Reflowing issues

[Emil]

I'm having some problems when reflowing this board and as I need to order more boards soon this is a good time to fix it.. I place the components by hand and reflow it in a toaster oven.

The two SOT-23 packages (Q1,Q2) sometimes rotates so that the one lead is off the pad. Could this be caused by the uneven amount of copper around the pads, causing the part to rotate as it reflows unevenly? (Pad 2 is part of a copper filled area with via stitching and pad 3 is connected to a wide trace while pad 1 is connected to a thin trace.)

The QFN-16 package (U1) appares to flaot a bit high on the thermal (center) pad so that some of the other pads are not soldered. I'm considering a few solutions:
-Remove the thermal pad, it's not needed for cooling.
-Leave the thermal pad as it is but get a new stencil with smaller holes in a grid pattern over the themal pad.
Are either of these good solutions?

I was suprised to find that the QFN-16 package takes a lot longer to place on the board than the DFN-10 2x2 and 3x3mm packages, which are surprisinly easy. I think part of it is the need to align it in both X and Y. Maybe longer pads would make it easier to visually align it?

[fcb]

Is your toaster oven under loop control, do you have a long enough soak stage before the ramp to TAL (time above liquidus)?

[Emil]

Is your toaster oven under loop control, do you have a long enough soak stage before the ramp to TAL (time above liquidus)?

No control (yet). I just turn it on, wait for all the parts to reflow + 10 seconds then off and open door after another 10 seconds.

The oven heats up quickly, so it's likely that the soak time is short.

[salbayeng]

Ah there's your problem, need more preheat. (Or possibly junky flux in your paste)
I use a toaster oven too, and some nice AIM leaded paste,  my method goes like this
(a) have the oven knob set at 200C to warm up for 10mins before hand
(b) put a thermocouple in a suitable hole on the pcb.
(c) put PCB into oven , wait for thermocouple to read 160C
(d) wind knob to maximum
(e) wait till thermocouple reads 205C (210C if you have chunky inductors)
(f) turn knob to off , open door, slide tray out, 
(g) let air cool to ~ 150C , then carefully put PCB on big chunk of aluminium
repeat with next PCB

You will also find a better result with hand prototypes, if you make all the pads a bit on the big side, (i.e. really long pads on IC's ) , this helps if you need to jam a 0603 or 1206 onto a 0805 footprint .
Useful tip:
If you use a mylar stencil, then shade over all the holes with a red texta, this helps with alignment, then when you squeegee the past in, it dissolves the red, and makes it slightly pink so you can see where you have pasted as it's pink. Magically when you go to reflow the board, all the joints turn bright red, so you can visually check while it's still in the oven. (The red comes off with alcohol)

[marshallh]

Do you have even heat distritubtion within your oven? (It must be a convection oven or you get dead spots)

I've never had problems like that even with some huge copper fills. The key is to keep the air moving to prevent thermal shadowing

[charlespax]

What king of solder paste are you using? Make sure it's leaded. I've had nothing but trouble reflowing with the solder-free lead-free stuff.

[Niklas]

I would suggest that you decrease the amount of solder paste on the thermal pad of the QFN. If it is not critical for the heat transfer, then maybe 50% coverage. Without a decrease, there is also a risk that the paste will be squeezed out and create a short to the surrounding terminals on the QFN package.

Temperature and timing is very dependent on your oven. I have been using a toaster oven with over and under heating that I scored from a flea market. With SnPb solder paste I put the PCB in the oven, turn the heat on to 150'C, wait for the thermostat to click, wait a minute to soak and then turn on full power until reflowed, open lid and shut off. Works also with multilayer PCBs with massive copper planes.

[Emil]

What king of solder paste are you using? Make sure it's leaded. I've had nothing but trouble reflowing with the solder-free stuff.

I use fresh Kester EP256 leaded solder paste.

[charlespax]

I use fresh Kester EP256 leaded solder paste.

Good, then you shouldn't have problems with the solder paste.

I wonder if it's those vias. Are the vias next to the SOT23 chips exposed or are they covered with solder mask (tented)? A photo of the bare PCB would be good if you have you to share.

[Kjelt]

I wonder if it's those vias.

Had some problems with vias to close to the component myself, the vias were "absorbing" some of the paste needed for the hidden cooling path under the chip, resulting in too little solderpaste on the actual cooling path contact. But making components spin Sounds more as if the temperature goes up too fast.

[charlespax]

That would be flux?   

I've had lots of trouble with the lead-free stuff. 

[marshallh]

Had 0 problems with lead-free paste here. I only use genuine Amtech paste.

There are alot of fakes/clones out there. Almost everything on ebay is guaranteed to be fake.

[IanB]

The two SOT-23 packages (Q1,Q2) sometimes rotates so that the one lead is off the pad.

Do you mean they tombstone? (They flip up on their side?)

This would be caused by the surface tension of the solder pulling against the part if one side melts before the other. The cure would be more even heating so that the solder on all the pads melts at about the same time. This could be helped by a longer preheat time, and/or a convection fan in the oven.

[T3sl4co1l]

I wonder if it's those vias.

Had some problems with vias to close to the component myself, the vias were "absorbing" some of the paste needed for the hidden cooling path under the chip, resulting in too little solderpaste on the actual cooling path contact. But making components spin Sounds more as if the temperature goes up too fast.

Besides speed or unevenness (the edges will heat up quicker, for one), you should keep traces entering pads from regular directions.  The one on the corner has two traces entering the pads from the same directions, which will tend to pull the part in that direction.  Entries should be balanced, so each trace 'pulls' the part in opposite directions and does net nothing.

Tim

[Emil]

I wonder if it's those vias. Are the vias next to the SOT23 chips exposed or are they covered with solder mask (tented)? A photo of the bare PCB would be good if you have you to share.

The vias are tented.

Not a bare board, but dont't think the components cover anything relevant: http://www.zeptobit.com/wp-content/uploads/2014/05/ZeptoMag_21_top_600.jpg

[Emil]

The two SOT-23 packages (Q1,Q2) sometimes rotates so that the one lead is off the pad.

Do you mean they tombstone? (They flip up on their side?)

This would be caused by the surface tension of the solder pulling against the part if one side melts before the other. The cure would be more even heating so that the solder on all the pads melts at about the same time. This could be helped by a longer preheat time, and/or a convection fan in the oven.

I have not had any problems with tombstoning. They rotate so that the leads are not centered on the pads. (Only the sot23 components). I'm on vacation now and don't have any pictures available, but i will take a picture and upload when I get home.

[salbayeng]

Ok , having looked at photo ,  you have managed to get the QFN packages attached and I would think that was more challenging than SOT-23.
The solder masking is done properly and should stop solder theft by the adjacent resistor.
There are some thickish traces but shouldn't create a problem if the reflow temperature profile is OK.
U4 seems to have slightly bigger pads (or is it an optical illusion?)

I would suggest
(a) drying the SOT-23 in an oven at 120C for several hours, possibly the pins are a little oxidised, and only two pins wet initially, and spin the part before the flux can do its job on the third pin.
(b) Slightly larger pads ( a bigger reserve of solder and flux), your pads are smaller than what I use.
(c) Longer or hotter preheat
(d) check the cleanliness of the PCB , are there thumbprints on the edges?
(e) stencilling/pasting issues : e.g. does the problem occur only on the PCBS on the (leftmost) edge of the panel

Have you checked Q1 and Q2 are being placed properly before reflow? 

[wraper]

I think that you have too big pads. You can make them smaller so part won't have a place where to move on the pad.
Edit: if I see correctly, compinent terminals are landing on pads edges, not in the middle. Therefore surface tension of melted solder on one can completely pull a part from another where solder is not melted yet. http://www.vishay.com/docs/84004/sot23.pdf
Also it seems that gap between resistors/capacitors pads is too big. Seems that their terminals land just on the edge of the pad and pad area don't even completely cover metallization of the part terminal .