Suggest next Teensy with Cortex M7

Status
Not open for further replies.
Nope, sorry, anything that's not on NXP's public website is covered by NDA. They will publish all the info (unlike certain Chinese chips...) but probably not until the chips are actually for regular sale.
So, I do not ask for more info on the chip (I can wait till October, or so)
I've read
https://blog.nxp.com/iot/i-mx-rt-series-the-crossover-between-applications-processors-mcus
http://www.nxp.com/docs/en/fact-sheet/I.MXRT1050FS.pdf
so I know the basic features, I'm interested in

but at which level of your priority list is a T4.0 prototype. I assume you will make a similar beta testing device as you did with T3.6.
 
I'm with the likes of jwatte and epyon in thinking that an MCU of 600MHz seems like overkill for all use cases I'd ever consider using an MCU for. I absolutely get that it's necessary for some but what percentage of the Teensy user base needs such speed? What cost in terms of ecosystem complication and reliability and rate of evolution will come with that move?

I think you'd be better off improving what you already have. I'm not just talking about things like Ethernet on the 3.5/3.6 but the whole Teensy ecosystem. The whole Teensy story told by the website for people trying to enter the world of Teensy could do with huge improvement - what it's all about, a coherent overview of all the products, the supporting software components, how I start, how I progress/troubleshoot, reference resources etc. Much of the documentation is woefully out of date - even the main Teensy Products page specification comparison table stops at 3.1. Many other pages don't progress beyond Teensy 2.0.

Perhaps I've got it completely backwards. Is PJRC making money from Teensies or are they making money from their own use of Teensies and we're just lucky enough to get cheap access to the great tools they've devised to get their own work done?

Having migrated from Arduinos, I think Teensies are a huge step forward but I think there's so much more that could be done to encourage and assist people making the transition.

Okay, now bringing it back to desired future hardware features, I'd prefer to see less peripherals on the boards. I would be happy if the main boards were limited to exposing the MCU's functionality. All of my 3.5s and 3.6s sport micro SD sockets that will never be used. I'd prefer it if such things were left to shields. One of the best things PJRC do for the maker community by providing Teensy boards is take away the electrical engineering pain of getting an MCU to interface with other things. Previously, I've disappeared down the rabbit hole of trying to provide good clean power and signal lines on a board design and after just about losing my sanity I have a huge amount of respect for those people who take care of such things so that I don't have to. In other words, don't underestimate just how far along the curve you bring us numpties just by providing good, clean, reliable access to the MCU's functionality.

I'd also like to cast a vote for quality over cost. Given the hardware cost relative to the labour cost in a project (1:100, 1:1000?), I'd far rather you spend a few extra pennies/dollars going for the best quality components you can get so that when a project is deployed the MTBF is higher and the rate of incidences of glitches and gremlins is lower.
 
don't think of 600mhz as overkill, lots of people are doing audio and video progressing now and video games that normally a raspberry pi was chosen over arduino to handle the task, then others with video walls of pixel strips to display which, all rely on cpu and memory usage.
 
Hi tonton81, note I said I consider 600MHz to be overkill for all use cases that I'd consider doing with an MCU - I acknowledged that other people will have different use cases but my main point was asking the questions: what percentage of users need 600MHz and would the majority of Teensy purchasers be better served with effort being directed elsewhere (and those are questions, not statements). I also wonder if the vast majority of Teensy purchasers are silent and the desired direction of development (that's for you alliteration lovers) is skewed by a very vocal minority. Regards,
 
don't think of 600mhz as overkill
Also not for me.
My application requires sampling of 4 ultrasonic audio channels at 375 kHz each, dual band downsampling to 37.5 kHz resulting in 8 channels at 37.5 kHz, followed by basic sound analysis. All this in real time. 600 Mhz with potential oversampling seems the minimum, but using 2 T4 instead of 4 T3.6 is much easier (less power, less communication bottlenecks)
 
@itinerantAlien: Well, the Teensy LC and 3.x will still be available, and you can use them. Nobody says you have to buy a 4.x

SD: Some time ago - and it still happens, just read the forum - there are many beginners who do not get SD over the slow SPI interface to work because they use "Arduino" SD breakouts with ominous level shifters. They all think, the Teensy is the problem. SDIO with a breakout is even more difficult. I am very glad that the new Teensies have this socket - even if I do not need it in any case. But when I need it - I just use it - without fiddling..
 
Last edited:
@FrankB, I'm sure we're both clear that it's not as simple as just opting out of buying the new board but I'll play along. PJRC are a very small team and if they devote resources to the development of a new board then they will have less resources to work on fleshing out their existing ecosystem. I'm not denying your, or anyone else's, need for such a board. I'm not denying your, or anyone else's, right to vote for such a board. I would ask that you respect my right to vote differently; that resources instead be spent fleshing out the existing ecosystem.

I have absolutely no issue with PJRC going ahead with the new board if and when they feel that's best for their business and their customer base overall. I'm grateful for having access to such great tools at such low prices and I also acknowledge that without the contributions of forum members the ecosystem would be much the poorer so I'm grateful to everyone who has made an effort. I just hope that PJRC give careful consideration of all their customers' views.
 
How about they do both? Work on the website and other less fun stuff and also develop a Teensy 4, you know walk and chew gum at the same time. I know I've slammed the website and online resources more than once but products are what sells and puts money in their pockets. Maybe they can temp hire a front end web developer to get the forum, website and online resources in place so that customers can really use it in intuitive way so PJRC can focus one the good stuff. I think the Teensy 4 (M7 put in Application Processor chassis) sounds amazing even if I don't have use for it yet. This would put PJRC in a position of real innovation in the Arduino world.

One thing that I hope is still on the blackboard is some type of "debug integration" into Arduino IDE. That would also really define PJRC as a leader in Arduino world and maybe some of the naysayers on Hackaday would respect the Arduino IDE a little more, probably not :)
 
I wasn't planning on jumping in here but from what I remember PJRC is only 3 or 4 people. Not many for what they have accomplished and what they are continuing to do. I am into robotics, machine vision, audio for those projects which can be very memory and computational expensive. These projects require something with the speed that new teensy promises. Whether the new Teensy 4 will be based on the exact chip Paul specified in the beginning of this thread, don't know. Personally, I would like to see the speed increase, memory increase with a capability to handle images or even a gpu on board :) and still with all the connections that Teensy is famous for. Since the T3.5 has come out I don't think I have touched any of my other Arduino boards - doing everything with the T3.5 now. Even though I still have a fondness for my Mega.

As for the web site it looks like Paul is trying to get it updated, and if he ever did a kickstarter type campaign for a web designer to overhaul the site I would definitely join in. Also, I haven't seen any of us volunteer to help PRJC overhaul the site, i.e., work on reorganizing and design so let's give them a break on this. It takes a lot of time and effort to do a website, I know, I used to develop them as a small part of my job.

I understand everyone's desires, and its good to let them to be know since there are a lot of good ideas that come out of them so please don't take me wrong here.

Anyway, this is just my two cents and no need to respond. Just had to get it out.

Respectfully.
 
@mjs513, by your comment "so let's give them a break on this" you seem to have taken a defensive stance on PJRC's behalf which makes me think you believe we're just sticking the boot into them. I'm most definitely not. As I stated before I'm incredibly grateful for everything in the ecosystem and to all who have contributed. But, if it is a business (and that's a genuine question, as I see PJRC do at least some consulting work so for all I know maybe the Teensy ecosystem is just a very generous gift to the maker community), then like all good businesses they will welcome constructive feedback. The feedback I gave, which was about a lot more than the website, was intended to provide some possible constructive ideas on how the business could be improved and if the business is improved then presumably more profit flows to PJRC, which I'm sure we can agree is a good thing.

As for volunteering to help improve things, I'd be happy to lend a hand. I don't have any experience with web design but I can help with content.

Regards,
 
As for volunteering to help improve things, I'd be happy to lend a hand. I don't have any experience with web design but I can help with content.

I would be willing to assist as well. I think others would also if PRJC desires it.
 
Last edited:
As I saw the SD/SDIO adapter on the T_3.5/3.6 - it was a natural fallout of what would otherwise be 'void/wasted' PCB space as extra pins and pads came off the K64/K66. It pulls out SDIO pins that can be used for other purposes - provides a very commonly used interface medium of high bandwidth and capacity. Others have proven those pins are still usable in other ways - and if you pulled SD adapter off - that space is still taken up by routing and other lines that render that area of the PCB off limits - it can easily be ignored - or used as native hardware better than an ad hoc shield could.

The new T_3.5/3.6 are not fully built out yet - but the work done has improved and expanded what works on T_3.2 and T_LC - they were not forgotten - they were enhanced at the same time the newer units added buses and RAM/FLASH presented to make use of.

To the degree possible I'd expect a T_4 to offer similar compatibility and enhancement to existing Teensy family - otherwise I doubt PJRC would see value in splitting efforts only to cannibalize existing proven hardware that is still better than the Arduino standard.

And yes ...
Bring on the megahertz. More power, more better.
 
Indeed there are only so many hours in every day. I'm weighing options and difficult choices...

The website does need a *lot* of work. The recent changes were meant to be a tiny first step. More such steps will be coming.

But no matter how successfully everything goes, I'm pretty sure PJRC won't ever manage to do anything about Hackaday naysayers. They always coulda dunnit better, probably using only a 555 timer. ;)
 
@Paul - My take on it, is you should do whatever projects you think are good for your company, especially ones that interest you, such that is keeps you excited and engaged.

I have been involved with some other smaller electronics/robotics companies, where the owners burned out and/or lost interest and have either folded, sold out, or tuned out. Example with one company I still do things with, overnight you have probably posted more responses to questions than that entire company has this entire year!
 
Went by Arduino.. yesterday to see 1.8.4 release notes - saw the Zero thing - Yikes - more ram/flash than a T_LC - more $$$ and less of most everything else. They seem to have the largish teenie sized one with SD onboard - but that isn't on the uno sized one it is supposed to 'work like'? The T_LC by comparison is impressive - especially with the level of T_3.2 interchangeability - and it is still cheaper than ZERO and much better. And that doesn't speak to interface libraries and support on the SINGLE buses ZERO is equipped with.

PJRC hardware is good or better - then there is the T_3.5 & 3.6! Evolve it as you can as you see the exciting need to for Audio or whatever as KurtE says to stay motivated.

The website is functional - minor improvements in ordering and linking could make it a good fixed resource. It seems I noted before :confused: - a Wiki would be the best complement to the FORUM with a more structured resource of detailed and current info i.e. whitepapers on usage and examples. Perhaps that is scary having it represent the state of Teensy in the hands of the mob or something?
 
Last edited:
But no matter how successfully everything goes, I'm pretty sure PJRC won't ever manage to do anything about Hackaday naysayers. They always coulda dunnit better, probably using only a 555 timer. ;)
Haha, maybe the website needs a 555 timer somewhere.
 
Here's the 555
Code:
const int triggerPin =  A1;
const int resetPin = 12;
const int threasholdPin = A3;
const int dischargePin = 10;
const int outPutPin =  13;

byte timerState = 0;

void setup() {
  pinMode(triggerPin,INPUT);
  pinMode(resetPin,INPUT_PULLUP);
  pinMode(threasholdPin,INPUT);
  pinMode(dischargePin,INPUT);
  pinMode(outPutPin,OUTPUT);
}
void loop() {
  if (digitalRead(resetPin)==0) timerState=0;
  else {
      if (analogRead(triggerPin)<1024/3)
                  timerState =0;
        //threashold overides trigger
      if (analogRead(threasholdPin)>1024/3*2)
        timerState =1;  
  }
digitalWrite(outPutPin,timerState);
if (timerState==0) pinMode(dischargePin,INPUT);
else pinMode(dischargePin,OUTPUT);
delay(1);
}
 
Here's the 555
Code:
const int triggerPin =  A1;
const int resetPin = 12;
const int threasholdPin = A3;
const int dischargePin = 10;
const int outPutPin =  13;

byte timerState = 0;

void setup() {
  pinMode(triggerPin,INPUT);
  pinMode(resetPin,INPUT_PULLUP);
  pinMode(threasholdPin,INPUT);
  pinMode(dischargePin,INPUT);
  pinMode(outPutPin,OUTPUT);
}
void loop() {
  if (digitalRead(resetPin)==0) timerState=0;
  else {
      if (analogRead(triggerPin)<1024/3)
                  timerState =0;
        //threashold overides trigger
      if (analogRead(threasholdPin)>1024/3*2)
        timerState =1;  
  }
digitalWrite(outPutPin,timerState);
if (timerState==0) pinMode(dischargePin,INPUT);
else pinMode(dischargePin,OUTPUT);
delay(1);
}

Any schematics to use the T3.6 as an oscillator?
something like this?
https://en.wikipedia.org/wiki/File:555_Astable_Diagram.svg
330px-555_Astable_Diagram.svg.png
would it work?
what about control pin?
 
Last edited:
LOL ok.. but guess what..some are using microcontrollers as gates (and/or/not...) or flip-flops, counters..

you'll need two teensy for an oscillator (the gate variant);-)

if you have enough teensys, maybe it's possible to create a cpu ??
 
Last edited:
@wmxz, that's the schematic I worked from coding it up, but didn't implement the control pin, just used 1/3 and 2/3 VCC since most classic 555 circuits don't use it either.

Unfortunately couldn't work out a way to abuse things enough to actually have it work on pins 1-8.

So two roughly equal resistors and a cap all series with cap going to ground, pin 10 is discharge and goes between the resistors, pins A&1A3(15&17)both go on the second resistor/cap junction for the classic blinky. Was surprised how simple the emulation code for it was.


@Frank B - have run into 2000s era PICs being used to emulate 1980s logic ICs in obsolete equipment. Only works from back in the day when user manuals had more pages of drawings than of operating instructions.
 
I'll take another 30 ADC's and a side of GPIO's, can I get extra ram with that?
For that matter I could use more of the pins that aren't broken out on the T3.6..... :rolleyes:
Teensy 3-6.JPG
 
tbh i'm still sticking with the teensy 3.2 until a proper rtos support is integrated (the teensy 3.2 has the freertos port for the arduino ide!)

some things i wonder about the mcu for the next gen teensy mcu: (maybe i can ask nxp directly for these questions)
why 'only' 512k ram, ideal would be to have 1k per mhz for me (so 600k or more)
what would the power consumption be of this mcu / mhz?
does it require a dual voltage (like 1.1v for core + 3.3v for logic)?
is it possible to execute from external sdram since it is mentioned to be support?
which packages are for the mcu? (qfn, tqfp, bga, ect)

now software questions for paul:
as you mightve noticed, i'm still developing on my teensy 3.2 because it has (unofficial) RTOS support, for the power of the T4 rtos support is a must basically, will this be integrated?

and hardware:
i know it takes quite some time to finish the hardware design, but i am planning to actually use something like this in a project of mine, will the bootloader ic available as well upon launch? (it all depends on the rtos support as well)
will the teensy 4 dev board have a external sdram IC for framebuffers and such (would be nice to have, since it's stated the mcu cost is bought down by 50%)?
will there be a reference board design uploaded? (for us to tinker with and design upon?)
 
Status
Not open for further replies.
Back
Top