SMT vs Through Hole: Where Each One Still Wins
SMT and through hole are not rivals. They are two ways to hold a part to a board. One is fast and dense, the other is strong and simple.
Every comparison page ends with a verdict. Surface mount wins, or through hole is still useful. That framing is why buyers keep asking the question.
The useful answer is smaller. Almost no production board is one or the other. It is a surface mount board with a handful of parts that had to go through a hole.
So the real question is not which process is better. It is which parts on this board need which one, and what the second process costs.
The Short Answer
Surface mount puts the part on pads and solders it with paste and heat. Through hole puts the leads through drilled holes and solders them on the far side, by wave or by hand.
Surface mount wins on density, on speed, on parts per square centimetre, and on small signal parts. Through hole wins wherever a part is pushed, pulled, plugged, unplugged or heavy.
That second list is short and stubborn. It covers the USB socket, the terminal block, the relay, the transformer, the big electrolytic and the connector that takes the cable.
Those parts exist on a board that is otherwise fully surface mount. So the line runs reflow first, then a second soldering step for the tall parts.
The decision, then, is per part and per board. Not per factory, and not per fashion.
What Each Process Physically Does
Surface mount starts with a stencil. Paste is printed onto the pads, the machine drops the parts, and the oven melts the paste in one pass.
There is no drilling for the parts. That single fact is behind most of the advantages listed below, because holes cost money and take space on every layer they pass through.
The process itself is worth reading properly once. The five steps that build a surface mount board run from the print to the test. This article will not repeat them.
Through hole is older and simpler to describe. A drill makes a hole, the hole is plated, the lead goes in, and solder joins the lead to the pad on the far side.
Doing that well is not simple at all. Drilling is a real share of the cost of a through-hole board, and every hole blocks the routing space on the layers it passes through.
That is the trade in one line. Surface mount buys density by giving up the mechanical grip of a lead through a hole. Through hole buys grip by spending board area and drill time.
Both end up as the same object. The assembly that comes off the line carries joints of both kinds, and the buyer’s drawing decides which is which.
Where Through Hole Still Wins
Start with the part that a person touches. A connector that takes a cable is pushed in and pulled out for the life of the product.
A surface mount joint under that load is a small pad of solder in peel. A lead through a hole is a pin in a socket of plated metal.
The reference entry on the topic is blunt about it. The description of how a lead is anchored puts connectors, relays and heavy parts on the through-hole side for exactly this reason.
The same logic covers anything with mass. A transformer, a big inductor or a large electrolytic capacitor carries its own weight in a drop test.
Then there is heat. A metal lead is a path. A power part in a through-hole package can dump some of its heat into the board through its legs. That is why a TO-220 still sits on a heatsink with its pins in holes.
Next is the socket. Anything that may need replacing or upgrading can sit in a socket only if it has leads, and that rules out most surface mount packages.
Last, and easiest to miss, is alignment. A part with leads cannot end up slightly crooked. Once the leads are through the holes, the body is fixed, and no amount of vibration moves it sideways.
Surface mount has the opposite property. A part can be placed slightly off and pulled straight by surface tension during reflow. That self-alignment is a gift for fine pitch and a small risk on a heavy part.
Where Surface Mount Wins
The speed is not a small difference. Placement machines are quoted at more than 136,000 components an hour, and no insertion machine is close to that.
Density follows. Parts can sit on both faces of the board, at pitches no leaded part could reach, with connections on all four sides of a package.
Routing follows density. A hole that passes through every layer blocks every layer. Remove the hole and the inner layers get their space back.
Small parts are also cheaper parts. Many surface mount components cost less than their leaded equivalents, and they are stocked wider.
Then the electrical side, which decides some designs outright. A lead adds inductance, and a smaller package radiates less. Above a certain speed, the surface mount version is not a preference but a requirement.
Shock and vibration also favour surface mount, for a physical reason. The parts are lighter and sit closer to the board, so there is less to swing.
What surface mount does not give is grip. The joint is the only thing holding the part on, and the joint is a thin layer of solder with a small footprint.
The Decision, As A Table
Seven questions cover almost every part on a real board. Answer them per part, not per board.
| Question | Favours | The physical reason |
|---|---|---|
| Is it plugged in and out? | Through hole | The load is peel on a pad, or a pin in metal |
| Does it weigh more than a few grams? | Through hole | Mass times drop equals a load on the joint |
| Must it be socketed or swapped? | Through hole | Most surface mount packages cannot be socketed |
| Does it need to shed heat to the board? | Through hole | A metal lead is a thermal path, a pad is thin |
| Is the pitch below 1 mm? | Surface mount | Leads need room; pads do not |
| Is the board space or weight critical? | Surface mount | No holes, both faces, smaller bodies |
| Is the signal fast or the loop tight? | Surface mount | No lead means less inductance and a smaller loop |
The table answers the engineering question. It does not answer the buying question, which is what the rest of a mixed assembly costs.
The Board Nobody Calls Mixed
The mixed board is the normal board. Reflow the small parts, then solder the tall ones.
If the board is double sided, the print, place and reflow cycle runs twice. The second pass heats the first side again, which is why the parts on the underside must survive two trips through the oven.
Then the through-hole parts go in, and there are three ways to join them. They differ in speed, in tooling and in how much a human hand is involved.
Wave soldering passes the whole board across a standing wave of molten solder. It is fast and it does not care how many joints there are.
It has two conditions. The solder reaches the underside of the board. The parts already on that side must be glued down, or they will float off when their paste melts.
And the wave reaches what it can see. A tall part on the far side can shadow the joints behind it. That is why a wave line is designed around part height and direction.
Selective soldering uses a small nozzle or a mini wave that visits one joint or one row at a time. Nothing else on the board is heated.
It is slower per joint and it needs a program. In return it removes the glue, the shadowing and most of the thermal load on the surface mount side. On a board with eight through-hole joints, that trade usually wins.
Hand soldering is the fallback, and it is the most expensive per joint in any volume. An operator, an iron, flux, and a joint that depends on that operator’s hand that day.
Hand work is right for a prototype, for a repair, and for the one connector that sits where nothing else can reach. It is wrong as the plan for a thousand boards.
There is a fourth route that removes the second soldering step entirely. It is worth its own section, because it is the one buyers rarely hear about.
Pin-In-Paste, And Its Arithmetic
Pin-in-paste prints the paste into the plated holes as well as onto the pads. The through-hole part is then placed like any other part, and the whole board goes through one reflow.
That removes a process, and it removes the hand that would have done it. The catch is the amount of paste, and the amount is a volume problem, not a setting.
Three volumes decide it. The volume of the plated hole, minus the volume of the pin that displaces solder. Then plus the volume of the fillet that has to form on both faces.
Then the paste itself. The paste is metal powder in flux, and most pastes are about half flux by volume. So only about half of the printed volume becomes solder.
That factor is what makes the arithmetic awkward. A worked example from a solder maker makes it concrete. On a 7 mil stencil, a square aperture needs a side of about 93 mil to deliver enough paste for one joint.
A circular aperture would need a radius over 50 mil, and at 100 mil pin pitch those deposits would overlap. The shape of the aperture is decided by the arithmetic, not by preference.
Where the stencil cannot deliver enough, a step stencil adds thickness locally, or a solder preform is dropped in to make up the difference.
Two practical limits remain. The process depends on how the paste is pushed into the hole, so print speed, print angle and hole diameter all matter.
And the fill cannot be seen. Paste inside a hole is invisible to the paste inspection machine, so a joint that looks right on the surface can be half empty inside.
That is the honest summary. Pin-in-paste is a good answer for a mixed board with a few through-hole parts. It is a poor answer when the paste volume cannot be delivered, or the fill cannot be checked.
Where The Money Actually Goes
Three cost lines move independently, and most comparisons collapse them into one sentence.
Drilling. Every hole is a tool, a machine minute and a bit of board area. A design that removes a hundred holes removes real cost.
Tooling and setup. A surface mount line needs a stencil, a program and a profile. None of that is per unit, so it hurts most on a small batch.
Labour. A through-hole joint made by hand is a person-minute. Fifty of them on a board is fifty person-minutes, and that number does not fall with volume.
That is why the two honest rules are not about which process is cheaper. They are about which cost line dominates this order.
On a small batch of a simple board, tooling dominates, and a design with few parts and generous through-hole spacing can be cheaper to build.
On a large batch, labour dominates, and any hand-soldered joint is the most expensive joint on the board. That is the sentence the comparison pages are reaching for, and it is only half of it.
The other half is that a mixed board pays for two processes at once. Stencil, program and profile for the surface mount side. Nozzle program or a hand station for the rest.
So the cheapest board is usually the one that needs the fewest processes, not the one that uses the fewest drilled holes.
How Each One Fails
The failure tells you what the process was never good at. Neither list is a reason to avoid the process, and both are reasons to put the right part on the right side.
| What you see | Which process | What actually happened |
|---|---|---|
| A connector pad lifts off | Surface mount | Peel load on a thin joint, from a cable pulled sideways |
| A joint cracks after potting | Surface mount | Resin and board expand at different rates through thermal cycles |
| A void inside a small joint | Surface mount | Flux gas trapped in a joint that is too small to let it out |
| The part next door moves | Surface mount | Rework heat reached a joint that was already finished |
| A plated barrel cracks | Through hole | Z-axis expansion pulls the copper through the hole |
| A joint is half empty | Both | Not enough solder reached the far side of the board |
| Two joints differ on one board | Through hole | Hand soldering, one operator, two different days |
The barrel crack is the one buyers underrate, because it appears months later. The expansion of the board through its thickness is what pulls that barrel apart. It is a property of the laminate, not of the solder.
The surface mount list reads like an argument for leads, and it is not. Each item has a countermeasure on the line. Load is handled by putting that one part through a hole.
Self-alignment is the countermeasure for placement error, a defined profile is the answer to voiding, and a rework station with a shield protects the neighbours. The catalogued drawbacks of the method are all process problems, and process problems get solved at the process step.
Hand soldering is the one item on either list with no process countermeasure. You can train the operator and you can inspect the joint. You cannot make two hands identical on two different days.
What To Put On The Drawing
The process is decided by the drawing long before the factory sees the board. Six items on that drawing prevent most of the arguments.
One. The assembly drawing itself, with a clear note on which parts are inserted and which are placed. Two arrows and a legend beat a paragraph.
Two. The package for every part on the bill of materials, written out. A part with two package options will be bought in whichever one is cheaper that week.
Three. The height limit on each side. A tall part on the underside changes what the wave or the nozzle can reach.
Four. The keep-outs around the through-hole joints. A nozzle needs room to land, and a hand iron needs room to approach.
Five. The test points. A probe needs a pad it can reach, and a through-hole lead on the probe side blocks it.
Six. The substitution rule. If a part may be replaced by a different package, say so and say who approves it. If it may not, write that too.
That sixth line is the one that decides a dispute. A specification sheet that a factory can build against names the package. A package change is then a change to the drawing, not a surprise at the end of the line.
Two more documents belong with it. The profile the oven is set to, which the second pass has to respect. And the workmanship standard the joints are judged by, which covers both kinds of joint in one document.
Then look at the line. An audit that walks the assembly route should show you where the second process happens. It should also show whether that happens at a machine or at a bench.
Send me the assembly drawing and the quotation you received for it. I will say which parts are being placed and which are being inserted by hand, whether the drawing leaves the package open, and where a second process was quietly added to the price. No cost, no obligation.
Check My DrawingQuestions Buyers Ask
Is surface mount always cheaper than through hole?
No. Tooling dominates a small batch, so a simple through-hole design can cost less to build. Labour dominates a large one, and then every hand-soldered joint is the expensive joint.
Why does my board need both processes?
Because a connector, a relay or a large capacitor carries mechanical load that a surface pad is not shaped for. Those parts stay leaded even on an otherwise fully surface mount board.
Can the through-hole parts be reflowed with the rest?
Sometimes. That is pin-in-paste, and it works when the stencil can deliver enough paste to fill the hole around the pin. If it cannot, a preform or a second process takes over.
Which process is stronger?
Through hole, for anything pushed, pulled or dropped. Surface mount is better under plain vibration, because the parts are lighter and sit closer to the board.
Is a through-hole board harder to repair?
Easier, usually. A lead comes out with a vacuum tool and goes back in with an iron. A fine-pitch part needs hot air and a shield so the neighbours stay put.
Can I mix them on both sides of the board?
Yes, and it costs. Each side that carries parts needs its own print, place and reflow, and any glued parts for wave soldering add a step of their own.
Questions & Comments
We read every one and reply within 24 hoursIf your drawing mixes placed and inserted parts, send it over. I will say which joints are hand work and where the quote hides them.