[Wolves] [OT] SMT soldering

[Chris Ellis]

On Sat, Mar 19, 2022 at 8:30 PM James Dutton <james.dutton at gmail.com> wrote:
>
> On Sat, 19 Mar 2022 at 19:34, Chris Ellis via Wolves <wolves at mailman.lug.org.uk> wrote:
>>
>> Hi James
>>
>> On Sat, Mar 19, 2022 at 2:52 PM James Dutton via Wolves
>> <wolves at mailman.lug.org.uk> wrote:
>> >
>> > Hi,
>> >
>> > I know this is not Linux, but I know some people on this list play with electronics.
>> > I have a component I wish to replace on a PCB.
>> > NCP1250
>> > It is a TSOP−6 package about 1.5 x 3 mm. It is a chip with 6 pins.
>> >
>> > I have not done surface mount stuff before.
>> >
>> > As it is so small, what do people suggest as options for going about doing it?
>>
>> That should be pretty manageable with just a standard soldering iron.
>>
>> The general trick with SMT hand iron work is rather counter-intuitive,
>> basically be messy and then clean up.  Desoldering braid and some flux
>> paste is very essential.
>>
>> Essentially, tack one lead into place to align it.  Then drag a solder
>> ball over the pins, don't worry about bridges.  Then clean up with the
>> desoldering braid.  Volia a clean looking SMT joint.
>>
>> The other option is to buy a hot air SMD rework station, you should be
>> able to get one on Ebay/Amazon for ~£30ish (tend to go by the name
>> 858D).  You can use this to reflow just the specific parts.  Easy to
>> remove the part.
>> Then add solder to each pad with an iron, cover in flux and use the
>> hot air to reflow.
>
>
> Hi Chris,
>
> Thank you very much for the advice and tips.

No worries.

> I have not needed or used a desoldering braid before.
> I have a heat gun (blower) but I think it is more suitable for plumbing than PCBs, It has wide nozzles that would heat the whole PCB!!!
> I was wondering how to do just one chip at a time so thank you for pointing me towards the 858D.

Yeh, you need something a bit more targeted and controlled to do hot
air rework on single chips.

>
> I am going to purchase some braid and the 858D before I continue.
>
> I had already replaced the triac, because it was an obvious short, but it did not fix the PSU.
> I have then removed some components to test them, and the rest I have tested in place.
> As it turns out, all the components that have lost their labels are faulty, so it is turning out to be quite easy to fault find, if I was starting again from new, I would just check the ones with damaged labels!!!
> It seems that SMT resistors are failing to become open circuit, and all the transistors/triacs fail to become shorts.
> Is that a general rule for SMT resistors, or are they specific to this PCB ?

In general resistors tend to fail open circuit, since their failure
mechanism is typically too much current flowing through them and
causing them to melt / burn.

I'm guessing that the mosfets shorted which then caused far too much
current to flow through something like a gate drive resistor.
Typically when the power mosfets fail they go short circuit and take
out the current paths that surround them.  If your lucky they might go
explosively open circuit.

Especially in switch mode designs, where the mosfet either ends up
shorting to ground or shorting high(er) voltage to the low side.

>
> It has been fun finding out about SMT. For example, in my testing I found a few components that appeared to be shorts, but on further investigation I found out what a label of "0" means, :-)
> In a through hole, one just uses a bit of wire, but for SMT, one uses actual components that look like resistors.

Yup, zero ohm jumpers are very handy, either for being able to select
circuit options at assembly or for jumping over tracks without needing
an additional layer.

>
> In case it helps others. With mains voltage and SMPS (PSU) , the voltage can be 340 Volts. But as the front end of the circuit is a rectifier, one can do all the testing with a 12V DC or 48V DC supply and the PSU should still function fine, and one can test with non-dangerous voltages. It does not necessarily provide the full amps at the output when you have it working, but it is much safer to work with and diagnose.

Yup, thats a pretty neat way to test things.  Especially if you set a
current limit on the bench supply.

Chris