On the subject of high-density connectors...
I really like the connectors of the Arduino Portenta.
I mentioned this previously and agree it would be an optimal solution for users who have requested high-density IO buses to be brought out.
For everyone who likes Arduino Portenta's high density connectors, I would like to ask you very specifically which other boards have you plugged your Portenta into so far? I especially want to hear about any experiences with custom made or non-Arduino boards which mate with Portenta's high density connectors.
It's the same solution reached by Arduino Portenta, pyboard, Rasp Pi compute module, many others.
While I don't want to nitpick minor differences, I do feel that high density connectors are not at all the same design as a card edge which mates with a SIMM or DIMM socket. Mating with a M.2 socket (eg, Sparkfun MicroMod) is probably pretty similar to a card edge for a DIMM socket.
Then there are the users like me who are willing to design a baseboard to bring out high density IO.
Have you designed any base boards for other products using the high density connectors.
it makes pluggable mounting to a baseboard much easier than it is now, with the mix of 2.54 & 1.27mm (ethernet) pin centers on the Teensy 4.1.
Having plugged Portenta into Arduino's first base board (the one with ethernet that's about the same size as Portenta), I do agree, it is a pretty nice experience. Or at least it is with Arduino's official products... and for the retail prices they're asking, you'd expect it to be a premium experience!
So far I'm not seen any 3rd party base boards for Portenta. Do any exist yet? Are any being commercially manufactured by 3rd parties? I'm particularly curious about the practical aspects of meeting the part placement tolerance for 2 high density connectors spaced relatively far apart for the boards to easily mate.
I know at one point Paul was not overly positive about connectors like the ones used on the now retired Intel Edison...
My opinion is really just an echo of the people I know who actually used Edison.
Back when Edison was on the market, I talked with many people about it. Lots of people who liked the concept of Edison but hasn't actually put it to any substantial use loved the high density connector idea. Pretty much everyone I talked with who actually did use Edison hated the reality of the high density connector.
This is exactly the problem with the high density connectors. For development boards that use them, the breakout boards are few, and they tend to be expensive.
High cost is indeed my main concern. Arduino can sell a $100 board and $50-80 accessories. PJRC can't. Or maybe we could, but it's a risk I don't wish to explore. And whether Arduino really can make $100 pricing work long-term in this market filled with agressively priced boards like ESP & Raspberry Pi is still unknown. It very well could end up like the Arduino Robot...
It would be simple enough for Paul to release a board that is routed to accept high-density connectors on its bottom side, and offer it with or without connectors installed.
Normally I don't talk much about the manufacturing side of Teensy. We use a local contract manufacturer for all the SMT soldering. Like all CMs, they're not usually eager to share too much detail of their internal process. But they are local and in pre-Covid times Robin & I could just drive over there and meet with them in person. We did have quite a lot of communication with them in the lead up to Teensy 4.0 and 4.1, which are the only Teensy models to have parts soldered on the bottom side. (The SD adaptor for Teensy 2.0 also has parts on both sides)
The way Teensy 4.0 & 4.1 are being manufactured is optimized for only 402 & 603 size parts on the bottom side. The clear feedback we've heard is putting as many of those 402-603 parts on the bottom and keeping all the complex parts on the top is optimum. They actually wanted *all* the 402s on the bottom, but they were pretty understanding that at least some parts like the crystal oscillator capacitors had to be on the top for solid engineering reasons. I'm pretty sure they could put complex parts like high density connectors on the bottom side. But it would almost certainly become a much more complicated (and thus expensive) process to manufacture.
From a business perspective, experimenting with different ways of manufacturing Teensy using a CM has an entirely different set of trade-offs than in-house manufacturing like Sparkfun & Adafruit do.
Whether only routing the PCB (leaving unpopulated locations on the bottom) is "simple" could be debatable. Arduino's website has a nifty tool to show you their PCB layout. Portenta is an 8 layer PCB. While they don't give the exact specs, if you look closely it's pretty clear they're using a "HDI" process with blind/buried laser drilled holes on the outer layers. Looking at my Portenta under a microscope shows completely flat vias, so my guess is they're also using filled & plated via process! I haven't look carefully at which combinations of layers are (likely) mechanical drilled. Seriously, just spend a few minutes looking at their PCB layout and try to figure out the PCB specs. Portenta has a mind boggling complex PCB layout. Maybe someone better than me at high density PCB layout could make more sense of it... but from the perspective of having made Teensy 4.0 and 4.1 which use "normal" 6 layer PCB process (mechanical drill through all layers), looking at Portenta's layout seems to be pretty much the exact opposite of "would be simple". But it does explain the high price. Each of those HDI spec things adds a lot of cost to the PCB, and it looks like they used all of them.
High density connectors also have the problem that they don't live long, if you use them repeatedly.
Intuitively this makes sense. Please don't interpret this question as skepticism, because I do agree. But I still would like to know if any hard data exists about the realistic lifespan and number of non-ideal human behavior mating cycles these connectors can be expected to still work reliably?
I.e. i don't have plans to do much with the Teensy MM. The connector is too much restriction (and it has less I/O, too).
I'll admit, I would a little skeptical when Sparkfun first told me of their MicroMod idea (or the Teensy-specific portion I was allowed to know). Since they're released, I've heard a lot of people express interest. It's still new in the market, so that may just be a honnymoon phase. But some of the comments I've heard seem like pretty realistic low-volume product plans.
I'm trying to keep an open mind, about both card edge modules and high density connectors. That's why I'm taking the time today to write this lengthy reply, even at this stressful time where I'm just barely managing to keep up with only some of the many new forum threads. These are important concepts and decisions for the future. Please know I am here and I am listening, even if I don't always manage to reply to everything.