Shop OBEX P1 Docs P2 Docs Learn Events
A Modest Proposal: Parallax Propeller M0 — Parallax Forums

A Modest Proposal: Parallax Propeller M0

Bill HenningBill Henning Posts: 6,445
edited 2014-05-23 14:40 in Propeller 1
Propeller M0 - cut down P16X64, could be called P4X22

(note to all - this is an idea for Parallax, for after P16X64, in order to capture new markets)

28 pin dip

Vss
Vdd
XIN
XOUT
/RST
BOE

22 I/O Smart pins (ideally drop BOE, get 23 I/O's)

use 2 for SER
use 4 for SPI flash

leaves 16 USER IO

4 cogs
128KB hub (4 system clock / 2 instruction cycle access to hub for each cog)

Almost exactly 1/4 of current design, means 4x as many die's per wafer, lot lower cost

3v3 only (to get rid of low vcore, causes more heat, so we drop fMax)

100MHz / 50MIPS

Should fit in 3.5mm x 3.5mm, so DIP-28 friendly

Target price: $1.50 in 1K quantity

Same instruction set as P16X64, so no new compiler work needed.

Parallax's answer to cheap ARM / PIC32 dip chips!

Comments

  • Oldbitcollector (Jeff)Oldbitcollector (Jeff) Posts: 8,091
    edited 2014-05-22 11:02
    Got a hack saw?

    propeller_dip_cut.jpg
    propeller_dip_cut.jpg
    458 x 279 - 25K
  • PublisonPublison Posts: 12,366
    edited 2014-05-22 11:10
    Didn't Ray already do this? :)

    http://forums.parallax.com/showthread.php/149255-P8X16A-DIP24-16xI-O-Propeller-now-working-)?highlight=Propeller
  • Bill HenningBill Henning Posts: 6,445
    edited 2014-05-22 11:33
    Gentlemen, we have proof positive that Ray has a hack saw...
  • jmgjmg Posts: 15,173
    edited 2014-05-22 15:20
    Bill Henning wrote: »
    3v3 only (to get rid of low vcore, causes more heat, so we drop fMax)
    The VCore still has to be there, but an on-chip regulator can give the illusion of 3v3.

    Some vendors can do Wide Vcc (1.8-5.5v) and not need a VCore decoupling pin, but for more powerful parts, most seem to bring out a VCore decoupling pins. (eg Cypress PSoC4000 does this )

    The $1.50 target price may be high, as the newest devices like PSoC4000start at 29c with Flash
    http://www.cypress.com/?id=5155

    This family comes in 
    QFN-24 24-pin 4 x 4 x 0.6 mm QFN with 0.5-mm pitch
QFN-16 16-pin 3 x 3 x 0.6 mm QFN with 0.5-mm pitch
SOIC-16 16-pin (150 Mil) SOIC
SOIC-8 8-pin (150 Mil) SOIC

    DIP as a target may be pushing back in time too far, but certainly 'easy SMD' is a viable package.
    Users can always mount, or buy mounted, on DIP headers for development.

    Those above are the 1.27mm pitch packages, and I think the TQFP32 @ 0.8mm is also hand and wave solderable.

    .. and for a more capable small uC module example, look for
    KIT_XMC_2GO_XMC1100_V1

    - Mouser list this at 1+ $6.75, with the Debug-USB link included on the DIP adapter board

    The XMC1100 start from 100 $0.95
  • User NameUser Name Posts: 1,451
    edited 2014-05-22 15:45
    jmg wrote: »
    DIP as a target may be pushing back in time too far

    I'm presuming Bill's reference here is NXPs LPC1114FN28. They made a huge and crazy decision when they stuck an M0 in a 600 mil 28-pin DIP. But many of us absolutely love it. Wish I had NXP's sales info to see just how many they are selling. All I can do is guess, and my guess is that it was a profitable decision for them, and probably sowed the seeds for higher volumes in their standard packages.

    I'd love to see a cut-down P2 in a DIP. I'd willingly pay a premium over the cost of the ARM Cortex part.
  • jmgjmg Posts: 15,173
    edited 2014-05-22 16:02
    User Name wrote: »
    ... Wish I had NXP's sales info to see just how many they are selling. All I can do is guess, and my guess is that it was a profitable decision for them, and probably sowed the seeds for higher volumes in their standard packages.
    .

    Looking at the modest disti stock levels, I'd say NXP are running the Wide DIP28 as a loss-leader, for what you rightly call higher volumes in their standard packages.
    - so that means Parallax have to support more than one package, which is costly.

    I prefer the Infineon approach, of a mass volume compact package and a low cost DIP module.
    - that $6.75 is a good price for Debug included Module.

    My only beef with that KIT_XMC_2GO_XMC1100, is they seem to have limited the XMC1100 to 3v3 purely because the Debug device is 3v3.
    Shame, but not a project killer.
  • Bill HenningBill Henning Posts: 6,445
    edited 2014-05-22 16:46
    jmg:

    I think 300mil DIP28 for education/hobbyists + SOIC28 for volume would be a good mix, it would also make a great smart peripheral for chips like the P16X64

    Thanks for the link, however that is a $0.29 eight pin SOIC device, with 2K sram, 16K flash, 5 I/O's... not comparable to what I am proposing. The DIP28 LPC1114 is closer. I'll look into the Infineon module.

    Having said that, I'll have to get some of those chips to play with!

    User Name

    Correct! I have a bunch of LPC1114's, including the DIP28 part. I really, really wish it was a 300mil package.
  • User NameUser Name Posts: 1,451
    edited 2014-05-22 19:20
    Bill Henning wrote: »
    I really, really wish it was a 300mil package.

    I had the same reaction initially. And we aren't alone. With time I got used to it. Now I sort of like the package. But if I manufactured boards like you do, I doubt I'd be so content.
  • LeonLeon Posts: 7,620
    edited 2014-05-23 01:19
    jmg wrote: »
    .. and for a more capable small uC module example, look for
    KIT_XMC_2GO_XMC1100_V1

    - Mouser list this at 1+ $6.75, with the Debug-USB link included on the DIP adapter board

    The XMC1100 start from 100 $0.95

    Here is the link to Infineon's web page:

    http://www.infineon.com/cms/en/product/evaluation-boards/KIT_XMC_2GO_XMC1100_V1/productType.html?productType=db3a304443537c4e01436ccecb5d154f

    There is also this board:

    http://www.infineon.com/cms/de/product/evaluation-boards/KIT_XMC11_BOOT_001/productType.html?productType=db3a30443b360d0e013b8f5163c46f62

    Which can use Arduino add-on boards.
  • Cluso99Cluso99 Posts: 18,069
    edited 2014-05-23 02:03
    The problem for Parallax is the cost for lowish volumes. They could quite happily do a P1 in a DIP28 0.6" without P8-15. Even if Parallax made these available with less margin because it's only a package cost (no R&D) there is no way you will see anywhere near those M0 prices unless you want to order 1M pieces in one go, and even then it will not be as cheap.
  • Bill HenningBill Henning Posts: 6,445
    edited 2014-05-23 09:30
    Mostly agreed.

    This suggestion was based on the following assumptions:

    - P16X64 will work great, and will have better than P8X32 sales
    - Parallax will want to enter new markets - specifically low cost high volume ones
    - Parallax wants to introduce a broader product line

    Of course if those assumptions are false, this suggestion has less merit.

    Additionally,

    - This would make a GREAT I/O expander for the P16X64!
    Cluso99 wrote: »
    The problem for Parallax is the cost for lowish volumes. They could quite happily do a P1 in a DIP28 0.6" without P8-15. Even if Parallax made these available with less margin because it's only a package cost (no R&D) there is no way you will see anywhere near those M0 prices unless you want to order 1M pieces in one go, and even then it will not be as cheap.
  • jmgjmg Posts: 15,173
    edited 2014-05-23 14:40
    Bill Henning wrote: »
    - Parallax will want to enter new markets - specifically low cost high volume ones

    Errm "low cost high volume" is not going to be DIP, which is why DIP is the wrong primary target.

    For that high-volume target, a key element will be small flash
    - I see Winbond are adding a USON 2 x 3mm package, which is much smaller than SO8, and could be more tolerable.

    With Flash such an important pairing, the new P-tiny part really should support XIP, and that can push down the RAM needed to what makes economic sense.
    ie smarter use of FLASH, can lower the offer price and PCB area.
Sign In or Register to comment.