Forum Rule: Always post complete source code & details to reproduce any issue!
Page 5 of 11 FirstFirst ... 3 4 5 6 7 ... LastLast
Results 101 to 125 of 254

Thread: Is there a market for a Teensy 3.1 48 pin ARM stamp?

  1. #101
    Senior Member
    Join Date
    Jun 2013
    Location
    So. Calif
    Posts
    2,828
    BGA...
    Out of the box thinking (uh-oh) ...

    Using tiny chip-o-board (die) with wire bonding... We're doing that on a project I'm on. Of course, much, much smaller than a leaded-package.
    The issue is product volume, and availability of K20 chips on wafers, pre-tested.
    We've used a a few local wire bonding shops for low volume prototypes. It can be affordable at low volumes.
    There's also ACF instead of traditional wire-bonding - Anisotropic conductive film, commonly known as ACF. We've started with one local company that does ACF.
    http://flipchips.com/tutorial/assemb...-introduction/

    But, it's probably easier to use BGA. Might need a very, very good PCB fab at those tolerances. Solder flux balls' sizes and the stencil have to be well done.
    Last edited by stevech; 06-15-2015 at 04:32 PM.

  2. #102
    Senior Member
    Join Date
    Nov 2012
    Posts
    412
    Now if someone could come up with a solution for the Teensy 3.1 and NOT talk about the future Teensy 3.x++ !!!!
    BTW ... there is a dead link for the Teensy 3.x++ on this forum board. It died last November.

    The 3.x++ is just vaporware until released and if it was released tomorrow, it would take 1 to 2 years for the software libraries to be thoroughly tested and debugged.

    As a customer, again, I am looking for a "present" clean solution for a 48 pin Teensy 3.1.

    From what I can ascertain from this topic, the company owner says NO due to no market (I disagree 100%) and other forum users came up with several great ideas.
    To be continued ...

  3. #103
    Senior Member
    Join Date
    Jun 2013
    Location
    So. Calif
    Posts
    2,828
    Just tossing around notions - on how to improve substantially, not just a trivial delta.

  4. #104
    Senior Member+ defragster's Avatar
    Join Date
    Feb 2015
    Posts
    8,805
    Quote Originally Posted by stevech View Post
    Just tossing around notions - on how to improve substantially, not just a trivial delta.
    Indeed - as noted - we should start a new thread for the T3.pp. Suffering the BG assembly would keep the size down to compatible width.

  5. #105
    Senior Member
    Join Date
    Jun 2013
    Location
    Torrance CA
    Posts
    395
    I've been watching this thread with some curiosity. I've often had to use the bottom pads of T3.x's in projects due to needing more GPIO etc.

    Given that Paul S has said there's no official PJRC interest, someone would have to do this as a Tindie etc. independent project. As an intellectual exercise I looked at the numbers.

    Based on other projects I've done, the base components (MK20, crystal, protection diodes, USB connector, discrete) + vendor shipping of the T3.1 are about $8.00 when purchased in qty (100) from Digikey & Mouser.

    The Mini54 from pjrc.com is $8.00 on the web site but I think in another post somewhere Paul said he'll do quantity discounts so let's assume $7.00 for that.

    100 PCBs @ 3cm x 7cm from ITeadStudio + shipping will be $75.21, so let's say $0.75 per board for that.

    A quick quote from smallbatchassembly.com says building 100 boards will be about $800, so that's another $8.00.

    Now we have to pay someone to test the boards and subcomponents etc., package the device for shipment, actually ship the product, email the invoice, etc. Let's assume $20/hour and all the above manual labor takes 15 minutes. That's another $5.

    So far we have $8 + 7 +0.75 + 8 + 5 = $28.75.

    There's three more things we'll need to pay for:

    1) Profit margin to make this project worthwhile
    2) Cost of shipping, padded envelope, antistatic bag, etc.
    3) Cost of the payment processing

    Assume #2 is passed off to the customer.

    For #1, from what I've read on reddit.com/r/hwstartups and other sites the rule of thumb is selling price should be 2.5x the base cost but that means our board will cost about $70 and I don't realistically think anyone will pay that. But let's say our theoretical builder wants to build these for goodwill more than profit and can live on $1.24/unit profit instead. So selling price will be $29.99.

    For #3, Tinde takes 5% of a sale and PayPal takes 2.9% + $0.30 so solving for X yields about $32.93.

    So, is there a market for 100 of these 48 pin Teensy 3.1 clones ~ $33 each?

    As much as I'd like a T3.1 with all the pins broken out I just don't think it would be worth a $13 premium. I might buy one just to expand my arsenal but can't see buying these consistently.

    But I realize I'm not 100% of the market, and if someone here wants to pursue this I'll be more than happy to help with building a BOM, designing the board, etc.

  6. #106
    Senior Member+ defragster's Avatar
    Join Date
    Feb 2015
    Posts
    8,805
    Quote Originally Posted by t3andy View Post
    link for the Teensy 3.x++ on this forum board.
    Can you re-open or post that link - I didn't find it - I started in Feb '15 as the 1.6.x stuff emerged and this was the first break I saw to discuss something since the LC and other updates took a break.

  7. #107
    Senior Member
    Join Date
    Jun 2013
    Location
    So. Calif
    Posts
    2,828
    It adds up, yes. Add warranty costs too.

    So is a board like Particle.io's Photon selling below cost? RPi 2 - bravely selling with low margins until volume got super high?
    https://store.particle.io/

  8. #108
    Senior Member
    Join Date
    Jun 2013
    Location
    Torrance CA
    Posts
    395
    Quote Originally Posted by stevech View Post
    So is a board like Particle.io's Photon selling below cost? RPi 2 - bravely selling with low margins until volume got super high?
    https://store.particle.io/
    I'm almost certain they're not selling for below cost. My guess is they're building those boards 1,000 or more at a time. Electronic components get much cheaper the more you buy -- for example the main MCU in the Teensy 3.1 is $7.40 if you buy them one at a time but is 40% less ($4.31) if you're buying 10,000 of them. PCB manufacture and assembly are similarly less expensive with larger batches....economies of scale etc.

    I picked 100 as the theoretical run for the enhanced Teensy 3.1 clone board because
    1. It's a reasonable initial capital outlay (you'll need to spend ~ $2,500 to get everything before you can start selling any)
    2. That's my guess at the size of the market

    If you could guarantee an order of 1,000 boards and could pay out of pocket for the upfront cost of the ~ $20,000+ in parts etc. you'd need to buy before you can sell your first board then you might be able to make them for about $25 each....

  9. #109
    Senior Member
    Join Date
    Jun 2013
    Location
    So. Calif
    Posts
    2,828
    particle.io's web site said they sold 17,000+ of gen 1 at $39.
    gen 2 starts July they say, at $19. Better MCU. Better WiFi (they changed from T.I.'s crappy CC3000 to Broadcom and it got cheaper!)
    I guess they bought gen 2 BOM parts assuming lower price will create more volume.

    So too the RPi and RPi 2 story.

    It's curious, this era of crowdfunding and all.

  10. #110
    Member
    Join Date
    Jan 2014
    Location
    Oslo, Norway
    Posts
    32
    Gen1?, Don't know how many Spark Core they sold, but it's the Photon they sold 17k of, I ordered mine in January, and first now they are in the mail somewhere.

  11. #111
    Senior Member PaulStoffregen's Avatar
    Join Date
    Nov 2012
    Posts
    20,171
    Don't forget to factor in reinvesting some of that revenue into ongoing product and software development. Especially the software!!

  12. #112
    Senior Member Constantin's Avatar
    Join Date
    Nov 2012
    Location
    In the yard with a 17' Dia. Ferris Wheel
    Posts
    1,408
    I think Potatotron hit it on the head with his extensive analysis. Given the likely cost of assembling a 48-pin Teensy, my suggestion is to focus on using the Tall-Dog designed breakout board on Tindie instead. The cost is about the same and there is no business risk. Heck, it's even got a Pogo pin for the reset pad!!!

    However, anyone with too much time on their hands is free to develop a new PCB, source the parts and assemble it all. I doubt it would be cost effective, even if done at offshore labor rates.
    Last edited by Constantin; 06-16-2015 at 07:30 PM.

  13. #113
    Senior Member
    Join Date
    Jun 2013
    Location
    So. Calif
    Posts
    2,828
    Tall-dog's thing is fine for a breadboard test/dev.
    But to build things used in real apps, not benchtops, it's my view that we need an interconnect using mating connectors, for expansion. Rather than or in addition to an option for pin headers.

  14. #114
    Senior Member+ defragster's Avatar
    Join Date
    Feb 2015
    Posts
    8,805
    Indeed - as noted - any new device should have modularity. But creativity like this on the T3.1 shows there are multiple ways to work with the under pins already - is that 51 I count for about a buck fifty? : https://forum.pjrc.com/threads/26071...-with-a-socket

  15. #115
    Senior Member
    Join Date
    Jun 2013
    Location
    So. Calif
    Posts
    2,828
    that is clever.
    works for a 1-of, or so.

  16. #116
    Senior Member
    Join Date
    Mar 2013
    Posts
    651
    My greatest grip with the Tall dog is if you screw up its a bear to disassemble.

  17. #117
    Senior Member Constantin's Avatar
    Join Date
    Nov 2012
    Location
    In the yard with a 17' Dia. Ferris Wheel
    Posts
    1,408
    I didn't say that the Tall dog rig is perfect... It's merely one way to get access to all pins (plus reset!) on a Teensy 3.1, especially for breadboarding.

    Would I use the Tall dog rig in a finished product? Unlikely! It's not IMO, meant for that type of application. I'd design my own Teensy clone instead...

    jimmayhugh's approach to getting to those header pins in the middle is pretty brilliant, given how difficult it is to accurately position a 14-point header so that it registers nicely with the other ones. The downside is the additional labor required to assemble the rig up front.

  18. #118
    Senior Member+ defragster's Avatar
    Join Date
    Feb 2015
    Posts
    8,805
    This is a huge uno sized board - and doesn't bring out the lower pins directly . . . but sockets would leave room or ways to fix that perhaps - for benchtop use at least

    @MichaelM just posted this link to a $10 option SparkFun saw fit to make: https://forum.pjrc.com/threads/28948...no-form-factor

  19. #119
    Senior Member
    Join Date
    Nov 2012
    Posts
    412
    This topic is a conundrum ... the definition of conundrum is a situation where there is no clear right answer or no good solution.

    The Teensy name was built on microcontroller(s) which are small and tiny.

    ARM chips are very small in footprint but carry a large number of GPIO.

    You can't have both. The Teensy name would have to change due to a larger footprint if one were to use all the GPIO the ARM chip(s) has to offer.

    The situation, right now, is the Teensy 3.1 is very small but many of the GPIO is on the bottom layer which is very costly and hard to connect.

    Last edited by t3andy; 06-22-2015 at 01:45 PM. Reason: addition

  20. #120
    Quote Originally Posted by t3andy View Post
    The situation, right now, is the Teensy 3.1 is very small but many of the GPIO is on the bottom layer which is very costly and hard to connect.
    No that's not the situation right now. The situation is that the option (that tindle breakout people keep posting) for getting those pins on the bottom layer is not elegant. It's not hard or particularly costly.
    There is a perfectly acceptable solution RIGHT NOW.

    Fab'n a new board would be costly and hard. No one is going to complain if you want to go ahead and do it though.

  21. #121
    Senior Member
    Join Date
    Jun 2013
    Location
    So. Calif
    Posts
    2,828
    If/until the 0.1 in. breadboard-compatible interconnect is augmented (not replaced) with a more dense connector - we are stuck with a low pin count ARM package and fewer on-chip peripherals, meaning more struggles to time-share the on-chip peripherals.

    But indeed, Teensy needs to be teensy.
    Last edited by stevech; 06-22-2015 at 06:37 PM.

  22. #122
    Senior Member+ defragster's Avatar
    Join Date
    Feb 2015
    Posts
    8,805
    Based on what PJRC showed in one benchmark of DUE and (<)ZERO - the T_3.1 has awesome power in Teensy space. Not ideal? - but with a few solutions it has more pins than it should for its size.

    I never found the link to the old T_3.1pp thread and no new thread yet . . . but getting the next thing designed/built would be better than over-extending the old one.

    Having even more power in marginally extended footprint ( 40-44 ) with a high density plug would be awesome - and it would ship on day one with software compatibility at 80-120% of a T_3.1. Maybe the bottom could put an SPI memory (instead of 3.1 pads) and SDIO flash footprint (under the extended end) with extra pins? Much larger and it would need a new name and not fit in the same project space as well as it does - though some of these could be cool names:

    Teensy Antonyms:
    astronomical, big, boundless, bulky, bumper, cavernous, colossal, considerable, cosmic, elephantine, enormous, extensive, formidable, giant, gigantic, good, goodly, grand, grandiose, great, gross, handsome, hefty, herculean, heroic, huge, hulking, immeasurable, immense, imposing, infinite, jumbo, king-size, large, largish, lofty, majestic, major, mammoth, massive, monolithic, monster, monstrous, monumental, mountainous, outsize, overgrown, oversize, planetary, prodigious, sizable, staggering, stupendous, substantial, super, titanic, towering, tremendous, vast, Vasty, voluminous, whacking, whopping.

  23. #123
    Senior Member
    Join Date
    Feb 2013
    Posts
    179
    Heh, the Teensy Jumbo.

  24. #124
    Senior Member
    Join Date
    Jun 2013
    Location
    So. Calif
    Posts
    2,828
    for me and others: 2nd SPI, SDIO, h/w floating point. And no added 0.1 in. pins on board edge.
    Jumbo-not!

  25. #125
    Senior Member
    Join Date
    Nov 2012
    Posts
    412
    I never found the link to the old T_3.1pp thread and no new thread yet
    https://forum.pjrc.com/threads/24633...ight=Teensy-LC

    Heh, the Teensy Jumbo.
    LargeDuino and if Paul could produce a 48 pin Teensy 3.1 maybe it could be called Teensy 3.1 LE (limited edition)

    The cost for a user to produce such a special ARM stamp would not be cost effective unless it was produced by PJRC.
    Last edited by t3andy; 06-22-2015 at 11:07 PM.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •