Propeller 2 Package Information ( PCB Footprint )

TubularTubular Posts: 3,836
edited 2013-03-13 - 22:58:48 in Propeller 2
Now that the design has been sent to the fab, is it possible to get a recommended PCB footprint, so we can make some PCBs to suit ?

thanks
Lachlan
«1

Comments

  • cgraceycgracey Posts: 12,179
    edited 2013-02-07 - 08:58:13
    Tubular wrote: »
    Now that the design has been sent to the fab, is it possible to get a recommended PCB footprint, so we can make some PCBs to suit ?

    thanks
    Lachlan

    We don't have an exact package yet, though I don't know if there's anything needed beyond "14mm-square-body TQFP-128".

    Here are some drawings by Practical Components of this type of package

    http://www.practicalcomponents.com/drawings/pdf/TQFP/A-TQFP128-14x14-1dwg.pdf
    http://www.practicalcomponents.com/drawings/pdf/TQFP/A-TQFP128-14x14-2dwg.pdf
    http://www.practicalcomponents.com/drawings/pdf/TQFP/A-TQFP128-14x14-3dwg.pdf
  • jmgjmg Posts: 14,176
    edited 2013-02-07 - 17:09:53
    The important number is the 0.4mm pin pitch.

    Have you checked if the P2 die will also bond in a MQFP80, 14x20mm, that has a lead pitch of 0.8mm, and may give a softer transition point ?
  • Wilt VineWilt Vine Posts: 6
    edited 2013-02-07 - 17:38:18
    What's a transition point?
  • TubularTubular Posts: 3,836
    edited 2013-02-07 - 17:50:54
    Chip, thanks for the drawing.

    Jmg, I think that's a very good idea if it could be done. I wonder if it will fit.

    Wilt, Jmg is referring to the way the existing Prop QFP package has a 0.8mm pitch, and many of us are used to hand or reflow soldering those. If the Prop2 could also have the same pin pitch (in that particular rectangular case), it eases the adoption process.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2013-02-09 - 23:19:50
    Schmartboard have existing boards that provide a solution for 0.4mm pitch. If that pitch is to become a real package, it may be best to consider the kind of packages that they have had the best success with.

    It may seem odd, but NOT having the leads stick out, may be better. Schmwartboards rely on the solder to flow under the package edge and connect with a natural viscosity of the liquid solder.

    Exposed leads at 0.4mm may just be a rather huge shipping and handling headache, as well as a construction challenge. In other words, consider a different approach to smaller pins. Retrain your soldering technique rather than try to hold back the future.
  • FredBlaisFredBlais Posts: 368
    edited 2013-02-10 - 06:10:51
    I don't think 0.4mm pitch is going to be a problem. I soldered a lot of these with a WESD51 station and a fat soldering tip. Consider drag soldering, and be sure to have some desoldering braid to remove some of the solder bridges that are going to show up.

    http://www.youtube.com/watch?v=wUyetZ5RtPs
  • pedwardpedward Posts: 1,566
    edited 2013-02-10 - 15:13:41
    I know .4mm has become the new standard, but the problem I see with it is reworkability. I've successfully reworked a .4mm connector, but what I've found is that even with 30GA jumper wire the wires are so close that it makes it really difficult to solder them. I've got a 20x stereo microscope for this, and .4mm is really hard.

    I really recommend .5mm as the tightest pitch. I would recommend using the QFN package for the .4mm pitch, maybe even BGA, and use .5mm for QFP.

    That's my 2 cents, having reworked .4mm, .5mm, .65mm, and .8mm.
  • Wilt VineWilt Vine Posts: 6
    edited 2013-02-10 - 19:38:06
    .4mm is of course the only sensible choice. These chips were supposed to be targeted for high volume commercial applications, no? Modern pick and place doesn't care about reworking. BGA is needed too.
  • pedwardpedward Posts: 1,566
    edited 2013-02-10 - 19:56:50
    Wilt Vine wrote: »
    .4mm is of course the only sensible choice. These chips were supposed to be targeted for high volume commercial applications, no? Modern pick and place doesn't care about reworking. BGA is needed too.

    Romney, is that you? Your reality distortion field is turned on again!

    EDIT: Now it's Crazy Busey! What's next, Nutz Nolte?
  • BigFootBigFoot Posts: 259
    edited 2013-02-12 - 19:13:21
    .4 mm lead spacing is not a problem but a stereo microscope really helps, .015 solder and
    a digitally controlled static controlled soldering iron with a .015 pointed tip is mandatory.

    It is also a good idea not to drink any coffee before soldering one of these chips down.

    Russ :)...
  • RaymanRayman Posts: 10,061
    edited 2013-02-14 - 07:57:48
    The package is the same as the SSD1963 that I'm using...

    Out of maybe 20 or so boards I've made so far, I think only one didn't require rework on this big chip...

    The worst part is that it's very difficult to see when pins are shorted together... I have to use a 10X loop to inspect all the pins...
  • AleAle Posts: 2,358
    edited 2013-02-16 - 09:43:32
    0.4 mm !!!! that is difficult to hand-solder :( 0.5 mm spacing would be much better.
  • SRLMSRLM Posts: 5,045
    edited 2013-02-16 - 09:55:23
    I have found that for anything smaller than 0.65 mm I need to use a stencil to make sure that the correct amount of paste is deposited (for reflow soldering).
  • Martin HodgeMartin Hodge Posts: 1,236
    edited 2013-02-16 - 11:13:08
    How about 0.4mm as the flagship product for commercial buyers & manufacturers. And a smaller run of larger pitch for hobbyists & prototyping.
  • pedwardpedward Posts: 1,566
    edited 2013-02-16 - 12:50:34
    After further research, the package Chip proposed above is a 14x14mm TQFP-100 with a .5mm pitch.

    That is the absolute minimum I'd recommend.

    FWIW, there was an article on HAD about an STM32F4, which is comparable in pin count. Funny thing is, it's 168Mhz and rated at 210 DMIPS (a contrived C based benchmark). The P2 will be the same package and 1280 MIPS :D

    One tactic that Parallax could take is to put the P2 die into a smaller pin count package, say, QFP-64 or PLCC-84, with reduced I/O, but easier for Hobbyists to handle.
  • TubularTubular Posts: 3,836
    edited 2013-02-16 - 14:27:28
    pedward wrote: »
    After further research, the package Chip proposed above is a 14x14mm TQFP-100 with a .5mm pitch.

    There are several pitches inside those drawing files, I think its the TQFP-128 with 0.4mm (else I have a board to revise!)
    pedward wrote: »
    One tactic that Parallax could take is to put the P2 die into a smaller pin count package, say, QFP-64 or PLCC-84, with reduced I/O, but easier for Hobbyists to handle.

    Jmg suggested (reply #3 above) the 14x20mm MQFP80 with a 0.8mm pitch, same as the current prop qfp, which makes some sense too. Prop2 certainly has plenty of I/O, many applications won't need it all.
  • pedwardpedward Posts: 1,566
    edited 2013-02-16 - 16:31:56
    Just did some arithmetic on power usage. If 1.5W is the max, then:

    1.17mW per MHz per COG. That means with all 8 COGs running, 9.375mW per MHz.

    It just so happens that the numbers are the same for MIPs, so it's 1.17mW per MIP or 651uA per MIP.

    5.2mA per MHz with all 8 COGs running, 651uA per MHz per COG.

    If using RCSLOW, and it's 20Khz like the Prop 1, that's 13uA per COG, 104uA for all COGs.

    If using RCFAST, and it's 12Mhz like the Prop 1, that's 7.8125mA per COG, 62.5mA for all COGs.

    Of course this is all speculative until they actually get silicon and can test it in the real world.
  • cgraceycgracey Posts: 12,179
    edited 2013-02-16 - 21:24:06
    Just so everyone knows, the Propeller 2 package is a 14mm x 14mm body 128-pin TQFP with 0.4mm pin pitch.
  • Martin HodgeMartin Hodge Posts: 1,236
    edited 2013-02-16 - 23:20:13
    That would be the innermost footprint in this image. (for comparison)
    800 x 461 - 37K
  • jazzedjazzed Posts: 11,803
    edited 2013-02-17 - 10:03:00
    I hate to say it, but 0.4mm won't work for me. 0.5mm pitch is fine.
    It's not about soldering skills. It's about PCB trace widths and clearances.

    A 0.4mm pitch device requires a trace and space design rule that is less than 6mil.
    This means I can't use cheap PCB services that need >= 7mil rules.

    With 0.5mm pitch there are no such problems.

    While I do not intend to make "developer" boards, I do have ideas for products, and 0.4mm spacing puts the current P2 plan off my radar.
  • Bill HenningBill Henning Posts: 6,445
    edited 2013-02-17 - 10:08:45
    I would also strongly prefer 0.5mm, however I get the impression that Parallax has already made a "cast in stone" decision to go 0.4mm

    I suspect that the 0.4mm package is less expensive.
  • TubularTubular Posts: 3,836
    edited 2013-02-17 - 12:08:06
    jazzed wrote: »
    I hate to say it, but 0.4mm won't work for me. 0.5mm pitch is fine.
    It's not about soldering skills. It's about PCB trace widths and clearances.

    A 0.4mm pitch device requires a trace and space design rule that is less than 6mil.
    This means I can't use cheap PCB services that need >= 7mil rules.

    Jazzed,

    The 0.4mm means you need 8 mil services, (0.2mm pad and 0.2mm spacing is ~8 mil each). So just about anyone offers this level.

    The board I just laid out uses 7 mil, which is also very widely available. There are just a couple of spots that need 6 mil, due to the Prop1 being included inside the Prop2 footprint temporarily (for testing; generally won't be required!), and they generally allow this

    But the point is you can do it with 8 mil track & space if you want.

    regards
    Lachlan
  • jazzedjazzed Posts: 11,803
    edited 2013-02-17 - 12:19:31
    @Lachlan,

    According to the body tests I did this morning, I get DRC errors between QFP pads with 0.4mm pin spacing. There are different parameters for pins and track/space, but I get consistency warnings on having different rules.
  • TubularTubular Posts: 3,836
    edited 2013-02-17 - 13:45:52
    Jazzed, you're technically correct. 0.2mm is 7.87 mil, so change the design rules to 7.8 rather than 8.0 mil and it should pass fine.

    The PCB fab houses seem to happily go a little bit below their minimum with no problems. I once had them phone up and complain at the 3 mil spacing level, which turned out to be an error in the ground pour rather than anything deliberate.

    Anyway my point is there are plenty of fabrication options, 8 mil (ok 7.8 mil) is easy to find, and you certainly don't need < 6 mil as you suggested. ExpressPCB's 3 for $51 service goes down to 6 mil.

    Just don't want people thinking it's impossible to do, thats all...

    cheers
    Lachlan
  • pedwardpedward Posts: 1,566
    edited 2013-02-17 - 15:40:29
    jazzed wrote: »
    @Lachlan,

    According to the body tests I did this morning, I get DRC errors between QFP pads with 0.4mm pin spacing. There are different parameters for pins and track/space, but I get consistency warnings on having different rules.

    Steve,

    OSH Park/Dorkbot PDX offers high quality boards for $5/inch in qty 3 prototype orders. Their minimum is 6 mil, and the quality is the absolute best you will get IMHO. I've used DipTrace and manually entered the specs for their boards, and it worked out great for stuff I've made.
  • jmgjmg Posts: 14,176
    edited 2013-02-17 - 20:18:51
    Tubular wrote: »
    Anyway my point is there are plenty of fabrication options, 8 mil (ok 7.8 mil) is easy to find, and you certainly don't need < 6 mil as you suggested. ExpressPCB's 3 for $51 service goes down to 6 mil.

    Yes, PCBs via virtually any PCB FAB should be simple with (nominal) 8 mil rules.

    I wonder if a special footprint is worth doing, ie with moved-fingers, tuned for manual assembly ?

    Such a footprint would have PADS extending well past the package-ends, and stopping early under the fingers, so that bridges close to the package are less likely, and the longer PAD finger gives a solder-thieving effect during cleanup ?
    PCB ends up a little larger than a mass-production design, but only slightly.
  • jmgjmg Posts: 14,176
    edited 2013-02-17 - 20:41:03
    Wilt Vine wrote: »
    I should hope that Parallax ignores the hobbyists and posers and produces this phantom chip in the smallest package possible. The die is already too big, and people here are asking for .8 mm? That's some serious retro there. I believe I have an Amiga from the late 80's with that kind of tech. Chips with pins. How quaint.

    The smallest package would be BGA, and that is not industrial,or moderate-volume, friendly. Fine for cellphones - but not the Prop 2 target market..

    So Gull-Wing wins, and a 0.4mm package is a good first choice there. (but it is not the smallest package possible).

    Vendors in Asia are providing new parts in a choice of pitch, so there is obvious customer demand for the better yields that a larger pitch device brings. Not everyone is rushing to 'finest possible'.

    Far from being retro, it is all about production yields, and sometimes on thickest-copper (high power) designs, a fine-pitch part can be a royal pain.
  • pedwardpedward Posts: 1,566
    edited 2013-02-18 - 01:16:16
    Wilt Vine wrote: »
    I should hope that Parallax ignores the hobbyists and posers and produces this phantom chip in the smallest package possible. The die is already too big, and people here are asking for .8 mm? That's some serious retro there. I believe I have an Amiga from the late 80's with that kind of tech. Chips with pins. How quaint.

    Dood, your trolling pseudo persona that you've registered is incredibly lame. I have no faith that Wilt Vine is your actual name, and none of your posts have been even the slightest bit helpful.

    In short, please go troll some other board, or return to your other Parallax forum account which will strip you of your anonymity and keep you in line.
  • BigFootBigFoot Posts: 259
    edited 2013-02-18 - 17:15:45
    .4 mm is not as bad as you might think, after you do a couple of them it gets easier. Rework is a problem though, I have
    found that you can cut the leads with a sharp exacto knife then remove the chip. Then you can flood the remaining stubs
    and lands with solder and carefully clean off the board with solder wick. The bottom line is to save the board, not the chip.

    When I have new prototype boards made, I ask for a thick solder coating, then I flood the lands with flux and solder down
    two pins to line up the chip. Then all you have to do is heat each land and the solder will flow to the chips lead.

    Our board house has a minimum of 7 mil traces and spacing and a minimum hole size of 15 mil's to qualify for the
    minimum standard cost. Be sure to make the power traces wider than usual for the P2, this will keep the voltage drop
    to a minimum.
  • pedwardpedward Posts: 1,566
    edited 2013-02-18 - 18:16:15
    FYI, the P2 has a bunch more power pins than the P1, each 8 pin group has dedicated power/ground pins. My understanding is that VIO is different than CIO, so you can have 3.3v IO and 1.8v core voltages.
Sign In or Register to comment.