PDA

View Full Version : Teensy 3.2



Paul
09-15-2015, 02:37 PM
Today we're happy to release Teensy 3.2.

http://www.pjrc.com/store/teensy32.html

Version 3.2 is a minor upgrade to Teensy 3.1. The main change is an improved 3.3V regulator, to allow Teensy to directly power ESP8266 Wifi, WIZ820io (W5200) Ethernet, and other power-hungry 3.3V devices.

We're specifying Teensy 3.2's power output at 250 mA and the maximum voltage input at 6 volts, due to PCB thermal dissipation limits. However, the actual regulator chip is capable of up to 10 volts input, and up to 500 mA output. These higher limits are intended to allow Teensy 3.2 to be more rugged when used with non-USB power sources which aren't well regulated 5 volt sources.

Teensy 3.2 is fully compatible with all shields and add-on boards designed for Teensy 3.1. It preserves the same size, pinout, and processor as Teensy 3.1.

The bootloader chip is also changed from Mini54 to KL02 (the same as Teensy-LC). A full schematic is available:

http://www.pjrc.com/teensy/schematic.html

vitormhenrique
09-15-2015, 03:08 PM
I wish you had included a 74LV1T125 on 3.2. I really like that feature on teensy-lc! But beeing a dirrect replacement for teensy 3.1 and that there is not much space left on teensy I can understand why not.

WMXZ
09-15-2015, 03:09 PM
Today we're happy to release Teensy 3.2.

http://www.pjrc.com/store/teensy32.html

Version 3.2 is a minor upgrade to Teensy 3.1. The main change is an improved 3.3V regulator, to allow Teensy to directly power ESP8266 Wifi, WIZ820io (W5200) Ethernet, and other power-hungry 3.3V devices.

We're specifying Teensy 3.2's power output at 250 mA and the maximum voltage input at 6 volts, due to PCB thermal dissipation limits. However, the actual regulator chip is capable of up to 10 volts input, and up to 500 mA output. These higher limits are intended to allow Teensy 3.2 to be more rugged when used with non-USB power sources which aren't well regulated 5 volt sources.

Teensy 3.2 is fully compatible with all shields and add-on boards designed for Teensy 3.1. It preserves the same size, pinout, and processor as Teensy 3.1.

The bootloader chip is also changed from Mini54 to KL02 (the same as Teensy-LC). A full schematic is available:

http://www.pjrc.com/teensy/schematic.html

ordered a couple of them.

riodda
09-15-2015, 04:44 PM
Today we're happy to release Teensy 3.2.

http://www.pjrc.com/store/teensy32.html

Version 3.2 is a minor upgrade to Teensy 3.1. The main change is an improved 3.3V regulator, to allow Teensy to directly power ESP8266 Wifi, WIZ820io (W5200) Ethernet, and other power-hungry 3.3V devices.

We're specifying Teensy 3.2's power output at 250 mA and the maximum voltage input at 6 volts, due to PCB thermal dissipation limits. However, the actual regulator chip is capable of up to 10 volts input, and up to 500 mA output. These higher limits are intended to allow Teensy 3.2 to be more rugged when used with non-USB power sources which aren't well regulated 5 volt sources.

Teensy 3.2 is fully compatible with all shields and add-on boards designed for Teensy 3.1. It preserves the same size, pinout, and processor as Teensy 3.1.

The bootloader chip is also changed from Mini54 to KL02 (the same as Teensy-LC). A full schematic is available:

http://www.pjrc.com/teensy/schematic.html

Many thanks for your precious work Paul ! may i put on the teensy 3.3 wish list a 12v capable regulator a canbus transcriver and i2c pullups ?
cheers !

Constantin
09-15-2015, 08:56 PM
Hey that's awesome!

So now that the Teensy 3.2 is released, do you guys have any plans for a Bed of Nails for the MKL bootloader chip? I'd love to use that bootloader chip, esp. if one chip can handle either a 3.2 or a LC MCU.

Also, for teensy DIY applications that do not envision USB host mode, I presume pin 4 on the USB connector can continue to be ignored and PTA6 pin on the KL02 and the PTA1 pin on the MK20 can be left floating? I presume this is the case but wonder if the two (MKL02/PTA6 and PTA1/MK20) can be left to float by themselves or if they require a connection between the two even if USB host mode is not wanted?

jonr
09-15-2015, 09:16 PM
Nice incremental update. Maybe those vias around the KL02 will make hacking in SWD easier.

Frank B
09-15-2015, 09:23 PM
Great.
I hope they will be available in european stores soon!

PaulStoffregen
09-15-2015, 09:38 PM
We shipped to several European distributors last week, so Teensy 3.2 should be appearing soon.

PaulStoffregen
09-15-2015, 09:47 PM
So now that the Teensy 3.2 is released, do you guys have any plans for a Bed of Nails for the MKL bootloader chip? I'd love to use that bootloader chip, esp. if one chip can handle either a 3.2 or a LC MCU.


Yes, this is planned.

This weekend I built the programmer, but unfortunately we had a mishap here which damaged the ZIF socket. I've got another socket on order yesterday. It's a special part that takes a few weeks to get.

This definitely is coming. The chip will be available in 2 sizes, the extremely small 3 mm QFN, and a large 7 mm TQFP with "easy" 0.8 mm pitch pins.

It will automatically detect if you've connected a MK20 or MLK26 and implement either Teensy 3.2 or Teensy LC. Later when we add a 32 bit Teensy++, the firmware will be updated to automatically detect the MK66 chip too.

Budmo
09-15-2015, 10:08 PM
The bigger button looks appealing! Can heat sink be added to improve voltage regulator chip thermal dissipation?

Constantin
09-15-2015, 10:20 PM
Thank you so much. I will revise the next turn of the board for the new chips. Progress has been so slow.... (argh!)

How about the question re: Pin 4 and separate floating pins vs. joining MKL02/PTA6 and PTA1/MK20 but leaving the joined pair floating?

PaulStoffregen
09-15-2015, 10:29 PM
Can heat sink be added to improve voltage regulator chip thermal dissipation?

Maybe, but it'd be quite a challenge.

Perhaps the most effective location would be on the back side of the PCB. Two vias are located close to the regulator's thermal pad, which provide an all-metal path from the silicon.

On top of the chip would be the other possible place. There you've got the plastic package between the heatsink and silicon, but it still might be pretty effective at cooling. A huge challenge would involve the height of nearby parts, like the PTC fuse and 2.2 uF capacitors. They're taller than the chip. The heatsink would need to be extremely small (3 mm) or a special shape to avoid those nearby parts.

As with all heat dissipation, force air flow makes a dramatic difference. Crazy as as a fan may seem, if you're really needing more than 250 mA, maybe something else using that power is also warming up?

If you don't need too much more, you might also just give it a try, with Teensy oriented in different position (possibly with well placed vents, if inside an enclosure) to allow a little natural convection air flow. The 250 mA number we're publishing is pretty conservative. The assumption is we can't predict a lot of factors about how people will actually use Teensy. Odds are good slightly higher current can work fine in your application. I personally tested about 1/3rd of an amp with 5.1 volts input, resting horizontally exposed to open but still air, with pins soldered and plugged into a breadboard. It got pretty warm to touch, but nowhere near the chip's rated temperature.

PaulStoffregen
09-15-2015, 10:35 PM
Also, for teensy DIY applications that do not envision USB host mode, I presume pin 4 on the USB connector can continue to be ignored


Yes, there's no need to connect pin 4 on the USB if you'll never use host mode.

Whether host mode software is ever published is also good question. It might happen someday... but with the anticipated 2nd USB host-mode port on the upcoming Teensy++ board, investing a lot of time into host mode support on the main USB port is looking less and less likely.



and PTA6 pin on the KL02 and the PTA1 pin on the MK20 can be left floating?


I would recommend keeping that connection to the MK20 chip. Today it's unused, but I can't guarantee it will always remain unused by future firmware.

defragster
09-15-2015, 11:58 PM
Sweet - more power! . . . order underway . . . adding SD adapters . . .

Just clicked this for no reason but, On the pins page this is missing? : http://www.pjrc.com/store/teensy32_pins.jpg

PaulStoffregen
09-16-2015, 12:51 AM
Yup, the picture was taken... still on the camera. Will get it onto the site tomorrow morning. It's time for a beer!

defragster
09-16-2015, 01:54 AM
Enjoy it - general/shared consensus seems to be you earned it :-)

Is this why OSH was out of Purple boards? Are they going to publish the 3.2 in Purple?

<edit>: order placed/shipped - now I have spare spares to last until the FPU version ships.

Frank B
09-16-2015, 09:56 AM
exp-tech (Germany) sells it:

http://www.exp-tech.de/mainboards/teensy?manufacturer=68

Theremingenieur
09-16-2015, 10:43 AM
... and I hoped to get a second integrated DAC :(

Cosford
09-16-2015, 12:24 PM
All I can say is; 'Neat'.

Constantin
09-16-2015, 03:49 PM
Yeah, it's pretty awesome, addressing the most common failure mode while keeping the same form factor. The new board is packed to the gills!

Spent some time this AM to design a new board using the MK20 and the MKL02. Got it down to 2 jumpers/0 Ohm resistors/or vias to make it work on a 2 layer board.

5087

I will be releasing a new set of libraries shortly with all bootloader chips (MiniTAN and MKL02). I have yet to put together the TQFP package for the MKL02, however..

defragster
09-16-2015, 04:45 PM
How about a topside heat sink? 14ga(2mm^2) copper wire flattened (1x3mm) and shaped to cross over the back edge of the USB shell? Thermal paste on chip contact and soldered to shell should be stable and low profile.
5088

With $3 shipping mine won't be here before Saturday.

>> opps: Expected Delivery Day:Friday, September 18, 2015 :( >> Saturday . . . .

pictographer
09-16-2015, 05:08 PM
This is a heat sink, not a heat pipe (https://en.wikipedia.org/wiki/Heat_pipe). A tiny heat pipe for this application would be awesome, though I don't know where we'd get them.

A heat pipe is hollow and has contents that change phase. Also the interior is typically shaped to provide increased surface area over a simple pipe either by grooves or sintered metal power.

defragster
09-16-2015, 05:52 PM
Dutifully corrected the nomenclature. My concern was if there was any risk in the design - and worth the trouble? With plastic shell and no nearby things to short, paste to mate and fill voids - don't want to make an antenna or such other issues. The sink mass seems significant enough and cheap/easy - especially if the usb shell gets involved. MarkII design rather than the vertical bend, the height difference might be better done with a flat top and the wire folded under to come down to the chip top. Flattened and shaped for contact - solder the side of bent under end to make sure the full bent piece won't fall free.
5089

Markus_L811
09-16-2015, 08:53 PM
The bootloader chip is also changed from Mini54 to KL02 (the same as Teensy-LC).[/URL]

Why did you change the bootloader chip?

Markus_L811
09-16-2015, 09:07 PM
exp-tech (Germany) sells it:

http://www.exp-tech.de/mainboards/teensy?manufacturer=68

Looks more like an T3.1 from a different angle.

OT.

Damn I know everyone shorts Teensy to "T" in the forum and the numbers is counting up but I'm feared a little bit the T800 has a bad reputation, so may Paul can make curve around the number.

Constantin
09-16-2015, 09:50 PM
Why did you change the bootloader chip?

Presumably, one bootloader chip to rule them all... as the MKL02 can be programmed to support all Teensy MCUs (3.2, LC, and the upcoming 3++).

The MKL02 is smaller than the 4x4mm version of the MiniTAN54, allows the retrofit of the external voltage regulator on the Teensy 3.x series boards.

Also, one less bootloader chip for Paul to maintain?

Constantin
09-16-2015, 09:52 PM
Looks more like an T3.1 from a different angle.

Yes, you are correct. You can clearly see the MiniTan54 bootloader chip with it's plethora of pins...

PaulStoffregen
09-17-2015, 12:07 AM
Why did you change the bootloader chip?

Mainly because the Mini54 is terrible. Back in early to mid 2012, when Teensy 3.0 was developed, the Mini54 and a NXP part were the only viable options. The NXP part was prohibitively expensive.

The KL02 chip is superior in every way, except it didn't exist until mid 2013. Even after it was "released", Freescale didn't ship production volumes for many months. I've been planning to switch to it for nearly 2 years. The change was delayed several times (and more Mini54s purchased) while other things, like the audio library & board, took priority.



Also, one less bootloader chip for Paul to maintain?

So much this!

The reality is I've strongly resisted adding features to the bootloader because the Mini54 is such a nightmare. Especially switching between clock speeds and power modes involves working around terrible bugs in the Mini54 silicon. Nuvoton's broken-English documentation leaves a lot to be desired, as does their reluctance to give honest answers to any questions about the shortcoming of their products. Some things about the Mini54 I even had to discover by reverse engineering.

Freescale's terse reference manuals won't win any writing awards, but they are complete and accurate and written in proper English. Freescale documents their errata. Freescale is also simply much, much better at designing processors and microcontrollers. Those little design decisions don't show up on a high-level comparison of tech specs and features, and most of things you'd take for granted and never appreciate... until you suffer using a not-as-well-designed part.


Presumably, one bootloader chip to rule them all... as the MKL02 can be programmed to support all Teensy MCUs (3.2, LC, and the upcoming 3++).

Yes, the new chip will allow supporting all board from a single chip with a single unified firmware.

The KL02 also will (hopefully) allow me to work on some really awesome new bootloader features in 2016 and beyond. Some of these are things I wanted to do all the way back before the initial release of Teensy 3.0 in September 2012.



The MKL02 is smaller than the 4x4mm version of the MiniTAN54, allows the retrofit of the external voltage regulator on the Teensy 3.x series boards.


Yes, the reduction from 16 to 9 square mm helped a lot. So did the change from 0.4 to 0.5 mm pitch. That and other improvements in the manufacturing process offset the cost of the new voltage regulator.

phangmoh
09-17-2015, 06:42 AM
Paul,

By adding LP38691 it would also mean the quiescent current will increase even when the Teensy goes to sleep which adds up like around 55 uA.
Is this correct?

duff
09-17-2015, 12:16 PM
Paul,

By adding LP38691 it would also mean the quiescent current will increase even when the Teensy goes to sleep which adds up like around 55 uA.
Is this correct?
I asked this too but never got response!

PaulStoffregen
09-17-2015, 12:23 PM
Yes, the regulator adds a quiescent current, regardless of Teensy's operating vs sleep modes.

PaulS
09-17-2015, 02:39 PM
Hi Paul,
Is the schematic correct? Can the VOUT33 pin also be used as an input to power the internal USB circuitry?
5097
Regards,
Paul

Constantin
09-17-2015, 03:55 PM
Is the schematic correct? Can the VOUT33 pin also be used as an input to power the internal USB circuitry?l

I'm not Paul, but I can confirm that yes, 3.3V at the Vout3.3 pin can be used as an input for 3.3V power and that powering it is essential to allow USB to function. I found this out the hard way when I bypassed the MK20 VR altogether (supplying 3.3V only to the VCC pins) and then having no means of communicating with the MK20 over USB.

If I remember the relevant section in the Freescale manual correctly, you can run the chip at a lower VCC voltage and still communicate via USB as long as 3.3V is supplied to the Vout3.3V pin (7). IIRC, it's one way the OEM allows designers in power-restricted applications to reduce power consumption. The MK20 uses less power at lower voltages, yet 3.3V is used for signalling over USB. Hence the allowance for two different voltage levels (VCC vs. VOUT). I'm not sure how lower voltages affect the ADC, etc. but presume they're primarily intended for sleep mode and similar applications.

So, if you want to put the MK20 on a further power diet, only power pin 7 when it's needed and to leave the USB circuits dormant when no USB cable is attached. I suppose the easiest way to do this is to have a small dedicated external voltage regulator attached to VUSB that does nothing but power pin 7 externally (plus a capacitor or two for stability, see the Freescale manual for details). Naturally, you wouldn't be able to use the internal MK20 voltage regulator under that scenario, so you'd end up with a minimum of 2 voltage regulators per board (one for USB, the other for the MK20).

jimmayhugh
09-17-2015, 09:00 PM
Hmmm...
It appears that the reset pad has been moved in the Teensy3.2, so my post (https://forum.pjrc.com/threads/26071-Using-all-Teensy3-x-pins-with-a-socket) needs to be updated to reflect the change.

PaulStoffregen
09-17-2015, 09:16 PM
Yeah, believe me, I of all people really wanted to keep those bottom side pads in the same locations. I ended up having to build a whole new set of pogo pins, just because a few of those moved. But it was unavoidable to cram the extra voltage regulator and large capacitor into the very limited space!

Budmo
09-17-2015, 09:31 PM
I received mine today and it appears to work fine. Did a quick hook-up to an esp8266 but haven't tasked it much other than typing AT+RST but it was great seeing some output from that in the serial monitor. My soldering skills are not so good and the 3.2 has a few extra little dots (like between GND and RX1) but I managed not to short anything. Thanks Paul!

5105

defragster
09-20-2015, 06:52 AM
Got mine today and loaded first sketch fine. Will TeensyDuino be modify IDE to note : Tools /Board /Teensy 3.1&3.2?

Fyod
09-20-2015, 10:37 PM
Just wondering if these'll be available at Oshpark? I always grab a couple Teensies when ordering prototypes.

Edit: just read in another thread that they are already shipping 3.2. Cool beans!

macaba
09-21-2015, 12:42 PM
Paul - thank you so much for exposing the USB data lines on larger pads.

May I ask what your intention is with those bottom square pads under the USB connector? With the addition of the data pads, do you have something in mind for this; perhaps this is the pattern for a particular connector?

johnnyfp
09-22-2015, 05:52 AM
The additional square pads under the usb socket is for a Capacitor needed for USB Host mode.

macaba
09-22-2015, 09:54 AM
The additional square pads under the usb socket is for a Capacitor needed for USB Host mode.

Thanks for the reminder; I had forgotten that little nugget of info.
I am still curious as to whether the addition of the USB data lines are placed to conform to any specific connector or whether it's literally just to make it much easier to solder hookup wires to.

dgarcia42
09-25-2015, 03:13 PM
Excellent! I have four on order and the datasheet open in another window. I think this'll be the first post fastled3.1 update!

EDIT: never mind on the library update, read the mcu as being a different model at first - easiest new platform port ever!

Windfreak
09-25-2015, 03:30 PM
Any idea of pricing on the new bootloader and when volume quantities will be ready?
Also, it would be nice to have a list of things to do/think about in a redesign from the mini54 to the new one..
Looks great Paul! Thanks.

Constantin
09-25-2015, 04:14 PM
Depends on which one you're using. The biggest MiniTAN54 needed more capacitors, the 5x5mm version is similar in needs to the MKL02 (1).

From a connection point of view, note the drop of the I2C lines (which aren't getting used in the Teensy 3.0 or 3.1 either).

The number of connections to the MCU does not change. I did end up using two jumper resistors to avoid creating vias on a 2-layer board. One is for power, the other carries the reset signal. See above.

Constantin
09-25-2015, 05:23 PM
This definitely is coming. The chip will be available in 2 sizes, the extremely small 3 mm QFN, and a large 7 mm TQFP with "easy" 0.8 mm pitch pins.

FWIW, I tried to find that larger chip your're referring to, and I presume it's either as-yet unreleased or not a MKL02Z32 series chip? I'm happy to wait to update the Eagle library but would love to do so before the new chip is released.

Windfreak
09-25-2015, 06:08 PM
I currently use the smallest minitan and would go with the smallest MKL.. Im planning a board rev in the next couple of weeks for other reasons, so if supply is good and price is better, I might as well bump the MiniTan off my board...

[QUOTE=Constantin;83107]Depends on which one you're using. The biggest MiniTAN54 needed more capacitors, the 5x5mm version is similar in needs to the MKL02 (1).

PaulStoffregen
09-25-2015, 07:29 PM
Work is continuing here. My hope is we'll have these chips for sale in a couple weeks. Pricing will be similar but slightly less than the Mini54.

The 3mm size uses 0.5 mm pitch pins. Our assembly vendor that solders the Teensy boards reports it's much better for manufacturing. If you can wait a few weeks, I hight recommend holding out for the new chip!

The replacement ZIF socket (for the one that was damaged a couple weeks ago) *still* hasn't arrived. Frustrating! It should be here any day....

The larger chip is a Freescale MKL04 part. Full details will be published when the chips go on sale.

I'm also working on a reference board which you'll be able to order from OSH Park. It uses the larger chip and 603 and larger parts (no tiny 402s) for easier hand assembly.

FWIW, no Teensy has ever used ezport mode, not even the original Teensy 3.0 released via Kickstarter three years ago.

Windfreak
09-25-2015, 07:51 PM
Thanks Paul. That schedule works for me. Ill design it in on my next rev.

Constantin
09-25-2015, 10:25 PM
Me too! Removed erroneous reference to EZPORT - apologies for that.

experion
09-26-2015, 09:16 PM
Work is continuing here. My hope is we'll have these chips for sale in a couple weeks. Pricing will be similar but slightly less than the Mini54.

The 3mm size uses 0.5 mm pitch pins. Our assembly vendor that solders the Teensy boards reports it's much better for manufacturing. If you can wait a few weeks, I hight recommend holding out for the new chip!

The replacement ZIF socket (for the one that was damaged a couple weeks ago) *still* hasn't arrived. Frustrating! It should be here any day....

The larger chip is a Freescale MKL04 part. Full details will be published when the chips go on sale.

I'm also working on a reference board which you'll be able to order from OSH Park. It uses the larger chip and 603 and larger parts (no tiny 402s) for easier hand assembly.

FWIW, no Teensy has ever used ezport mode, not even the original Teensy 3.0 released via Kickstarter three years ago.

Hi, got 4 of the teensy 3.2. has a friendlier feel to it. Think it may have something to do with the button.

Paul, is the Minitan bootlader compatible with the new teensy 3.2? I mean since the new boot-loader is not ready for shipping yet.

PaulStoffregen
09-26-2015, 10:03 PM
Paul, is the Minitan bootlader compatible with the new teensy 3.2? I mean since the new boot-loader is not ready for shipping yet.

No, the Mini54 implements only Teensy 3.0 and 3.1.

To implement Teensy 3.2, you'll need the new chip. It was supposed to be selling by now, and it is indeed coming in a matter of a few weeks. I know waiting can be painful, but if you can, I'd recommend holding out a few weeks. It really is coming soon.

experion
09-27-2015, 02:48 PM
No, the Mini54 implements only Teensy 3.0 and 3.1.

To implement Teensy 3.2, you'll need the new chip. It was supposed to be selling by now, and it is indeed coming in a matter of a few weeks. I know waiting can be painful, but if you can, I'd recommend holding out a few weeks. It really is coming soon.

Thanks for your prompt answer. You have accomplished something really great with the Teensy 3 and above, nothing appears to come close of similar size.

We designed a custom PCB based on the teensy 3.1 and the Teensy audio board. The audio board was needed because Teensy audio library does not allow a second audio input. When this is done, it would make a lot of difference in the use of teensy for audio small audio devices. It would definitely make a difference to us.

My enquiry is:

(a)Can you make the teensy 3.2 part list available on the Teensy site. maybe as a kit...

(b) Not sure if this is already available, is it possible to provide the Teensy 3.1 and 3.2 Eagle files. This helps avoid potential errors in the design of custom PCB., especially on BOM and footprints.
There was an issue that arose because the Teensy 3.1 schematic had the 16MHz crystal without capacitors. However, my designer added the capacitors because a crystal guideline advised to use capacitors. This lead to some confusion with the PCB house and sorting out the BOM.

Hence why it would be great to get a clearer instruction on using the teensy 3.1 and 3.2 for custom design.51715172

(c) If it possible to add the second audio input to the Audio library without the teensy audio board?

Constantin
09-27-2015, 03:24 PM
I don't speak for Paul, but I know he doesn't use Eagle. I have published the eagle files for the processor and the mini tan boot loaders. I believe frank also published a known good eagle layout file with all the connections. IIRC, his uses a two layer board, whereas the Teensy 3.0-3.2 are based on a four later board.

Depending on how much coffee Paul wants to consume, it is possible that he'll accomplish the 3++ with a 4 layer board but my guess is that 6 layers will be required to route all the necessary BgA connections.

Elsewhere, people have figured out most of the components used. The schematics are a great place to start, however.

PaulStoffregen
09-27-2015, 06:53 PM
I have published the eagle files for the processor and the mini tan boot loaders. I believe frank also published a known good eagle layout file with all the connections.

Did I make links to all these, from here?

http://www.pjrc.com/teensy/eagle_lib.html

If there are more I haven't linked, please let me know and I'll update the page.




Depending on how much coffee Paul wants to consume, it is possible that he'll accomplish the 3++ with a 4 layer board but my guess is that 6 layers will be required to route all the necessary BgA connections.

Yes, it's going to be 6 layers.

I recently got a quote from our PCB vendor. In volume, 6 layers is only a modest increase in cost over 4 layers. In fact, the estimated increase in PJRC's cost for a year's worth of Teensy++ 3.x PCBs (just the extra cost of increasing from 4 to 6 layers) will probably end up lower than what we spend on coffee over a year!

But then, I do drink quite a lot of coffee and Robin buys Illy medium roast beans....

Frank B
09-27-2015, 07:46 PM
I believe frank also published a known good eagle layout file with all the connections. IIRC, his uses a two layer board, whereas the Teensy 3.0-3.2 are based on a four later board.
Not me....

dpharris
09-28-2015, 06:36 PM
Is there an Eagle library for the 3.1 and 3.2?
Thanks
David
PS -- my 3.2s are ordered from Oshpark ... the day after your announcement!

Theremingenieur
09-30-2015, 07:30 PM
Ordered two of the 3.2 today at Diigiit Robotics Europe :)
I will already build two prototypes with slightly reduced functionality before I switch over to the MK66 whose 2nd DAC is needed for complete operation. But this will already allow me to check and optimize my input circuitry. If everything works as expected, I'll publish a new FreqMeasure library (which I will call PeriodMeasure because it seems more logical to me), allowing input capture on multiple FTM channels.

I feel like apart from the 2nd DAC, the MK66 will be some overkill since I'll make no use of the FPU. I learned handling that with pseudo-integers and am up to now satisfied with the result.

sumotoy
10-02-2015, 01:52 AM
ordered 3 some min ago at PIJRC shop, but noticed only now that it's actually out of stock! :(

PaulStoffregen
10-02-2015, 05:16 AM
Opps, that's a mistake. There's *plenty* of Teensy 3.2 in stock!

sumotoy
10-02-2015, 06:12 AM
Ah, feel much better now... :D

Theremingenieur
10-03-2015, 01:09 PM
Mine arrived today! :)

Constantin
10-10-2015, 02:08 PM
Did I make links to all these, from here?

FWIW, here is my latest Eagle library for the bootloaders (MiniTan and MKL02/MKL04) as well as all MCU footprints. Additionally, I have come up with new footprints of the boards themselves. This attachment covers the Teensy 3.x and LC series.

8595
8596

I will try to update all my previous footprints. The board files cover a number of variants, from all-pins and pads to simple DIL packages for folk who need the center 'tunnel' for GPIO traces.


In fact, the estimated increase in PJRC's cost for a year's worth of Teensy++ 3.x PCBs (just the extra cost of increasing from 4 to 6 layers) will probably end up lower than what we spend on coffee over a year! But then, I do drink quite a lot of coffee and Robin buys Illy medium roast beans....

No wonder you manage to answer questions at all times of day! :)

cartere
10-10-2015, 02:55 PM
Speaking of coffee

My latest project, will control grinder when finished.

5257

Constantin
10-10-2015, 04:01 PM
That's great! Can it brew it and bring it to you too? :)

defragster
10-11-2015, 08:44 PM
Paul: Still on the SparkFun Teensy 3.2 page (https://www.sparkfun.com/products/13736) - they missed the whole point of the supply upgrade on the T_3.2:
Also, it can provide system voltage of 3.3V to other devices at up to 100mA.

KurtE
10-13-2015, 11:24 PM
I know that it may be a semi-niche product usage, but I wish that I could use either the LC or 3.2, directly powered by a 3s lipo battery. This is one reason why I know that at least a few people are instead using something like an Adafruit pro-trinket (http://www.adafruit.com/products/2000) and I would hate to have to back to an Atmega...

For example one might want to plug the Teensy into a Robotis Dynamixel servo chain like an AX-12, and have the Teensy listen on the Dynamixel Serial chain (Half duplex 1mbs) and either monitor some sensors and/or control some devices...

defragster
10-13-2015, 11:46 PM
It looks like the trick to LiPo power on the Trinket is this shield (http://www.adafruit.com/products/2124). Like the onehorse addon for Teensy (https://www.tindie.com/products/onehorse/lipo-battery-charger-add-on-for-teensy-31/) it powers from a LiPo that it charges from USB when plugged in. There a higher charger current version now too - neither are mass produced, nor quite as big and nor as cheap.

Of course a few mA may go farther on the lower power trinket. That Trinket Shield wouldn't mate to a Teensy - but seems it should wire up okay?

KurtE
10-14-2015, 12:14 AM
Thanks defragster: But if I am not missing something those are to connect up a 1S lipo (3.7v), the AX Servos typically run on 3S lipo or about 11.1vs, which is higher than the VR on the Teensys are rated for. Looking at the Pro Trinket, it looks like it can handle up to 16v.

Obviously I can run a Teensy with a small external BEC, but that adds weight and complexity. Again just sort of a wish list type of thing

defragster
10-14-2015, 12:27 AM
KurtE - you are right on both: "Up to 16V input". That is easier with lower current draw. PJRC DeRates the T3.2 due to the heat for anything near 10v its LDO might take when taking advantage of the current near/over 250mA it can produce. In your post I saw LiPo - and the Trinket is advertised with that LiPo addon. I didn't under stand you meant LiPo 'pack' (3s?) to get the voltage needed for the AX-12 servo 'thing' (I didn't recognize that for what it was).

macaba
10-16-2015, 12:12 PM
FWIW, here is my latest Eagle library for the bootloaders (MiniTan and MKL02/MKL04) as well as all MCU footprints.

This really does not matter but I figure you might want to fix this in a future version:

"This is a footprint for the MK20dx256VHL7"
should be
"This is a footprint for the MK20dx256VLH7"

akkotaire
10-26-2015, 04:39 PM
I understand the regulator on the teensy 3.2 is rated 10V max. But I need to supply the setup with a 3s lipo (so 11.1V to 12.6V). Since I will be using very little current could it be ok ?
Thanks

PaulStoffregen
10-26-2015, 05:05 PM
First, make sure you've cut the VUSB-VIN trace. That last thing you want is for 11 volts directly from a battery to connect to a USB cable and damage an expensive computer!

The voltage regulator is Texas Instruments part LP38691 (http://www.ti.com/product/LP38691). If you read the datasheet, section 7.1 " Absolute Maximum Ratings" on page 4 says "V(max) All pins (with respect to GND)" is -0.3 to 12 volts. The first footnote explains:


(1) Absolute maximum ratings indicate limits beyond which damage to the component may occur. Operating ratings indicate conditions for
which the device is intended to be functional, but do not ensure specific performance limits. For ensured specifications, see Electrical
Characteristics. Specifications do not apply when operating the device outside of its rated operating conditions.


Of course, TI recommends only using it up to 10 volts.

Also, please consider the Teensy 3.2 PCB is capable of dissipating about half a watt of heat from the regulator. At half a watt, it runs pretty hot, but still well within the regulator's capability.

If you have 12 volts input and 3.3V output, that's 7.8 volts across the regulator. If Teensy draws 35 mA, the result will be 0.3 watts. Even small increases in current on the 3.3V line result in pushing closer to the 0.5 watt maximum recommendation.

So in terms of your question "could it be ok", anything over 10 volts is pushing your luck and going over 12 volts is very risky, and even Teensy's normal current draw puts you close to the maximum heat the PCB can dissipate, so there's very little margin to power anything else from that 3.3V power.

I certainly wouldn't sell a product built this way, nor would I install such a thing in an unattended difficult-to-service location... but for a one-time project with an urgent deadline (something like a trade show demo), I might risk it. There's no question this is risky, so you'll have to make your own judgment call.

Theremingenieur
10-26-2015, 05:06 PM
Just try it out... A teensy 3.2 costs only 20 bucks... ;-)

I don't know how much current the teensy itself draws, but if there are no peripherals to source from the internal voltage regulator at the same time, you should have some headroom.

Edit (after seeing what Paul has posted in the meantime while I was writing):
Why not put an ordinary 7806 or 78L06 in-between?

akkotaire
10-26-2015, 05:40 PM
Thanks for the detailed reply, I will plan to add a regulator in between.
For a student 20$ makes a difference ;-)

Theremingenieur
10-26-2015, 05:49 PM
For a student 20$ makes a difference ;-)

Please accept my apologies, I didn't mean to sound arrogant in whatever way. When I wrote about "only" 20 bucks, it was comparing to the Mafia-duinos, which are less powerful and more expensive, at least here in France.

akkotaire
10-26-2015, 05:57 PM
Please accept my apologies, I didn't mean to sound arrogant in whatever way. When I wrote about "only" 20 bucks, it was comparing to the Mafia-duinos, which are less powerful and more expensive, at least here in France.

Ne vous inquiétez pas ! Ça n'a pas été mal pris.

riodda
11-19-2015, 02:34 AM
I got mine from the last order from Oshpark and it's alredy on my Racecar tyre temperature sensor prototype !
I wonder Paul if after teensy LC you may consider doing an HC or pro version, i had a look on the K20 sub family and i've seen that a dual can channel version of K20 is avilable (MK20FX512VLQ12 price goes from 3$ to around 6$ in quantities according to Freescale) and for people like me using this hardware in the automotive market would be great to have a second can line.
Maybe a complete board is not even enecessay maybe just the support in the booloader and in teensyduino is enough then if you need the second can line you will just build your own board and buy the bootloader chip.

Fyod
11-19-2015, 03:15 AM
^ https://forum.pjrc.com/threads/24633-Any-Chance-of-a-Teensy-3-1


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.

Great times ahead for us CAN BUS geeks.

pix-os
11-19-2015, 04:12 AM
sounds pretty good, things i like about that chip mostly: MCU clock, FPU, SRAM amount and double the DAC, double the fun! :D

stevech
11-19-2015, 04:42 AM
sounds pretty good, things i like about that chip mostly: MCU clock, FPU, SRAM amount and double the DAC, double the fun! :D

Do we know which package? Peripheral suite varies by pin count of package. 100 pin?

defragster
11-19-2015, 06:03 AM
https://forum.pjrc.com/threads/24633-Any-Chance-of-a-Teensy-3-1?p=79140&viewfull=1#post79140

WMXZ
11-19-2015, 06:21 AM
Do we know which package? Peripheral suite varies by pin count of package. 100 pin?

and
https://forum.pjrc.com/threads/24633-Any-Chance-of-a-Teensy-3-1?p=83933&viewfull=1#post83933

Markk
12-12-2015, 04:06 PM
The chip will be available in 2 sizes, the extremely small 3 mm QFN, and a large 7 mm TQFP with "easy" 0.8 mm pitch pins.

Hi

I know it's probably a bit OT to ask about this here but has anybody any experience with "homemade reflow oven" + QFN?

I've successfully made a custom build Teensy 3.1 with various chips down to 0.5mm pitch and 0603 parts. But all pins were visible/outside so far.

Also I had to remove quite a few bridges, so I'm a bit anxious about the hidden pins of QFN packages.

Hints about getting rid of bridges also welcome (it seems there is too much solder on the small pads, but barely enough on the big ones).

Thanks!

_Markk

johnnyfp
12-12-2015, 06:18 PM
All the time. Trick is not to put too much solder paste. But if you do then you can use a clean iron (absolutly no solder on the iron, so wipe it on a sponge again and again) and then just touch the bridge on the qfn. It should wick onto the iron.
If that doesnt work, try using a fine wire solder wick and gently and quickly tap the iron onto it. Becareful not to take to much solder away as you may disconnect the pin and then have to add solder again.
Finally if that doesn't work, you'll need some liquid flux usually in a dab pen form and dab the qfn, then get a hot air gun and yet the part till its lose and you should be able to gently move it in to position and hopefully de bridge.

Goodluck.

Markk
12-12-2015, 07:03 PM
All the time. Trick is not to put too much solder paste. ... Goodluck.

Thanks for the answer but this is about using a DIY reflow oven with a solder paste stencil. I would never finish my design with hand soldering ;-)

5786

I hope I'll soon find time to start document the project. It's a Teensy 3.1 custom build + MCU controlled MPPT solar charger + weather station (wind speed, direction, rain, temp, humidity) + USB host with G3 modem stick and software PPP TCP/IP stack + simple audio amp/headphone + SD card + (the most important thing) gamma spectrometer with low power high voltage generator. It's my first design in electronics. Second PCB revision. And it's mostly working ;-) Third revision is in progress and I'd hate to introduce new problems...

-Markk

johnnyfp
12-12-2015, 08:02 PM
That was using a reflow oven. If you put to much pressure when squidging the solder paste through the stencil you'll get to much paste on the pads. Then when you reflow you may get bridges. What i described is how to fix after reflow.
Ive made a teensy+lipo charger+esp8266 micro board that uses these small qfn and sometimes when it comes out the oven it bridges because of inconsistent pressure when applying the paste through the stencil.

Fyod
12-12-2015, 08:21 PM
I find that most stencils (depends on the thickness) allow you to apply the correct amount of paste for qfn chips. It makes nice little balls when finished and I've never had too much solder.
Except I heat my boards with a hot air soldering gun.

Markk
12-13-2015, 08:28 AM
That was using a reflow oven. If you put to much pressure when squidging the solder paste through the stencil you'll get to much paste on the pads. Then when you reflow you may get bridges. What i described is how to fix after reflow.

Ah, I completely misunderstood! Now that I reread after a good night of sleep it was obvious. Apologies.

Thanks for the tip about pressure because I obviously got it wrong, thinking I had not enough pressure for the stencil to really touch down on the BCP fully and therefore not scraping off enough. Will try lighter pressure next time.

About fixing after reflow thank you for the tips too. Judging from this, it seems QFN is no less likely to develop bridges than ordinary chips but also no more difficult to remedy...

BTW for others who may read this, in order not to repeat my mistake:

If you have solder bridges behind the legs of a chip (talking about outgoing pins, not QNF) you sometimes have to apply even more solder to really gobble it all up and only then wick it off. I destroyed a chip trying desperately to wick behind the legs, before I found this out >:-}

_Markk

veng1
12-13-2015, 04:50 PM
Ah, I completely misunderstood! Now that I reread after a good night of sleep it was obvious. Apologies.

Thanks for the tip about pressure because I obviously got it wrong, thinking I had not enough pressure for the stencil to really touch down on the BCP fully and therefore not scraping off enough. Will try lighter pressure next time.

_Markk

Practically speaking, you can't apply too much pressure to the squeegee. Automated screen printers apply a substantial amount of pressure. What can happen is that the stencil bows up and allows too much past to flow sideways under the stencil leaving excess paste on the pads. It usually works well to tape tape the stencil to another piece of PCB material that is the same thickness as the target so it stays flat. Usually just under one side is enough but for the first few tries, it won't hurt to surround the target board with PCB material on all sides and tape the stencil to that. After you do it a few times, one side will probably be enough.

The thickness of the stencil also determines how much paste is left and 4-6 mils is fairly standard. Too much paste usually a bigger problem than too little.

It takes a little practice but you can always clean the board off and try it again.

Markk
12-13-2015, 08:27 PM
Thanks veng1


What can happen is that the stencil bows up and allows too much past to flow sideways under the stencil leaving excess paste on the pads.

That's exactly what I thought. So on the second revision I pressed down quite hard. No change.


The thickness of the stencil also determines how much paste is left and 4-6 mils is fairly standard.

I ordered with Seeed and its a 0.12mm stencil (http://www.seeedstudio.com/service/index.php?r=stencil), so ~4.7 mil.

As I said I pressed down quite hard but it still looks way too thick:
5795

Perhaps it would help to reduce the cream layer outlines in relation to the smd copper in Eagle?
5794

Any suggestions welcome.

Thank you very much!

_Markk

Fyod
12-13-2015, 09:27 PM
I use the trick with old boards around the stencil too. Also makes it much easier to place the new board(s) correctly.

Markk: That thickness looks good to me.
If its only the processor that's bridging, I would look at the chip part library and make the creams smaller. Or reduce them globally as you're showing.
The problem with thinner stencils could be warping etc.

veng1
12-13-2015, 09:45 PM
Normally the stencil house does a size reduction on the pads for small pads as the paste should be smaller than the pads, then a tombstone reduction where bigger SMT pads have a trapezoidal shape to pull the solder away from the center of the part to give good fillets and then on large pads like a voltage regulator or thermal pads a technique called a window pane is used to allow the flux gasses to escape.

potatotron
12-14-2015, 12:18 AM
I ordered with Seeed and its a 0.12mm stencil (http://www.seeedstudio.com/service/index.php?r=stencil), so ~4.7 mil.

That does seem too thick. When I've done QFN I've always used 3 mils stencils (0.075 mm). It's surprising how little solder is actually needed.

Edit -- Attached is a picture of the last time I made a custom Teensy board.

5800

veng1
12-14-2015, 12:43 AM
There are two variables, the stencil thickness and the pad reduction.

I tend to use a thick stencil because I want the largest components I can get away with. I'll use a 0805 even if I could use an 0603 simply because it allows more process headroom independent of the fact that 0805 R's and C's cost more.

I then find a bigger stencil reduction allows the print to be offset more and still get a decent print but I'm coming from the direction where I want to max out a screen printer at 180 prints/hr, not by hand printing. It's not about how many prints you can get per hour, its about how many good prints you can get per hour. If 3 mils work you you, you should use it. If it ain't broke, don't fix it.

potatotron
12-14-2015, 01:07 AM
I tend to use a thick stencil because I want the largest components I can get away with. I'll use a 0805 even if I could use an 0603 simply because it allows more process headroom independent of the fact that 0805 R's and C's cost more.

Sorry for derailing this topic but how do thinner stencils require smaller parts? I haven't had any issues using e.g. 1206 LEDs or battery holders like this https://www.digikey.com/product-detail/en/BU2032SM-HD-G/BU2032SM-HD-GCT-ND/755451 with 3 mil stencils.

Constantin
12-14-2015, 01:23 AM
AFAIK, it's the other way around - parts with small pitches require thin stencils to prevent bridging. The downside is that thinner stencils (like the great mylar stuff that Osh stencils uses) will not last as long or be as durable. Also, I found that stainless stencils from the likes of itead tend to run on the thicker side. Great stencils but relatively thick, leading to somewhat common bridging for chips with high pitch densities such as the MK20.

veng1
12-14-2015, 04:46 AM
Agreed. Thinner stencils give less paste thickness and will reduce bridging although somewhat of the same effect can be had be doing a stencil reduction where the stencil apertures are made smaller than the component pad thus also applying less paste.

Normally the thickness of a stainless steel stencil can be specified. I like to use SIPAD for prototypes as that includes the paste applied to the board, reflowed flat, tacky flux applied and then covered with a peel off film. Then the stencils they supply tend to work the same way in volume production. http://www.sipad.com/

With the SIPAD process, the film is pulled off and then the parts can be placed on the pads and they will stick fairly well to the flux. It isn't cheap because it includes the price of a stainless steel stencil plus them applying the paste but it works far better than doing it with a manual squeegee. If the component feature sizes allow hand printing, I wouldn't use them due to the price.

Constantin
12-14-2015, 01:58 PM
Interesting process and company. If I understand it correctly, you send them the boards along with the cream gerbers, and they apply a tacky paste and flux, seal the thing, and send it back. Then you unseal the board, apply the components and reflow?

veng1
12-14-2015, 02:48 PM
That's correct.

It allows machine applied paste accuracy for someone that doesn't have a screen printer. It also allows parts to be picked up and relocated if one is careful which is almost impossible with wet paste and small component features.

I also find it is possible to convince a contract manufacturer to run some boards down their automated reflow oven at no charge if the boards they are running are using the same solder profile and they don't have to do any set-up. They won't do that if they need to set up a screen printer as that is quite a bit of work and while they are doing the set-up, their line is down and they aren't making any money. In contract manufacturing it is key to keep the line flowing.

Constantin
12-14-2015, 04:12 PM
That's really nifty. I couldn't see a pricing sheet there, could you give me a ballpark re: cost? Perhaps normalized across a typical set of boards on a $/in^2 basis, for example?

veng1
12-14-2015, 09:39 PM
That's really nifty. I couldn't see a pricing sheet there, could you give me a ballpark re: cost? Perhaps normalized across a typical set of boards on a $/in^2 basis, for example?

I think you really need to get a quote. They are more than willing to do that. The last couple I had done were paid by my clients and I'd be guessing but I'm thinking maybe $500 to get set up. I can't get a screen printer set up at a contract manufacturer for twice that much less get any panels printed.

Because there is the cost of making a stainless steel stencil and set-up, there is going to be a minimum charge. Then there is a cost per panel to do the actual process which is heavily dependent on volume so the setup is amortized over multiple boards.

Like most volume processes, it is less expensive to do multiple boards per panel because most of the charge is based per stencil wipe. If I recall, they probably use a standard squeegee that will probably print a panel about 17" wide but you should confirm that. So they can probably print about 17" square and a panel that is less will cost essentially the same amount, so it pays to use big panels and then break them apart.

Markk
12-14-2015, 10:26 PM
Thanks everybody for the feedback.


There are two variables, the stencil thickness and the pad reduction. I tend to use a thick stencil because I want the largest components I can get away with.

Can you share your pad reduction parameters or are they a trade secret :-) ?

Thanks,
_Markk

veng1
12-15-2015, 03:58 PM
There aren't so much trade secrets as experience with the target production process. If you are going to a contract manufacturer, the easiest thing is to ask them the size boards they want to process. And their preferred stencil thickness. I've tended to err on the side of too much solder because I wanted to make sure I had maximum heat transfer but 3-6 mils, probably on the lower side. Typically the maximum placement area is 18" X 18 " although very few houses will work with boards that big. Because needed to process large LED display panels, I had two machines back to back with offset board stops to handle 24" boards. Also ask the board house for the edge clearance because there is a limit to how close components can be put to the sides of the boards and miss the conveyors. After you know what is a suitable board size, send the Gerbers to the PCB manufacturer. If you know what panel size you want, tell them or say something like:

Panelize the attached Gerbers for maximum yield with a panel size not to exceed ?? X ?? inches.
Send full size Gerbers back as panelized for making a solder paste stencil.

Then send the Gerbers for the panel to the stencil house and tell them:

Please use the attached Gerbers for a ?? mil thick stainless steel stencil to fit a standard 26" X 26" frame (or whatever size the contract manufacturer needs)
Reduce all pads by 10%.
Perform tombstone reduction on all 2 pad components such as R's and C's. (also whatever size the contract manufacturer needs)
Windowpane all large thermal pads.

Then when you get the boards back, inspect the solder joints with a magnifier and decide if the fillets look good. I'm sure there are pictures that can be found with Google to show good fillets or but the better practice is to use a copy of IPC-610. In any event, the fillets should be concave, not convex.

After that, you just keep doing it until it's good enough.

macaba
12-16-2015, 10:59 AM
Thanks for sharing your experience veng1; I learnt something from your post.

veng1
12-16-2015, 02:15 PM
Glad to help.

Wish I could say more but it is surprising how much of an art soldering is even today.

npashine
03-10-2016, 12:14 PM
Hi all,
Is there any way to cut KL02 from teensy 3.2 after flashing or any way to power down the KL02.

Constantin
03-10-2016, 01:11 PM
IIRC, The boot loader chip powers down unless you press the prog button to erase the teensy mk20 flash to prepare it for a new upload. Only the first few beta test boards featured a non sleeping mkl02.

Can that chip be removed? Sure, I did that with one of the LC's when I needed to harvest a MKL-02 before they were released as standalone chips. However, you risk damaging the rest of the board and likely will damage the MKL02 as well. MCUs can only handle so many heating cycles before they become less reliable. FWIW, my recycled MKL02 worked. I heated it with a hot air gun followed by rapping the board on a hard surface causing the chip to fly off. I then fluxed the daylights out the thing and used braid and a spotless, tinned tip to clean up the contacts.

Now that MKL02 chips are available, I don't see the benefit of harvesting old chips. Too much risk, too little return, unless your time is free, enthusiasm is high, and you don't mind the potential issues associated with a fried MKL02.

npashine
03-22-2016, 07:21 AM
Hi all,

can I remove bootloader after once burned the program, as I have to run a single program and don't need any changes.
And also, any separate bootloader is available so that I can reconnect and burn the program.

defragster
03-22-2016, 07:47 AM
There is no bootloader stored in flash to remove. The Teensy scheme uses a stand alone programming chip on the board that facilitates the programming.

npashine
03-22-2016, 07:49 AM
Hi,

What is the use of MKL02 and can i remove it after flashing to reduce cost.

Constantin
03-22-2016, 11:42 AM
Yes you can remove it after flashing. However, before you go down that path, I suggest you read up on the issues associated with that on store page for the chip.

Bottom line: if removing the boot loading chip is what makes the difference between a potentially profitable design and a loss-maker (at the pilot production stage) then your product needs to be rethought. You are much better off in the first generations of alpha and beta testing to be able to quickly reprogram your boards. If your design makes it into high volume production, it may make most sense to preprogram the MCUs in a rig or the board as a whole (as Paul does).

I toyed with the idea of external programming boards but then reconsidered. For one, it complicates programming, introducing more failure points. Secondly, there is the issue of how to support Paul and his efforts. It's much easier from an accounting point of view to simply include the boot loader chip and be done with it. If you want amazing tech support, you have to pay for it.

Windfreak
06-27-2016, 01:20 PM
Id have to second Constantin. My product with the MKL02 in it here:
https://windfreaktech.com/product/microwave-signal-generator-synthhd/
I put the hooks to remove the chip and have ISP programming, but so far have been building with the MKL02. Im not even really planning to remove it. We sell a few to Universities so it probably just helps my bottom line to give people the power to be "Aurduino Compatible" and "easily" program their own firmware.
A big BUT.. But my BOM and selling price are high compared to the price of Paul's chip which helps out a bunch. Obviously, if you are going for low ASP and volume you might want to remove it quickly.

fenestron
06-30-2016, 10:06 PM
Today we're happy to release Teensy 3.2.

http://www.pjrc.com/store/teensy32.html

Version 3.2 is a minor upgrade to Teensy 3.1. The main change is an improved 3.3V regulator, to allow Teensy to directly power ESP8266 Wifi, WIZ820io (W5200) Ethernet, and other power-hungry 3.3V devices.

We're specifying Teensy 3.2's power output at 250 mA and the maximum voltage input at 6 volts, due to PCB thermal dissipation limits. However, the actual regulator chip is capable of up to 10 volts input, and up to 500 mA output. These higher limits are intended to allow Teensy 3.2 to be more rugged when used with non-USB power sources which aren't well regulated 5 volt sources.

Teensy 3.2 is fully compatible with all shields and add-on boards designed for Teensy 3.1. It preserves the same size, pinout, and processor as Teensy 3.1.

The bootloader chip is also changed from Mini54 to KL02 (the same as Teensy-LC). A full schematic is available:

http://www.pjrc.com/teensy/schematic.html

hi Paul,

i just received my teensy 3.2. i am a new bee with this board and i want to design my custom teensy3.2 board.
i will use your schematic to do this but i wonder that after installing all the component,
is there any need to upload firmware into the main processor and/or KL02 to program the processor with the arduino IDE.

Theremingenieur
06-30-2016, 10:16 PM
The KL02 is needed every time you want to upload firmware. It is a hardware replacement for the boot loader which "eats", when realized in software, a part of the processor's flash memory, like it happens for the Arduino boards. The KL02 does handle the boot loading thing, and the whole flash memory of the Teensy is free for your programs.

Ben
06-30-2016, 11:53 PM
is there any need to upload firmware into the main processor and/or KL02 to program the processor with the arduino IDE.

The MKL02 can not be programmed by yourself if you want to use the teensyduino ecosystem, it's firmware is a trade secret of PJRC. You can buy pre-programmed MKL02 ICs directly from PJRC: https://www.pjrc.com/store/ic_mkl02.html
On this page you'll also find useful information for designing your own board.