The use of SMD caps on hand soldered PCB's

[Rolo]

I'm want to work more with SMD components for my small scale hand soldered projects. Just installed KiCad and the learning curve has started 
Looking at other people's projects I noticed that they often use the conventional radial caps in projects where the rest is SMD. What could be the reason for this, is it that SMD caps are difficult to hand solder ? Or is it the price, availability, quality or brand choice ? I'm talking about small values, like 1uF to 220uF.

[BrianHG]

For home made 2 layer PCBs, and with power usually on the other side & these caps are for power decoupling, it is the convenience of having a wire from to bottom to solder join the 2 sides.  And electrolitics cost less for values above 10uf and are easy to salvage.  Well, at least it used to be this way for me in the beginning all those years ago.

[aylons]

1uF to 220uF are not small values for PCB capacitors, specially SMD.

In this range (specially above 10uF), electrolytic capacitors are more common (read cheaper), and PTH electrolytic are easier to sold than SMD ones. I don't see much of a preference for axial, bur Radial single-ended indeed seems to be more common, probably because it lends to a easier board layout.

EDIT: I misread radial for axial in the the first post, corrected answer to fit

[rstofer]

If you are using a DIY PCB, you probably need parts with leads to make up for the lack of plated thru holes.  Or you wind up adding jumper wires at the vias.

If you look at the Arduino Uno R3, you will see a couple of electrolytic capacitors that are surface mounted.  I haven't tried these.  I have just used SMD Tantalums.

I prefer SMD within the limits of my toaster oven.  I don't even try reflow soldering components to the back side.  If I have some decoupling caps on the back, I will hand solder them.  I have also had good luck with smaller boards by just cooking them on a hotplate.

[ovnr]

Hand-soldering SMD electrolytics is a pain in the ass, for what it's worth.

[Rolo]

To clarify, by home made I mean home designed, hand soldered, low volume. The PCB's are made by the well known suppliers. (OSHpark, Seeed, ect).

[sentry7]

Just installed KiCad and the learning curve has started 

Kind of off-topic, but have you used any other kind of EDA? I moved from ExpressPCB quite recently, and I must say that with all the flaws of KiCad, it's still 1000% better than what I was using before. KiCad has a massive online presence and community, and on their forums, you usually get a response within the next few minutes. Have fun with it!

[aylons]

To clarify, by home made I mean home designed, hand soldered, low volume. The PCB's are made by the well known suppliers. (OSHpark, Seeed, ect).

Still, the problem with SMD electrolytic is that the pads are almost entirely beneath the capacitor, making it very tedious to solder. Go PTH if doing electrolytic by hand.

Sent from my LG-D802 using Tapatalk

[mariush]

I think a lot of people are going with through hole because the footprint is much smaller, and because the capacitors usually have a higher lifetime rating.
For example, some SMT capacitors have 2000h @ 105c while the equivalent through hole capacitors (of smaller diameter but slightly taller) could be 5-6000h @ 105c

Also, they tend to be a bit more expensive compared to through hole capacitors.

I've been thinking lately of making a fully smt board and needed some capacitors for a couple of linear regulators ( i think i chose lm2941 but not 100% sure right now) that need a capacitor on the output with esr between 0.1ohm and 1 ohm , so using ceramics was not the best solution.  Sure, i could have used a ceramic resistor in series with a ceramic capacitor but it's ugly.

So anyway, was looking at something like 15uF-100uF and the cheapest smt capacitor i could find was a 22uF 16v at around 0.14$ in 1000pcs with 1000h@105c from Panasonic or Kemet (I don't remember now) while I could buy a 100uF 16v Rubycon through hole electrolytic for 0.09$ even in 100pcs quantity... and that series was rated for 3-5000h@105c
Even smt polymers from Panasonic (33uf/16v or something like that) with around 150mOhm esr were about 20 cents in quantity, and they were less tall than smt electrolytics.

// sure I only looked at Digikey and didn't look thoroughly. I'm sure somewhere I could have found a cheap smt electrolytic suitable for use with a linear regulator, but I think it's correct to say with through hole there's more options and often the parts are available in volume or it's easier to use multiple different capacitors (different diameters, heights, only lead pitch and lead thickness matters with through hole)

[Rolo]

Yes, I did a few small boards using FreePcb. Nice little program but support stopped a while ago. It's PCB only so no integration with a schematic drawing. I only make 2 layer, small boards, more like modules I use in various projects.
Now I'm working on a VFD display and took the challenge of making the filament PSU in KiCad.

Work in progress :


 

[DaJMasta]

Hand-soldering SMD electrolytics is a pain in the ass, for what it's worth.

I think this is a lot of the reason, but it's also not necessarily the worst.  If you slightly tin the pad (like tacking down an IC) and then you have access to the little bit of the tab coming out from the base, they are reasonable to hand solder, though through hole electros are definitely easier.  The real trouble comes with if you don't consider this when laying out your board and hide some access to the cap leads by putting it next to another cap, for example, or you populate your board out of order.  You can always do everything but the through hole stuff and then just do through hole parts, but when your SMD stuff has some height, like connectors or electrolytic caps, you have to pay much more attention to physical orientation, tight work spaces, and assembly order when hand soldering.

If it's only 2 layer, it's sometimes convenient to have the power on the backside, anyways, so the through hole pads save you some vias.

[mclemens1969]

My homemade stuff is all SMD. I think it's pretty easy to put on the SMD caps. I purposely use 0805 instead of 0603 parts (or smaller) just to keep it easy to solder. I probably should use all 1206 since I am not space constrained !

All you do is put a blob of solder on one of the pads, hold the cap down with tweezers and melt that end. Then finish the other end. Bathed in flux for great results.

[hans]

SMD electrolytics are very vulnerable components. If you drop your product it's likely the first to snap off, in addition to some questionable inductors (although these may be more vulnerable because they are the highest component..  when dropped they will take the full hit).
TH variants are much more securely mounted. Same goes for connectors and other components you can identify as vulnernable.

Hand soldering SMD caps is not that hard, given that you oversize the SMD pads sufficiently. Especially if you're working in a confined space that's nice for beginners. For reflow its not necessary.
In my experience, hand soldering e.g. 5x3.2mm (or 3.2x2.5mm) oscillator packages is more of a pain than electrolytics. Oversizing those pads for handsoldering is almost a must, unless you want to float your oscillators quite literally.

What your capabilities are at home made primarily is a function on your courage/experience I suppose. I do 0402 and QFN (with the via through PCB trick) at home with little tools, just a soldering iron with fine and normal tip, fine solder, perhaps some flux and some good eyes.
If I will do QFN on more projects I will probably get some hot air stuff.
If I want to BGA, I probably will look for some reflow oven solution. But at that point things like board houses capabilities and allocated budget for hobby to order at said board house is also an issue.

The photo by blueskull is a nice demonstration what can be done with a complete toolset though.

[ovnr]

Blueskull's board is quite nice, except for the two SO8/SO8W chips that haven't reflowed fully. Needs a tiny bit more time in the oven, methinks.

[Dubbie]

There are quite a few others that haven't reflowed properly. Some on the top right

[jbb]

SMD electrolytics do tend to be a pain.

SMD MultiLayer Ceramic Capacitors (MLCCs) are often really good.  But two warnings:

• I hear that package sizes 1206 and up have a real risk of cracking (may short circuit) if hand soldered, because one end gets much hotter than the other.
• For X type dielectrics (e.g. X5R, X7R), the capacitance drops off a LOT with DC bias .  Like 50% or worse, so check the datasheet.  This can cause distortion of AC waveforms, o X5R etc. caps aren't very good for filters. Moving to a large package usually helps.

For precision applications use C0G or NP0 dielectrics.  THey have lower temperature coefficient and much lower sensitivity to DC bias.

SMD Film capacitors are available, but are often more expensive than equivalent through-hole.

[james_s]

Familiarity probably. I routinely hand solder 0603 capacitors and other parts without issues. Haven't had any trouble with SMD electrolytics either assuming the pads protrude some. I hate them for other reasons though and avoid using them if at all possible. I've spent too much time cleaning up damaged boards full of leaky SMT electrolytics.

[obiwanjacobi]

Slightly bigger pads, practice SMD soldering on some scrap boards and use plenty of flux (and cleaner afterwards).
Should not be a problem.

[2c]

[flochlandla]

MLCCs, especially those with class-2-dielectric (and that is what we are talking about for buffer caps) are very sensitive to bias voltage.

A capacity is specified for 0V Bias and the higher your voltage, the lower the differential capacity (which is the effective value at your working point).

I really want to encourage you to take a look to the TKD homepage (https://product.tdk.com/en/search/capacitor/ceramic/mlcc/characteristic/). The search interface at the bottom of their page allows you to select a capacity range at a certain DC voltage.
Example #1: You select a 10µF input capacity for a USB-device (up to 5.25V). You find a model with 6V3 rating at size 0805 and think you are done but then - surprise, surprise - at 5V the capacity is only 2 (!) µF.
Example #2: Let's take a more extreme case: You need a 100uF model operating at 12V - there are no MLCCs on the market that provide such a high capacity at that voltage. The best thing you can get is a 22uF/16V with size 2220 (5.0x5.7mm²).

Tantalum caps and electrolytic caps (and also class-1-dielectrics) do not have that bias-problem. However, the price for their stability is either a high ESR for small package sizes (e.g. Tantalum 0603 with up to 6 Ohms ESR) or a bulky device. As long as you have no problems with the component height you may simply use Electrolytic caps as regulator input caps, etc.

Let's take the previous example: 100uF @ 12V as SMD-component - for good measures you select a capacitor with 16V rating and and ESR<500mOhm. So let's have a look to Digikey:
* MLCC 100uF @ 12V not available, so 5x22uF/16V in size (5 times 5.7x5.0x3.1mm³, <10mOhm ESR): 5x0,77€= 3.85€
* Aluminum Electrolytic Cap: 100uF/16V (dia=5.0mm, h=6.1mm, 360mOhm ESR): 0.12€
* Aluminum-Polymer 100uF/16V (dia=6.3mm, h=6.0mm, 25mOhm ESR): 0.19€
* Tantalum 100uF/16V (6.0x3.2x2.8mm³, 200mOhm ESR): 0.36€
* Tantalum-Polymer 100uF/16V (7.3x4.3x2.0mm³, 50mOhm ESR): 0.61€

As you can see MLCCs don't even make sense as they are horribly expensive (almost 30x more - you can't ignore that in mass production).
If the component height is no problem you should simply chosse the conventional Aluminum-Electrolytic cap.
If you require a lower ESR take the Aluminum-Polymer-Cap.
If you have small space, take the Tantalum. If you require small space+low ESR, take the Tantalum-Polymer.

You see, there are several parameters to consider when decicion for a certain capacitor is done.

[Rolo]

Thanks for all the info, it's very helpfull. I just discovered that KiCad has two type of SMD footprints, one standard and a with the text "handsolder" in it, that gives the bigger pads. Nice !

[mmagin]

For me, hand assembling something, if it's MLCC or solid tantalum, I'll go for SMD.  If it's electrolytic I'll go for radial leaded.  A lot of the bigger SMD electrolytics are just like a radial package except on a plastic carrier as far as I can tell.  There's not much obvious advantage except in automated assembly, and to get the same mechanical strength, you better solder it well.