using DIY reflow oven; can multiple passes harm components or the board?

[linux-works]

I am new to pcb reflowing.

I borrowed a friend's DIY oven (which is known to work) and I'm about to try loading my first set of parts on a 2 layer (fairly simple) pcb.

I'd like to try this out in stages and see how well it goes.  but I'm not sure how bad it would be if I tried to do 2 or even 3 passes on the board.  what I'm thinking of is to solder all the simple parts (caps and resistors) first and see how well that goes.  then, try for the smd chips.  leave off the expensive smd chips and test the board to see if I can load firmware to it (arduino based using smd atmega 328).  if it takes firmware ok, then I can finish the board with the final baking of the more expensive parts.

obviously I will not solder electrolytics and plastic .1" header pins and any other plastic parts (I wonder, how tolerant are smd leds to reflowing?).

can you guys advise me on this?  once I get the hang of it and gain confidence, I will likely load the board fully and do the baking just once, but I'd like to take it in stages and if I get it wrong on the first try, I don't want to waste good chips and have to throw the whole thing away, etc.

the board was done in a cheap china pcb house (pcbway) and I've hand soldered this board in previous builds and it holds up well, but it IS a budget pcb service and so I'm not sure if I would weaken the solder mask or even the traces/vias if I did multiple bakes.

advice, please?

TIA

[Gary.M]

How does your friends diy reflow oven control temperature and the profile? I can tell you from experience when building an oven and calibrating it that too much heat and several passes will make a mess of even a very good laminate when temperatures aren't tightly controlled.

[linux-works]

my friend used a pretty common arduino system and a thermocouple (that I placed right on the pcb, touching an unused pad).  he was able to solder a pretty high density chip that few would attempt to solder by hand.  he said it worked the first time.

for the heck of it, I tried my first pass on the atmega 328, some .1uF and some 10k's so that I could try loading the bootloader into it.  it worked!  was able to BL it and then download a sketch.

I did not have a stencil and just painted on some leaded paste (MG chem. brand).  after I brushed away the residue solder balls that were left over from the process, the job looked ok visually, at least.

next time I get boards, I do plan to get a metal stencil but I didn't with this board run.

perhaps this board will be a test board.  I'll do one more placement of the remaining smd parts, do one more bake and then place the rest of the TH parts by hand.

just to boot the system, I did have to hand solder the 6pin ISP connector and also the 6 pin 'ftdi' inline connector, as well as a TH 16mhz resonator.  after I BL'd the system, I unsoldered the resonator (it does not seem like that would be a good thing to send thru the oven, but maybe its ok, I don't know yet) and I did unsolder the .1" headers (isp and ftdi) since those have plastic spacer insulators on them and those did not look like they'd be happy to be baked.  I'll resolder the 6pin inline ftdi connector since this needs to be able to take software updates; but the isp connector is no longer needed once you do the initial BL.  eventually, I'll get some pogopins setup to do this and avoid having to solder any connectors on just to test the cpu system and get it initially loaded.   (I wonder if the resonator would also work on a small remote board that connects via pogo pins, saving me from having to temp-solder it and then unsolder later on?)

all the parts that I'm using (that are smd) are digital parts and nothing is analog.  the analog parts are all TH and those will get sockets and be done by hand at the final stage.

before I start doing this en-masse, I'll test the hell out of this first-time baked board and see if it works as well as my hand-soldered boards.

thanks for the advice, guys.  here's what the board looks like when its done and in its case (this one was hand soldered by me a week or two ago):



as you can see, its not a 'difficult' board and nothing is very dense (the cpu is the 'hardest' part on the board to hand solder) and its just a simple 2 layer board with reasonably thick traces and nothing too challenging imho.  hopefully it won't suffer for going thru the oven two times.  and I probably won't be doing 2 passes once I get the hang of it and build confidence.

[linux-works]

oh, and I still am wondering how well the plastic smd leds will do when going thru the oven?

are those ok to bake (just 1 pass) - or should they always be done by hand?  the little green ones have a bit more plastic 'lens' to them than what I normally see in smd leds, so maybe those should not be baked?

[salbayeng]

The simple solution is to use leaded solder paste, this gives you longer effective oven time because your peak oven times are lower.
So 3 passes leaded is about the same as one pass of unleaded.
Use the quickest / lowest profile,  expect some of the SMD electros and large coils to need touch up with a soldering iron, these big components need longer / hotter profiles usually, (having a longer preheat may be helpful).
In general you will get a better result spending longer at preheat (160C) and shorter at peak, 30sec at (205C)
Might be helpful also to wipe the pcb with some liquid flux between passes.

If you have a glass door on your unit, you can just put a blob of solder paste in each of the 4 corners of the PCB. Watch through the window, when all 4 blobs melt, open the door and turn oven off.

One issue with multiple passes is to do with large copper areas and thick tracks, if you have not crosshatched these copper fills, then they are likely to form blisters (due to gas evolution). (almost certain for 3 unleaded passes).

3 passes with LED's with leaded solder should be OK, just be careful not to bump them when hot (the clear plastic will slide off!!)

For double sided , I use a piece of aluminium flywire to rest the PCB on, it catches any parts that fall off, (that's pretty rare), and supports the bigger parts,  but be careful not to bump the PCB until it has cooled to < 170C

[DerekG]

I borrowed a friend's DIY oven (which is known to work)

As your friend is familiar with his oven, why not have him on-hand to help you during your first run?

[KL27x]

It's more expensive than a reflow oven, but a stereomicroscope did more for me than a reflow oven.

I don't want to waste good chips and have to throw the whole thing away, etc.

For one, nothing is going to go wrong (in one reflow, anyway) that can't be fixed with more time and effort. Second, I think for the SMD IC's, you may find hand-soldering an attractive option. Personally, anything with exposed pins on it, I can solder just fine. And IC's in particular, better/faster than paste/reflow when inspection and rework is included. When I order stencil I leave off SMD IC and do those by hand. For same reason as you, I put the expensive IC's on there after all the other parts are on there. In case I don't use all the boards, or in case I have firmware revision in the interim.

I personally doubt multiple passes through oven will adversely affect most parts. But check datasheet. SMD ceramic resonators are very sensitive, for instance.

[linux-works]

I borrowed a friend's DIY oven (which is known to work)

As your friend is familiar with his oven, why not have him on-hand to help you during your first run?

that won't help much.  he's a real novice and he's also out of town for the next few weeks.

btw, things are now worse on this board that I just tried today.  first bake seemed quite hopeful.  I put the cpu on and not much else (just support circuitry for it).  I was able to bootload it and then arduino load it.  so I did a 2nd pass with almost all the rest of the parts and the cpu still said hello on the serial bus, but i2c seems quite dead (i2c_scan shows NO devices) and even a gpio led seems to not light up when I digitalWrite() to it.

perhaps the atmega328 just cannot do more than 1 bake.

I have no control over the temp profile unless I download new arduino code to the oven system.  I don't have the code handy and don't want to overwrite what my friend has in his oven.  I think it should be an ok profile as-is, but perhaps that could be a problem, too.

[linux-works]

The simple solution is to use leaded solder paste, this gives you longer effective oven time because your peak oven times are lower.
So 3 passes leaded is about the same as one pass of unleaded.

I am using leaded paste.  I made sure NOT to get the lead-free crap (lol).

[salbayeng]

The M328 will withstand 3 passes through a normal leaded oven,  I've run ~1000 PCB's with M328 with two passes ,  and then reworked several of those, never a problem.
I use a $20 Kmart toaster oven for prototyping at home, and the only time I've ever burnt M328 chips was when the big oven broke down , and we were pushing panels through the toaster oven, it was outside in the fresh air, and we forgot about two panels, so they were about 20mins at 220C, the PCB's were visibly darker, and the overlay was badly yellowed, and the solder was all grey, even then 5 of the M328's still worked and only 3 died, (we swapped out all 8 to be on the safe side, all the other components, including a blue LED were OK.

It's possible your DIY oven has too much infra-red heat, this causes black components to get hotter than they should.

For the toaster oven the following method is used. (This may be helpful to you)

(a) if there are no visible pads near the glass door, put some dobs of solder paste on the two corners in the front
(b) place the boards on the rack (with flywire underneath) , insert a thermocouple into any available mounting hole
(c) position the PCB's with about a 1" gap from the door (it's a bit cooler near the door)
(d) ensure you have a baffle under the PCB (i.e. the drip tray that comes with the oven)
(e) set oven timer to 5mins (in case you forget!), set temperature knob to "200C", turn on BOTTOM element only, this will give convection heating only, and will slowly warm up the oven and it's contents
(f) wait a max of two minutes in the 150-160C zone for preheat (this oven only does about 170C on the 200C setting, you may have to tweak the temperature knob.
(g) wind the temperature to maximum, turn BOTH elements on , (now you have IR heating from top, and convection from below)
(h) now be ready to turn off power and open door when
      no reflow (you have missed it?) and temp goes past 210C
      you see reflow, count 10 secs, and if now over 205C
      you see reflow, count 10 secs, and if under 205C, wait for 205C or  30secs maximum.
(i)  pull PCB's from oven at 160C, and place on large aluminium (or steel) sheets to cool.
Using the above method , I get about 0% to 2% of joints that are greyish (not quite reflowed) that need touch up, and that's about right for prototyping, as you may have to get in there with the hot air gun on several occasions later.

--------other things----
You may have a problem with solder balls, these can get stuck under the chip and give odd symptoms. Without a stencil, you invariably put too much paste on,  try putting the M328 on the PCB first, then just run the thinnest line on the pcb all the way around the chip so you are just touching all pins; if you have some , place a drop of liquid flux on top of the m328, it should run out to all the legs.

PCB's are sometimes a bit "greasy" , try wetting a paper towel with liquid flux, and give it a good rub all over, this will give the copper pads that little extra tack, so the solder paste is better behaved when you dispense it.

Prototyping kit components (and probably many hobbyists parts) don't have good storage conditions, moisture pickup can make for problems like "head in pillow" and pop-corning , you need to keep your parts bagged, warm and dry.

Hope that's of some help.
Bob.