Any Chance of a Teensy ++ 3.1?

[linuxgeek]

What if the board came with, or for a nominal price, a high-density to breadboard headers?
You get the possibility of getting at more capabilities, and have the ease-of-use of plugging it all into a breadboard which I think improves the development of the libraries.
I think a lot of us have multiple teensys, but might need only one of those HD->breadboard cables. It allows many of the advanced users in here to try something quickly to help a user, improve a library.

 

[instrumentek]

complete newbe suggestion, what about implementing finer pitch through hole pins all the way around but then with each teensy 3.x++ provide another pcb that goes from the finer spacing to the larger 2.54 spacing for the bread boarding guys. Then when implementing the prototype to a pcb just design around the finer pitch on the actual teensy.

 

[Donziboy2]

complete newbe suggestion, what about implementing finer pitch through hole pins all the way around but then with each teensy 3.x++ provide another pcb that goes from the finer spacing to the larger 2.54 spacing for the bread boarding guys. Then when implementing the prototype to a pcb just design around the finer pitch on the actual teensy.

It would become more costly for those just getting into the Teensy.

 

[WMXZ]

@Paul,
any decision on mcu footprint: BGA or QFP?
This was asked before, but I do not remember having seen an answer.

 

[Donziboy2]

@Paul,
any decision on mcu footprint: BGA or QFP?
This was asked before, but I do not remember having seen an answer.

Pretty sure he is leaning towards BGA if he intends to keep the same teensy footprint. The QFP is like 20mm which would not fit inside the 2 main rows of headers.

edit
If they are both pin compatible it may make DIY'ers lives easier if they decide to build the ++ into a design to use the QFP.

 

[PaulStoffregen]

Pretty sure he is leaning towards BGA if he intends to keep the same teensy footprint.

Yes, indeed only the BGA fits. Even then, it's 13 mm which comes right to the edges of the pads on both sides. Headers or sockets can solder to the bottom side, but if you want to solder them on the top side, you either need to shave away some of the plastic where the chip is, or accept that they just can't mount flush with the top side of the PCB due to the 13 mm chip width.

Of course you can use either the BGA or LQFP version of the MK66FX1M0 chip if you design your own board. When Teensy3++ is released, the bootloader chip will be updated to auto-detect either MK20DX256 or MK66FX1M0.

 

[defragster]

My #318 post was a quick hack on a twice hacked hack in MSPaint. If the 1.27mm pins have any value I showed two approximations. The central double gives 18 pins allowing it to fit inside the edges. As far as outside edges they could be placed half on each side on the end. That end with 1.27mm pins could then float over the breadboard end, and the main portion would have balanced .1" pins for a stable board.

Though as central board, crossing through pins that may make them unroutable on the far side so it may require 90° rotation (or split in two) to allow routing. Ideally though it needs to allow placing a top end mount micro SD flash device?

I understand Paul's issue with a HiDen connect - it will add a second part or a dollar to his retail and another dollar to any mating board that is also not user solder-able. Assuming the 1.27mm can be soldered by those needing the extra pins.

Edited my post #318 manually removed lame first exemplar. That only adds 0.6" not 1" as it assumed a higher density takeoff to avoid a naming conflict with LARDuino for such a huge board.

 

[mxxx]

I will be reading this thread, even if I don't answer everything. Please understand Teensy3++ will be a breadboard format board, likely 48 pins, where the left-side tries to be as compatible as possible with the 28 outside pins of Teensy 3.1 and Teensy LC. Alternate form factors or abandoning Teensy 3.1 compatibility are not up for discussion. A high density connector is unlikely, but not totally out of the question.

sorry to bring this up again, but i'm still wondering what speaks against two-row connectors? routing?

it seems to me to most obvious way to maintain compatibility both in terms of pinout *and* size/form factor (+/- a few mm width), as well as bring out lots of pins. for example: olimex H405, except for the massive jtag connector and clunky usb socket, it's actually fairly 'teensy'

 

[Frank B]

sorry to bring this up again, but i'm still wondering what speaks against two-row connectors? routing?

it seems to me to most obvious way to maintain compatibility both in terms of pinout *and* size/form factor (+/- a few mm width), as well as bring out lots of pins. for example: olimex H405, except for the massive jtag connector and clunky usb socket, it's actually fairly 'teensy'

Yes, I also think that this is the best option.
It is also breadboard compatible (by not using the second row, or using it (pinheader) upside-down).

Edit:
And it can be sawed-off if size matters (is it possible to add many small holes between the rows to make it easier?)

 

[defragster]

underPads on the four layer 3.1 are subject to failure from abuse (physical stress, solder heating)- won't going to 6 layer likely make the 'abuse' threshold lower? Of course through hole pins void all layers.

 

[jasper]

For breadboarding, one would probably need to solder a flat ribbon cable to the side -- maybe reuse an old IDE cable?.

Tried this it's very hard to do and unreliable. the wire in those ide cables is hair thin.
we need proper pin holes to solder onto our boards , or some standard connector.

Our soldering lady can solder all the pins on the teensy 3.1 in 30 seconds. We pay her $2.00.
the only good thing about the 0.1in pins is their large size provides structural and thermal stability which seems to be the key to quick soldering.

 

[Nominal Animal]

the wire in those ide cables is hair thin

The 80-wire ones ("PATA cable") sure are, but I've got some thicker 40-wire ones (old style IDE cables) that have multiple stranded wire, which are not nearly as bad.

That said, I'm pretty happy with the number of pins in Teensy 3.x, actually. I don't need the UARTs, but i2c and SPI are always in short supply. An edge connector would please me more, because then it'd be easier to reuse Teensy for different projects -- or in my case, make it less likely I'll burn another one by my lack of soldering skills. Or maybe I just need to get a big bag of ZIF sockets, and shut up.

Our soldering lady can solder all the pins on the teensy 3.1 in 30 seconds.

I'm orders of magnitude slower than your soldering lady.

For the 2.0++, I use a modded chip socket on the top side, so that I can use the socket as an impromptu breadboard. It also stabilizes it quite a bit, without increasing the volume taken by the board at all. (The top of the USB socket is level with the top of the socket.) Makes for effortless prototyping. For the 3.0 and 3.1, I haven't found a socket that could be modded to work like that.

 

[PaulStoffregen]

but i'm still wondering what speaks against two-row connectors?

Breadboards, mostly, but also a desire to keep a small form-factor.

For 32 bit "Teensy", I want the smoothest possible migration path between 3 different products. Similar form factor is a top priority.

Dual row headers have been successful for Raspberry Pi and Beaglebone. I do believe this may be worth considering for a future (likely 2017) product, which would probably abandon form-factor compatibility. Such a product would be called something else, maybe not even "Teensy", but certainly not a name like "Teensy++ 3.1", implying it's just a bigger version of "Teensy 3.1".

And it can be sawed-off if size matters

LOL, yeah, we're going to tell people to saw the edge off... what could possibly go wrong?!

underPads on the four layer 3.1 are subject to failure from abuse (physical stress, solder heating)- won't going to 6 layer likely make the 'abuse' threshold lower? Of course through hole pins void all layers.

I'm pretty sure bottom-side pads will be equally fragile on 4 or 6 layer boards. For Teensy 3.0 (and 3.1), the choice was between making the signals available on SMT pads versus not exposed at all.

Yes, through-hole pads are much worse for PCB routing. Even the large bottom-side pads also create a huge routing obstacle, since they greatly limit where vias can be placed, which impacts routing decisions on all layers. Just placing 14 pads on the bottom side of Teensy 3.0/3.1 was incredibly difficult.

But maybe a very fine pitch though-hole connector could work, like 1.0 or 1.27 mm pitch?

 

[jbliesener]

But maybe a very fine pitch though-hole connector could work, like 1.0 or 1.27 mm pitch?

Does that indicate that you're inclined towards defragster's suggestion?

View attachment 4831

If so, I'd vote for 1.27mm pitch to facilitate hand-soldering. Or two rows with 1.27mm pitch and 2.54mm distance between them.

Then again, a dual-row socket with 1.27mm distance between the rows and would give a mechanically more robust solution than two "single" rows...

 

[PaulStoffregen]

Does that indicate that you're inclined towards defragster's suggestion?

One of many options I'm exploring involves pads to add a fine-pitch through header, for perhaps 6-10 extra signals, instead of bottom-side pads like Teensy 3.1. Obviously though-hole pads take up a lot of PCB real estate on all layers. If this happens, which is a pretty big "if", it can't be for a large number of signals. My main hope would be access to the 6th serial port, 4th I2C, and maybe the 3rd SPI, and maybe a couple extra signals so 16 bits of at least one port are exposed.

A high density Hirose connector with nearly all I/O, similar to Intel Editon & Tinyduino, is not something I'm seriously considering.

Significantly different form factor than Teensy 3.1 is also not an option, at least not for this upcoming product. I understand it's fun to talk of such things, but it's really not an option for the next product.

The question I tried to ask was about prioritizing... pretty much the opposite of this conversation about high density connectors and form factors. Really, I'm fine with off-topic and wandering conversation. My only concern is establishing unrealistic expectations for an upcoming product.

 

[WMXZ]

The question I tried to ask was about prioritizing... pretty much the opposite of this conversation about high density connectors and form factors. Really, I'm fine with off-topic and wandering conversation. My only concern is establishing unrealistic expectations for an upcoming product.

So, it is all about possible applications. I guess, you already synthesized the multi-year forum contributions and know the different usages of TLC and T3.1. I have the feeling these applications will continue to be important, but new possibilities could arise and should be addressed by the design.

I, myself, I'm mainly interested in high-performing digital audio processing using DSP instructions on multiple MCUs. I will always use external ADC's , so high-speed data I/O and board-to-board synchronization is important for me. So far, I can all do with T3.1 @144 MHz (no luck with 168 MHz) but a higher clock would allow higher performing DSP.

 

[Frank B]

LOL, yeah, we're going to tell people to saw the edge off... what could possibly go wrong?!

Don't tell it

But, someone who is able to solder e.g. the RCT-Crystal is able to do this too, especially when an aid exists (row of small holes). *

OSH-Park delivers boards, where you have to do something similar.
But ok, i think you're right. Probably that's not ok for people with two left hands.


*If not... then you sell some more

 

[mxxx]

Significantly different form factor than Teensy 3.1 is also not an option, at least not for this upcoming product. I understand it's fun to talk of such things, but it's really not an option for the next product.

The question I tried to ask was about prioritizing... pretty much the opposite of this conversation about high density connectors and form factors..

my apologies, but that's why i brought up the dual-row thing. if a/the priority is to maintain backwards compatibility with existing projects and products, i'd just guess this would in fact be the least disruptive approach; at the same time, it's a way to accommodate a lot more pins while not relying on overly exotic connectors. it's impossible to say of course but my thinking is/was: an additional ~25 mm in one direction (increased length) is likely to cause more issues (protruding, occluding, touching other components, etc) than < 10 mm in the other (increased width).

the dual rows per se may not be breadboard-compatible but in this day and age, i find it difficult to imagine a lot of people actually breadboard anything remotely so complex and in need of all those pins. it's too easy and cheap now to just come up with a pcb. fewer and fewer parts can be breadboarded to begin with, because they're SMD. and it's easy enough to accommodate all sorts of changes on a breadboard, so all that should be fairly low priority.

 

[MuShoo]

@mxxx, I guess you and i go about things very differently - I'd much prefer to breadboard something first, and then make a PCB once I've confirmed the breadboard schematic. That's what breakout boards are for. Yes, PCBs are pretty cheap, but I still can't design a schematic and get a PCB made of it within an hour. Breadboarding is there so that you get your PCBs right the first (or more likely the second) time.

I'll include the initial breadboard design for one of my products:


That's actually the SIMPLER version of this device. The MORE complex my design is, the more likely I am to breadboard it first, because the more likely there are to be little inconsistencies that won't present themselves readily until actual in their physical configuration.

I'm not worried at all about 'backwards compatibility' in terms of hardware shape. If your design requires a teensy of a specific size, use the one you originally design it for. If your design requires the processing power of the 3++, design your hardware/case design around that requirement. If you require the processing power of the 3++ in the smallest space possible, put the electronics of the 3++ directly onto your PCB (this is what I do with my current device).

 

[mxxx]

true ... most certainly we do.

and that's fair enough. though i'd certainly disagree about the shape/size/compatibility question (more in principle, as it doesn't really cause me sleepless nights). if that wasn't an issue, why bother making something that's 'teensy', it could be any size; why bother keeping the pinout, people can re-design around any pinout; the smallest size will of course always be putting the electronics directly on a pcb, but for hobbyists (like me) being able to use a dev board tends to mean a couple of headaches less. and it's not like i'd breadboard the teensy first.

 

[Frank B]

@mxxx, I guess you and i go about things very differently - I'd much prefer to breadboard something first, and then make a PCB once I've confirmed the breadboard schematic. That's what breakout boards are for. Yes, PCBs are pretty cheap, but I still can't design a schematic and get a PCB made of it within an hour. Breadboarding is there so that you get your PCBs right the first (or more likely the second) time.

I'll include the initial breadboard design for one of my products:
View attachment 4849

That's actually the SIMPLER version of this device. The MORE complex my design is, the more likely I am to breadboard it first, because the more likely there are to be little inconsistencies that won't present themselves readily until actual in their physical configuration.

I'm not worried at all about 'backwards compatibility' in terms of hardware shape. If your design requires a teensy of a specific size, use the one you originally design it for. If your design requires the processing power of the 3++, design your hardware/case design around that requirement. If you require the processing power of the 3++ in the smallest space possible, put the electronics of the 3++ directly onto your PCB (this is what I do with my current device).

It's possible to use breadboards (I use them) with one row of pinheaders on the top, and the other on the backside - Better than non-accessible pads or hd-connectors.
( But I like those pads too )

But Paul already said that dual-row is out of the game, so this discussion is fruitless

@Paul:
Everyone want other functions/pins, and we can't have them all - Someone will always complain. I, personally have no idea why someone needs all these UARTs or more than two I2C - but i don't care much about it. Some functions make more sense than others..
You want to make a versatile product, so i bet at the end we get a good compromise from you.

 

[MichaelMeissner]

my apologies, but that's why i brought up the dual-row thing. if a/the priority is to maintain backwards compatibility with existing projects and products, i'd just guess this would in fact be the least disruptive approach; at the same time, it's a way to accommodate a lot more pins while not relying on overly exotic connectors. it's impossible to say of course but my thinking is/was: an additional ~25 mm in one direction (increased length) is likely to cause more issues (protruding, occluding, touching other components, etc) than < 10 mm in the other (increased width).

I don't think Raspberry Pi is a good model for dual rows, since the pins are all on one side. I find such things to be unstable physically, and you have to use mounting screws to provide stability (or don't stabilize it).

Beaglebone Black is better in that you have dual rows on each side, and you could mount a shield without having mounting screws.

the dual rows per se may not be breadboard-compatible but in this day and age, i find it difficult to imagine a lot of people actually breadboard anything remotely so complex and in need of all those pins. it's too easy and cheap now to just come up with a pcb. fewer and fewer parts can be breadboarded to begin with, because they're SMD. and it's easy enough to accommodate all sorts of changes on a breadboard, so all that should be fairly low priority.

I dunno, I'm one of those people that still only uses breakboards and generic protoboards. I find I often change pin assignments as I modify the code. Unless you could do overnight pcb production with quantity 1 boards in the $1 range, I don't see how custom pcb's would work during development. Sure, once you have the basic design down, then I can see the use of custom pcb's.

I suspect we are just doing different levels of teensy programming.

 

[PaulStoffregen]

I, myself, I'm mainly interested in high-performing digital audio processing using DSP instructions on multiple MCUs.

I'm also really interested in this.

Really, my hope is to get this product released, and then be able to spend about 1 year on really working on software and the website documentation.... kinda like the first several months after Teensy 3.0 (where I got to create OctoWS2811 and ported many libs) and the 10+ months after Teensy 3.1 (where I worked mostly on the audio library). Most of 2015 seems to have been consumed by working on hardware... first Teensy-LC, then a lot of mundane manufacturability stuff on existing boards, and now Teensy3++. There's so much more I want to do on the software side, and some of the really awesome audio stuff is going to need the extra memory and speed of this board.

I also want to try some ambitious ideas for adding networking layers into the Arduino system, which is the big reason why I want to go with the chip that has ethernet. That too will likely turn into an absolutely huge software project. Nobody in the Arduino or microcontroller world seems to be really working on such things, despite all the IoT hype. Seems their answer to complexity of networking is to just use Linux.

 

[PaulStoffregen]

I suspect we are just doing different levels of teensy programming.

Yes, this so very true.

Teensy is used by many people with an incredibly wide range of skills for a tremendous variety of projects.

 

[mxxx]

I don't think Raspberry Pi is a good model for dual rows, since the pins are all on one side. I find such things to be unstable physically, and you have to use mounting screws to provide stability (or don't stabilize it).

Beaglebone Black is better in that you have dual rows on each side, and you could mount a shield without having mounting screws.

oh, i didn't mean to advertise raspberry pi as a model; see above, i referred to the olimex stm32 boards. anyways, i most certainly didn't mean to turn this into a do-people-use-breadboards discussion, so let's just not mention dual rows again. frankly, i don't care much; i'm usually served pretty well by just the outer pins of the 3.x boards. fwiw, the main thing i'm looking forward to is the additional SPI port(s). (edit: even though that i think that's something silly to say; as in: i like to do X, that's why Y is important.)