Any Chance of a Teensy ++ 3.1?

Status
Not open for further replies.
I'd buy one of those. It seems the K26 spec I saw included those core things.

EK701 said:
- For a ++ product, a 50% larger form factor is ok, as long as all the pins are broken out in a manner that makes them easy to use
Click to expand...

50% would take 28 pins to 42 pins, a good improvement (especially with NVRAM and place for a SD Flash) - Even without new under pads - or an SMD connector.
 
a blast from the past... mister Gerber and the e-Myth. Got to listen to him about 20 years ago. Has some good points re franchising and growing a one man shop.
 
No Cortex-M7 makes me sad, but I see the dilemma. There are STM32 and Atmel SAM options with most of the peripherals, but the ADC's are 12 bit at best, and usually only a single DAC. Not to mention the work involved switching the code base over to a new series. Really is a shame Freescale doesn't have a K27 or something along those lines.

Maybe it is possible that they make a pin-compatible M7 later and you can go from a Teensy3.1++ to a Teesny3.2++ without much effort.
 
Last edited:
It would be great if the next teensy had a pinout corresponding to 3.1 to accomodate the "shields" users here have made. Can bus is a must!
I wouldn't mind if it was say 50% longer to accomodate more pins on the sides and belly or microsd holder.
 
Hows the teensy ++ coming on? DroneScan is now fully committed to Teensy 3.1 for our onboard system but now we need another UART serial port because the drone we are using interfaces through serial, we need 2 serial barcode scanners and we need another serial for our ground station communications. We are using i2c for all of our sensors and SPI for the LCD display. PWM output and input for servos and radio controlled stuff, plus a few other pins for blinking lights and buzzers. so we are full house , and now just need another UART!!!!

pcb.jpg
 
After *much* thought & consideration, I'm 90% confident we're going to use the Freescale MK66FX1M0 chip for a Teensy++ 3.x product.

Freescale's website has complete specs and documentation for their MK66FX1M0 chip. Some highlights: 180 MHz Cortex-M4F (single precision FPU), 1M Flash (with 8K cache), 256K RAM, 4K EEPROM, 2 USB ports, ethernet, SDHC, 2 ADCs (many pins muxed), 2 DACs, 3 SPI, 4 I2C, 2 CAN, 6 serial, 20 PWM, touch sensing.

We'll almost certainly skip the first generation of Cortex-M7 chips and then consider making a Cortex-M7 board when the next generation arrives. As much as I've wanted to go with Cortex-M7, every indication is the first generation will offer only modest performance improvement that will likely come as a high cost increase. I believe the "sweet spot" for Cortex-M7 will arrive farther in the future, as the semiconductor vendors move to under-90nm silicon processes for microcontrollers. In the meantime, this new MK66FX1M0 chip will let us get a high-end Teensy done relatively quickly. My goal is to get this done by Christmas. Of course, everything always takes longer than expected, so please don't consider this as hard deadline. It's more of an optimistic goal. ;)

My plan is a 48 pin form factor board, where the left-side 28 pins exactly replicate the existing Teensy 3.1 & LC form-factor. The extra space on the right hand side will be mostly used for a SD card and some sort of connector for a 2nd USB port.

These chips have about 90 available I/O pins. Only about half will be able to come to the breadboard-friendly edges. I haven't decided what we'll do with the others. A high density connector is unlikely. I'll probably try to bring as many useful signals as I can to bottom side pads, similar to Teensy 3.1. Only one thing is absolutely certain: difficult compromises have to be made!

The K66 chip has an ethernet mac. I intend to bring all the RMII pins to the edges, so an ethernet shield can be made. Good software support for the on-chip ethernet port may take a very long time, but I do want that to be a long-term option, so we're almost certainly going to use the K66 chip instead of the K26.

One minor downside is that none of these newer chips offers 5 volt tolerance.

The other change is power consumption. For speeds above 120 MHz, the chip uses a special "high speed run" mode, which consumes quite a bit more power. It's looking like approx 90 to 100 mA. If the 2nd USB port is used, quite a bit of extra current is required for 480 Mbit/sec speed.
 
Thank you, Paul. Now we know what we can expect.

PaulStoffregen said:
My goal is to get this done by Christmas. Of course, everything always takes longer than expected, so please don't consider this as hard deadline. It's more of an optimistic goal. ;)
Click to expand...

That's fine. Note, though, that you didn't promise any specific year for Christmas... ;)

One minor downside is that none of these newer chips offers 5 volt tolerance.
Click to expand...

Uh, that might be a bummer. Maybe you could add diodes and 5V output for at least some pins?

Two USB ports call for a host mode, can we hope for that? Which brings me to the next question about how you plan to handle the VIN/VUSB issue we discussed last week? Jumper to cut, jumper to solder or something entirely different?

I guess there won't be any castellated holes, right?

Any plans for a debugging connector?

Oh, and will there be a Kickstarter campaign like there was for T3? I'd be more than happy to contribute...
 
Last edited:
PaulStoffregen said:
After *much* thought & consideration, I'm 90% confident we're going to use the Freescale MK66FX1M0 chip for a Teensy++ 3.x product.

Freescale's website has complete specs and documentation for their MK66FX1M0 chip. Some highlights: 180 MHz Cortex-M4F (single precision FPU), 1M Flash (with 8K cache), 256K RAM, 4K EEPROM, 2 USB ports, ethernet, SDHC, 2 ADCs (many pins muxed), 2 DACs, 3 SPI, 4 I2C, 2 CAN, 6 serial, 20 PWM, touch sensing.

We'll almost certainly skip the first generation of Cortex-M7 chips and then consider making a Cortex-M7 board when the next generation arrives. As much as I've wanted to go with Cortex-M7, every indication is the first generation will offer only modest performance improvement that will likely come as a high cost increase. I believe the "sweet spot" for Cortex-M7 will arrive farther in the future, as the semiconductor vendors move to under-90nm silicon processes for microcontrollers. In the meantime, this new MK66FX1M0 chip will let us get a high-end Teensy done relatively quickly. My goal is to get this done by Christmas. Of course, everything always takes longer than expected, so please don't consider this as hard deadline. It's more of an optimistic goal. ;)

My plan is a 48 pin form factor board, where the left-side 28 pins exactly replicate the existing Teensy 3.1 & LC form-factor. The extra space on the right hand side will be mostly used for a SD card and some sort of connector for a 2nd USB port.

These chips have about 90 available I/O pins. Only about half will be able to come to the breadboard-friendly edges. I haven't decided what we'll do with the others. A high density connector is unlikely. I'll probably try to bring as many useful signals as I can to bottom side pads, similar to Teensy 3.1. Only one thing is absolutely certain: difficult compromises have to be made!

The K66 chip has an ethernet mac. I intend to bring all the RMII pins to the edges, so an ethernet shield can be made. Good software support for the on-chip ethernet port may take a very long time, but I do want that to be a long-term option, so we're almost certainly going to use the K66 chip instead of the K26.

One minor downside is that none of these newer chips offers 5 volt tolerance.

The other change is power consumption. For speeds above 120 MHz, the chip uses a special "high speed run" mode, which consumes quite a bit more power. It's looking like approx 90 to 100 mA. If the 2nd USB port is used, quite a bit of extra current is required for 480 Mbit/sec speed.
Click to expand...


Sounds more like a Teensy 3.9 then a ++ :p
I will say the more pins the better, yes you can add expander IC's but they will never be as fast as onboard IO's.
Maybe have some extra pins/wire holes on the side that go up vs down toward a breadboard?
sOxTFLY.jpg


P.S.
Google is failing me when I do a search with the PN :/ Could you toss us a link Paul.
 
The other change is power consumption. For speeds above 120 MHz, the chip uses a special "high speed run" mode, which consumes quite a bit more power. It's looking like approx 90 to 100 mA.
Click to expand...
Is current draw significantly less at 120 MHz or below?
 
There's pretty much zero chance any Teensy++ design will make the SDRAM or FlexBus features usable. Far too many pins are needed, and some of them mux with really important functionality.
 
I found normal softserial on normal arduino was good at sending but poor at receiving serial so I didn't try it again.
Will try again on teensy. Most things work better on teensy.
3.3v is a problem for us.
all of our i2c sensors are 5v. Our serial scanners are 3.3v but 5v tolerant
our led display is 5v
Our serial comms is 5v but we could switch to 3v xbee
Our pwm servos are 5v

Maybe make a 3v side and a 5v side of the board?

How's your RF pcb design skills? Can you include a wireless chip on board? They are pretty cheap now. Just have a spot where we can connect an antenna.

something with this kind of functionality:
WiFi Module - ESP8266
https://www.sparkfun.com/products/13678
 
Last edited:
Lots of WiFi w/ARM, or WiFi + ARM.

Opinions alert:
ESP8266 is way too garage-shop for me. Read their docs (I can read only the parts in English and that is terse).

Particle.io Photon is what I'm fiddling with in this realm. Their servers and IDE choices are great. Uses Broadcom WiFi. They tried and dumped T.I.
This is the future, not more DIY libraries; Get libraries from chip vendors. I'm using the ST "HAL" libraries, and despite claims, I find them complete and reliable. To include DMA/Interrupts in all peripherals, even pin change, PWM tables, UARTs, SD cards, SDIO, ADC, DAC, I2S, timers, GPIO, etc. uSD card w/SDIO 4 bit (DMA based), rather than SPI, and with FATFS was just a plugin, no struggles. No new code. Freescale is way behind in comparison.

Wrapper APIs for "Wiring" and selected Arduino compatibility has been done by many in ARM-land. Teensy 3 has some amazing AVR source code compatibility via an incredible effort in C++ operator overloads, etc. But 8 bit/AVR compatibility can be left behind, IMO, for the kind of users interested in ARM processors.
 
Last edited:
PaulStoffregen said:
My plan is a 48 pin form factor board, where the left-side 28 pins exactly replicate the existing Teensy 3.1 & LC form-factor. The extra space on the right hand side will be mostly used for a SD card and some sort of connector for a 2nd USB port
Click to expand...

so left side = where the usb connector is? so that'll consume another ~25 mm or so on the right ...

any chance you'd consider parallel pins rather than the elongated form factor? (as in Donziboy2's image above? or stm32 disco, nucleo, etc?). mainly worried about drop-in replacement/compatibility when space is tight; i'd guess that would be less of an issue in less cases if teensy3++ was a bit wider, rather than longer. sd card and second usb port would be nice, but why not add them off board when needed?
 
Sounds good, esp. i like the on board SD-slot.
More RAM + more SPI is great, and the bigger cache will give a good speedup.


Ethernet: Nowaday everyone uses WLAN on cheap Breakoutboards, i'm not sure if this is really needed. But the device costs nothing, so why not.
Paul, did you see the errata+workaround regarding USBReg ?
 
Frank B said:
Ethernet: Nowaday everyone uses WLAN on cheap Breakoutboards,
Click to expand...

There is a sudden resurgence of interest in IoT and while the term is over-used, it has created an expectation that connectivity is pretty much required of most projects now.

The less external connections and pins the better, if the comms board can be onboard its 2 less power connectors i need to worry about and 2 less RX/TX pins that need to be exposed, plus maybe it frees up a UART because those things often use a uart port for comms.

I don't really care which exact chip is used to connect to the wifi, i would like an easy way to connect, either as a network client or as a network server serving up a web page, the ESP8266 offers that kind of functionality with little effort from the programmer, so if we can include all that, fantastic, and from a marketing perspective you are then at the same level as the intel galileo and edison IoT offerings.
 
K66 pin wish list

Concerning the 90%prob K66 T3.x, and knowing that compromises have to be made, and different users have different requirements, I put forward my essential I/O requirements, that I plan to use simultaneously

- all I2S ports (MCLK, RX_BCLK, RX_FS, RXD0, RXD1, TX_BCLK, TX_FS, TXD0, TXD1), i.e all four data channels plus all clocks
- one SPI port
- 2 to 3 UARTS
- 5 to 8 digital lines
 
PaulStoffregen said:
After *much* thought & consideration, I'm 90% confident we're going to use the Freescale MK66FX1M0 chip for a Teensy++ 3.x product.

Freescale's website has complete specs and documentation for their MK66FX1M0 chip. Some highlights: 180 MHz Cortex-M4F (single precision FPU), 1M Flash (with 8K cache), 256K RAM, 4K EEPROM, 2 USB ports, ethernet, SDHC, 2 ADCs (many pins muxed), 2 DACs, 3 SPI, 4 I2C, 2 CAN, 6 serial, 20 PWM, touch sensing.
Click to expand...

Fantastic, a very logical decision. The MK66FX1M0 package options seem very challenging, I wish you good luck with that.

I see the Ethernet block shows IEEE1588 support which is exciting; I'm happy to help beta test any functionality in this area.
It probably won't affect your design, as there will be an ethernet shield, but please keep in mind anything required for PoE functionality.
 
Last edited:
Awesome.

[EDIT] I've just re-read the original post again and realised that this has already been covered, so this post is a bit redundant.


One request is that, somehow, more pins are brought out to an easily use-able form. Whether that's castellated edges, or more through hole pins etc.
To maintain Teensy'esque form factor, maybe, as above, a second row of pins on each side would be viable. Keep T3.1/LC style pinouts on the inner rows, then bring out more pins on the outer. If you have the PCB space, could then also bring out additional pins to pads underneath too.

With the insane amount of hardware peripherals on that chip, I think it would be a shame to restrict it's usability by only bringing out a subset of the pins to be easily useable.
 
Last edited:
PaulStoffregen said:
My plan is a 48 pin form factor board, where the left-side 28 pins exactly replicate the existing Teensy 3.1 & LC form-factor. The extra space on the right hand side will be mostly used for a SD card and some sort of connector for a 2nd USB port.
Click to expand...
regarding the onboard SD card slot. Side mounted is good. Of interest would be the potential for accessibility to the card if the Teensy is embedded in a project. That is to say, would the format allow for the sd card to be proud of the teensy, allowing access to the sd card and the usb slots if the teensy is within a housing, such as thin walled project boxes.
 
Status
Not open for further replies.
Back
Top