managers the robotics park site will more than double next years..

So prepare to see a decadent greedy uncompetitive made in USA..
being flushed out by highly competitive and capable made in CHINA...

some folks around the world missed the train a decade ago..
playing around with marketing people will just not solve a damn thing in this scenario

Paul

[floobydust]

I agree the usual IoT MCU/module manufacturers are asleep at the wheel for years now. Look at the Espressif milestones and they are on the Shanghai stock exchange.
A small freight train: ESP32-WROOM-32ue from Digi-Key USD $3 giving 520KB SRAM, 4MB FLASH, dual-core 80-240MHz, Bluetooth, WiFi, FCC certs, and a load of peripherals and firmware.
Compare with LM555 USD $2 and prepare to bow down to your new overlords.

They're notorious for dumping products in order to stomp out other manufacturers. They have a long term strategy.
If the US Government sets sights on Espressif there are no alternate products even close, to use for a manufacturer.

I had a stand-alone ESP8266 (station mode, LAN only) that intermittently every few days or week would generate router security [WAN access refused] messages in the logs.
There was no reason at all for it do anything but spit out a webpage on local request. This is what got me wondering if Espressif complies with the chinese constitution and their IoT products carry the same risk as huawei products, although they are state-owned and have military ties.

[ebclr]

It's already set, no way back, nobody has the industrial infrastructure China has,  The only option will be joining then....., USA can't made even a chair no more

[Smokey]

Not sure I agree with all that China praise (in this case).  Just because you can make something cheap by cutting corners like outsourcing documentation and support to your users doesn't mean its better, and sure doesn't make it dominant.
 
If I'm in the gambling mood I'll go to Las Vegas.  I don't want to risk my business or reputation with a roll of the "wow look how cheap that Chinese thing is on aliexpress!" dice.

[PKTKS]

Not sure I agree with all that China praise (in this case).  Just because you can make something cheap by cutting corners like outsourcing documentation and support to your users doesn't mean its better, and sure doesn't make it dominant.
   

Couple of decades ago may be cheaper the primary consideration..

That is definitively no longer the case ..
They are getting  better every day..

And keeping good value for the money unlike products that rely on depreciation and brand or maketology (marketology) ideas which only apply to a very small market segment

Paul

[Smokey]

Revisiting.  I'm now sure how long this has been up, but they have a longevity commitment now:
https://www.espressif.com/en/products/longevity-commitment

Looks like 12 years.
ESP32 = 2016 + 12 = 2028
ESP32-S2 = 2020 + 12 = 2032
ESP32-S3 = 2021 +12 = 2033

Can anyone comment on the current state of the ESP-IDF dev tools?
https://www.espressif.com/en/products/sdks/esp-idf

[bookaboo]

I haven't used a wide variety of toolchains, mostly MPLABX and (was then) Atmel Studio but fwiw....

I can say that ESP-IDF works reasonably well for me, we've built WiFi and Bluetooth products with it which are in production with no issues.
They can integrate into various platforms (Eclipse and I believe VS), we used VSC, which is basic but fine. JTAG is possible I think on the entire ESP32 range, on the S3 it's direct over USB which is nice. No live variables though I read the S3 is hardware capable for this.
Documentation.... not as detailed as Microchip, but it's approached from a "here's an example" sort of way, which I don't mind. Everything's abstracted, so very little or no playing direct with registers.

It has it's quirks, clean build takes forever and sometimes is required at random. Also be prepared for many functions including common ones to change when the IDF version changes but overall if you need BT/WiFi it's hard to beat the bang for buck.