Forums

Blackfin PLCC

Started by csb June 30, 2003
Al Clark wrote:
> > Reworking a BGA is not as easy as just removing and resoldering. The > contacts are solder balls. > > Our supplier will place a BGA for $65 ea after the board was been > profiled ($80). Reballing is $125 and this does not count replacing the > reballed part on the board. If you are dealing with a very expensive > part, you might want to reball the part, but usually we throw away the > part and replace with a new one.
$125 to reball a chip? There aren't many chips which even cost that much, so I guess this pricing is a low volume thing :-) Why does cleaning the board cost so much? You can do it quite easily with flux loaded woven copper matting.
> Prototype BGA sockets are very expensive as well (maybe $1K).
They aren't just expensive - they screw up the signal quality, too.
> In production, BGAs are not difficult. They tend to self align very well.
Yes and no. They suffer the same problems as other surface mount parts. If you glue spot them you have alignment difficulties. If you let them float free some get shaken completely out of place before they reach the reflow area. Someone told me years ago how glue spotting caused them alignment problems with QFPs. This seemed odd, until I realised he was talking about using cheap labour to hand glue spot and mount parts in a robot free zone. Why would anyone even try to glue spot, when assembling that way? Strange. It seems recently most of the cheapest labour places have finally gone over to full robot assembly. A lot of QFPs still go into toys by hand, though. It still amazes me to see the speed with which a skilled worker can mount those things.
> The biggest issue is PCB layout especially when the pitch is .8mm or > smaller. This makes vias very small (often microvias are needed which is > not something every pcb house can provide and they are definitely more > expensive).
It amazes me how much PCB construction has improved in the last few years. 1980's PCBs would definitely have been pretty useless for today's fine pitch parts.
> BGAs are certainly a reality and if you want to build DSP boards you > better get used to them. Of course, there are suppliers that would love > to save you the trouble ;-)
If you want to build practically anything above the simple MCU level you'd better get used to them. Regards, Steve
Steve Underwood <steveu@dis.org> wrote in
news:bduvmg$pm1$1@hfc.pacific.net.hk: 

> Al Clark wrote: >> >> Reworking a BGA is not as easy as just removing and resoldering. The >> contacts are solder balls. >> >> Our supplier will place a BGA for $65 ea after the board was been >> profiled ($80). Reballing is $125 and this does not count replacing >> the reballed part on the board. If you are dealing with a very >> expensive part, you might want to reball the part, but usually we >> throw away the part and replace with a new one. > > $125 to reball a chip? There aren't many chips which even cost that > much, so I guess this pricing is a low volume thing :-) > > Why does cleaning the board cost so much? You can do it quite easily > with flux loaded woven copper matting.
Where did cleaning come in? Reballing is a process that remakes the BGA contacts. Profiling determines how fast the board heats, etc. for correct reflow. You only do it once for a given assembly type. If you did 6 identical boards you wou profile one and place 6 (cost = $80 + 6*$65)
> >> Prototype BGA sockets are very expensive as well (maybe $1K). > > They aren't just expensive - they screw up the signal quality, too. > >> In production, BGAs are not difficult. They tend to self align very >> well. > > Yes and no. They suffer the same problems as other surface mount > parts. If you glue spot them you have alignment difficulties. If you > let them float free some get shaken completely out of place before > they reach the reflow area.
Unless you have parts on both sides of the board, parts are not glued. When I hand solder a QFP, I place a small amount of solder on a corner pad. I then heat this pad with the QFP pins absolutely perfectly aligned on the board. I then flip the board 180 degrees, and solder the opposite corner. The rest of the pins are then soldered in a kind of controlled bridging approach. FLUX IS YOUR FRIEND!!! My technician uses a x-acto knife to help slide the part, I use tweezers. In all cases, this is done under a microscope.
> > Someone told me years ago how glue spotting caused them alignment > problems with QFPs. This seemed odd, until I realised he was talking > about using cheap labour to hand glue spot and mount parts in a robot > free zone. Why would anyone even try to glue spot, when assembling > that way? Strange. It seems recently most of the cheapest labour > places have finally gone over to full robot assembly. A lot of QFPs > still go into toys by hand, though. It still amazes me to see the > speed with which a skilled worker can mount those things. > >> The biggest issue is PCB layout especially when the pitch is .8mm or >> smaller. This makes vias very small (often microvias are needed which >> is not something every pcb house can provide and they are definitely >> more expensive). > > It amazes me how much PCB construction has improved in the last few > years. 1980's PCBs would definitely have been pretty useless for > today's > fine pitch parts.
Its been a long time since I laid out pcbs on a light table with tape. I remember when plated thru holes were troublesome. Now even a "down & dirty" house can etch 5/5 multilayer boards (5 mil min trace, 5 mil clearance).
> >> BGAs are certainly a reality and if you want to build DSP boards you >> better get used to them. Of course, there are suppliers that would >> love to save you the trouble ;-) > > If you want to build practically anything above the simple MCU level > you'd better get used to them.
What's a DIP? This term seems vaguely familiar..... -- Al Clark Danville Signal Processing, Inc. -------------------------------------------------------------------- Purveyors of Fine DSP Hardware and other Cool Stuff Available at http://www.danvillesignal.com
On Thu, 03 Jul 2003 00:02:18 +0800, Steve Underwood <steveu@dis.org>
wrote:
> >It amazes me how much PCB construction has improved in the last few >years. 1980's PCBs would definitely have been pretty useless for today's > fine pitch parts.
No kidding. However, sometimes the reduction in part count from high-integration ICs can simplify things somewhat. Several years ago at a previous employer we had acquired a particular model of a competitor's product and had taken it apart and were playing around with it. Their rf hardware was a total joke and about 10x more expensive than it needed to be; somebody really, really paranoid about crosstalk had apparently contributed to the design. When we got to the digital boards, though, I was very impressed. There were only a very small number of big ICs and very few connections between them. Now this, I thought to myself, is how a board should be built. Later in a staff meeting management was asking about our evaluation of this design and I said, "The digital section could've been laid out with a crayon." A few people thought I was denegrating it but a few of our more cognizant executives understood that this was a significant compliment. Eric Jacobsen Minister of Algorithms, Intel Corp. My opinions may not be Intel's opinions. http://www.ericjacobsen.org
Al Clark wrote:
> Steve Underwood <steveu@dis.org> wrote in > news:bduvmg$pm1$1@hfc.pacific.net.hk: > >>Al Clark wrote: > > Where did cleaning come in? Reballing is a process that remakes the BGA > contacts. > > Profiling determines how fast the board heats, etc. for correct reflow. > You only do it once for a given assembly type. If you did 6 identical > boards you wou profile one and place 6 (cost = $80 + 6*$65)
OK, a confusion of terms. I understand what you mean by profiling now.
> Unless you have parts on both sides of the board, parts are not glued.
Are you refering to QFP or BGA? QFPs are routinely glued by the robots so the board is less susceptible to shaking as it passes through (maybe) several other robot and shakey conveyer belt steps. I wish people wouldn't use that concrete like red glue, though. It makes changing QFPs a real pain. Perhaps that is the intention!
> When I hand solder a QFP, I place a small amount of solder on a corner > pad. I then heat this pad with the QFP pins absolutely perfectly aligned > on the board. I then flip the board 180 degrees, and solder the opposite > corner. The rest of the pins are then soldered in a kind of controlled > bridging approach. FLUX IS YOUR FRIEND!!!
In production environments the girls (its usually girls) tend to use a soldering iron with a bit the whole width of one side of the QFP. In 4 goes (i.e. one on each side) they can put a QFP in place very accurately. Messing around with corner pins is just for people like us, who don't do this often enough [actually one in a while is often enough for me :-\ ]
> My technician uses a x-acto knife to help slide the part, I use tweezers. > In all cases, this is done under a microscope.
I can't image being able to solder with a microscope. I can't remember seeing someone try. Doesn't the lens get in the way? Even low power microscopes come quite close to the object. Doesn't it fog with flux, too?
> Its been a long time since I laid out pcbs on a light table with tape. I > remember when plated thru holes were troublesome. > > Now even a "down & dirty" house can etch 5/5 multilayer boards (5 mil min > trace, 5 mil clearance).
They always seem to get them bubble free too. Most places used to give you a few bubbles a few years ago.
> >>>BGAs are certainly a reality and if you want to build DSP boards you >>>better get used to them. Of course, there are suppliers that would >>>love to save you the trouble ;-) >> >>If you want to build practically anything above the simple MCU level >>you'd better get used to them. > > What's a DIP? This term seems vaguely familiar.....
Cheese DIP? Lets go for a DIP? They seem familiar..... Regards, Steve
"Steve Underwood" <steveu@dis.org> wrote in message
news:bdvu5a$nqq$1@hfc.pacific.net.hk...
> I can't image being able to solder with a microscope. I can't remember > seeing someone try. Doesn't the lens get in the way? Even low power > microscopes come quite close to the object. Doesn't it fog with flux, too? >
Hello Steve, I have had the priviledge of soldering under a microscope. And I know quite a few others who do it that routinely with TQFPs. The trick is to use a scope with a very long focal length objective. This will give you several inches of working space. This trick has been exploited for years in macrophotography so as to have more working room. The 6x objective gives somthing like 4 or 5 inches of working room. The scope I used is set up so you look horizontally into the unit and your work is underneath the unit. This takes a little getting use to. A few months back I was was at a client's where they do quite a bit of work with BGAs, and they had a neat optical camera that could see three rows of balls on a soldered BGA. The head of the camera was about a 1/4 inch wide and would be manouvered about by having the board on a translation table. Their production engineer showed me a bunch of stuff about their manufacturing capability, and I was duly impressed. I had done a BGA design for them, and they were building it. BGAs are not for the home assembling environment! I've heard many claims that BGA soldering has better reliability than TQFP soldering. Of course this assumes one possesses a full production type of setup. Clay
"Clay S. Turner" <physicsNOOOOSPPPPAMMMM@bellsouth.net> wrote in
news:T7MMa.5048$Sy1.78@fe02.atl2.webusenet.com: 

> > "Steve Underwood" <steveu@dis.org> wrote in message > news:bdvu5a$nqq$1@hfc.pacific.net.hk... >> I can't image being able to solder with a microscope. I can't >> remember seeing someone try. Doesn't the lens get in the way? Even >> low power microscopes come quite close to the object. Doesn't it fog >> with flux, too? >> > > Hello Steve, > > I have had the priviledge of soldering under a microscope. And I know > quite a few others who do it that routinely with TQFPs. The trick is > to use a scope with a very long focal length objective. This will give > you several inches of working space. This trick has been exploited for > years in macrophotography so as to have more working room. The 6x > objective gives somthing like 4 or 5 inches of working room. The > scope I used is set up so you look horizontally into the unit and your > work is underneath the unit. This takes a little getting use to.
I use a microscope that has a range of 7x - 30x. The microscope is mounted on a boom stand. With a bit of practice you learn exactly where your hands are under the scope. There is a real incentive for learning this skill quickly since the tip of the soldering iron is maybe 700 degrees F. Soldering is best done with the microscope set to 7-8x. Sometimes, I inspect the connections at 20-30x, especially if I have identified a problem. Clay is right that the focal length should be long. There are optical systems for this purpose aimed at electronics that are very nice, I use an old Baush & Lomb which is OK. My tech uses an Olympus that is better.
> > A few months back I was was at a client's where they do quite a bit of > work with BGAs, and they had a neat optical camera that could see > three rows of balls on a soldered BGA. The head of the camera was > about a 1/4 inch wide and would be manouvered about by having the > board on a translation table. Their production engineer showed me a > bunch of stuff about their manufacturing capability, and I was duly > impressed. I had done a BGA design for them, and they were building > it. BGAs are not for the home assembling environment! I've heard many > claims that BGA soldering has better reliability than TQFP soldering. > Of course this assumes one possesses a full production type of setup. > > Clay > > >
The house I use for BGAs have many years experience as a rework center for Silicon Graphics. They have individual screen stencils for most BGAs which is great for prototyping. You can equip a bench for 1-2K dollars (or euros) to solder most SMT parts. BGAs are SUBSTANTIALLY more expensive. I can't see ever having our own equipment for this purpose. -- Al Clark Danville Signal Processing, Inc. -------------------------------------------------------------------- Purveyors of Fine DSP Hardware and other Cool Stuff Available at http://www.danvillesignal.com
Steve Underwood wrote:
> > Al Clark wrote: > > Steve Underwood <steveu@dis.org> wrote in > > news:bduvmg$pm1$1@hfc.pacific.net.hk: > > > >>Al Clark wrote: > > > > Where did cleaning come in? Reballing is a process that remakes the BGA > > contacts. > > > > Profiling determines how fast the board heats, etc. for correct reflow. > > You only do it once for a given assembly type. If you did 6 identical > > boards you wou profile one and place 6 (cost = $80 + 6*$65) > > OK, a confusion of terms. I understand what you mean by profiling now. > > > Unless you have parts on both sides of the board, parts are not glued. > > Are you refering to QFP or BGA? QFPs are routinely glued by the robots > so the board is less susceptible to shaking as it passes through (maybe) > several other robot and shakey conveyer belt steps. I wish people > wouldn't use that concrete like red glue, though. It makes changing QFPs > a real pain. Perhaps that is the intention! > > > When I hand solder a QFP, I place a small amount of solder on a corner > > pad. I then heat this pad with the QFP pins absolutely perfectly aligned > > on the board. I then flip the board 180 degrees, and solder the opposite > > corner. The rest of the pins are then soldered in a kind of controlled > > bridging approach. FLUX IS YOUR FRIEND!!! > > In production environments the girls (its usually girls) tend to use a > soldering iron with a bit the whole width of one side of the QFP. In 4 > goes (i.e. one on each side) they can put a QFP in place very > accurately. Messing around with corner pins is just for people like us, > who don't do this often enough [actually one in a while is often enough > for me :-\ ] > > > My technician uses a x-acto knife to help slide the part, I use tweezers. > > In all cases, this is done under a microscope. > > I can't image being able to solder with a microscope. I can't remember > seeing someone try. Doesn't the lens get in the way? Even low power > microscopes come quite close to the object. Doesn't it fog with flux, too? > > > Its been a long time since I laid out pcbs on a light table with tape. I > > remember when plated thru holes were troublesome. > > > > Now even a "down & dirty" house can etch 5/5 multilayer boards (5 mil min > > trace, 5 mil clearance). > > They always seem to get them bubble free too. Most places used to give > you a few bubbles a few years ago. > > > >>>BGAs are certainly a reality and if you want to build DSP boards you > >>>better get used to them. Of course, there are suppliers that would > >>>love to save you the trouble ;-) > >> > >>If you want to build practically anything above the simple MCU level > >>you'd better get used to them. > > > > What's a DIP? This term seems vaguely familiar..... > > Cheese DIP? Lets go for a DIP? They seem familiar..... > > Regards, > Steve
There are some pretty nice skinny DIPS around. Jerry -- Engineering is the art of making what you want from things you can get. &#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;
Al Clark wrote:
>
...
> > I use a microscope that has a range of 7x - 30x. The microscope is > mounted on a boom stand. With a bit of practice you learn exactly where > your hands are under the scope. There is a real incentive for learning > this skill quickly since the tip of the soldering iron is maybe 700 > degrees F. Soldering is best done with the microscope set to 7-8x. > Sometimes, I inspect the connections at 20-30x, especially if I have > identified a problem. > > Clay is right that the focal length should be long. There are optical > systems for this purpose aimed at electronics that are very nice, I use > an old Baush & Lomb which is OK. My tech uses an Olympus that is better. >
... I use a binocular erect-image microscope with about 2.5" clearance. I bought the body on eBay for about $40, refurbished it and mounted it, using an old 1.25" eyepiece focuser meant for a Newtonian telescope. A pair of 10X WF about doubled the cost. It's also great for removing splinters. It has two powers, about 7.5 and 15. It's good enough. jerry -- Engineering is the art of making what you want from things you can get. &#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;&#2013266095;
On Wed, 02 Jul 2003 23:33:58 -0400, Jerry Avins <jya@ieee.org> wrote:

>Steve Underwood wrote: >> >> Al Clark wrote: >> > Steve Underwood <steveu@dis.org> wrote in >> > news:bduvmg$pm1$1@hfc.pacific.net.hk: >> > >> >>Al Clark wrote: >> > >> > Where did cleaning come in? Reballing is a process that remakes the BGA >> > contacts. >> > >> > Profiling determines how fast the board heats, etc. for correct reflow. >> > You only do it once for a given assembly type. If you did 6 identical >> > boards you wou profile one and place 6 (cost = $80 + 6*$65) >> >> OK, a confusion of terms. I understand what you mean by profiling now. >> >> > Unless you have parts on both sides of the board, parts are not glued. >> >> Are you refering to QFP or BGA? QFPs are routinely glued by the robots >> so the board is less susceptible to shaking as it passes through (maybe) >> several other robot and shakey conveyer belt steps. I wish people >> wouldn't use that concrete like red glue, though. It makes changing QFPs >> a real pain. Perhaps that is the intention! >> >> > When I hand solder a QFP, I place a small amount of solder on a corner >> > pad. I then heat this pad with the QFP pins absolutely perfectly aligned >> > on the board. I then flip the board 180 degrees, and solder the opposite >> > corner. The rest of the pins are then soldered in a kind of controlled >> > bridging approach. FLUX IS YOUR FRIEND!!! >> >> In production environments the girls (its usually girls) tend to use a >> soldering iron with a bit the whole width of one side of the QFP. In 4 >> goes (i.e. one on each side) they can put a QFP in place very >> accurately. Messing around with corner pins is just for people like us, >> who don't do this often enough [actually one in a while is often enough >> for me :-\ ] >> >> > My technician uses a x-acto knife to help slide the part, I use tweezers. >> > In all cases, this is done under a microscope. >> >> I can't image being able to solder with a microscope. I can't remember >> seeing someone try. Doesn't the lens get in the way? Even low power >> microscopes come quite close to the object. Doesn't it fog with flux, too? >> >> > Its been a long time since I laid out pcbs on a light table with tape. I >> > remember when plated thru holes were troublesome. >> > >> > Now even a "down & dirty" house can etch 5/5 multilayer boards (5 mil min >> > trace, 5 mil clearance). >> >> They always seem to get them bubble free too. Most places used to give >> you a few bubbles a few years ago. >> > >> >>>BGAs are certainly a reality and if you want to build DSP boards you >> >>>better get used to them. Of course, there are suppliers that would >> >>>love to save you the trouble ;-) >> >> >> >>If you want to build practically anything above the simple MCU level >> >>you'd better get used to them. >> > >> > What's a DIP? This term seems vaguely familiar..... >> >> Cheese DIP? Lets go for a DIP? They seem familiar..... >> >> Regards, >> Steve > >There are some pretty nice skinny DIPS around. > >Jerry
I used to have some PALs that were DIPs... ...actually think I still do... ;) Eric Jacobsen Minister of Algorithms, Intel Corp. My opinions may not be Intel's opinions. http://www.ericjacobsen.org
Jerry Avins <jya@ieee.org> wrote in news:3F03A9C0.8C7949FA@ieee.org:

> I use a binocular erect-image microscope with about 2.5" clearance. I > bought the body on eBay for about $40, refurbished it and mounted it, > using an old 1.25" eyepiece focuser meant for a Newtonian telescope. A > pair of 10X WF about doubled the cost. It's also great for removing > splinters.
Which of course points to the other common practice of fine hand work under a microscope: Surgery. Which I'll leave to folk with steadier hands than mine! -- Kenneth Porter http://www.sewingwitch.com/ken/