Class 4 · Soldering and Fabrication
4 h · 3 at the bench 00:00Solder is not glue. It is an alloy that wets clean hot metal and refuses dirty cold metal, and every bad joint you make is one of those two words.
Why this class is four hours long#
Because soldering is a motor skill, and motor skills are acquired by repetition under correction, not by explanation. You will make somewhere between forty and eighty joints today. The first ten will be poor, and that is the schedule working correctly.
There is a second reason. Termination quality is the single largest cause of intermittent faults in live performance. Not design errors, not component failures: joints and strain relief. A technical director who can look at a terminated connector and say "that will fail in about two months" is protecting a production in a way that no amount of theory does.
What is actually happening#
Solder joins metals by wetting them: the molten alloy forms a thin intermetallic layer with the copper surface. It is closer to brazing than to gluing, and it needs three things at once.
- Clean metal. Oxide is not wettable. Copper oxidises in air within hours, faster when warm. This is what flux is for: it chemically strips the oxide at soldering temperature and briefly protects the surface while the solder flows.
- Enough heat, in the right place. Both parts must reach solder melting temperature. Not the iron: the joint. Heating the solder instead of the joint is the classic beginner error and it produces a joint that looks finished and is not connected.
- Time short enough not to damage anything. Two to four seconds for most joints. Heat travelling into a connector melts its insulator and lets the pin move. Heat into a semiconductor damages it. Heat into PVC cable insulation shrinks it back from the joint.
Those three pull against each other, and every soldering technique is a way of getting all three at once.
The order of operations that works#
- Set the iron to temperature and let it stabilise. 340 to 370 °C for leaded solder, 370 to 400 °C for lead-free. A hotter iron for a shorter time damages less than a cool iron for a long time.
- Clean and tin the tip. A bright wetted tip transfers heat; a dull oxidised one does not, and almost every "my iron is not hot enough" complaint is a dirty tip.
- Mechanically secure the parts first. Solder is a poor mechanical fastener. Hook the wire, crimp the tags, or use a clamp.
- Heat both parts together with the tip touching both.
- Feed solder into the joint, not onto the iron. When it flows, it will run toward the heat and form a smooth fillet.
- Remove the solder, then the iron. Do not move the joint until it has solidified.
- Inspect. A good joint is shiny, concave and smooth, and you can see the shape of the wire through it.
The order of operations that works#
What the failures look like#
| Fault | Appearance | Cause | How it presents in a show |
|---|---|---|---|
| Cold joint | Dull, grainy, lumpy | Not enough heat, or moved while cooling | Intermittent, worse when warm |
| Dry joint | Solder balled up, not wetted | Oxide, no flux, or dirty surface | Works on the bench, fails under vibration |
| Too much solder | Round blob hiding the wire | Feeding solder rather than heating | Hides a bad joint underneath it |
| Bridge | Two pads joined | Too much solder, tip too broad | Dead short, often immediate |
| Overheated | Brown flux, lifted pad, melted insulation | Too long on the joint | Pin moves in the connector, fails later |
| Cooked insulation | Insulation shrunk back, strands exposed | Heat travelled up the wire | Short to the shell, intermittent |
Strain relief, which is the actual subject#
Here is the sentence to remember from this class:
A cable fails where the flexing is concentrated.
Strain relief, which is the actual subject#
Everything in cable-making is about spreading that flex out or moving it somewhere the conductor is not.
The failure sequence is always the same. The cable is pulled or bent repeatedly at one point. The copper strands work-harden there. One strand breaks. The remaining strands carry the same current in a smaller area and run hotter. More break. Resistance rises, the joint gets warm, and the connection becomes intermittent long before it becomes open.
That intermittent phase is where the money is lost, because it is a fault that appears during a performance and disappears when you test it.
The rules that follow#
- The clamp must grip the outer jacket, never the conductors. If the strain reaches the solder joint, the joint is the strain relief, and solder is brittle.
- Leave the conductors slightly longer than the shortest path. A gentle service loop inside the shell means the strain lands on the clamp rather than the joints, and it lets you re-terminate once without shortening the cable.
- Heatshrink over the joint, not over the strain relief. Heatshrink is insulation and a little mechanical support. It is not a clamp, and a joint whose only support is heatshrink will fail.
- Match the cable to the gland. A cable too thin for its clamp is not gripped. A cable too thick is crushed, and crushed insulation is a short in eighteen months.
- The bend radius rule: never tighter than eight times the cable diameter, and never at the connector at all.
XLR, 3-pin and 5-pin#
Pin 1 screen, pin 2 hot (positive), pin 3 cold (negative). For DMX on 5-pin: pin 1 screen, pin 2 data minus, pin 3 data plus, pins 4 and 5 unused in almost every rig.
Order of operations matters and half of today's errors are here. Put the shell and the boot on the cable first. Everybody forgets this once, and having to unsolder three joints teaches it better than being told.
Strip the jacket about 25 mm. Do not nick the conductors. Separate and twist the screen, sleeve it, and tin all three lightly. Solder into the cups with the cable held in a vice. Check no strand has escaped, seat the insert, and tighten the clamp onto the jacket.
Then test it. All five steps from Class 2, including the flex test. A cable you made and did not test is not finished.
Speakon#
No soldering for the common types: the conductors go into screw or spring terminals. The failure mode is different and worth knowing — a strand escaping the terminal and touching the adjacent pole, which on a loudspeaker output is a short across an amplifier that can deliver a hundred amps. Twist, insert fully, tighten, and tug each conductor before you close the shell.
RJ45 for network and DMX-over-IP#
RJ45 for network and DMX-over-IP#
T568B is the ordering used almost everywhere in this industry: orange-white, orange, green-white, blue, blue-white, green, brown-white, brown. Either standard works as long as both ends match, and the reason B dominates is convention rather than physics.
Two things that are not obvious and that cause real faults:
Untwist as little as possible. The twist is what rejects interference, and untwisting 30 mm at the connector undoes a surprising amount of it. Keep it under 13 mm.
The pairs are not in numerical order for a reason. Pins 3 and 6 are one pair, which is why a cable that "looks right" but has 3 and 6 split across two pairs will pass a continuity test and fail at 100 Mbit/s over any distance. A cheap cable tester that only checks continuity will call that cable good. This is precisely the Class 2 lesson about proving the right thing.
Crimping, which is most of what you will actually do#
Soldering gets the attention. In the field, on a production, in a rack, most terminations are crimped, and a good crimp is more reliable than a good solder joint under vibration.
A correct crimp is a gas-tight cold weld. The die deforms the barrel and the conductor together until the metal flows and there is no air path between them. It is not a squeeze that holds the wire mechanically; it is a joining process, and that distinction is why the tool matters more than the technique.
The tool is the whole subject. A ratcheting crimper with the correct die for the terminal will not release until it has completed the cycle, so every crimp gets the same force. A general-purpose plier-type crimper applies whatever force your hand had that afternoon. The industry uses ratcheting tools because consistency is the property being bought, and a cheap crimper is the single worst false economy on the kit list.
The four kinds you will meet#
The four kinds you will meet#
| Type | Where | What goes wrong |
|---|---|---|
| Bootlace ferrule | Stranded wire into a screw or spring terminal | Not used at all, so strands splay and one escapes |
| Ring and spade | Bolted studs, chassis earths | Crimped over the insulation instead of the conductor |
| Insulation-displacement (IDC) | RJ45, ribbon cable, Socapex inserts | Wrong conductor size for the slot |
| Machined pin | Multipin connectors, Socapex, CEE | Pin not seated, so it pushes back on mating |
Bootlace ferrules are the one to internalise. Any stranded conductor entering a screw terminal should be ferruled. Without one, the screw crushes and cuts strands, the strand count carrying current falls, resistance rises and the terminal heats. It costs a few cents and about four seconds, and its absence is visible in half the racks you will ever open.
Inspecting a crimp#
- The wire barrel grips conductor; the insulation barrel grips insulation. Both, separately. A crimp that has caught the insulation in the wire barrel is a connection to nothing.
- You should see a little conductor past the barrel, one to two millimetres. None at all means the wire was short of the end; a lot means it was too long and is unsupported.
- The tug test. Pull it, properly. A correct crimp on stranded wire will break the wire before it releases the barrel. If it slides out, every crimp you made that day is suspect.
- No solder on a crimp. Solder wicks up the strands and creates a hard point exactly where the conductor needs to flex. Crimp or solder, never both.
At the bench#
Three hours and twenty minutes. This is the most hands-on class of the course.
Block A — Fifty joints on scrap (45 min)#
Not connectors. Wire to tag, wire to wire, wire to pad, on scrap board. The technician inspects every tenth joint and you keep going until the inspection passes twice in a row.
The metric is consistency, not beauty. Ten identical acceptable joints is a pass; two beautiful ones and eight poor is not, because a cable is only as good as its worst joint.
Block B — Destructive testing (30 min)#
Make four joints deliberately badly: one cold, one dry, one with too much solder, one with the wire moved during cooling. Then break each one and look at the fracture surface under a loupe.
Then measure each one's resistance, and measure it again while flexing. The cold joint will read fine and fail on flex. Write that down: a bad joint often measures perfectly at rest, and this is why visual inspection and the flex test both exist.
Block C — Build three cables (75 min)#
- One 3-pin XLR audio cable, 2 m.
- One 5-pin DMX cable, 2 m, with 110 Ω cable, not microphone cable.
- One RJ45 patch lead, T568B both ends.
Each fully tested with the five-step procedure, logged, labelled and signed. These are yours and you will use them in Class 8 and Class 10, so make them well.
Block D — Inspect and condemn (30 min)#
Twenty cables from the store. Inspect each and decide: pass, repair, or condemn. Write one sentence of justification for each.
Then the technician gives their verdict and you compare. Disagreements are the lesson. Most people condemn too little, because a cable that is currently working feels like a cable that is fine.
Numbers from this class#
| Quantity | Value |
|---|---|
| Iron temperature, leaded solder | 340–370 °C |
| Iron temperature, lead-free | 370–400 °C |
| Leaded eutectic solder melting point | 183 °C |
| Lead-free (SAC305) melting point | 217–220 °C |
| Time on a typical joint | 2–4 seconds |
| Jacket strip length for XLR | about 25 mm |
| Maximum untwist at an RJ45 | 13 mm |
| Minimum cable bend radius | 8 × cable diameter |
| XLR pin 1 | Screen |
| XLR pin 2 | Hot, positive |
| XLR pin 3 | Cold, negative |
| DMX 5-pin, pin 2 | Data minus |
| DMX 5-pin, pin 3 | Data plus |
| DMX cable impedance | 110 Ω |
| T568B first pair | orange-white, orange |
| RJ45 pair on pins 3 and 6 | green-white, green |
| Acceptable joint resistance | under 0.05 Ω |
↑↓ move · ↵ open · esc close