What would you like to see in The Next Chip?

[ESP_Angus]

proper documentation

Can you please elaborate a bit on the parts of documentation you have found to be less than proper? We'll do our best to improve them.

[samsam]

BTW - do you intent to continue the approach you took with ESP32-S2 and make ESP32-B2 or sort of - ESP32 with BT only (without WiFi)?
What are the plans regarding new versions of Bluetooth implementation?
Thanks

[PaulNi]

Following on viewtopic.php?f=2&t=11593&start=10

I wish for compatibility with new ethernet PHYs t1 and t1s to be considered on the development stages. Technically, t1 and t1s phys are transparent to the MAC layer, but... there are few things here and there which are better to be tested on the development stage for compatibility, like MAC layer performance when shared medium is involved, or how upper layers handle the jabber condition on PHY level.

I believe, at least 100base-t1's use cases like car infotainment, and edge network for low speed WiFi will be of interest to Espressif.

[DL88AI88]

Hi there,

after two year anniversary of questioning what we like to see in the next chip I am starting to ask myself when will I see the Next Chip?
Sure, you released a single core chip design recently (based on Tensilica's LX7?) which is good for some applications, but...
what about a dual core chip?
Esp32 is becoming of age and you stated years ago, that you do not want to rest on your laurels.
I am just teasing a bit, please do not mind, for I am a great fan of your esp32 dual core system and can't wait to get its successor in my hands.
Hopefully based on Tensilica LX7 .

Keep on doing your great work.

Cheers

DL88AI88

[ESP_Sprite]

PaulNi: Noted.

DL88AI88 & samsam: Don't worry; we have more than the S2 in the pipeline, and we're certainly not abandoning the dual-core architecture that the ESP32 has. Also, while I'm not quite sure which parts of BT upcoming chips will implement, we will be following the latest standards with that. In general, assume that the time between the ESP32 and the ESP32-S2 is a anomaly; we want to release chips within a smaller timeframe than that.

[pataga]

I hope you're giving some attention to the micro-controller and hard real-time aspect as well. For me a really drool-worthy module would have the architecture features of the Teensy 4 along with communications hardware.

Scalable instruction clock with fine resolution, independent of peripheral clocks.
Fast gpio - the Teensy 4 can bitbang a clock at 150MHz with C code
low power consumption 100mA @ 600MHz
low interrupt latency
64bit floating point FPU

[earlyAdopter]

Bump! Waiting on response from ESP team on additional CAN controller, and reliability info (FIT/MTTF) numbers required for use in industrial applications.

FYI link to what FIT means: https://electronics.stackexchange.com/q ... lculations

A 2nd Canbus controller would be very useful for automotive applications.

CAN Controller
I did not see ESP_Sprite mention of the CAN controller in the 30pages here. I hope we've kept the CAN controller. And as mentioned above, do consider adding another controller i.e. if that ship hasn't sailed already.

And as a nice to have, make that a CAN-FD controller which supports up to 5 MBaud.
Eg: https://www.nxp.com/docs/en/fact-sheet/ ... TRLFUS.pdf

Reliability/Quality
Do consider putting out reliability & quality information for the hardware. By that I mean publish the Mean Time To Failure (MTTF) or Failure In Time (FIT) information. I am sure(hope) such testing was done for the ESP32. This is much needed in IoT applications in the industrial/automotive environment. On that note, I'd be very grateful if this info is already available for the ESP32 and anyone can point me to that.

High Temperature Operation
Hope the chips continue to be made for the -40°C to +125°C temperature range.

Increased Encryption Bits
Be nice if the chip directly supported storage of hardware encryption keys. Say about 16 keys, each being 256-bits. If I recall correctly, I think the current number of bits available for the user are 768 bits.

Memory Protection Unit (MPU)
Is there going to be support for MPU in the next chip? Be nice to be able to physically guard/restrict application software to the MPU and from accessing any other memory outside defined interfaces.

[cbrieske]

5ghz WiFi
EMAC with 1588 supports (Maybe RGMII?)

[rudi ;-)]

how you think about WiFi 6 ?

[PaulNi]

how you think about WiFi 6 ?

Just any more complete support of WiFi is better. You will be competing on the completeness of support of the WiFi stack and flawlessness of its functioning with all upcoming contenders. We already see a problem with 802.11b only devices, or devices that had 802.11g and 802.11n support, but had software not configured for use with them. We also had a client who had a device with a buggy Atheros 802.11N implementation that only agreed to connect to Atheros APs

You probably will not gain much with beamforming in devices where you physically can't put proper antennas, and the same goes with 60Ghz tech. Power consumption on them is monstrous, and you will be competing with proper AP SoC makers in that field. You just have to watch closely how compatibility landscape changes, so you will not be left out on any upcoming mainstream standard/standard extension support.

Coming back to the topic of antennas. @Rudi, are there any upcoming changes to the analogue side RF performance? Saying that it's very hard to get proper antenna performance in physically small devices, is there any hope for something like a MEMS antenna tuner on package?

In mass manufacturing, I found it's very hard to get you antenna tune to be stable as a lot of things happening during manufacturing can detune it. The factory getting a new PCB supplier, flux residues, minute changes to passives over the years by suppliers, weather during reflow in non-moisture controlled factories, changes in device casing by third parties, and even greasy hands of assembly line workers.

It's very easy to have your antenna to be detuned in use: user holding the device by the antenna, device being stuffed into walls, furniture, crevices in vehicle frame, and effects of moisture on PCB over time (had terrible experience making bluetooth devices on phenolic.)