SMD parts and breakage

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hello

In the process of transitioning many parts on my amp design to SMD and am polling the audience here about the quality of connections and any potential issues related to doing so.. I have in the last year finally caved in to using more and more SMD parts and am glad to say so for the time. As I have been changing the layouts and footprints, I am finding that there are some SMD capacitors that are actually flexible in that they resist the cracking and breakage in these areas. I am also looking to change my resistors into the Mini-MELF type and, in all this process, began to wonder if there is much of a risk in failure of any components beyond the through hole level of joint failures. My board is nearly 10" square and imagine that the larger the board, the more the issues. Often, when heating for SMD, if the entire board isn't kept the same temperature, it will warp and have issues and wonder what the experiences of others is here. The transistors are now pretty much required to be SMD, whereas everything else is still widely available TH. Worth it to SMD resistors and caps? Or ?

Thanks for input
S
 
I suggest you carefully read the data sheet for the parts you plan to use. Many surface mount capacitors will crack from the thermal shock when soldered with an iron. The cracked parts will cause trouble intermittently and can really be difficult to find. I have had good success using a toaster oven for reflowing a whole board.
 
Well it depends whether you making many boards or a one off assembly and how many components are on the board. In a manufacturing environment the board will go through a number of phases during the soldering process including pre-heating.

Some hobbyists use a toaster oven but you need you be careful about how much heat the board will be subjected to as this can cause the fibreglass PCB to delaminate. I've seen technicians with poor soldering skills whilst using hot air tools blister a PCB.

You can solder by hand with a good quality temperature controlled soldering iron using low melting point solder such as the types with 2% silver.
 
You would be surprised how much punishment SMD parts can take.

All the failures I have seen were attributed to human assembly techniques such as soldering one end of a ceramic capacitor and then pressing down on the other end whilst soldering, this will cause the ceramic to crack.

If this is a one off PCB then hand soldering will be preferable as solder paste can be a messy affair. However some SMD packages such as BGA (Ball Grid Array) you will have no choice but to use a solder paste.
 
MSSOP also leaves you little choice.

I configured a Black and Decker toaster oven into a nice reflow oven -- Controleo 3
Controleo3 DIY reflow oven

The issue I have with MSSOP is the need of radiating heat off the device from it's "belly" -- see the Texas Instruments data sheet for the THS3121 cfb amplifier. I've tried using plated through hole vias with a custom aperture and still have failures.

Frankly, I wind up using through-hole electrolytic capacitors -- probably because I have a huge stock --
 
Thanks Nigel
I've been doing stencils and paste and have been really enjoying it so far. I have a box of waste pcbs from all the practice and would consider myself perhaps average by now being that it's really only been about a year into SMD for me. A year ago I was th as much as possible and now SMD all the way if possible.
 
Thanks for all the above responses
I think my main concern is that I will most certainly use solder paste and heat the entire board to do the job. I just know that every component has a different temperature coefficient of expansion and am mostly concerned that the parts will be of a difference to the entirety of the PCB. Just have to try it out and see. May not reveal anything for years.
 
Thanks for all the above responses
I think my main concern is that I will most certainly use solder paste and heat the entire board to do the job. I just know that every component has a different temperature coefficient of expansion and am mostly concerned that the parts will be of a difference to the entirety of the PCB. Just have to try it out and see. May not reveal anything for years.

You can buy SMD ovens that do the proper profile of heating and cooling down.
 
Had a go at soldering a 64 pin TQFP package today.
I put some solder down using a syringe but it went down a bit thick.
I then used a heat gun to melt the solder.
Lots of big blobs everywhere !
I had to wick off the excess solder.
Then I noticed one of the pins was only half way on the pad.
I blamed the chip then blamed pcb maker.
Then when I checked part outline it was out by about half a pin !
I hadn't quite got my measurements right from datasheet.
I did measure it against a print out but it didn't see far out.
Obviously far enough out to mess things up on the real pcb.

So fixed it and getting new pcb's made, luckily only $5 from JLCPCB.
I was already getting another pcb made so p+p was virtually free.
 
Fun stuff aye
Usually shipping is the most expensive part of jlcpcb. I try to order multiple pcbs each time. I've also noticed that for small parts, using jlcpcb, the pads spacings can be narrow, so I usually make the pads smaller than spec on the layout so they come out correctly in real life. Any experience with stencils? My last order I set different apertures for the stencil vs the actual pad and apparently they didn't use that paste layer I provided. Not sure how to deal with that.

S
 
I use drag soldering for fine pitch parts. The only time I use paste is for BGA and LGA parts, where you can’t access the join becuase it’s under the chip.

Here’s a quick YouTube guide to drag soldering:

YouTube

The key to successful drag soldering is good temperature control (I use a Metcal HF iron), plus heaps of flux. Never rely on the flux that’s in the solder wire.

Much better to use drag soldering than muck around with paste and stencils, and then watch components tombstone or otherwise shift out of alignment during reflow.
 
Also I once built a stack of data acquisition boards were we’d stuffed up the footprint for an ADC. From memory we’d used a footprint with 0.65mm pad pitch rather than 0.635mm (0.025”). The result was that if you lined up the pins at one end, at the other end it was about half a pin out. We were only making half a dozen protos, so I just lined it up in the middle and made it work. We definitely fixed it on the production run.
 
Care is needed if you try to solder TQFP packages.
If not using a stencil it is easy to get too much solder on the pins.
I tack down one corner of the TQFP and melt the solder until the package is exactly square then I tack opposite corner.
I use a SMD solder paste syringe and put on a tiny line of solder along the pins.
I then gently slide the soldering iron along to melt the solder.
I then wick off the excess solder.
Troublesome small shorts can be got rid of wit ha quick dab of the iron.
The trick is to put down flux, then lay the wick along the pads then move the soldering along the wick to draw up the solder.
If you move the wick along the pins it can bend the pins if you are too rough.

I designed a TQFP package adaptor pcb with through hole pads to connect to a motherboard. The advantage of this was I could probe the through hole pads for shorts between adjacent pins.
 
I have recently made the transition in the other direction: from SMD to through-hole :) I started playing with electronics about 8 years ago and realized that SMD 0603 passives are doable but frustrating. For a while a did 0603 then moved to 0805 and stayed in 0805 territory for years. Now I do things with a mixture of 1206 and through-hole. I did 100 pin 0.5mm pitched TQFP a few times and I was surprised how easy it is. But THT is always easier and I just prefer the looks of boards with THT components. My sister said it reminds her of a miniaturized industrial complex. SMD, on the other hand looks like a colony of ants :) I only recently started caring about the looks of things. Guess I am changing.

If size is an issue and I have to make something as small as possible, I might go back to full SMD. But right now I am staying in THT land. Of course, chips that are only available in SMD will stay.

And I never tried QFN packages. I am almost afraid of them, but I bought a hot-air station a few weeks ago, and I might try something with TQFN. A sort of a challenge, I guess.
 
On the QFN note

I made a UPS power supply for the Raspberry Pi to control the power cycling and to prevent SD corruption using both MLF-8 and UQFN-20 parts. It definitely took some practice to get them to solder right the first times. I suppose it could be possible to do them without a stencil or the like, but, what I discovered is that you must get the correct scraping of the solder paste from the stencil in order to not have bridging of the pins. It honestly took many a go around to get good enough at it to where now, I can make and bake a dozen in about a half hour. Reliably. If you use a hot plate and a lower temp solder paste, you can poke the parts lightly while they are at temperature and they will wick their positions. Also, in my boards, there's enough pad showing to get the solder iron in there to soak up any additional solder that may be bridging. Interestingly, I've only had one bridging under the part, whereas the other problematic practices usually bridged on the sides of the parts. I've got it figured out now. Somewhere along the line, I knew I was going through the growing pains of SMD and paid the prices, now feel quite confident.

My latest and perhaps last rendition of my 'Ghost' amplifier will be mostly SMD and should be finalized and made in a few months. I'll get some good pics and dispense any info and results that I've learned along it's way back here.. I'm very satisfied with my results in my prototype and confident going forward with SMD parts.

I also moved up to 0805 for ease of reading, handling and placement :)

S
 
Last edited:
I soldered another 64 pin TQFP package today.

First job is to get the chip aligned to the pads on the pcb.
Given they are on 0.5mm pitch its not easy.
I got it as close as I could then solder tacked one corner.
I checked it with my magnifying glass and it seemed about right.
I tried to use less solder this time so there is less to wick off.
I still had some quite big blobs of solder but they mostly wicked off ok.
A couple of solder shorts were very hard to deal with.
So I looked up the circuit and found these were on unused pins so just left them.
A bit unprofessional but once I installed the TQFP pcb onto the motherboard pcb it worked fine.
On this TQFP pcb I made the 0805 pads a little larger so easier to solder by hand.
I had a little trouble soldering the small discrete SMD as they wouldn't tack very well.
So I changed my soldering iron bit and suddenly it was much easier.
The old bit was a little tarnished so was not holding solder very well.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.