fgervais001
Member
I would prefer if you would not link to my images without my permission, could you remove it please?
Future Teensy features & pinout
- Thread starter PaulStoffregen
- Start date
PaulStoffregen
Well-known member
I must agree, seeing the AC mains wires in that photo makes me cringe. I'll admit, I've used similar wiring in LED projects, but always on the back side of a display hidden away from where people could easily touch. For display at events like Maker Faire, I've added a plastic frame around those wires, even when they're not easily accessible.
You're probably not interested in this feedback or reworking your already-built project, but seriously from a human safety perspective, you really should use this sort of power supply. It should be "IEC Class II" rated, which is usually indicated by this symbol of 2 squares.
IEC Class II means all paths from the AC mains to output have 2 layers of insulation. Especially for an installation where unattended children could touch the wires, skimping on this safety feature is pretty unwise.
Regarding the use of M.2 E key connectors, a Sparkfun Micromod M.2 product is on the way, as discussed elsewhere on this forum. It will use the same 1062 chip as on Teensy 4.0 & 4.1. How successful that product is will probably influence future choices about M.2 format boards.
You're probably not interested in this feedback or reworking your already-built project, but seriously from a human safety perspective, you really should use this sort of power supply. It should be "IEC Class II" rated, which is usually indicated by this symbol of 2 squares.
IEC Class II means all paths from the AC mains to output have 2 layers of insulation. Especially for an installation where unattended children could touch the wires, skimping on this safety feature is pretty unwise.
Regarding the use of M.2 E key connectors, a Sparkfun Micromod M.2 product is on the way, as discussed elsewhere on this forum. It will use the same 1062 chip as on Teensy 4.0 & 4.1. How successful that product is will probably influence future choices about M.2 format boards.
fgervais001
Member
Thank you for the feedback, I understand you concerns.
Nice, I wish you the best of luck with your upcoming products.
Nice, I wish you the best of luck with your upcoming products.
I might be wrong here but from my point of view a big factor will be the pricing. As an M.2.processor only module (like it's done with MicroMod) will not work without* an carrier it should be cheaper than a comparable*full Teensy. If it costs the same or even more it's somewhat hard to find a good reason to use it. I know that in the case of the MicroMod Sparkfun will call the the prices. It's just that even if the MicroMod version will not been a success doesn't necessary mean that the community doesn't like to have an easy to use processor only solution. I know there is an ongoing*discussion about a bootloader chip but to be honest even if it would be accessible*right now the requirements*to use it with a custom project are just to high for the kind of projects I am involved. But a processor module that could be attached to a much easier to build carrier board would hit the sweet spot.
Kind of hard to compete on price/cost with 100 mil pin headers.
Personally, I'd love to have a low cost compute module available but without ethernet, meh. And micromod has a hard limit on gpio count.
Pin headers are fine. I would not insist on anything like an M2 connector for a processor module.
With a processor module I meant a Teensy striped*down to the bare minimum to allow it to run. Even the USB port, LEDS, buttons etc need to be on the custom carrier in this case if they are needed at all. But to be honest I don't know how much price reduction this would allow.
A micromod processor is pretty close to that. SFE sells them - RP2040 for $12, SAMD51 for $15. I'd take short odds on the 1062 one being $15.
Frankly, give me a Teensy4.0+ ("double wide" with 80 pins including ethernet, host USB and lcd pins) for $20 and I would be happier than a pig in sh!t. I think that would open up more market opportunities for me. I've sold >500 boards that take a T4.1 and require some assembly, I'd imagine a $20 T4.0+ with lots more pins would increase that, maybe significantly.
I think they'll probably be closer to $20 to cover the MKL02 and Paul's royalties. Hopefully we'll know soon.
PaulStoffregen
Well-known member
We're not even able to make the normal width with 48 pins (Teensy 4.1) having USB host & ethernet at $20. Growing the PCB larger isn't going to make it cheaper!
Many of the parallel LCD pins conflict with other uses, no matter the size and cost of the PCB. That's why it wasn't brought out to a connector or any other way.
One can but hope!
I'd be happy with the just the lines to the phy brought out to headers. Don't know if that has timing/routing issues. I assume (incorrectly?) that Host USB goes straight into the 1062. I was also thinking the SD card didn't need to be on the core module though that is a relatively low cost item.
And for the record, I was thinking a 40 pin double wide.
DavidWH1968
Active member
DO you mean a double wide PCB, or maybe a double row of pins at each side (yielding 80 pins) ? Either way, @PaulStoffregen is right that the PCB will get larger, and that will increase the cost per module. The double row of 100mil pins would, however minimize the cost increase some.
Last edited:
If you mean that the "MSRP" of a module with twice as much area would be double, that completely ignores the cost of the various components (chips, connectors, ...) which are likely to dominate the pricing. Other factors like special processing and shipping can add up quite a bit, as well. In the boards I make and sell, PCB (bare) costs are about 20% of the the partially assembled boards. For fully assembled, closer to 10%. It is not a factor to ignore but clearly not even close to the dominant one in pricing a fully or partially assembled board. In the case of the Teensy 4, the 1062 is likely the largest cost factor. In the case of my boards, connectors are.
By "double wide", I meant 2 rows of 2x20 pins on each side. So, in the case of a Teensy-like product, .9" wide - 29% wider.
Your statement was ambiguous so I sought to disambiguate. Plus, the bare PCB cost doesn't drive the end price very much (as I pointed out) - a lesser argument point IMO.
When talking about future teensys, that doesn't automatically mean newer processors. The 1062 is a fantastic processor and I would be happy to have more choices based on that. In my earlier post, I called the new board a T4.0+, implying a 1062.
As to routing more signals, I don't have insight into the T4.0/4.1 routing so it's not clear that implies more space. The T4.1 with more lines didn't have to get any wider than the T4.0 though that is only 8 more lines.
When talking about future teensys, that doesn't automatically mean newer processors. The 1062 is a fantastic processor and I would be happy to have more choices based on that. In my earlier post, I called the new board a T4.0+, implying a 1062.
As to routing more signals, I don't have insight into the T4.0/4.1 routing so it's not clear that implies more space. The T4.1 with more lines didn't have to get any wider than the T4.0 though that is only 8 more lines.
DavidWH1968
Active member
Bare PCB cost can drive the price significantly if you have to use a 6 layer PCB instead of a 4 layer board. It shouldn't be necessary to have too many actual connectors, apart from the SD slot, on the board at retail. The current method on the 4.1 of making the footprints available for extra RAM and flash seems to work fairly well. The same can be done with a parallel SDRAM footprint and other optional connections that many developers may or may not want to use.
The big concern over the added footprints is that the more complex they are, obviously, the more real estate that will be consumed. And the higher speed connections (ie: 200MHz SDRAM, or a 100+MHz video signal) would require more signal routing area for length matching. This type of thing could lead to a need for more layers on the PCB, resulting in possibly significantly higher cost when manufacturing the PCB itself.
Personally, I would be willing to pay the extra money if needed to have ready access to a more capable development board with good flexibility in how I use it. The T41 comes close to this, but falls short in the external memory access with the horribly slow PSRAM. (Still far better than nothing though!)
Of course, @PaulStoffregen could just opt for a much larger board and give us a jack of all trades device, but that is already present in the RT1170 EVK from NXP that is designed to be a jack=of-all-trades board. And Arduino has already proven that boards priced in the $75 and up range get a very limited audience.
The big concern over the added footprints is that the more complex they are, obviously, the more real estate that will be consumed. And the higher speed connections (ie: 200MHz SDRAM, or a 100+MHz video signal) would require more signal routing area for length matching. This type of thing could lead to a need for more layers on the PCB, resulting in possibly significantly higher cost when manufacturing the PCB itself.
Personally, I would be willing to pay the extra money if needed to have ready access to a more capable development board with good flexibility in how I use it. The T41 comes close to this, but falls short in the external memory access with the horribly slow PSRAM. (Still far better than nothing though!)
Of course, @PaulStoffregen could just opt for a much larger board and give us a jack of all trades device, but that is already present in the RT1170 EVK from NXP that is designed to be a jack=of-all-trades board. And Arduino has already proven that boards priced in the $75 and up range get a very limited audience.
PaulStoffregen
Well-known member
FWIW, Teensy 3.5, 3.6, 4.0 and 4.1 are all 6 layer PCBs. Vias are standard all-layer drilled.
I believe Arduino Portenta uses 8 layers with HDI features like blind laser drilled vias.
I believe Arduino Portenta uses 8 layers with HDI features like blind laser drilled vias.
DavidWH1968
Active member
Thanks for the clarification. I was under the impression they were 4 layer boards.
Is there anything new on the RT1176 front, such as decisions regarding expansion footprints, pin layout, etc? (Not trying to push on this because I know you have alot on your plate with updating Teensyduino support for Arduino 1.8.15, and soon the Arduino 2.0 IDE.)
Hi
I use Teensy for audio experiments, and have built a few small synths and some effects units. Currently I'm trying my hand at digital guitar effects in a stomp box format. To me the two-core versions of the 1170 seems like a dream come true. I'm all for simplicity in programming to enable fast prototypes and having one slow core for handling the 'UI' (buttons, rotary encoders, LEDs and displays etc.) and presets and one fast for for real-time audio processing is perfect instead of having to implement cooperative multitasking to do both on one core.
To me the perfect 1170-Teensy would be:
* Two core 1170 model
* Single board with MCU, Codec, basic analog buffers/protection on one board. Codec connected to the high performance core.
* RAM: solder pads for QSPI chips like on the Teensy 4.1
* Flash: solder pads or SD connector
* 44.1KHz/16 bit stereo codec is enough for my purposes. High sampling frequency would be a bonus as it can simplify handling aliasing. To me there is no need for increasing resolution to 24 bits.
* Another nice bonus would be a Codec with some DSP power like some of the AKM chips, but all the stuff above would be more important. Low end-to-end latency is also more important.
/MJ
I use Teensy for audio experiments, and have built a few small synths and some effects units. Currently I'm trying my hand at digital guitar effects in a stomp box format. To me the two-core versions of the 1170 seems like a dream come true. I'm all for simplicity in programming to enable fast prototypes and having one slow core for handling the 'UI' (buttons, rotary encoders, LEDs and displays etc.) and presets and one fast for for real-time audio processing is perfect instead of having to implement cooperative multitasking to do both on one core.
To me the perfect 1170-Teensy would be:
* Two core 1170 model
* Single board with MCU, Codec, basic analog buffers/protection on one board. Codec connected to the high performance core.
* RAM: solder pads for QSPI chips like on the Teensy 4.1
* Flash: solder pads or SD connector
* 44.1KHz/16 bit stereo codec is enough for my purposes. High sampling frequency would be a bonus as it can simplify handling aliasing. To me there is no need for increasing resolution to 24 bits.
* Another nice bonus would be a Codec with some DSP power like some of the AKM chips, but all the stuff above would be more important. Low end-to-end latency is also more important.
/MJ
DavidWH1968
Active member
It seems to be pretty much a done deal about the dual core RT117x MCU, after all that's what we're all patiently waiting for. The audio buffers don't seem too likely, but we've been putting Teensy's on breakout boards for that purpose. The big one to me is that the RT117x series MCU's have got a full 32-bit, 200MHz, external memory interface built in to them, so I'm personally hoping to end the reliance on slow QSPI memory in favor of at least 16-bit parallel SDRAM and Flash, or ideally two SDRAM footprints for adding 32-bit external memory. I'm hoping, but not counting on these things because there are already a lot of "must have's" like the audio, video, serial, DAC, etc.
Too many of these good things and we could end up with a mini-ITX form factor when what we really want is small for prototyping.
Agreed, the 'real' memory bus is a biggie. My point was more to have the option to expand memory. The QSPI chips are faily easy to solder by hand, but I do not know if the same goes for 32-bit memory. On the other hand: having some external memory pre-soldered to the board might actually result in a smaller board than a board with the option for adding your own. I don't know how that would affect the price of the board, but I guess a single SDRAM chip in bulk quantities does not add much to the total price.
I do think that audio is quite big niche of what teensies are used for, so having an audio-specific board with codec integrated on the board would still make a lot of sense - to me
DavidWH1968
Active member
There will also always be resellers (like ProtoSupplies) that will sell the Teensy with optional parts already soldered on the board for you. I personally have both a reflow oven and a hot air station so I'm not concerned about putting a little solder paste on the pads to attach a TSSOP SDRAM or NAND flash. I know, however, that many (maybe most) do not have access to either a reflow oven or a hot air station. If you are good at soldering, though, even TSSOP can be hand soldered fairly quickly. Not much different from hand soldering a QFP144.