Author Topic: Altering the paste mask for correct paste coverage on large pads+reflow ovens  (Read 1838 times)

0 Members and 1 Guest are viewing this topic.

Offline AlfBaz

  • Super Contributor
  • ***
  • Posts: 1520
  • Country: au
I read in this post by free_electron about doing a lattice over large pads to reduce paste coverage.
I know how to change the expansion on all pads with drc and on individual pads but I can't help but feel that on large pads you would need to distribute "paste drops" to ensure a good spread of solder
Does anybody know how this is done?
« Last Edit: February 12, 2013, 02:18:30 AM by AlfBaz »

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 4786
  • Country: us
    • SiliconValleyGarage
Re: Altering the paste mask for correct paste coverage on large pads
« Reply #1 on: February 08, 2013, 02:32:16 AM »
Exactly. That's what the lattice does. On a large square padyou would place a 3x3 grid of smaller squares.

For example. An 13 mm by 13 mm pad could have nine 3x3 mm squares of solder paste.

You define this in the pcb library. Not in the pcb design ! ( you can do it there too but that is cumbersome )

Draw your pad as normal in the properties inspector set paste mask expansion to a negative value larger than the pad size. So if you have a 13 by 13 pad : set the paste mask expansion to -14
Now, select the paste mask layer. Place fills.  P-F. And simply draw where the paste needs to go. So you would draw a 3 by 3 and copy that in the grid you want.

Once done you can simply place this component. As the fills belong to the pcb library they are not affected by the drc rules you set in the design. You can still apply paste rules to pads and the 'normal components' will abide by that rule. You latticework will not as it consists of fills.
If you have other manuasl paste fills in the pcb you can create rules there again. The ones pulled from the library can be filtered for ( filter for free objects. The ones pulled from library are then not affected ). So , doing this in the pcb library gives you fine control. You have three levels of rules.
The pad rules, the free fill rules and the fill rules for the ones coming from library.

On the question how much pullback you need : that depends on many factors. Thickness of the stencil , flux/ solder ratio of the paste used, weight of the package ( to prevent floating) Thermal holes in the big pad that may wick part of the solder. You need to talk to the assembly people. There are some rules of thumb and those guys will know which way to shift the coverage ratio to get consistent good soldering.
Professional Electron Wrangler

Offline AlfBaz

  • Super Contributor
  • ***
  • Posts: 1520
  • Country: au
Re: Altering the paste mask for correct paste coverage on large pads
« Reply #2 on: February 08, 2013, 08:18:14 AM »
Thank you very much free_electron, your mention of this in the other thread stuck in my head, and in yet another thread you discuss the lack of need for convoluted temp controll in a home made toaster oven, followed by this gem are as though I have gained experience without doing anything  :)

Offline free_electron

  • Super Contributor
  • ***
  • Posts: 4786
  • Country: us
    • SiliconValleyGarage
Re: Altering the paste mask for correct paste coverage on large pads
« Reply #3 on: February 10, 2013, 03:22:23 AM »
In a toaster you have no control. Your process is not stable, even from cycle to cycle. Never ever compare what you do in a toaster to a 100k$ reflow oven with 12 zone heating. That machine needs 24 hours to become stable after power on. But it will guarantee you perfect process control each and every time. And then you can start thinking about how to 'up' the yield by playing with solder deposits... In a toaster you won't see that as the process masks off completely what you are trying to do.( you may see some improvement. In certain cases, but it won't be universally applicable )
Professional Electron Wrangler

Offline AlfBaz

  • Super Contributor
  • ***
  • Posts: 1520
  • Country: au
Re: Altering the paste mask for correct paste coverage on large pads
« Reply #4 on: February 11, 2013, 11:35:13 AM »
From what I can see a toaster oven, or any food oven straight from the shop would have many limitations but as a project that goes beyond simple electronic control of an element and a fan, I believe we could address some of the issues in an attempt to create a repeatable process that closely follows the reflow profile we desire.

By far the biggest obstacle I can see is the uniform distribution of heat throughout the board

Given that a board will have an unknown distribution of components and that the individual components may have different mass and heat transfer coefficients, sections of the board will heat up at different rates. During the pre-soak cycle, that rate of temperature rise is the critical parameter to prevent flux in the paste from splattering.

Temperature sensitive components will have to have their reflow profile studied along with the solder paste’s profile to come up with a worst case profile that still prevents any thermal shock.

During the soak cycle we want the flux to do its job and boil off. If the board has not achieved uniform heat then it is conceivable that the flux will be at different stages of its process prior to reflow. Too short a soak time may have “heavy” parts of the board where the flux has not activated. Too long and we may have “lighter” parts of the board begin to oxidise or have contaminants otherwise “spoil” the joint during reflow.

The same non-uniformity of heat can create problems during the reflow cycle where the solder may be in its liquid state for too long or too short a period giving us either brittle joints or cold joints, respectively.

Forced convection should alleviate this non uniformity to an extent as it increases the efficiency of heat transfer, but the placement of the fan in most ovens is at the back. To me this would be akin to reflowing components with a hot air gun from the side of the board. Placing the fan at the top would be a good start but I was thinking that a series of duct like structures from the rear fan to the top and bottom of the oven might be simpler. Diffusion of the air flow in some manner at the top and bottom would be desirable in an attempt to have hot air land evenly over the board.

We would also have to carefully regulate the air flow so as to no blow small components off their pads or possibly prevent the surface tension of the solder to pull components into alignment. The latter problem could easily be solved by stopping the fan just prior to the end of the reflow cycle.

Design for manufacturability can be another tool used to aid even heat distribution. If the design of the board will allow flood filling of the board with copper pours, this will help in distributing the heat throughout the board due to coppers good heat conductivity.

With regards to your comments about “upping yields” by adjusting solder deposits, are you saying that it’s not that critical on one offs? I am very inexperienced in this field and thought that too much solder paste on large thermal pads under IC’s may cause the chip to “float” causing problems at the pins and that this would happen no matter what the reflow process would be.




Offline free_electron

  • Super Contributor
  • ***
  • Posts: 4786
  • Country: us
    • SiliconValleyGarage
Re: Altering the paste mask for correct paste coverage on large pads
« Reply #5 on: February 11, 2013, 05:12:11 PM »
-snip- lots of stuff


With regards to your comments about “upping yields” by adjusting solder deposits, are you saying that it’s not that critical on one offs? I am very inexperienced in this field and thought that too much solder paste on large thermal pads under IC’s may cause the chip to “float” causing problems at the pins and that this would happen no matter what the reflow process would be.


On the'lot's of stuff. That is exactly why the real machines used in industry use a 'static air mass'. THere is no air flow over the board ( or quasi none) the board is shoved in a pool of stagnant air that is held at a specific temperature. THe oven has multiple zones. the conveyor moves a t fixed speed and the 'time' the board remains at a certain temperature is determined by the length of the zone. Thats why these machines are 2 to 3 meters long ( Or more. I've seen reflow ovens that are 5 to 6 meters long ). They have an area where the air ( or nitrogen in some cases ) is heated and then flows down through a baffle. this creates a constant pool of air at a fixed temperature. There is virtually no airflow so no parts will blow away.

Now, on the one-offs : if the part doesn't 'flow or it floats you simply rework it. Not something you want to do in mass production. there you want the whole board correct in one shot , time after time. any rework is extra time and cost.
In mass production they tune the paste deposition and zone settings of the oven by running multiple boards ( sometimes hundreds ) till they got it right. On large boards they use sacrificial boards stuffed with thermocouples and a recorder that gets sent through the oven to see the profile at various points of the board. This allows them to fine-tune the whole thing.

if you are doing a one-off you are not going to sit there first on a  bare board with thermocouples figuring that out. you wing it.
There is a perfect solution for one-off's though and that is using a vapor phase reflow oven. Those machines are guaranteed to produce a perfect soldering profile over and over but they come at a cost and their construction makes them difficult to use for mass production as they can only do 1 board at a time and do not have a conveyor ( experiments have been done but largely failed due to the physics involved in vapor phase soldering )

A vapor phase machine is basically a turkey fryer with a cooling ring around the top. You take a deep vessel and put a heating element at the bottom. Around the top of the vessel you put a couple of turns of hollow pipe where you pump a cold liquid through. Now , you pour a measure of liquid with a known evaporation point in the bottom and heat it up until you get a cloud of steam. This is where the physics kicks in. Water , at a standard atmospheric pressure of 1 bar boils at exactly 100 degrees C. Now matter how much heat you crank in that heating element at the bottom : the steam will always bee 100 degree C. you either raise pressure to increase boiling temperature , or you use a different liquid.

the vapor phase ovens used to use some nasty chemicals which were very toxic , especially when inhaled .... but since the creation of' Liquid Teflon' in the mid 80'2 this problem has been solved. The chemical used is called Galden and made by Solvayt. Galden is a totally inert liquid that boils around 190 degrees to 240 degrees depending on the formulation you buy. Galden is available in temperature grades . Steps are about 3 degrees C. You can drink that stuff, it passes straight through your body. Can't be absorbed , can't be destroyed by acid.  It is basically a PTFE just like Teflon.

So what do we do : we heat a puddle of Galden until it evaporates in a cloud at a static temperature (determined by the physics) the heat makes the cloud rise. as it approaches the cooling zone it drops below vapor phase and condenses on the cooling rings and drops back down into the puddle of liquid where it is re-heated.
So you essentially make a heat-pump using a cloud of vapor.

Now all you have to do is lower your board in the CLOUD ( not in the liquid ! ). since the board is cold at first the Galden condenses on the board and drips back off the bottom but int the process it disposes of its heat into the board. The board and parts act as heat sink. When the board has fully warmed up to the temperature of the Galden cloud, the Galden evaporates off the board as well. there is no risk in overheating as the cloud temperature is a constant. all you have to do is lower the board in the cloud : see it get 'wet' with Galden , wait for the board to become dry at which point all solder becomes liquid and then gently lift it out of the cloud , through the cooling zone. done

Galden is eutectic in nature just like solder. It toggles from liquid to vapor at 1 and only 1 fixed temperature. There is no 'zone'. it is a very 'thin' liquid, it flows better than water and has a very low molecular weight. This makes sure that a puddle of Galden cannot push a part aside. it simply doesn't weight enough. It is a very strange liquid. It is also used as a coolant in various high-tech applications ( a coolant is basically a heat transfer liquid ) like ion implanters where it replaces freon. When i was in working in the waferfab we had to stop using freon and switch to Galden. I remember the little weird 'balls' it was packaged in. Like a cube but with rounded edges and a spout on the side. Bloody expensive stuff. probably has come down in price now - nope just checked. Still at well over 1000$ a bottle )  http://www.pchemlabs.com/subcatagoryb.asp?PID=Galden-Heat-Transfer-Fluids

So these machines produce consistent soldering results
problem is :

- one board a time and you need to wait for the cycle to complete
- time to lower and raise the board
- you contaminate the Galden with flux remains.... but you can 'scrub' the Galden to regenerate
- Galden is an extremely expensive liquid. To the cost of over 1000$ per bottle. You may as well use mouse-milk ...

These ovens are used a lot in the manufacturing of military , avionics , space and medical devices. simply because of the excellent process control and 'endpoint' detection ( board dry = solder liquid without risk of overheating. there is also no problem with oxidation once the flux has evaporated as the whole board is submerged in a cloud of Galden. There is no Oxygen around... so the solder or copper does not oxidize once flux has evaporated.

http://www.asscon.de/e/pages/products/laboratory.html

http://www.lesker.com/newweb/fluids/heattransfer_galden_ht.cfm?pgid=0

http://www.solvayplastics.com/sites/solvayplastics/EN/specialty_polymers/Fluorinated_Fluids/Pages/Galden_PFPE.aspx
Professional Electron Wrangler

Offline AlfBaz

  • Super Contributor
  • ***
  • Posts: 1520
  • Country: au
Thats very interesting info!
I had heard about these vapour phase ovens but had never gotten around to finding out what they were.
As far as I can tell there are three main ways to heat something up; radiation like IR heaters, conduction, where if you heat/cool one side of an object and not the other the temperature differential between one side and the other causes the conduction of heat across the temperature grade (such as a frypan on a stove) and finally convection which is similar to conduction except it mainly happens through a gaseous medium such as air.

From how you are describing the process it sounds as though the main transfer mechanism is conduction to start with and convection toward the end.
Aside from the constant temperature benefits due to the fixed boiling point of that liquid it would seem that they are maximising heat contact with all parts of the board.

You also talk about the set up of industrial reflow ovens by monitoring temperatures on the boards during trial runs. The thought had crossed my mind that to get the best process in a home oven you would have to measure multiple points on the populated board to see how bad the temperature differentials are across the board, but given that most of the time it may be an expensive one off this would not be very practical.

One thing that I could do is create several sample boards with varying degrees of thermal mass gradients, put these through the process and plot multiple key points on the test pieces to see what worse case temperature differences would be. I could then try different things to minimise these differences or possibly find that they are not large enough to be of concern.

Here’s one of the things I was contemplating but I know squat about thermodynamics so I don’t know if this will do anything. Thought it might be worth while trying out though.

Instead of just using the (usually) one element that comes with the oven install three more, similarly rated ones so you end up with one on the top, bottom and sides. Then we have four elements trying to heat a given space to the same temperature as one element. At this stage, have we reduced the thermal inertia of the load, the load being the chamber and the boards?

I understand that the thermal inertia wont physically change but if we draw an analogy of say pulling a load with only one engine and then again with four engines, in the latter case the load will have less influence on our control of it.

Now, you may be thinking, as I have, what do we do to prevent overshoot? Do we also need a cooling system?

Well these are my lines of thinking... You can cool something down by applying heat! This may sound like an oxymoron but if we have a heat source like a blow torch that has a temperature of say 1000 degrees and apply it to something that is at 2000 degrees we are creating an energy differential that will try to reach equilibrium.

In the same way if we have elements that are sitting at 200 degrees with TRIAC triggered AC at a phase angle of say 90 and we reduce the firing angle to 135 then the elements will be forced to output energy that we are putting in and in effect reduce temperature quicker than it would if we simply turned the elements off.
Am I loosing the plot or is there some merit in this?


« Last Edit: February 12, 2013, 02:11:50 AM by AlfBaz »


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf