First time soldering a SMD, tips/tricks?

[aphex]

So i've now got a few amps under my belt, i'm moving on to more involved models. The parts should be arriving today for a Mini^3, which requires soldering of two SMD op amps.

Any tips and tricks on how to best accomplish this?

 

[DisgruntledVirus]

use a soldering iron

 

[SphinxnihpS]

flux

 

[slackwarelinux]

dead bug mount through hole parts on the smd pads to save time.
trust me, it will be easier.

 

[Demon-Xanth]

I learned at my last job interview about 10 years ago. Literally. Just got "certified" recently.

Put some solder on one pad
Heat it up and place the part on the pads, lining it up
remove the heat and wait a second (letting go of the part too early may cause tombstoning, and you need to heat it up again to put the part back down)
Apply solder to the other end
Clean up the first side

Once you get used to it, unless you're doing really fine pitched things (ie: QFP-100 and smaller) parts SMT is easier. Just take your time.

 

[aphex]

Thanks guys. I think my biggest problem right now might be my iron, I have a Weller SP23 25-watt iron, but since its their basic kit there aren't all that many tips available for it. Need to get a digital soldering station one of these days.

 

[Modelworks]

Besides the tack one corner method you can also use superglue to glue the part in place then solder.
And get a good magnifying light/lens setup

You can also just solder all the pins , covering them all with solder so there is one blob over all the pins on one side, then use desoldering braid to remove the extra.

 

[MrDudeMan]

Originally posted by: Modelworks
Besides the tack one corner method you can also use superglue to glue the part in place then solder.
And get a good magnifying light/lens setup

You can also just solder all the pins , covering them all with solder so there is one blob over all the pins on one side, then use desoldering braid to remove the extra.

This is some of the worst advice possible. There is no reason to use super glue and it's also a really bad idea if you are working with anything involving high frequency signals or sensitivity to parasitic capacitance. Blobbing solder on pins allows excessive heat to wick inside the part. You want as little heat to go through the pin as possible to protect the part. The pins are thin so they have quite a bit of thermal inertia compared to the copper traces on the board, but blobbing solder across pins negates this and will affect the long term reliability of the part.

I worked as a soldering tech in college for over a year doing surface mount soldering daily. Get a good pair of right angle tweezers, fine pitch solder, and don't use a really small/sharp tip. The only things you need to use flux to solder are surface mount ICs with several pins - QFP package or anything similar. For resistors, capacitors, etc. you need to just hold it in place and tack a side down.

Keep in mind that solder really isn't all that conductive compared to metal, meaning you don't want to make connections with solder. You want metal on metal contact and to only have solder there to keep it in place. In terms of heat, less is more. If you are working with really sensitive components, adding too much heat can change the properties of the silicon and can really change the operation of some devices. You don't have to be super paranoid, but keep in mind that almost everything you will solder in the surface mount world can and will break with too much heat for too long.

If you are using flux, you need to clean it off with a flux removing compound or it will oxidize and weaken your solder joints. Typically I will solder everything that is in a waterproof package (most things are), spray on the flux remover, and then wash the board in warm soapy water. At that point, I will solder non-waterproof components like electrolytic capacitors and remove the flux from those as well. Before I turn it on, I blow compressed air over it and make sure it's all out from under the pads of the ICs. This is a pretty typical sequence of events for surface mount soldering.

 

[Jeff7]

But aren't professional circuits made with wave-soldering? Glue parts down, then flood the board's surface with liquid solder. Seems to work pretty well.

 

[LTC8K6]

I successfully soldered a new Q9 on to a Radeon 9800. 8 tiny little pads.

You need a low watt soldering iron with a very small tip, a very steady hand, patience, and a lot of light to see well.

 

[ra1nman]

Originally posted by: Jeff7
But aren't professional circuits made with wave-soldering? Glue parts down, then flood the board's surface with liquid solder. Seems to work pretty well.

I work in contract manufacturing and they use solder paste (which works like the how it sounds). It's printed on to the pcb board while a smt machine places the part. When the board goes thru the oven, the paste melts and holds the part. The wave machine is primarily used for larger thru-hole parts like connectors and headers.

 

[PottedMeat]

Originally posted by: ra1nman

Originally posted by: Jeff7
But aren't professional circuits made with wave-soldering? Glue parts down, then flood the board's surface with liquid solder. Seems to work pretty well.

I work in contract manufacturing and they use solder paste (which works like the how it sounds). It's printed on to the pcb board while a smt machine places the part. When the board goes thru the oven, the paste melts and holds the part. The wave machine is primarily used for larger thru-hole parts like connectors and headers.

Is there a mask for each layer, then they squeegee the paste on? Some older PCBs have the SMT components stuck on with a tiny glob of glue - do they do that anymore?

 

[ra1nman]

Originally posted by: PottedMeat

Originally posted by: ra1nman

Originally posted by: Jeff7
But aren't professional circuits made with wave-soldering? Glue parts down, then flood the board's surface with liquid solder. Seems to work pretty well.

I work in contract manufacturing and they use solder paste (which works like the how it sounds). It's printed on to the pcb board while a smt machine places the part. When the board goes thru the oven, the paste melts and holds the part. The wave machine is primarily used for larger thru-hole parts like connectors and headers.

Is there a mask for each layer, then they squeegee the paste on? Some older PCBs have the SMT components stuck on with a tiny glob of glue - do they do that anymore?

Very rarely we see boards that have smt parts glued on. Usually those customers have boards found in environments were there is alot of vibration and shock (typically aeronautics or military vehicles).

We don't produce our own boards, we receive them from a board house which manufactures them to our specifications. The boards are anywhere from 5 to 35 layers thick with each layer having it's own set of traces. Kinda like you mentioned, a machine applies paste on to a screen with holes cut in specific locations. The excess is then squeegeed off leaving the locations that need solder behind.

 

[Modelworks]

Originally posted by: MrDudeMan

Originally posted by: Modelworks
Besides the tack one corner method you can also use superglue to glue the part in place then solder.
And get a good magnifying light/lens setup

You can also just solder all the pins , covering them all with solder so there is one blob over all the pins on one side, then use desoldering braid to remove the extra.

This is some of the worst advice possible. There is no reason to use super glue and it's also a really bad idea if you are working with anything involving high frequency signals or sensitivity to parasitic capacitance. Blobbing solder on pins allows excessive heat to wick inside the part. You want as little heat to go through the pin as possible to protect the part. The pins are thin so they have quite a bit of thermal inertia compared to the copper traces on the board, but blobbing solder across pins negates this and will affect the long term reliability of the part.

I worked as a soldering tech in college for over a year doing surface mount soldering daily.

Maybe you should go back to school and learn more about the actual field you are talking about. I didn't work as a tech in college, I worked as an engineer for Cray, GE and Sandia.
We used the method I described on a regular basis.

Typical college kid response, takes a class and thinks he knows everything.
Come back when you have a couple decades real world experience.

 

[Modelworks]

Originally posted by: ra1nman

Originally posted by: PottedMeat

Originally posted by: ra1nman

Originally posted by: Jeff7
But aren't professional circuits made with wave-soldering? Glue parts down, then flood the board's surface with liquid solder. Seems to work pretty well.

I work in contract manufacturing and they use solder paste (which works like the how it sounds). It's printed on to the pcb board while a smt machine places the part. When the board goes thru the oven, the paste melts and holds the part. The wave machine is primarily used for larger thru-hole parts like connectors and headers.

Is there a mask for each layer, then they squeegee the paste on? Some older PCBs have the SMT components stuck on with a tiny glob of glue - do they do that anymore?

Very rarely we see boards that have smt parts glued on. Usually those customers have boards found in environments were there is alot of vibration and shock (typically aeronautics or military vehicles).

We don't produce our own boards, we receive them from a board house which manufactures them to our specifications. The boards are anywhere from 5 to 35 layers thick with each layer having it's own set of traces. Kinda like you mentioned, a machine applies paste on to a screen with holes cut in specific locations. The excess is then squeegeed off leaving the locations that need solder behind.

I remember a friend making a home version of the machine you describe, he used a silk screening like process to apply the paste.

 

[MrDudeMan]

Originally posted by: Modelworks

Originally posted by: MrDudeMan

Originally posted by: Modelworks
Besides the tack one corner method you can also use superglue to glue the part in place then solder.
And get a good magnifying light/lens setup

You can also just solder all the pins , covering them all with solder so there is one blob over all the pins on one side, then use desoldering braid to remove the extra.

This is some of the worst advice possible. There is no reason to use super glue and it's also a really bad idea if you are working with anything involving high frequency signals or sensitivity to parasitic capacitance. Blobbing solder on pins allows excessive heat to wick inside the part. You want as little heat to go through the pin as possible to protect the part. The pins are thin so they have quite a bit of thermal inertia compared to the copper traces on the board, but blobbing solder across pins negates this and will affect the long term reliability of the part.

I worked as a soldering tech in college for over a year doing surface mount soldering daily.

Maybe you should go back to school and learn more about the actual field you are talking about. I didn't work as a tech in college, I worked as an engineer for Cray, GE and Sandia.
We used the method I described on a regular basis.

Typical college kid response, takes a class and thinks he knows everything.
Come back when you have a couple decades real world experience.

I'm also an engineer, but thanks. I provided two valid reasons why you shouldn't do what you were talking about, but what did you offer other than rude, sarcastic responses? Also, being an engineer offers no credibility to this argument as anyone can learn to solder. Note my post where I didn't use that as validation. The companies you worked for mean nothing either. I work at Intel. Big deal?

I've been soldering for over 10 years, but only did it professionally for a year. Just because I didn't do it at work the entire time doesn't mean I didn't use over 3 miles of solder on my own projects, most of which are surface mount.

Maybe you just didn't work in an industry with highly sensitive parts, but I was working with drivers for piezoelectric motors going on satellites and any IC soldered out of spec had to be thrown out. At over $50 a pop, you are allowed few mistakes and considering the motors had to move in steps of < 10nm, it was quite obvious when one wasn't working. Glue or residual flux on the board changed the parameters enough to be noticed. Sorry about your "dumbass college kid argument" not working, but maybe you could try again with some actual information instead of only your ego.

P.S. Your advice was bad because using glue can and will effect certain circuits. Offering that as general advice is piss poor at best, and I'd think someone with "decades" of experience would know that.

 

[MrDudeMan]

Originally posted by: LTC8K6
I successfully soldered a new Q9 on to a Radeon 9800. 8 tiny little pads.

You need a low watt soldering iron with a very small tip, a very steady hand, patience, and a lot of light to see well.

A very small tip makes it harder, not easier. You should be able to do it with a thicker tip using less heat and less time. If you can get it to flow across the pins of an IC, it looks exactly like the part was reflowed and this is very hard to do with a small tip.

 

[Modelworks]

Originally posted by: MrDudeMan
o.

P.S. Your advice was bad because using glue can and will effect certain circuits. Offering that as general advice is piss poor at best, and I'd think someone with "decades" of experience would know that.

Your idea that a person soldering a opamp is going to effect the circuit because he uses glue, is hilarious. Or that heating the pins for longer than an instant will destroy it is funny as well.
These parts can take 200C for over 20 seconds before any damage begins to happen.
Look up reflow oven and the temps involved.

I've seen many an engineer that can quote from the books and then starts scratching their head when the real world doesn't do it that way.

It is because you might have to sterilize a room when you are a doctor doing an operation , that doesn't mean you need to when you are placing a bandaid on a cut.

As for areas I worked in, it was prototyping on things like interconnects on supercomputers running at gigabit speeds long before half the world even knew of the term.

The guy is building a small portable amp, not the next GPS system for missile defense.
I suppose you would have him purchase a hot air station as well ?

 

[aphex]

If this isn't the ultimate nerd fight, I don't know what is. Were talking about soldering and people get all hot and bothered lol

Anyway, so the electronic parts store I usually used is closed until monday, so all I have at my disposal this weekend (which I'd like to get to work on it if possible) is what I have now and rat shack. I have a 25w Weller SP23 Iron and some 63/37 solder, but its a bit thick (0.32). I was over at rat shack earlier and saw they had 62/36/2 solder at a very tiny .015" diameter.

 

[Modelworks]

Originally posted by: aphex
If this isn't the ultimate nerd fight, I don't know what is. Were talking about soldering and people get all hot and bothered lol

Anyway, so the electronic parts store I usually used is closed until monday, so all I have at my disposal this weekend (which I'd like to get to work on it if possible) is what I have now and rat shack. I have a 25w Weller SP23 Iron and some 63/37 solder, but its a bit thick (0.32). I was over at rat shack earlier and saw they had 62/36/2 solder at a very tiny .015" diameter.

LOL, yeah it is.
I just get upset when people try to make something more complicated than it need be.

What you have will work fine.
Use the fine solder if you like, I actually prefer the larger solder since it contains more flux making the solder flow better.
I have seen people do fine work with a lot less

 

[randay]

get the digital station. it helps to know exactly how hot you are.
and as with many things, practice makes perfect.

oh also magifying glasses help!

disclaimer: i am not a pro-soldering-god, please dont flame me.

 

[bobsmith1492]

Wow what a thread.

It's not hard; iron+solder+tweezers is all you need; flux is good if you mess up at some point.

My procedure:

1. Melt a tiny amount of solder onto a corner pin.
2. Place the part properly aligned holding it with tweezers.
3. Touch the soldered pad with the iron.
4. Release tweezers; solder the other pads (opposite corner first - realign while melted if needed).

The worst thing I've had to solder was a 0.4mm pitch 64-pin ethernet chip. The pins are like hairs; the hardest part was aligning it though. A wave tip does wonders...

 

[PottedMeat]

Originally posted by: bobsmith1492
Wow what a thread.

It's not hard; iron+solder+tweezers is all you need; flux is good if you mess up at some point.

My procedure:

1. Melt a tiny amount of solder onto a corner pin.
2. Place the part properly aligned holding it with tweezers.
3. Touch the soldered pad with the iron.
4. Release tweezers; solder the other pads (opposite corner first - realign while melted if needed).

The worst thing I've had to solder was a 0.4mm pitch 64-pin ethernet chip. The pins are like hairs; the hardest part was aligning it though. A wave tip does wonders...

I usually do that - one difference - I tin some of the pads instead of the pin. If the solder blobs, suck up excess with braid. All with 2 $10 40W hakko pencil irons. I hate how TI/Maxim/etc. are moving towards those leadless chip scale packages / bga for their newest most interesting stuff.

 

[bobsmith1492]

Originally posted by: PottedMeat

Originally posted by: bobsmith1492
Wow what a thread.

It's not hard; iron+solder+tweezers is all you need; flux is good if you mess up at some point.

My procedure:

1. Melt a tiny amount of solder onto a corner pin.
2. Place the part properly aligned holding it with tweezers.
3. Touch the soldered pad with the iron.
4. Release tweezers; solder the other pads (opposite corner first - realign while melted if needed).

The worst thing I've had to solder was a 0.4mm pitch 64-pin ethernet chip. The pins are like hairs; the hardest part was aligning it though. A wave tip does wonders...

I usually do that - one difference - I tin some of the pads instead of the pin. If the solder blobs, suck up excess with braid. All with 2 $10 40W hakko pencil irons. I hate how TI/Maxim/etc. are moving towards those leadless chip scale packages / bga for their newest most interesting stuff.

Whoops, I meant pad, not pin.

BGAs get tricky... the tech at work did some using a hot-air gun but I bet those are expensive!

 

[chorb]

I just got to see a BGA rework station in action today, crazy bit of machinery.