Surface Finish on PCB and Manufacturing

[Isaac]

Hi, guys,

I'm designing a 2 layer PCB and I'm requesting quotes on various combinations of parameters and all that to get the best setup for the lowest cost based on my aplication. My boards will have QFPs as the most critical components in respect to automated soldering process.

I know that the surface finish on the PCB pads and exposed copper have a effect on the reliability of the final result of the automated soldering process, but how much will it be critical to choose bare copper over HASL, or HASL over ENIG for a given application? I mean, for all Through Hole parts that's not a real benefit on choosing ENIG over HASL and, if the boards are assembled quickly and the PCBs don't stay still on a shelf or box just bulding up oxidization on the exposed pads, that's not a real benefit on choosing bare copper over HASL.

One can imagine how cost effective would be to save a few cents on the bare PCBs just based on decisions like this...

Well, I think it's a great subject to be discussed and can became some kind of reference for other users...

[georges80]

All my designs in the past decade+ have been SMT.

All boards have been HASL for SMT devices that have leads out of their sides and I've never had an issue hand assembling or when machine assembled.

With BGA (high count parts) or lots of DFN/QFN the assembly houses I've used will typically run ENIG since the pads are very 'flat'.

HASL is cheap, ENIG obviously adds some cost due to the gold. I've had a few boards that I hand assemble (solder paste/oven etc) done with ENIG and it makes it a easier to reliably solder QFN/DFN parts without needing to touch up.

Personally, I would never consider bare copper for any board (even prototypes).

So, I consider HASL or ENIG as the only options (for my applications). Choice between the two is then based on whether I have BGA/DFN/QFN parts in the design and if so then ENIG is my preference.

cheers,
george.

[calexanian]

Gold actually presents some issues when soldering. If a pad is to be soldered, do not plate it with gold has been common place. The solder takes longer to wet to the gold plating and can be problematic. We have never had any serious issue with it, but we are not up in the many thousands at a time quantity that some others may be in. I will say this with certainty. On our TQFP packages the standard plated pads flow way better than the gold ones. Less flux is required also.

[calexanian]

Adding to that. Many companies offer selective gold plating for contacts and other pads that will not be soldered later. We gold plate, plug, and tent our vias on some of our boards for low leakage.

[Isaac]

Gold actually presents some issues when soldering. If a pad is to be soldered, do not plate it with gold has been common place. The solder takes longer to wet to the gold plating and can be problematic. We have never had any serious issue with it, but we are not up in the many thousands at a time quantity that some others may be in. I will say this with certainty. On our TQFP packages the standard plated pads flow way better than the gold ones. Less flux is required also.

This shows that, in some cases, ENIG is just overkill and a waste of money. Making the right decisions for lowering costs is part of the work. Sometimes, a really importante one.

All my designs in the past decade+ have been SMT.

All boards have been HASL for SMT devices that have leads out of their sides and I've never had an issue hand assembling or when machine assembled.

With BGA (high count parts) or lots of DFN/QFN the assembly houses I've used will typically run ENIG since the pads are very 'flat'.

HASL is cheap, ENIG obviously adds some cost due to the gold. I've had a few boards that I hand assemble (solder paste/oven etc) done with ENIG and it makes it a easier to reliably solder QFN/DFN parts without needing to touch up.

Personally, I would never consider bare copper for any board (even prototypes).

So, I consider HASL or ENIG as the only options (for my applications). Choice between the two is then based on whether I have BGA/DFN/QFN parts in the design and if so then ENIG is my preference.

cheers,
george.

Yep, bare copper is a dumb decision, sometimes. But I can imagine that it can be used in some applications with good results. And I think it really depends on the manufacturing processes and time to assemble and solder the PCB.

[Alphatronique]

HI

Personally all my board was EING ,yup cost bit more  ,but it trouble free ,and may stocked for long time before use

also please note that some PCB shop begin to use cheaper allow for HASL that contain nickel (Asia)
and nickel may cause issue whit SAC-305 and some type of flux

as for save cost ,PCB was one factor but one that often under looked was to work closely whit your assembler
you may this way cut assembly cost by 2 or even 3  and most of time assembly cost was the dominant price

running myself a PCB assembly shop i see it every day

Best regard
Marc L.

[Ton]

As already mentioned gold in a solder joint is not a good thing, some of the gold from the pad gets dissolved into the solder during liquidus.
This mixing of gold into the solder makes et more brittle, there by increasing the probability for broken solder joints

for a typical ENIG surface the gold layer is between 4um and 6 um thick, and the amount of gold dissolved into the solder joint is very small, this does not really weaken the joint much, but if you add gold to just around 10 - 12 um then you have enough gold in the mixture to have a severely weakened solder joint. (luckily gold is expensive and PCB manufactures keep the layers as thin as possible)

so don't solder on galvanic gold this kind of pads are for edge connectors and spring connections only.

I have been involved with a compagny where running 5-20 million PCB's trough the solder oven was the norm. and they always used OSP (organic surface protection) this was the only thing that could keep the BGA's from popping of the PCB when some stress was introduced into the PCB

HASL was not an option due to all the BGA components, some even stacked - big problems with co-planarity of the HASL surface

so for general Electronics stuff, where the chassis of the equipment can keep stress away from the PCB then ENIG is a reasonable bet, with HASL also ok if you don't have to place parts with co-planarity issues.

If you can handle OSP, then go for it, but be aware of the demands regarding shelf life and humidity.

for production runs, don't go to cheap on the PCB manufacturer field failures are much more expensive 

[Alphatronique]

Hi

interesting last time i try OSP got lot of issue  ,but that long time ago

main issue was that if solder not cover all of the copper up to the solder mask
it let exposed copper that naturally fail later 
since lead-free wetting was less that whit Sn/Pb it not help ...

also 2 side board need 2 reflow run but OSP go away after fist one , need to assemble quick and whit white glove

so need really huge production for handle this

as for EING  i have board whit big FPGA on it that qualified vibration test and MTBF > 20 year
that board was assembled for GE and it bolt it into huge power transformer that vibrate at 60/50Hz  100% duty cycle

[Ton]

Well they did use conformal coating quite a bit, so that could be the reason, and yes

production line setup was typical:  SMD paste, SMD placement, solder oven, flipping, SMD paste, SMD placement, solder oven . then conformal coating  and of to assembly.

anyway what fits a portable device company, might not fit other places.

I just remember that there was quite a lot of experiments ongoing to prevent the BGA's from developing broken solder joints, and the OSP method won.

[Alphatronique]

Hi

yup you right what work for one not work for all 
but totally agree that for cell phone ,and most consumer mass market OSP may very attractive
but for outdoor industrial  no sure ...

that wly most important was to talk for PCB and assembly FAB
and not only talk to sale rep since it normally a machine to said "yes no problem" to every question ;-)
but talk to staff it knot how made it live easier  ,save time and finally made you save money

[Isaac]

Yes, I agree with that. Making a good relationship with your assembler would help finding the best solution for you and, as a result, bring your costs down.

[phil]

Just be careful if you plan on selling kits or have the user solder on extra parts: HAL is usually lead free and can become a problem when hobbyists solder it with 60/40
In that case, better pay the few cents extra to get a gold finish. Better to have some gold in a 60/40 joint than the lead free crap.

Phil

[Araho]

One important thing to remember when it comes to deciding your finish is also the shelf life of the PCB before components are mounted and soldered. If you're going straight from PCB production to assembly then ENIG probably won't make a difference, but if they are stored for a few months (e.g. a kit or other use) prior to assembly and soldering, then the ENIG will do a much better job of protecting from oxidation and other annoying effects.

I can probably dig up the paper where I read about this if you want it.

[Isaac]

I can probably dig up the paper where I read about this if you want it.

Please, do.

[free_electron]

ENIG is used if your boards need along shelf life before they will be assembled.
if you will go fast from board to assembly you could do tinflash or even OSP.
but those boards have a very shirt shelf life and need vacu-pack !
osp degrades with moisture and tinflash .. well it corrodes like the pest if you touch it with hands.
tinflash has a shelf life of max 2 weeks.
OSP in dry-pack can do a month. OSP is cheapest.

[calexanian]

ENIG is used if your boards need along shelf life before they will be assembled.
if you will go fast from board to assembly you could do tinflash or even OSP.
but those boards have a very shirt shelf life and need vacu-pack !
osp degrades with moisture and tinflash .. well it corrodes like the pest if you touch it with hands.
tinflash has a shelf life of max 2 weeks.
OSP in dry-pack can do a month. OSP is cheapest.

I am glad we are in Fresno. Nice and Dry.. We have some boards what are just tin plated and have had them sitting on the shelf for years. They assemble just fine. But then again we use RA flux which will pretty much clean anything. Actually today I ran a small batch of 8 boards with Kester 331 water wash organic core. It flows a bit slower but a water bath is sure friendlier than defluxing the RA flux. The smell of the 331 sucks though. Organic core has always been nasty. We tried going no clean but our boards go out in an agricultural setting and the fumes from the liquid chemicals find their way in and react with the flux and you wind up having to deflux anyways. Since we started plugging and tenting our vias however we have had much less leakage problems. We used to use Criterion Circuits and then Sigma Circuits and those boards were fantastic. Like perfect every time. Now we use FXM marked boards. Some Chinese company.. Not so good. They work, but you get gaps in the silk screen and hand touch ups in the solder masks, things like that every now and then. I looked into switching to Logan, but we are too low rent for them nowadays. For a while we were using King Circuits but they started farming out stuff out and it was just not working out.