Hi all,
For those who haven't seen, Parallella got funded on KickStarter - a total of just under $900k raised. For those who would like to hear more, I'll be talking about the project at OSHUG on 15th November.
As part of my reward package, I'll receive 8 Epiphany III (16-core) bare chips in December. I have neither the skills or facilities to use these, so I am open to any OSHUG suggestions for putting them to a good use. I'm sure some of you have both the skills and facilities to do something creative with them.
Best wishes,
Jeremy
Hi Jeremy,
On 29 October 2012 15:16, Jeremy Bennett jeremy.bennett@embecosm.com wrote:
Hi all,
For those who haven't seen, Parallella got funded on KickStarter - a total of just under $900k raised. For those who would like to hear more, I'll be talking about the project at OSHUG on 15th November.
As part of my reward package, I'll receive 8 Epiphany III (16-core) bare chips in December. I have neither the skills or facilities to use these, so I am open to any OSHUG suggestions for putting them to a good use. I'm sure some of you have both the skills and facilities to do something creative with them.
The Adapteva website says these are "324-ball 15x15mm flip-chip BGA" [1] but doesn't mention the pitch? Perhaps it's implicit — my knowledge of BGA devices goes as far as appreciating that they are difficult for hobbyists to work with. In any case, if it were 1mm or higher it seems you could get a small four layer board done fairly cheaply [2], and perhaps this could just do breakout and be used to hook one up to an FPGA dev board. However, from a quick calculation it looks like the pitch has to be 0.8mm or under.
Wondering if Omer, Sukkin, or Al have any ideas, as they've far more experience with this sort of thing ...
As a last resort we could try the "quit whining and solder" technique :o)
http://eds.dyndns.org/~ircjunk/images/bga-adapt2.jpg
Best,
Andrew
[1] http://www.adapteva.com/products/silicon-devices/e16g301/ [2] http://siliconexposed.blogspot.de/2012/07/bga-process-notes.html
With these classes of chips one normally has to use machine placement, as hand placement can't easily be relied upon to be accurate enough. One normally outsources this rather than trying to make the prototypes at home, after placement most are x-rayed in order to check bonding and proper alignment as a pure visual inspection is not good enough. In terms of pads/balls 0.8mm pitch was common on BGA but it varies and newer flip chips can go down to 0.4mm pitch, but without datasheets and land patterns its difficult to make any concrete assumptions apart from the obvious machine placement recommended. The packages are not what I would coin as home assembly friendly, but that's to be expected with the number of pins we are talking about here.
regards Al
On Tue, Oct 30, 2012 at 8:25 AM, Andrew Back arback@computer.org wrote:
Hi Jeremy,
On 29 October 2012 15:16, Jeremy Bennett jeremy.bennett@embecosm.com wrote:>> Hi all,
For those who haven't seen, Parallella got funded on KickStarter - a total of just under $900k raised. For those who would like to hear more, I'll be talking about the project at OSHUG on 15th November.
As part of my reward package, I'll receive 8 Epiphany III (16-core) bare chips in December. I have neither the skills or facilities to use these, so I am open to any OSHUG suggestions for putting them to a good use. I'm sure some of you have both the skills and facilities to do something creative with them.
The Adapteva website says these are "324-ball 15x15mm flip-chip BGA" [1] but doesn't mention the pitch? Perhaps it's implicit — my knowledge of BGA devices goes as far as appreciating that they are difficult for hobbyists to work with. In any case, if it were 1mm or higher it seems you could get a small four layer board done fairly cheaply [2], and perhaps this could just do breakout and be used to hook one up to an FPGA dev board. However, from a quick calculation it looks like the pitch has to be 0.8mm or under.
Wondering if Omer, Sukkin, or Al have any ideas, as they've far more experience with this sort of thing ...
As a last resort we could try the "quit whining and solder" technique :o)
http://eds.dyndns.org/~ircjunk/images/bga-adapt2.jpg
Best,
Andrew
[1] http://www.adapteva.com/products/silicon-devices/e16g301/ [2] http://siliconexposed.blogspot.de/2012/07/bga-process-notes.html
oshug mailing list oshug@oshug.org http://oshug.org/cgi-bin/mailman/listinfo/oshug
PS I am speaking to some folks to see if we can get some help building a prototype with these
regards Al
On Tue, Oct 30, 2012 at 9:11 AM, Alan Wood folknology@gmail.com wrote:
With these classes of chips one normally has to use machine placement, as hand placement can't easily be relied upon to be accurate enough. One normally outsources this rather than trying to make the prototypes at home, after placement most are x-rayed in order to check bonding and proper alignment as a pure visual inspection is not good enough. In terms of pads/balls 0.8mm pitch was common on BGA but it varies and newer flip chips can go down to 0.4mm pitch, but without datasheets and land patterns its difficult to make any concrete assumptions apart from the obvious machine placement recommended. The packages are not what I would coin as home assembly friendly, but that's to be expected with the number of pins we are talking about here.
regards Al
On Tue, Oct 30, 2012 at 8:25 AM, Andrew Back arback@computer.org wrote:
Hi Jeremy,
On 29 October 2012 15:16, Jeremy Bennett jeremy.bennett@embecosm.com wrote:>> Hi all,
For those who haven't seen, Parallella got funded on KickStarter - a total of just under $900k raised. For those who would like to hear more, I'll be talking about the project at OSHUG on 15th November.
As part of my reward package, I'll receive 8 Epiphany III (16-core) bare chips in December. I have neither the skills or facilities to use these, so I am open to any OSHUG suggestions for putting them to a good use. I'm sure some of you have both the skills and facilities to do something creative with them.
The Adapteva website says these are "324-ball 15x15mm flip-chip BGA" [1] but doesn't mention the pitch? Perhaps it's implicit — my knowledge of BGA devices goes as far as appreciating that they are difficult for hobbyists to work with. In any case, if it were 1mm or higher it seems you could get a small four layer board done fairly cheaply [2], and perhaps this could just do breakout and be used to hook one up to an FPGA dev board. However, from a quick calculation it looks like the pitch has to be 0.8mm or under.
Wondering if Omer, Sukkin, or Al have any ideas, as they've far more experience with this sort of thing ...
As a last resort we could try the "quit whining and solder" technique :o)
http://eds.dyndns.org/~ircjunk/images/bga-adapt2.jpg
Best,
Andrew
[1] http://www.adapteva.com/products/silicon-devices/e16g301/ [2] http://siliconexposed.blogspot.de/2012/07/bga-process-notes.html
oshug mailing list oshug@oshug.org http://oshug.org/cgi-bin/mailman/listinfo/oshug
On 30/10/2012 08:25, Andrew Back wrote:
The Adapteva website says these are "324-ball 15x15mm flip-chip BGA" [1] but doesn't mention the pitch? Perhaps it's implicit — my knowledge of BGA devices goes as far as appreciating that they are difficult for hobbyists to work with.
I wouldn't be put off this. There are many back-street workshops that routinely remove, re-ball and resolder BGAs on laptop motherboards, without damaging any of the adjacent devices. Worst case, you could employ one of these. PCB manufacturing houses often do assembly, the board house I use would cost you about £20 each to put 8 devices down professionally. I have a SMD preheater and hot air rig which I do QFN and small BGA in the "shack" at the bottom of the garden, so I follow the "get on an solder it" brigade.
As for testing, I would stick with a JTAG boundary scan (what it was designed for) rathet than x-raying as x-rays can damage certain technologies.
HTH, Jason
Hi Jason,
On (30/10/2012 14:53), Jason Flynn G7OCD wrote:
On 30/10/2012 08:25, Andrew Back wrote:
The Adapteva website says these are "324-ball 15x15mm flip-chip BGA" [1] but doesn't mention the pitch? Perhaps it's implicit — my knowledge of BGA devices goes as far as appreciating that they are difficult for hobbyists to work with.
I wouldn't be put off this. There are many back-street workshops that routinely remove, re-ball and resolder BGAs on laptop motherboards, without damaging any of the adjacent devices. Worst case, you could employ one of these.
You can also watch people on Youtube fixing their Xbox problems by 'reflowing' chips using tea candles, that seems to work too :)
As you've mentioned, these are 'back-street' methods and I don't know about you but worrying about whether my BGA has been reflowed properly or not is the last thing I want while bringing up a complex design.
That said, I could get access to our SMT lab at the University which has the necessary equipment (sans a paste station but we can improvise) to deal with this if anyone wanted to go ahead with this. We could have another OSHUG trip to Canterbury for some SMT fun and beer!
PCB manufacturing houses often do assembly, the board house I use would cost you about £20 each to put 8 devices down professionally.
That is a cracking deal, would you mind sharing the details of that board house, off the list if you'd prefer?
As for testing, I would stick with a JTAG boundary scan (what it was designed for) rathet than x-raying as x-rays can damage certain technologies.
Good point.
Cheers, Omer.
On Wed, 2012-10-31 at 15:45 +0000, Omer Kilic wrote:
Hi Jason,
On (30/10/2012 14:53), Jason Flynn G7OCD wrote:
On 30/10/2012 08:25, Andrew Back wrote:
The Adapteva website says these are "324-ball 15x15mm flip-chip BGA" [1] but doesn't mention the pitch? Perhaps it's implicit — my knowledge of BGA devices goes as far as appreciating that they are difficult for hobbyists to work with.
I wouldn't be put off this. There are many back-street workshops that routinely remove, re-ball and resolder BGAs on laptop motherboards, without damaging any of the adjacent devices. Worst case, you could employ one of these.
You can also watch people on Youtube fixing their Xbox problems by 'reflowing' chips using tea candles, that seems to work too :)
As you've mentioned, these are 'back-street' methods and I don't know about you but worrying about whether my BGA has been reflowed properly or not is the last thing I want while bringing up a complex design.
That said, I could get access to our SMT lab at the University which has the necessary equipment (sans a paste station but we can improvise) to deal with this if anyone wanted to go ahead with this. We could have another OSHUG trip to Canterbury for some SMT fun and beer!
Hi Omer,
That certainly opens up some possibilities. OSHUG in Canterbury sounds fun - I missed the last one.
I spoke to Adapteva yesterday, and these are 0.8mm BGA chips. All the design stuff should become available in coming weeks. They intend doing the board design fully in the open.
Best wishes,
Jeremy
On 31/10/2012 15:45, Omer Kilic wrote:
That is a cracking deal, would you mind sharing the details of that board house, off the list if you'd prefer?
Sure. I've previously used PCBTrain. http://www.pcbtrain.co.uk/quote.php?set=assembly The online calculator shows £126 ex VAT & P&P for 8 assemblies with 1 BGA on each on 15 day turnaround.
I'm not affiliated with PCBTrain but they've been reasonably reliable for me (although I did have a couple of unpopulated PCBs with microscopic short circuit that I needed to sever).
I'm sure there are others out there which would do the job too but I've not seen many assembly houses that state anything other then "contact us for a quote" so please look around too.
Hi Jason
I am not sure that you are using their online assembly quoting tool in the way they envisage, it is designed for complete board assembly and they make their money by the number of components, all of which have to be supplied in PnP ready packaging except Rs and Cs which they give you for free from their own reels (its cheaper for them to do so) because the assembly charge for each is a magnitude more than their value!
Their terms basically indicate they reserve the right to charge what they like if the assembly requirements are unusual, so you might want to get that quote in writing ;-)
In my experience these sort of individual difficult package placements (as opposed to complete board assemblies) are more expensive, cheapest I have seen for this sort of placement in the past is about £40 + vat per chip.
"There are many back-street workshops that routinely remove, re-ball and resolder BGAs on laptop motherboards, without damaging any of the adjacent devices. Worst case, you could employ one of these."
This sounds like 'shenzhen' talk to me, certainly in the U.K. these sorts of workshops are not prevalent, most that operate on these sorts of chips charge for their skills at market value (unless you can pull in some favours) . I also would recommend with give these chips the respect they deserve when it comes to both design and assembly as they are as common as 'rocking horse sh*t'.
Omer the Canterbury ideas an excellent one.
The more important question however is what sort of boards do we actually want to make with these chips, its a far more interesting conversation?
regards Al
As for
On Wed, Oct 31, 2012 at 6:12 PM, Jason Flynn G7OCD flynnjs@yahoo.com wrote:
On 31/10/2012 15:45, Omer Kilic wrote:
That is a cracking deal, would you mind sharing the details of that board house, off the list if you'd prefer?
Sure. I've previously used PCBTrain. http://www.pcbtrain.co.uk/quote.php?set=assembly The online calculator shows £126 ex VAT & P&P for 8 assemblies with 1 BGA on each on 15 day turnaround.
I'm not affiliated with PCBTrain but they've been reasonably reliable for me (although I did have a couple of unpopulated PCBs with microscopic short circuit that I needed to sever).
I'm sure there are others out there which would do the job too but I've not seen many assembly houses that state anything other then "contact us for a quote" so please look around too.
oshug mailing list oshug@oshug.org http://oshug.org/cgi-bin/mailman/listinfo/oshug
-
Alan Wood -
Andrew Back -
Jason Flynn G7OCD -
Jeremy Bennett -
Omer Kilic