T sub g

Alright gentlemen and ladies of the forum, I again seek knowlage on what is probably a well known fact. I have spent a good part of the morning with various IPC standards and weighty tomes such as the Electronic Packaging and Interconnection Handbook and still the answer eludes me. When a multilayer board is supported at the edges on a conveyor in reflow or wave solder and it exceeds the T sub g, (I don't know how to do suffixes here and Glass Transition Temperature is such a long phrase), it seems it will sag. This causes warped boards, overflow in the wave and other problems with components. My question is does the board tend to return to normal flat condition as it cools, in other words does it have "memory". Or will it remain in the bowed orientation. This question also pertains to bare boards. If we receive them warped beyond spec., will baking them above Tsub g and allowing or forcing them to relax and then cooling be permanent or will they slowly return to there "natural" warped condition. I appreciate the transfer of knowlage that transpires here more than I can say. Thanks for any help. Gary

reply »

| Alright gentlemen and ladies of the forum, I again seek knowlage on what is probably a well known fact. I have spent a good part of the morning with various IPC standards and weighty tomes such as the Electronic Packaging and Interconnection Handbook and still the answer eludes me. | When a multilayer board is supported at the edges on a conveyor in reflow or wave solder and it exceeds the T sub g, (I don't know how to do suffixes here and Glass Transition Temperature is such a long phrase), it seems it will sag. This causes warped boards, overflow in the wave and other problems with components. My question is does the board tend to return to normal flat condition as it cools, in other words does it have "memory". Or will it remain in the bowed orientation. This question also pertains to bare boards. If we receive them warped beyond spec., will baking them above Tsub g and allowing or forcing them to relax and then cooling be permanent or will they slowly return to there "natural" warped condition. | I appreciate the transfer of knowlage that transpires here more than I can say. Thanks for any help. | Gary Gary, Darned good questions and obviously important. Glass transition temperature (correctly abbreviated as Tg), as you know, is the temperature at which the resin (epoxy, polyimide, etc.) begins becoming "liquid" again. Actually, the epoxy in laminate and core material is fully cured ("C" staged) while in the prepreg it is in the "B" stage awaiting relamination to molecularly x-link and become cured. The key to dimensional stability, as warp and twist, primarily involves two things. One is the way reference, or power, planes are arranged. We all know the importance of having "balanced" multilayer constructions regarding copper foil thickness, or weight, about the center line of the board in the Z axis. Equally important is the choice of core and prepreg materials in MLB constructions. First, heavy glass styles are more dimensionally stable. That is a 7628 glass style (very thick at about .007" while containing relatively very little resin at about 43% resin to glass ratio) is significantly more stable than say a 1080 glass style (about .0025" thick having up to 70% resin content). Part of the equation is that boards with lower resin to glass ratio materials generally exhibit less warp and twist than do others - both after relamination and after solder processing when Tg is exceeded. For MLB's requiring thinner cores and prepregs (higher layer counts - over 6 layers), it is important to design constructions that take advantage of minimal resin content as well. For example, if we design a board with all .005" thick dielectrics (as a ten layer, .062" thick as an example) it is important not to use all 1080 glass styles as too much resin will be part of the package. It is advisable to use a single ply of 106 material (about .002" thick with a 65% resin content) combined with a single ply of 2113 material (.003" thick but with a resin content of about 58%) thus reducing the overall resin content. This is called an unbalanced construction providing improved dimensional stability in all axes including reducing warp and twist. Another factor invloved here is using homogeneous glass styles (known as 60's yarn meaning in the warp direction there are 60 yarn bundles, made up of 7-9 micron diameter glass fibers). With this type construction, together with unbalanced designs, we have a natural "nesting" of like fiber bundles mechanically "interlocking" thus minimizing warpage. This is where the "memory" you speak of comes in. With all this, and properly managed relamination processes (flat plattens and tooling plates, evenly distributed press thermal properties, the specified press cycle for the chosen material, and about twenty other things) we have flat boards to within 1.5% or even down to +, or - .005"/" when designs and everything else comes together. Regarding Tg being exceeded - typically this is not an issue as the time is brief. It can be with regular epoxy formulations at say 120C Tg. With modified epoxies reaching to 180C and beyond, only a few seconds will not violate Tg excessively. Concerning memory - it is there unless acted upon by excessive mechanical force as weight or pressure. During bake cycles, boards should be placed in stacks in ovens and heated to remove moisture at temperatures not exceeding Tg's. For assembled boards, some type support is advised for this consideration. For more help, or confusion, I will have a new article in PC Fab Magazine within a few months. If you can hardly wait for this incredible event, call or email me, Enjoy, Earl Moon

reply »

| | Alright gentlemen and ladies of the forum, I again seek knowlage on what is probably a well known fact. I have spent a good part of the morning with various IPC standards and weighty tomes such as the Electronic Packaging and Interconnection Handbook and still the answer eludes me. | | When a multilayer board is supported at the edges on a conveyor in reflow or wave solder and it exceeds the T sub g, (I don't know how to do suffixes here and Glass Transition Temperature is such a long phrase), it seems it will sag. This causes warped boards, overflow in the wave and other problems with components. My question is does the board tend to return to normal flat condition as it cools, in other words does it have "memory". Or will it remain in the bowed orientation. This question also pertains to bare boards. If we receive them warped beyond spec., will baking them above Tsub g and allowing or forcing them to relax and then cooling be permanent or will they slowly return to there "natural" warped condition. | | I appreciate the transfer of knowlage that transpires here more than I can say. Thanks for any help. | | Gary | Gary, | Darned good questions and obviously important. Glass transition temperature (correctly abbreviated as Tg), as you know, is the temperature at which the resin (epoxy, polyimide, etc.) begins becoming "liquid" again. Actually, the epoxy in laminate and core material is fully cured ("C" staged) while in the prepreg it is in the "B" stage awaiting relamination to molecularly x-link and become cured. | The key to dimensional stability, as warp and twist, primarily involves two things. One is the way reference, or power, planes are arranged. We all know the importance of having "balanced" multilayer constructions regarding copper foil thickness, or weight, about the center line of the board in the Z axis. Equally important is the choice of core and prepreg materials in MLB constructions. | First, heavy glass styles are more dimensionally stable. That is a 7628 glass style (very thick at about .007" while containing relatively very little resin at about 43% resin to glass ratio) is significantly more stable than say a 1080 glass style (about .0025" thick having up to 70% resin content). Part of the equation is that boards with lower resin to glass ratio materials generally exhibit less warp and twist than do others - both after relamination and after solder processing when Tg is exceeded. | For MLB's requiring thinner cores and prepregs (higher layer counts - over 6 layers), it is important to design constructions that take advantage of minimal resin content as well. For example, if we design a board with all .005" thick dielectrics (as a ten layer, .062" thick as an example) it is important not to use all 1080 glass styles as too much resin will be part of the package. It is advisable to use a single ply of 106 material (about .002" thick with a 65% resin content) combined with a single ply of 2113 material (.003" thick but with a resin content of about 58%) thus reducing the overall resin content. This is called an unbalanced construction providing improved dimensional stability in all axes including reducing warp and twist. | Another factor invloved here is using homogeneous glass styles (known as 60's yarn meaning in the warp direction there are 60 yarn bundles, made up of 7-9 micron diameter glass fibers). With this type construction, together with unbalanced designs, we have a natural "nesting" of like fiber bundles mechanically "interlocking" thus minimizing warpage. This is where the "memory" you speak of comes in. | With all this, and properly managed relamination processes (flat plattens and tooling plates, evenly distributed press thermal properties, the specified press cycle for the chosen material, and about twenty other things) we have flat boards to within 1.5% or even down to +, or - .005"/" when designs and everything else comes together. | Regarding Tg being exceeded - typically this is not an issue as the time is brief. It can be with regular epoxy formulations at say 120C Tg. With modified epoxies reaching to 180C and beyond, only a few seconds will not violate Tg excessively. | Concerning memory - it is there unless acted upon by excessive mechanical force as weight or pressure. During bake cycles, boards should be placed in stacks in ovens and heated to remove moisture at temperatures not exceeding Tg's. For assembled boards, some type support is advised for this consideration. | For more help, or confusion, I will have a new article in PC Fab Magazine within a few months. If you can hardly wait for this incredible event, call or email me, | Enjoy, | Earl Moon Earl, That was great. It just saved my butt. We had a fab come in for one of our customers. The customer is not satisfied with the fab shop because of twisting issues on the bare boards. Of coarse, the fab shop insists they cannot see a problem and they're not doing anything wrong. When I showed my customer this article he jumped up and down. The guy was pumped. He called his fab shop and sure enough. They are using 1080 on a .062" 10 layer fab. My customer said the fab guy was tripping over his toungue and trying to come up with a million justifications for doing what he did. In the end, the fab supplier said he'd try out the single ply 106. So...... Kudos, Earl. Nice work. Much thanks, Justin and My Customer

reply »

| | | Alright gentlemen and ladies of the forum, I again seek knowlage on what is probably a well known fact. I have spent a good part of the morning with various IPC standards and weighty tomes such as the Electronic Packaging and Interconnection Handbook and still the answer eludes me. | | | When a multilayer board is supported at the edges on a conveyor in reflow or wave solder and it exceeds the T sub g, (I don't know how to do suffixes here and Glass Transition Temperature is such a long phrase), it seems it will sag. This causes warped boards, overflow in the wave and other problems with components. My question is does the board tend to return to normal flat condition as it cools, in other words does it have "memory". Or will it remain in the bowed orientation. This question also pertains to bare boards. If we receive them warped beyond spec., will baking them above Tsub g and allowing or forcing them to relax and then cooling be permanent or will they slowly return to there "natural" warped condition. | | | I appreciate the transfer of knowlage that transpires here more than I can say. Thanks for any help. | | | Gary | | Gary, | | Darned good questions and obviously important. Glass transition temperature (correctly abbreviated as Tg), as you know, is the temperature at which the resin (epoxy, polyimide, etc.) begins becoming "liquid" again. Actually, the epoxy in laminate and core material is fully cured ("C" staged) while in the prepreg it is in the "B" stage awaiting relamination to molecularly x-link and become cured. | | The key to dimensional stability, as warp and twist, primarily involves two things. One is the way reference, or power, planes are arranged. We all know the importance of having "balanced" multilayer constructions regarding copper foil thickness, or weight, about the center line of the board in the Z axis. Equally important is the choice of core and prepreg materials in MLB constructions. | | First, heavy glass styles are more dimensionally stable. That is a 7628 glass style (very thick at about .007" while containing relatively very little resin at about 43% resin to glass ratio) is significantly more stable than say a 1080 glass style (about .0025" thick having up to 70% resin content). Part of the equation is that boards with lower resin to glass ratio materials generally exhibit less warp and twist than do others - both after relamination and after solder processing when Tg is exceeded. | | For MLB's requiring thinner cores and prepregs (higher layer counts - over 6 layers), it is important to design constructions that take advantage of minimal resin content as well. For example, if we design a board with all .005" thick dielectrics (as a ten layer, .062" thick as an example) it is important not to use all 1080 glass styles as too much resin will be part of the package. It is advisable to use a single ply of 106 material (about .002" thick with a 65% resin content) combined with a single ply of 2113 material (.003" thick but with a resin content of about 58%) thus reducing the overall resin content. This is called an unbalanced construction providing improved dimensional stability in all axes including reducing warp and twist. | | Another factor invloved here is using homogeneous glass styles (known as 60's yarn meaning in the warp direction there are 60 yarn bundles, made up of 7-9 micron diameter glass fibers). With this type construction, together with unbalanced designs, we have a natural "nesting" of like fiber bundles mechanically "interlocking" thus minimizing warpage. This is where the "memory" you speak of comes in. | | With all this, and properly managed relamination processes (flat plattens and tooling plates, evenly distributed press thermal properties, the specified press cycle for the chosen material, and about twenty other things) we have flat boards to within 1.5% or even down to +, or - .005"/" when designs and everything else comes together. | | Regarding Tg being exceeded - typically this is not an issue as the time is brief. It can be with regular epoxy formulations at say 120C Tg. With modified epoxies reaching to 180C and beyond, only a few seconds will not violate Tg excessively. | | Concerning memory - it is there unless acted upon by excessive mechanical force as weight or pressure. During bake cycles, boards should be placed in stacks in ovens and heated to remove moisture at temperatures not exceeding Tg's. For assembled boards, some type support is advised for this consideration. | | For more help, or confusion, I will have a new article in PC Fab Magazine within a few months. If you can hardly wait for this incredible event, call or email me, | | Enjoy, | | Earl Moon | Earl, | That was great. It just saved my butt. We had a fab come in for one of our customers. The customer is not satisfied with the fab shop because of twisting issues on the bare boards. Of coarse, the fab shop insists they cannot see a problem and they're not doing anything wrong. When I showed my customer this article he jumped up and down. The guy was pumped. He called his fab shop and sure enough. They are using 1080 on a .062" 10 layer fab. My customer said the fab guy was tripping over his toungue and trying to come up with a million justifications for doing what he did. In the end, the fab supplier said he'd try out the single ply 106. So...... Kudos, Earl. Nice work. | Much thanks, | Justin and My Customer

Justin, Ah, you flatter me - my boy. All in five minutes work. Seriously, glad to help one and all. However, we must temper our delight in new knowledge with an understanding not everyone understands some of this stuff and we don't want to "waste" otherwise qualified suppliers. This type stuff has been around for some time. The problems associated with it are few but important. I can't go into all of them in this forum (my article will) because of time and space. I will keep you posted and, again, if there are specific questions please email or call me. Earl Moon

reply »