Samplefinger
Offline
God Member
Karma: 8
Posts: 822
ALWAYS ASK FOR THREE. One to use. One to lose. One to abuse.
|
 |
« on: March 14, 2013, 11:10:52 am » |
OSH's DRU minimums are 6 mil trace width / 6 mil spacing / 13 mil drill size / 7 mil annular ring. In routing a 100 pin BGA part I have found that to meet these rules on a 2 layer board I have to make my pads 12 mils. The default for this part is 16 mils. So this gives me pause and I ran it by the guy who supports DipTrace over at their forum and he thinks it is not a good idea either. The fact is, with these minimums, there doesn't seem to be any way to route the innermost pins on a 2 layer board without getting the pads down to 12 mils. Are there PCB manufacturers that have smaller DRUs than the ones I stated above that OSH Park uses? This was the fanout I was able to get with OSH's minimums plus making the pads 12 mils. It is the best I can do and the autorouter does worse. 
|
|
|
|
|
Logged
|
Latest Sampling Scores: ATXMEGA64A3U-MH x3, ATXMEGA256A3U-MH x3, SST38VF6404-90-5C-EKE x3, SST38VF6402-90-5C-EKE x3, PGA870 x3, THS770006 x3
|
|
|
|
Global Moderator
Boston area, metrowest
Offline
Brattain Member
Karma: 249
Posts: 16566
Available for Design & Build services
|
 |
« Reply #1 on: March 14, 2013, 11:58:24 am » |
Itead goes smaller. I normally use 12mil drills with them.
|
|
|
|
|
Logged
|
|
|
|
|
Global Moderator
Boston area, metrowest
Offline
Brattain Member
Karma: 249
Posts: 16566
Available for Design & Build services
|
 |
« Reply #2 on: March 14, 2013, 12:50:57 pm » |
PCB capabilities for IteadStudio (need conversions from mm to mil to mm: 10 mil = 0.254mm .15mm = 5.9mil, call it 6 mil)
Layers 1 - 4 Material FR-4
Board Dimension (max) 500mm X1100mm Board Dimension (min) 10mm X10mm Outline Dimension Accuracy ± 0.2mm >> 7.9mil Board Thickness 0.40mm--2.0mm >> 15.7mil to 78.7mil Board Thickness Tolerance ± 10% Dielectric Separation thickness 0.075mm--5.00mm >> 3 mil to 197 mil Conductor Width (min) 0.15mm( Recommend>8mil ) >> 6mil Conductor Space (min) 0.15mm( Recommend>8mil) >> 6 mil Outer Conductor thickness 35um Inner Conductor thickness 17um--100um Copper to Edge 0.25mm,0.5mm >>10 mil, 20 mil (why 2 numbers?) Plated Component,Plated via Diameter(Mechanical) 0.3mm--6.30mm >>12mil to 248 mil Plated Hole Diameter Tolerance(Mechanical) 0.08mm >> 3 mil Unplated Hole Diameter Tolerance 0.05mm >> 2 mil Hole Space(min) 0.25mm >> 10 mil Hole to Edge 0.4mm >> 16 mil Annular Ring(min) 0.15mm >> 6 mil Aspect Ratio 8:01 Solder Resist Type Photosensitive ink Solder Resist Color Black ,Green, White, Blue ,Yellow Solder Resist Clearance 0.1mm >> 4 mil Solder Resist Coverage 0.1mm >> 4 mil Plug Hole Diameter 0.25mm--0.60mm >>10 mil to 24 mil Selective Finish HASL, ENIG
|
|
|
|
|
Logged
|
|
|
|
|
Colorado
Offline
Edison Member
Karma: 41
Posts: 1260
Reviving dead brain cells with Arduinos.
|
 |
« Reply #3 on: March 14, 2013, 01:54:57 pm » |
I can confirm iTead's minimums as well. I have done several 6mil traces and 10mil drills boards through them.
|
|
|
|
|
Logged
|
|
|
|
|
Samplefinger
Offline
God Member
Karma: 8
Posts: 822
ALWAYS ASK FOR THREE. One to use. One to lose. One to abuse.
|
 |
« Reply #4 on: March 14, 2013, 02:39:51 pm » |
I can confirm iTead's minimums as well. I have done several 6mil traces and 10mil drills boards through them.
I'm not seeing how 6 mil traces beat 6 mil traces but 10 mil drills certainly beat 13 mil drills. My only concern is seeing this: http://hackaday.com/2012/09/11/bed-of-nails-and-accuracy-in-pcb-manufacturing/Do you think this was an outlier?
|
|
|
|
|
Logged
|
Latest Sampling Scores: ATXMEGA64A3U-MH x3, ATXMEGA256A3U-MH x3, SST38VF6404-90-5C-EKE x3, SST38VF6402-90-5C-EKE x3, PGA870 x3, THS770006 x3
|
|
|
|
Global Moderator
Boston area, metrowest
Offline
Brattain Member
Karma: 249
Posts: 16566
Available for Design & Build services
|
 |
« Reply #5 on: March 14, 2013, 03:03:31 pm » |
I see 64 pads around the perimeter. Where do the other 36 go? Pads on the bottom layer? Looks like their might be a hint of another layer, kinda dark on my moniter, like dark brown on black.
|
|
|
|
|
Logged
|
|
|
|
|
Samplefinger
Offline
God Member
Karma: 8
Posts: 822
ALWAYS ASK FOR THREE. One to use. One to lose. One to abuse.
|
 |
« Reply #6 on: March 14, 2013, 03:10:22 pm » |
Any comments about pad size? In DipTrace the default pad size for the Atmel 2560 CBGA is 16 mils. Is 12 mils a show stopper? How is that determined, manufacturer specified or CAD vendor specified or somethign else?
|
|
|
|
|
Logged
|
Latest Sampling Scores: ATXMEGA64A3U-MH x3, ATXMEGA256A3U-MH x3, SST38VF6404-90-5C-EKE x3, SST38VF6402-90-5C-EKE x3, PGA870 x3, THS770006 x3
|
|
|
|
Global Moderator
Boston area, metrowest
Offline
Brattain Member
Karma: 249
Posts: 16566
Available for Design & Build services
|
 |
« Reply #7 on: March 14, 2013, 03:46:09 pm » |
Pad size for the balls should come from the spec sheet of the part you are using. Pad size around the outside is your call for mating to traces for the rest of the board. Go as wide as you can while leaving 8-10mil clearance in between to meet DRU rules for trace clearance/seperation distance.
|
|
|
|
|
Logged
|
|
|
|
|
UK
Offline
Edison Member
Karma: 45
Posts: 2244
What a host of balls she had seen: gaity, the brass buttons...
|
 |
« Reply #8 on: March 14, 2013, 04:00:16 pm » |
From the datasheet for the ATmega640/1280/1281/2560/2561
In the details for the 100C1 CBGA:
Ball diameter: Min: 0.35mm (13.78 mil) Typ: 0.40mm (15.75 mil) Max: 0.45mm (17.72 mil)
It doesn't actually give a recommended landing pattern as such, unfortunately, like a lot of other data sheets do, so it's pretty much guesswork what the size should be.
16 mil is a good size as it's pretty much the "typical" ball size. 14 mil would be acceptable I guess at a pinch.
I wouldn't go below 14 mil - that's just asking for trouble.
|
|
|
|
|
Logged
|
|
|
|
|
Samplefinger
Offline
God Member
Karma: 8
Posts: 822
ALWAYS ASK FOR THREE. One to use. One to lose. One to abuse.
|
 |
« Reply #9 on: March 14, 2013, 04:12:00 pm » |
From the datasheet for the ATmega640/1280/1281/2560/2561
In the details for the 100C1 CBGA:
Ball diameter: Min: 0.35mm (13.78 mil) Typ: 0.40mm (15.75 mil) Max: 0.45mm (17.72 mil)
It doesn't actually give a recommended landing pattern as such, unfortunately, like a lot of other data sheets do, so it's pretty much guesswork what the size should be.
16 mil is a good size as it's pretty much the "typical" ball size. 14 mil would be acceptable I guess at a pinch.
I wouldn't go below 14 mil - that's just asking for trouble.
16 mil is exactly what DipTrace provides on that pattern so they agree with you. I guess I will try again with 14 mil pads and see what I can get it to do.
|
|
|
|
« Last Edit: March 14, 2013, 04:14:20 pm by JoeN »
|
Logged
|
Latest Sampling Scores: ATXMEGA64A3U-MH x3, ATXMEGA256A3U-MH x3, SST38VF6404-90-5C-EKE x3, SST38VF6402-90-5C-EKE x3, PGA870 x3, THS770006 x3
|
|
|
|
|