Shop OBEX P1 Docs P2 Docs Learn Events
Square IC sockets? — Parallax Forums

Square IC sockets?

sumdawgysumdawgy Posts: 167
edited 2014-10-15 07:40 in General Discussion
I am looking at [URL=" http://www.ti.com/product/tca6418e/samplebuy"]TCA6418E- I2C Controlled 18 Channel GPIO Expander[/URL]

But It would be best if I could find the (or a) socket it seats in. anyone know where I should look? So far, I'm having no joy.

the pdf for the pin layout is here (also hyperlinked on the page).
«1

Comments

  • Heater.Heater. Posts: 21,230
    edited 2014-09-23 08:27
    That chip is in a ball grid array (BGA) package. It is very, very small 1.6mm from corner pin to corner pin! I very much doubt you will find a socket for such a thing. If you are lucky you will find a break out board with one on it from Sparkfun, Adafruit or somewhere.

    I'm sure others here can suggest more hobbyist friendly port expanders. Even just good old shift registers in DIP packages.
  • sumdawgysumdawgy Posts: 167
    edited 2014-09-23 08:28
    Here's a image of the pins and for the curious I didn't crop out the pinout sheet.

    Untitled.jpg
    449 x 431 - 63K
  • sumdawgysumdawgy Posts: 167
    edited 2014-09-23 08:29
    Heater. wrote: »
    I'm sure others here can suggest more hobbyist friendly port expanders. Even just good old shift registers in DIP packages.

    yea I still have 2-3 of those from an earlier project altho I wanna go i2c for the interrupt feature.
  • SRLMSRLM Posts: 5,045
    edited 2014-09-23 08:30
    I don't think you'll find a socket for that. Your best bet will be to make a breakout board and try soldering it. You'll probably want to get 10+ with an expected yield somewhere around 25%.
  • sumdawgysumdawgy Posts: 167
    edited 2014-09-23 08:33
    Heater. wrote: »
    That chip is in a ball grid array (BGA) package. It is very, very small 1.6mm from corner pin to corner pin!

    yea kinda why I'm hoping to find a socket out there. I was thinking of making one with a dremel and tiny springs.
    Just thinking about it makes my eyse cross. (oh god, I've gone dyslexic)
  • Heater.Heater. Posts: 21,230
    edited 2014-09-23 08:39
    Give up the hope. The only way you are going to use that chip if is if someone has made a break out board for it. Or if you design your pwn break out board and get it made at OSHPark or some other board house.

    As SRLM says you may find soldering it a bit hard and waste a bunch of chips and boards whilst trying.

    Even if you succeed the end result is a break out board that will be as big as many of the other port expander DIP package solutions.
  • sumdawgysumdawgy Posts: 167
    edited 2014-09-23 08:40
    still searching a bit i did stumble across this daughter board. http://www.datasheetarchive.com/dl/Datasheets-SL7/DSASL00129002.pdf
  • Heater.Heater. Posts: 21,230
    edited 2014-09-23 08:41
    Which has got what to do with the chip you are discussing?
  • sumdawgysumdawgy Posts: 167
    edited 2014-09-23 08:49
    Heater. wrote: »
    Which has got what to do with the chip you are discussing?

    seems to have the adressing ability i was looking for and MIGHT make a replacement,

    HOWEVER, more on topic I found this ..http://www.emulation.com/catalog/sockets/development/compression.cfm

    which leads me to a more elegant solution....
    setup pcb with pads in proper positions

    I could protoype such a pcb then print it and etch it.
    mount a "picture frame" to hold the chip in position followed by a top wich applies the compession to make a stable connection.
  • Duane DegnDuane Degn Posts: 10,588
    edited 2014-09-23 08:57
    Heater. wrote: »
    I'm sure others here can suggest more hobbyist friendly port expanders. Even just good old shift registers in DIP packages.

    I agree with Heater on this. '595 for increasing outputs and '165 for increasing inputs. Communicating (and soldering/socketing) these chips is as easy as it gets.
  • Heater.Heater. Posts: 21,230
    edited 2014-09-23 09:04
    If you get to the point that you have made a PCB for this chip you might as well continue and solder the sucker down to it. As noted before sometimes this will fail. Never mind.

    I suggest you get a free sample. When you see how small it is you will soon give up the idea:)

    You may even need a multi-layer PCB to allow routing of all the connections to those pins.
  • tritoniumtritonium Posts: 543
    edited 2014-09-23 09:37
    Hi
    The 23017 gives 16 I/O and has interrupts, comes in SPI and I2C varieties, in a 28 pin dip package and is as cheap as err... chips
    Dave
  • PublisonPublison Posts: 12,366
    edited 2014-09-23 09:37
    Heater. wrote: »

    You may even need a multi-layer PCB to allow routing of all the connections to those pins.

    Agreed. If you made a board or found a break out board, you will need an oven to bake it. You can't solder BGA by hand.

    Use the DIP versions or SOIC versions.
  • Mark_TMark_T Posts: 1,981
    edited 2014-09-23 09:41
    You normally need 4/6 layer board with microvias for BGA chips. For soldering I am told it is possible
    with a hotplate and USB microscope and flux pen. The microscope to check the alignment before heating,
    for which fiducial marks on diagonal corners (in copper, not screen-printed!) are invaluable. Positional
    accuracy is 100 microns or less perhaps, hence the microscope (a hand lens is subject to parallax error)

    I'm tempted to say "that way madness lies". Still if you manage it then ordinary SMT is a doddle! I personally
    find DFN/QFN quite stressful enough.
  • TubularTubular Posts: 4,702
    edited 2014-09-23 14:27
    I think there are much easier solutions, but...

    If you don't need the central 8 GPIO, you could perhaps treat it like a QFN, have elongated pads that extend well beyond the perimeter of the chip, and use hot air to see what happens.

    It'd be a ~0.4mm pitch, good practice for Prop 2. You'd want to try and connect that inner ground pad though, if possible.
    396 x 391 - 4K
  • YanomaniYanomani Posts: 1,524
    edited 2014-09-23 17:20
    Aside from the many concerns of having to solder those tiny things in place, I've found a couple of links that seem to be very interesting to you all.

    This company does the socket.
    In fact, if they didn't have it, they'll make it for you.
    BUT (here comes the devil's share, as usual), they are truly very expensive and you still has to deal with the problem of doing a breakout board to solder them.

    There is another possibility:

    This company has a solution very close to your needs, although at a .5 mm pitch (look at), BUT, they offer to help you finding whatever you need.
    I believe that's possible that they even agree to make it, according to your own specs.
    They also have an assembly service, to solder the parts on the boards, for you.
    Sure they don't sell peanuts, but you can get what you want.

    Hope it helps a bit.

    Yanomani
  • JonnyMacJonnyMac Posts: 9,105
    edited 2014-09-23 19:06
    I'm sure others here can suggest more hobbyist friendly port expanders.

    The MCP23017 is very easy to use. It's a much bigger chip (physically), though.
  • sumdawgysumdawgy Posts: 167
    edited 2014-09-23 20:06
    Heater. wrote: »
    You may even need a multi-layer PCB to allow routing of all the connections to those pins.

    yea that bothered me too.
  • sumdawgysumdawgy Posts: 167
    edited 2014-09-23 20:10
    Mark_T wrote: »
    I'm tempted to say "that way madness lies". Still if you manage it then ordinary SMT is a doddle! I personally
    find DFN/QFN quite stressful enough.

    well I'm thinking that i would need some 40 puts per floor alone....a lot of 8 bits would be needed.

    I dunno I'm hoping to find a fairly reliable hack.

    BUT if i must at least it's gonna be i2c (but with a lot of addresses.)
  • sumdawgysumdawgy Posts: 167
    edited 2014-09-23 20:13
    Tubular wrote: »
    I think there are much easier solutions, but...

    If you don't need the central 8 GPIO, you could perhaps treat it like a QFN, have elongated pads that extend well beyond the perimeter of the chip, and use hot air to see what happens.
    . You'd want to try and connect that inner ground pad though, if possible.

    elegant, even if it does blow off some pins...
    Well it's late just got off work gonna sleep on it.
  • sumdawgysumdawgy Posts: 167
    edited 2014-09-23 20:16
    JonnyMac wrote: »
    The MCP23017 is very easy to use. It's a much bigger chip (physically), though.

    the size kinda makes it easier to work with... and I DO like that it's 16bit

    THANK YOU...I have some thoughts to try in the AM...... if they don't pan out...then I WILL go this way.:thumb:
  • sumdawgysumdawgy Posts: 167
    edited 2014-09-24 10:20
    Heater. wrote: »
    That chip is in a ball grid array (BGA) package. It is very, very small 1.6mm from corner pin to corner pin! I very much doubt you will find a socket for such a thing. If you are lucky you will find a break out board with one on it from Sparkfun, Adafruit or somewhere.

    Well I DID find a source. Test socket factory

    Aries Electronics is a leading manufacturer of a broad range of interconnect products and Correct-A-Chip® technology.

    reading thier site.....Aries Has Mastered These Manufacturing Processes... (blah blah blah...WAIT..) "Assembly of Solder Balls on Ceramic PC Boards & Connectors"

    They also have quite a few more common ones on hand. Of course the rep for my location was on lunch during my call but I DID find that the thier 6.5mm max bga was configurable to my spec. The advantage here is mounting. Thier disclaimer is heat deforming the solder balls and causing failure. (fortunately for my purposes, there's no heat involved & next to no current.)

    So untill I speak to ...(Well,I got his Email) my rep. . . I can't say if the price is in the hobby range. If they've not built one yet...gonna have to pay prototyping fees.

    I am going to look around for a few other places but I imagine the prices will be similar unless I luckilly find one that already has some xbga-n25 sockets sitting on a shelf.

    Untitled.jpg
    901 x 336 - 113K
  • Heater.Heater. Posts: 21,230
    edited 2014-09-24 13:45
    You are looking at spending 10 or a 100 or a 1000 times more than the cost of the chip in order to be able to use the chip.

    Give it up and go for one of the "normal" solutions suggested in the posts above.

    Unless, that is, you are evaluating the chip to build the next gizzmo that you are going to sell in the hundreds of thousands or millions.
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2014-09-24 14:13
    sumdawgy wrote: »

    I suppose you have all the time in the world with your free chip samples but your $300 socket will still need mounting too. Do you know how illogical it is to pursue this course?

    Test sockets are always $$$+ but as Heater mentioned that's not a problem when you are building squillions. What was that you say? At a hobby price. ROTFL
  • Heater.Heater. Posts: 21,230
    edited 2014-09-24 14:25
    Once you get passed the attraction of the free samples it seems you can actually buy these chips from Digikey in 1 off quantities at $2.49 each.

    For sure a lot of other port expanders can be had for that price which are actually usable.
  • TubularTubular Posts: 4,702
    edited 2014-09-24 14:40
    What we really need is to create the 'Pyramid of Pogo', with an array of pogo pins starting at 0.1x0.1" pitch but angled inwards towards 0.3x0.3mm pitch.
  • sumdawgysumdawgy Posts: 167
    edited 2014-09-24 17:08
    Do you know how illogical it is to pursue this course?

    Test sockets are always $$$+ but as Heater mentioned that's not a problem when you are building squillions. What was that you say? At a hobby price. ROTFL

    Well I wish someone had at least pointed me at test sockets in the first place ; ]

    yea it may be a bit dreamy but until reality slaps you down... it might still be feasible.
    But I agree a $300 price would be right out of hobby price.
    But it's still a great dream.. the size makes a 60 output board smaller.
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2014-09-24 17:27
    sumdawgy wrote: »
    Well I wish someone had at least pointed me at test sockets in the first place ; ]

    yea it may be a bit dreamy but until reality slaps you down... it might still be feasible.
    But I agree a $300 price would be right out of hobby price.
    But it's still a great dream.. the size makes a 60 output board smaller.

    I think if anyone suggested "test sockets" they might have been slapped down, that's the reality. The only reason for these BGA packages is for automated volume production as they also require fine pitch multi-layer boards otherwise how do you route them? I avoid them as I like to stick to double-sided PCBs in small volumes (100s, 1000s).

    But how would a big bulky socket make the board smaller??? Even the DIP version of the chip would be a lot smaller than the socket. All those pogo pins and ZIF mechanism etc takes a lot of space, these things were never designed to be dedicated to just one per chip. They aren't even really suitable for burn-in evaluation of a chip as you don't have the circuit characteristics that you would if it were mounted in tight on the board as it is designed to.

    These sockets are fine for "testing" which really means either pre-programming or configuration, but the handling of each chip introduces both much more expense and also can increase the failure rate. I can buy a MEMS oscillator chip for under a buck but if it's handled by the distributor to program it my special requirements it costs three bucks. BTW, they are only tiny little 4 pin DFNs, not BGAs.
  • sumdawgysumdawgy Posts: 167
    edited 2014-09-24 19:02
    I think if anyone suggested "test sockets" they might have been slapped down, that's the reality. The only reason for these BGA packages is for automated volume production as they also require fine pitch multi-layer boards otherwise how do you route them? I avoid them as I like to stick to double-sided PCBs in small volumes (100s, 1000s).

    But how would a big bulky socket make the board smaller???.

    lol at the slap down
    & yea I want to dream But even in that test socket ..if you note the picture....a few posts up....the test socket is still a bit smaller than a US quarter coin.

    The prop's footprint is small ..but my accessory board(s) will dwarf it. (This is even before I add relays.)
    I suppose I could mount it in an old ATX case.
  • Peter JakackiPeter Jakacki Posts: 10,193
    edited 2014-09-24 19:38
    sumdawgy wrote: »
    lol at the slap down
    & yea I want to dream But even in that test socket ..if you note the picture....a few posts up....the test socket is still a bit smaller than a US quarter coin.

    The prop's footprint is small ..but my accessory board(s) will dwarf it. (This is even before I add relays.)
    I suppose I could mount it in an old ATX case.

    If that is the same socket then it's still huge, 3D wise. Much bigger than any easy to solder DIP, QFP or even QFN etc.

    Now if you want to dream in that fruitless and pointless manner, go to bed. If you want to dream up stuff that really makes a difference, then count the costs, weigh the pros and cons, and get into it, then tell us after, we may perhaps even applaud you. But never ever consider going to such great trouble and expense using your freebies just because they are freebies, as if that were some noble quest, it just makes you look very silly, and if you continue to be obsessed with "tilting at windmills" then don't be surprised when reality slaps you in the face and knocks you to the ground!

    As a "Sancho", I will just look down and mutter something about " no one could have made any mistake about it but one who had something of the same kind in his head." :)
Sign In or Register to comment.