Teensy 4.0 (hypothetical) pin assignments

Yes Theremingenieur, I agree. Too bad about the lack of DAC interface. The ADC's are only 12-bit resolution so that is going in the wrong direction as well.

Paul, after reviewing the manual I wanted to ask you to consider the possibility of bringing out enough signals to support the Camera Serial Interface (CSI). From the IMXRT1050RM manual, Table 19-1 defines the external signals, and the bottom paragraph of page 796 defines the signals required to support 8-bit, 10-bit, 16-bit and dual 8-bit connections. If I am reading the ALT-4 column of your table correctly, you are already bringing out most of the required signals for 8-bit operation. The only signal I think you would have to add is CSI-PIXCLK. See Sections 19.4.2, 19.4.3 and 19.4.4 for the three modes of operation.
Thanks -W
 
I know this may be a little pre-mature since T4.0 isn't off the drawing board but wanted to throw this out there. Will be any plans for redesigning the prop-shield for the T4.0 down the road?
 
mjs513 said:
I know this may be a little pre-mature since T4.0 isn't off the drawing board but wanted to throw this out there. Will be any plans for redesigning the prop-shield for the T4.0 down the road?
Click to expand...

Well the chip not having a DAC may put a damper on some of the uses of the prop shield (i.e. having a mono amplifier and flash memory to play sounds).
 
Well the chip not having a DAC may put a damper on some of the uses of the prop shield
Click to expand...
Was sitting having coffee and thought came across my mind - about pinouts vs propshield - thinking ahead for a very future project :). I just looked around and maybe a DAC could be added but that would require a lot of workarounds I think not having one on board. Thinking out loud again :)
 
some must haves

SDCard
RTC
1-2 - I2C could live with one, but perhaps two would be nice but isn't the point of i2c to be able to run many sensors on one bus? I've had multiple sensors with 2-3 328ps in a i2c network, 2 hw i2c ports is of course amazing to have, but I consider a luxury. However having 2 enables running more than one big item quickly, IMUs, Screens, etc. So perhaps 2 is the min.
1-2 - SPI bus (i hardly use it, 1 would be fine)
1- CANbus
1-2 serial ports UART TX/RX at least one for GPS
A fair amount of PWM - Analog pins but if I had to balance these I'd go with more PWM over analog.

I'd be willing to trade *some* ram space for ram speed, but 256/512 is a really good size. If that could be fast 256/512 or 512/1mb would be amazing and completely worth extra cost.

Important things to me...

5v tolerant is ... extremely nice to have, I'd love to see two versions 3.3 and 5v. But build costs yea... maybe not possible. It's just nice to be able to run everything at 5v.

Floating point horsepower and memory is the most important bits to me and what I do. I do lots of maths for flight, GPS that sorta thing.

If I need more PWM I'll just use a TLC59711 if I just had too.

As a side note, 2.4Ghz RF (not wifi, but wifi would be...meh ok) is important to me, as the things I do are mostly in the air and having RF onboard would be absolutely killer. But I know this would mostly require a longer board footprint so it'll be worth the wait.

But honestly, your designs are amazing and what do I know. LOL T4.0 will be amazing either way. 600Mhz will really be an asset for what I do. As long as it has the general basics of microcontrollers it's gonna be awesome.
 
i doubt rf or wifi would be included on the onboard production due to size constrain and having effect on price point. You should have your own boards for that. For alot of people, 1-2 uarts isn’t enough, plus it wouldn’t make sense to have less than the old gen arduinos. 5V tolerancy is diminishing over time, most if not all current MCUs are at 3.3V now and even moving to 1.8V ...
 
tonton81 said:
i doubt rf or wifi would be included on the onboard production due to size constrain and having effect on price point. You should have your own boards for that. For alot of people, 1-2 uarts isn’t enough, plus it wouldn’t make sense to have less than the old gen arduinos. 5V tolerancy is diminishing over time, most if not all current MCUs are at 3.3V now and even moving to 1.8V ...
Click to expand...

I understand on all points.

Leaving 5v sure wipes out a lot of things just working together without having to have multiple voltages. say 3.3 for the MC and 5v for a screen. But yea, I hear ya.
 
Godfear17 said:
some must haves
Click to expand...

Have a look at mjs513's earlier posts that gives you the best guess so far of what we could expect, if Paul does not carries out 'last minutes' changes.
In the end it is all space management, as the form-factor is fixed and not negotiable.
 
This is not really a thread to request features. This is (or was, time is mostly past now) about which 24 of the 97 pins to bring out to breadboard friendly pins, and with others beyond those might come to bottom-side pads if space permits.

The available feature set is already fixed. Decisions about features are limited to choosing among the features that exist on the iMXRT's available pins.

Believe me, I'm also disappointed NXP didn't put DACs into this chip. Neither is there a radio for wireless communication. Features like 5V tolerance might have been nice too, but it's just not in the silicon.

NXP has already made the chip. They've actually made 2 revisions, which is highly unusual. Semiconductor masks & setup are so expensive (many millions for this sort of chip) that almost any problems are "solved" by just removing those features from the documentation. But the first version had serious limits on the power supply range, so they made a second revision to fix that and about half the errata of the first version. They are planning to make more variants of the chip, so the future is likely to bring other possible features. But for the chip we can buy now (or very soon), the silicon is fixed. We have to take what we get (from NXP) and like it.

This first board isn't going to have any extra chips, beyond the microcontroller, flash memory, bootloader and required power supply.

As I explained in the first message, this thread isn't about discussing future boards, and it's *definitely* not about how to get more pins than fit in the 1.4x0.7 inch form factor! It's about the trade-offs that need to be decided with the chip that's available now.
 
Last edited:
No doubt it'll be an interesting board.

2 things, could you list a feature list and pinout of what it will be?

2nd, which chip are you using? mainly wanting to know what memory it'll have.

ok 3...
3rd, will this be a Kickstarter like before? Kind of sad I couldn't get in on the last one. I've heard about teensy for so long, and never looked into it more. Can't believe that. lol
 
I think, as Paul said for the first release its going to be the 1052. Maybe later versions will go to the 1060. Similar to the T3.x progression. Just a guess.
 
Apologize my ignorance. Will the i.MX RT1052 (or 1060, if or when implemented) perform better than 3.6 in terms of double precision floating point processing, apart from being much faster?
I mean, will be this unit "just" 4-5 time faster than the Teensy 3.6 in double math processing?

Apart from this, thank you for your work Paul and Mike and all guys from PRJC.
Looking forward for a unit to replace ours Teensy 3.6 in our systems (already have hundreds).
 
LeD said:
Apologize my ignorance. Will the i.MX RT1052 (or 1060, if or when implemented) perform better than 3.6 in terms of double precision floating point processing.
Click to expand...


So I was looking up the coremark benchmark results for the 1050/1060 line and it's bench is just under Intel Core 2 Duo 1.2Ghz (200marks slower) and 1000marks (a full 1/3rd quicker) above Raspberry Pi Zero to give you a reference for it's raw power. Like you, floating point horsepower is life and I think this thing will be a beast at it. I mean what else are we going to use at these pricepoints and footprint? Who wants to use a rPi when you have a Teensy 3.x/4.x options now?

Will be an absolute game changer for sure. I think 3.5 with 5v is the killer game changer but the new school 3.3v 3.6/4.0 will really draw a lot of people only needing a fast MC.
 
LeD said:
Apologize my ignorance. Will the i.MX RT1052 (or 1060, if or when implemented) perform better than 3.6 in terms of double precision floating point processing, apart from being much faster?
I mean, will be this unit "just" 4-5 time faster than the Teensy 3.6 in double math processing?

Apart from this, thank you for your work Paul and Mike and all guys from PRJC.
Looking forward for a unit to replace ours Teensy 3.6 in our systems (already have hundreds).
Click to expand...
Note, the Teensy 3.5/3.6 only support single precision in hardware. All double precision is done via software emulation.
 
To confirm … from the 1050 RefMan:
1.3 Features
The i.MX RT1050 processors are based on ARM®Cortex®-M7 Platform, which have the
following features:
Single ARM Cortex-M7 with:
• 32 KB L1 Instruction Cache
• 32 KB L1 Data Cache
• Single-precision and double-precision FPU (Floating Point Unit)
Click to expand...
 
MichaelMeissner said:
Note, the Teensy 3.5/3.6 only support single precision in hardware. All double precision is done via software emulation.
Click to expand...
Thank you, it was my worrying, to have another super-fast chip on what to use double precision computation (necessary for my flight control algorithms) with software emulation that slows a lot big matrices operations.
I did it the same on 3.6, but mathematics optimization was a blood bath.
 
One thing that isn't about pin assignments, but it might be useful in future Teensy's is having the correct resistors so that so-called smart chargers can recognize that the Teensy can accept 2.1 amps instead of 500 mA. I was buying a power strip with USB ports, and I was noticing that all of the power strips with USB ports were listed as being smart. Similarly in the past for USB charger batteries, I could find non-smart USB connections with a little digging, but it seems harder to find them these days. I realize that Teensys would probably might still be limited to Full speed (12 Mbit/s, USB 1.0 speed), but it would be useful to for smaller neopixel setups to know that I can get 2.1 amps instead of 0.5.

I don't know how hard this would be to add to the design. I was under the impression that in the simplest case, it was just two specific resistors on the D- and D+ pins.

<edit>
I read parts of the RT1050 datasheet, and it looks like on pages 3281-3282 the chip does have some basic support for chargers. If the magic configuration bits are properly set, it could draw 0.9A through 1.8A, which at least is a step up from 100mA worst cast scenario.

Obviously, it would have been nice if the chip could support USB 3.0 transfer speeds and higher amperage. Note, I would expect the Teensy 4.0 to have a fuse like the Teensy 3.x/LC's have for power it itself consumes including the 3.3v outputs. But for things like neopixels or servos that want to draw power directly from VIN, it is useful to get as much power as possible.
 
Last edited:
I've only skimmed through this article at DigiKey. It is primarily on USB 3.1 power capabilities and implementing them with a USB-C connector, up to 20V at 5 Amps is possible. It just touches on the battery charging of USB 2.1. Limiting to 5V it is 1.5 Amps or 3 Amps depending on the connector used.

https://www.digikey.com/en/articles...c-and-using-power-delivery-for-rapid-charging
USB Type-C employs pull-up resistors (for a DFP) and pull-down resistors (UFP) on the CC1 and CC2 pins. The pull-up resistor (Rp) determines the DFP’s current provision capacity. A fixed value pull-down resistor (Rd) on the UFP forms a voltage divider with Rp. By sensing the voltage at the center tap of the voltage divider, a UFP can detect a DFP’s advertised current (Figure 4).
Click to expand...
The article references
http://www.ti.com/lit/wp/slyy109/slyy109.pdf
from TI that references all sorts of docs at the end including
http://www.usb.org/developers/powerdelivery/
at USB. It links to
http://www.usb.org/developers/docs/devclass_docs/
which has many relevant docs... The docs under the Battery Charging heading tell what is needed for each connector type.

For a small device using higher voltage and current USB 3.1 power capabilities, take a look at the TS-80 soldering iron from Miniware.

<edit> On the PTC fuse, a tap before it for high current loads.
I think this is for a later Teensy 4.
<2nd edit>The more I read on this, it looks like additional hardware beyond resistors is required for non USB C connectors. It does look like it could be implemented on a small board you plug into the USB port, and it does the power negotiations before it routes the D+ and D- lines to the Teensy USB port. That board would need it's own MCU, or some pins, including an ADC line or two, from the Teensy. It would also require software on the Teensy.

Anyways, I've learned a bit that I will need in the future.
 
Last edited:
PaulStoffregen said:
If that were done, something would need to replace the 500 mA PTC "fuse".
Click to expand...

Yep, if it can't be done without blowing the budget (either build cost or real-estate on the board), it can't be done.

I was just saying that being able to pull 1A (or 2A preferably) of power from a single cheap commercial battery (i.e. USB charger batteries that increasingly limit their output) might be useful to some users, particularly in a small space constrained package that does not take much extra space other than what the Teensy already takes up.
 
We pause this conversation for a comment from the resident crank.

USB 3 energy and current levels are problematic. And the new IEC62368-3 is designed to rein in some the the weird stuff now found in USB 3 cables and connectors.

To put it succinctly, USB3 can be a fire hazard. Have seen some problems in the industrial world with USB safety. TC108 and national bodies are writing stuff to mitigate; and governments are starting to take notice.

Although, have noted that the Klingon High Command has rescinded all energy, current and voltage limits for data cables.
 
You must log in or register to reply here.
Back
Top