Future Teensy features & pinout

I can see an embedded application with high level graphics, the firestick for example.
If I'm honest, I only want to abuse the GPU. A far away friend has a robot with NXP chips which have a GPU (RT595, might be wrong). He offloads his limb movement planning when there is high obstacle density (hard planning problem). His robot can "sense" attacks and dodge automatically. I am jealous. He said it wasn't possible to exploit the 1062 in the T4 like the other chips with dedicated 2D vector acceleration. So I never bothered to try or investigate.
But at what point should we just get a pc?
Yeah, we simply don't have the weight, power, or monetary luxury for anything more than small cheapish microcontrollers. Our fighting robots are knee to thigh tall RoboOne like bipedal humanoid robots, they have various requirements, cost, weight, power, low pit-time (limbs and parts need to hot-swap-able), ect.

I also desire to see what a GPU would do to the current graphics libraries like TGX and T-COMPUTER. It would be cool to see a Windows 3.1/DOS emulator like the ESP32 FabGL project.

If only it had a DAC...
:D All 1170's have a 12bit DAC! So that's exciting for the future. After making 6 projects with the T4.1, I too wished for a DAC on the T4.
 
Sigh... well it was great for a moment, the T4 has a 12 bit ADC. And it turns out, on quick perusa, it seems that \all of NXP's M7 line have 12 bit ADCs, though some do also have DAC(s). So it seems so far, that the choice is the slower 16 bit ADC and 12Mb/s USB, or accept the 12bit ADC to get the 480Mb/s USB and of course the 1 usec/12 bit for the T4 ADC compared to the 2.5usec/12 bit for the T3 ADC.

I am working a product development effort where the customer insists on 16bit ADC for "marketing". Sigh.. again.

For my own instrumentation work, I sometimes really do need 16 bit ADC and a >/= 12bit DAC. The slower ADC and USB for the Teensy 3 have been limiting in what experiments I can do with it.
 
Hi,

I saw that the IMXRT1170 has a MIPI DSI output. Whould be nice to have it exposed, a display is very often needed in applications....and MIPI seems to be the state of the art...RGB and others require a lot of digital lines...

Thank you

Torsten
 
Sigh... well it was great for a moment, the T4 has a 12 bit ADC. And it turns out, on quick perusa, it seems that \all of NXP's M7 line have 12 bit ADCs, though some do also have DAC(s). So it seems so far, that the choice is the slower 16 bit ADC and 12Mb/s USB, or accept the 12bit ADC to get the 480Mb/s USB and of course the 1 usec/12 bit for the T4 ADC compared to the 2.5usec/12 bit for the T3 ADC.

I am working a product development effort where the customer insists on 16bit ADC for "marketing". Sigh.. again.

For my own instrumentation work, I sometimes really do need 16 bit ADC and a >/= 12bit DAC. The slower ADC and USB for the Teensy 3 have been limiting in what experiments I can do with it.

You have the daisy seed that has a 480MHz processor and 16bit ADC’s and also a DAC I think. Maybe I’m not allowed to say it on this forum though.
 
Hi,

I saw that the IMXRT1170 has a MIPI DSI output. Whould be nice to have it exposed, a display is very often needed in applications....and MIPI seems to be the state of the art...RGB and others require a lot of digital lines...

Thank you

Torsten

+1

Filler added since there's apparently a ten character minimum in this forum.
 
Hi,

I saw that the IMXRT1170 has a MIPI DSI output. Whould be nice to have it exposed, a display is very often needed in applications....and MIPI seems to be the state of the art...RGB and others require a lot of digital lines...

Thank you

Torsten

I’ve been toying with the RT1176-EVK for going on a year, and not only can I confirm it supports MIPI DSI output, but it also includes a lovely display out-of-box. Furthermore, the MCU not only includes a GPU and a vastly improved pixel processing pipeline, but includes low-level instructions for building modern UIs, including workspaces, windowing, icons, typefaces, and more.

It’s a huge leap from the 106X series in myriad ways not listed here, but we’re also talking about a medium form-factor EVK (and a markedly higher pin count)… all of which is to say were a future Teensy to use this chip, assuming maintenance of the eponymous form factor, many of the chip’s features would not be available on board.

The only possible way access could be provided to all features is by exposing all pins externally for hypothetical peripherals designed to take advantage of them. I have the vaguest recollection this was mentioned as a no go, but take this with a grain of salt; Paul is the only b̴a̴n̴d̴ voice that matters, and I wouldn’t prod due to speculation/wishful thinking.

Time will tell.

P.S. It’s worth noting that, even if many features were to remain unavailable, an RT1176 Teensy would be extraordinarily powerful. I’ve a mind to develop such a board, but it would be an untenable endeavor with no intention of profit, not to mention I have no interest in competing with PJRC. It would be much more gratifying to work with rather than against, and likely an invaluable learning experience
 
Distributors are showing a small number of the 800 MHz industrial temperature range parts arriving near the end of 2021. But beyond those, all appearances look like 1 GHz 1170 chips won't exist in volume production until much later. How much later, I don't know. You can check out the various distributor website and see the same info I see. It's pretty depressing...

Hello Paul,
is any chance to require a production of Teensy 4.1 mounting the Industrial version of iMXRT1062 (so the iMXRT1062C version) instead of the Consumer version (iMXRT1062D)?
Obviously, such a change wouldn't be requested for a single board but at least for 100 of them.
 
This has probably been mentioned, but it would be awesome to have a Teensy 4.1 (5?) with builtin WiFi option like the Pico W.

-Mark
 
I hope this is the correct place to post this.

What I need is a replacement for the TeensyLC. I have quite a few projects that do not need high end processing, but make heavy use of Paul's USB drivers such as 'keyboard' and 'joystick'. The Teensy 4.0 will do the job, but it would be nice to have a cheaper, lower performance product to knock a few dollars off the price.

While it would be nice to have a replacement for the Teensy 3.5 (i.e. maximal exposed GPIO), that was an edge case that I have only used in one design. I can use the 4.0 or 4.1 for all the others.
 
What I need is a replacement for the TeensyLC. I have quite a few projects that do not need high end processing, but make heavy use of Paul's USB drivers such as 'keyboard' and 'joystick'. The Teensy 4.0 will do the job, but it would be nice to have a cheaper, lower performance product to knock a few dollars off the price.

I haven't heard any discussion of replacements for LC and 3.x. Perhaps the supply of the older chips will eventually recover to some degree. Availability of all Cortex M0/M3/M4 seems to be limited, i.e. not just NXP Kinetis, so I don't know what one could use as a low-cost platform.
 
Well, I would consider the 4.1 to be a replacement for the 3.6. The 3.5 has access to more GPIO pins than either, but my use was an edge case. I've only built three copies of that project.

If you look at the PJRC store, you'll see some 'real talk' about how the entire 90nm process is spooling down and supply may simply be dead soon. While it blows my mind that companies would shutdown manufacturing while it is still in demand, that is how the silicon industry works. The purpose of my post is to put a bug in a designer's ear that I will buy an LC replacement.
 
Teensy 4.0 with ESP32-C3 Wifi Capability

Teensys.jpg

Above is a Teensy 4.0 with a LilyGo T-01C3, which is a minimalist ESP32-C3, placed at the end to show what could be possible.
As you can see combined it is shorter than a T4.1.

The T-01C3 schematic is shown here.
The T-01C3 only brings out 5 GPIO pins, 3 of which would be needed to program it, so we would need to bring out more pins, possibly
something like this, but without the USB capability as that would not be needed.

I would suggest that a combined Teensy 4.0 and a T-01C3 would be a fantastic product. It would give Teensy a WiFi capability together with multi-processor capability.
Let's call it the Teensy T4.0E32WiFi. Gosh that's a mouthfull!!

The ESP32 could be programmed via a T4.0 Uart onboard connection. I would suggest that a number of the GPIO pins would be connected to Teensy I/O pins. This would give the option of commanding the ESP32-C3 via toggling pins or writing to the UART port.

I have a project at the moment where I use one Teensy pin to wake up an ESP32-C3 and command it to send a WiFi (ESP_Now) message to a Sonoff Basic2 to tell it to energise it's relay (or not) to turn on/off a valve.

If it was intended to make the board the same dimensions as T3.5/T3.6/T4.1 then a number of the pins which are on the bottom of the T4.0 could be brought out to the sides and let's say 8 of the 15* ESP32-C3 pins would be brought out to the side. The other 7 would be used for inter processor communications.

The way I envisage the combination working is that the ESP32 would provide the WiFi interface together with any web pages communicating the results/being commanded by the Teensy.

I hope all that makes sense to you. The ESP32-C3 is a low cost implementation of the ESP32 which should make it a commercially viable product.

ESP are starting to approach Teensy capability, already having announced 400MHz products, and one of the shortcomings of the Teensy is lack of WiFi capability. This will close that gap.

EDIT: * 15 (actually 16) on low cost chip, others have 22.
 
Last edited:
We're doing something similar with a product that we're developing (STM32 for processing and an ESP32 for wifi). We use one USB port and set the STM32 to act like a USB to UART bridge when we want to program the ESP32. Works fairly well, if not time consuming to have to load the bridge software on the STM32 first, flash the ESP32, then flash the actual software on the STM32. I'm sure something more elegant could be developed if it were targeted as a development board.

The ESP32 are fast, but their documentation is super crummy and their floating point performance leaves a lot to be desired. Which is why we're using an STM32 co-processor on our product - to do the heavy computational lifting and using the ESP32 as just a wifi head.
 
We use one USB port and set the STM32 to act like a USB to UART bridge when we want to program the ESP32. Works fairly well, if not time consuming to have to load the bridge software on the STM32 first, flash the ESP32, then flash the actual software on the STM32.
I envisage a small piece of code hidden away in the Teensy which provides this function. I am assuming that it is not much more than passing through from one uart to another and controlling 1 pin.
 
I envisage a small piece of code hidden away in the Teensy which provides this function. I am assuming that it is not much more than passing through from one uart to another and controlling 1 pin.

Maybe two pins (I have to push two buttons when uploading code to the ESP32). But, yeah, super simple. Would be really cool if it was integrated with the Teensy bootloader so you could flash teensy code over wifi, either on a local network or an image hosted on a server.
 
Maybe two pins (I have to push two buttons when uploading code to the ESP32). But, yeah, super simple. Would be really cool if it was integrated with the Teensy bootloader so you could flash teensy code over wifi, either on a local network or an image hosted on a server.

Adding WiFi to the Teensy is for sure one of the most important features. I also hope that it’ll be integrated.
 
@onehorse made a T_3.1 add-on board for the 8266 years back. Mounted to Teensy, shared enough pins to program and communicate. A couple of those were beta tested here - esp32 has come a long way since then.

@FRDM4236 has been working with ESP32's PICO and Now S3's using bare chips and the newest using the modules. If anyone remembers the original FRDM4236 was a castellated board in the layout of a T_3.6 that mounted a T_4.0 ... so it was format of a T_4.1 just barely before the T_4.1 released. Current WIP is trainer4edu.com/nanoboy/nanoboy_s3. It is meant to be stand alone ESP32-S3. Though perhaps a FRDM42S3 would be a good fit? Emailed link of p#589 to @FRDM4236. Another cool thing is use of LiPo power to these boards including charge circuit. Would be a full redesign to mate to a T_4.0 - but it could be done. Might also work with T_4.1 - though antenna clearance and overall size might have to grow depending on how it connects.

Did some attempt to see how fast the ESP32's were with 240 MHz PICO's - forget the numbers but they were not near a T_3.6 in 'loops per second' even combining both cores. Maybe a bit faster than a T_3.2 ...

Note: @onehorse and @FRDM4236 both previously used simple wire antenna that is as effective as alternate - and perhaps more 'routable/mountable' in use? Using one of the smallpico's the WiFi scanner was as good or better than ones with 'other' antennas - it picked up neighbor's devices 200 ft next door when 'other' did not.

<edit> This probably worthy of a new thread?
 
I am not sure that wifi, or ethernet, or any such need to be a part of the base board. But, it is certainly nice to have the pins available to connect those interfaces.

While we're at it, what would be helpful for some of my projects, is bluetooth, as in a simple way to connect to a cell phone or laptop and not reliant on or in conflict with a local network

But again, this is a basically an MCU on a carrier and that is an important product. It should not try to be a general purpose pc.
 
Would be a full redesign to mate to a T_4.0 - but it could be done.
I don't see it as a full re-design as most of the pins needed have already been brought out from the MCU to the bottom of the T4.0.
Signal wise all that would be needed is to re-route those pins to the ESP32 on the end of the board.
I appreciate that the 3.3V regulator and it's circuit may need to be revisited as the ESP32 can take quite a large chunk of current when transmitting.
An alternative could be for the ESP32 to have it's own 3.3V regulator.
Yes the ESP32 would have to be laid out but there are any number of examples out there for inspiration (I didn't say Copy!!).
 
Should I re-post #589 in another thread?

Seems like a good plan. Interest and capability go way back and the need is still there.
UNO R4 just came out with an ESP Combo unit using a new ARM M4 that runs at T_LC speed with half the RAM/FLASH of a T_3.6 ... at 5V
 
While we're at it, what would be helpful for some of my projects, is bluetooth, as in a simple way to connect to a cell phone or laptop and not reliant on or in conflict with a local network

Besides WiFi, the ESP32 can use be used for bluetooth/ble. At least I have heard it can be used for either.

-Mark
 
Besides WiFi, the ESP32 can use be used for bluetooth/ble. At least I have heard it can be used for either.

-Mark
It depends on which ESP32 chip you are using. Some only support Wifi. Some support traditional bluetooth. Some support BLE bluetooth. And presumably there are chips that support both. Most ESP32 chips use an ARM chip as the underlying processor, at least one (ESP32-C3) uses MIPS instead Arm. Some ESP32's have 2 cores, some only have a single core.
 
Back
Top