中文English

How to Attach Thermocouples for Reflow Profiling | Five Methods Differ by 9 °C in Testing, and the Risk of Tape Lifting Components

On the same board, in the same oven, with the same set of parameters, the reflow profiles measured by two people can differ by several degrees. The difference often lies not in the instrument but in how the thermocouple is attached.

This page sets out the actual effect of the attachment method on the readings, the differences among five common methods, a specific risk on high-density boards that can damage the board, and how to make profiles measured at different times comparable with each other.

1. The difference caused by the attachment method is larger than the error of the thermocouple itself

A comparative study that applied five attachment methods simultaneously on the same pad (KIC, published in Global SMT & Packaging in March 2025) gives the key figure: across the heating and cooling stages, the readings of the five methods differed by up to about 9 °C. Of these, attachment with high-lead solder read lower temperatures, and polyimide tape read higher.

To put this 9 °C in context: the tolerance classes of the K-type thermocouple itself are ±1.5 °C or ±0.004|t| for Class 1, and ±2.5 °C or ±0.0075|t| for Class 2. The difference caused by the attachment method is several times larger than the tolerance of the thermocouple itself, and it also eats up a considerable share of the margin in the solder paste’s process window.

This means two things:

  • Do not mix attachment methods within one measurement. Mixing them creates false temperature differences between different points on the board, and you cannot tell whether that is the real temperature distribution or an effect of the attachment method.

  • For profiles to be compared across batches, the attachment method must be the same. Changing the attachment method is the same as changing the measurement system, and the baseline accumulated before cannot be carried over directly.

2. Five common attachment methods

Attachment methodReading tendencyRepeatabilityMain limits
High-temperature solder (high-lead, high-melting-point alloys such as Sn10/Pb88/Ag2)Reads lower in the heating stageIn the study, stayed within the thermocouple’s ±1 °C toleranceRequires soldering skill; space is limited on high-density boards; lead-containing materials must be managed separately
Polyimide tapeReads higher in the heating stageCan stay within the ±1 °C toleranceThe tape is an insulator and forms a thermal resistance between the thermocouple and the measurement point; on high-density boards there is a risk of lifting components (see next section)
Aluminum foil tapeBetween the twoFairThe same risk of lifting components on high-density boards
Thermally conductive adhesive (0.92 W/mK grade in the study)Close to the actual solder joint temperatureGoodNeeds curing time; removal takes more work
Low thermal conductivity adhesive (0.19 W/mK grade in the study)Close to the actual solder joint temperaturePerformed most consistently in the study: all 20 trial runs within ±1 °C, and no detachmentNeeds curing time

It is worth noting that the adhesive with the lower thermal conductivity actually performed most consistently. This reminds us that the purpose of the attachment material is not to “conduct” heat to the thermocouple, but to hold the thermocouple steadily against the measurement point, while changing that point’s initial thermal behavior as little as possible.

Another practical recommendation, from a solder paste manufacturer, advocates soldering directly onto the solder joint with high-melting-point solder as the first choice, and explicitly does not recommend polyimide tape, on the grounds that it is itself an insulator and its readings easily become inaccurate. The two sources agree in their assessment of tape: it can be used, but you need to know that it reads high, and that it is not an accurate method.

3. A specific risk of tape on high-density boards: it can lift components

This point is rarely mentioned in catalogs and general tutorials, but the same study states it clearly: neither aluminum foil tape nor polyimide tape is suitable for high-density boards populated with components, because during the stage when the solder paste goes liquid, the pull of the tape can lift components off the board.

The mechanism is not hard to understand: at the moment the solder joints melt, the component is held to the pad by nothing but the surface tension of the molten solder. At that point, even a little shrinkage or curling of tape stuck nearby is enough to carry the component away or shift it.

The cost of this risk is not an inaccurate measurement, but ruining the very board used for the reflow profiling — and that board is usually a specially prepared, representative sample. By the time it is discovered, it has often already run through the whole oven. When profiling high-density boards, adhesive or high-temperature solder should be considered first as the attachment method.

If conditions on site allow just tape, at a minimum do two things: avoid areas dense with components, and do not put tape over the thermocouple bead — the bead is what actually measures the temperature, and covering it means measuring the temperature of the tape.

4. How to route the wires: three often-overlooked details

  • The thermocouple wire itself is a heat dissipation path. As the wire runs out from the measurement point, it carries away part of the heat. So near the measurement point, the wire should run a short distance flat against the board along the isothermal direction before being led away, rather than standing up vertically through the temperature gradient.

  • Keep the lengths of the wires as equal as possible, and bundle the wires with sleeving. Wires of uneven lengths easily tangle inside the oven and catch on the conveyor mechanism, and they also make the routing impossible to reproduce from one run to the next.

  • Keep the profiler body far enough from the board. The industry’s practical recommendation is to leave at least one board length of distance — the profiler has considerable heat capacity, and being too close skews the temperature of the nearby area.

What these three things have in common is that they change “the measurement system’s own interference with the site”, not the instrument’s accuracy. However accurate the instrument, if the act of measuring itself changes the object being measured, the numbers do not represent the real process.

5. Reproducibility: how to make this profile comparable with the last one

The value of a reflow profile lies in “comparing with the last one” and “comparing with the process window”. To be able to compare, the measurement conditions must be recorded and replicated. We recommend keeping these five things for every measurement:

Item to recordWhy
Attachment method and material lot numberChanging the material is the same as changing the measurement system. The study shows the methods differ by up to 9 °C
The position of each thermocouple (with photos)Temperatures at different positions on the same board differ greatly. Without a record of positions, the next profile cannot be matched with this one
A table matching channel numbers to positionsWhen reviewing the profile afterward, you need to know which line represents which component
Carrier/fixture configuration and board orientationHeat capacity and shielding change the profile; the same board placed the other way round is different
Oven settings and the actual oven temperature at the timeThe set value does not equal the actual value. The oven’s condition drifts with the cleaning cycle and the load

Of these five items, photos are the easiest to skip, and also where problems often arise. “Which pin of U3 was it attached to last time?” — this question usually cannot be answered from written records alone, and if it cannot be answered, this profile cannot be compared with the last one. The cost of taking one photo is far lower than rerunning a measurement.

6. Putting the attachment method back into the whole measurement chain

The error in reflow profiling comes from a whole chain, and the attachment method is just one link in it:

  • The tolerance of the thermocouple itself — classified as Class 1/2/3 under IEC 60584-1, with the error varying with temperature.

  • Extension wire and connectors — the wrong material or oxidized contacts both introduce error.

  • The profiler’s channel accuracy and sampling rate — sampling too slowly misses the peak.

  • The attachment method — the part discussed on this page, measured at up to about 9 °C.

  • Choice of position — what is measured is the temperature at that point, not the temperature of the whole board.

Looking at these items together, you will find that if you spend money upgrading to a higher-grade thermocouple without standardizing the attachment method, the improvement may be eaten up by the variation of the attachment method. Standardizing the attachment method and the recording process first is usually more effective than changing the wire, and costs almost nothing.

7. How to read a reflow profile: once you have the profile, check these 5 numbers first

The first six sections dealt with how to measure the temperature accurately. After the profile is measured (the reflow temperature profile, in English “Reflow Profile”, also often written in Chinese as “reflow soldering temperature profile” or “reflow oven temperature profile”), these are the 5 numbers to look at: check each one first against the solder paste manufacturer’s datasheet, then against the component specifications as the upper limit.

Figure 1: The 5 check points on a reflow profile (schematic, not representing the actual values of any solder paste). Horizontal axis: time →; vertical axis: temperature →; reference line: liquidus (about 217 °C for SAC305). Check points: 1 ramp rate; 2 soak (optional); 3 time above liquidus (TAL); 4 peak; 5 cooling rate. Schematic: it marks just the positions of the 5 check points; the shape of the curve and the coordinates do not represent the actual values of any solder paste or oven.

Table: How to check the 5 numbers (using Indium8.9HF/SAC305 lead-free solder paste as the example)

Check pointWhere to read it on the profilePaste manufacturer’s recommendation (acceptable range in parentheses)Component classification conditions (J-STD-020, lead-free)What the datasheet says
① Ramp rateAverage slope from room temperature to peak1.0–1.5 °C/s (0.5–2.5)3 °C/s or less (counting just the segment from the liquidus at 217 °C to the peak)This range can reduce solder balling, beading and hot slump. The two columns measure different segments (paste column: room temperature to peak; component column: 217 °C to peak), so do not compare them directly.
② SoakThe plateau around 140–160 °C; optional140–160 °C, 20–60 s (140–170 °C, 30–120 s)150–200 °C, 60–120 sShortening or eliminating the soak helps reduce HIP (Head-in-Pillow) and graping; soak time is also related to BGA/CSP voiding. The manufacturer recommends starting with a linear profile without a soak.
③ Time above liquidus (TAL)Seconds the profile stays above liquidus (about 217 °C for SAC305)45–60 s (30–100)60–150 s (above 217 °C)Enough time is needed for good wetting and reliable solder joints. The right-hand column gives the conditions of the component heat-resistance classification test; the standard notes they are for classification, not a prescribed board assembly profile; if your process is longer, hotter or steeper than this, please confirm with the component manufacturer.
④ Peak temperatureThe peak of the profile (based on the value measured by the thermocouple)230–260 °C (230–262)245–260 °C, depending on package thickness and volume (260 °C for thickness <1.6 mm; 245 °C for >2.5 mm and volume ≥350 mm³)The datasheet gives just a range. It must be both above the manufacturer’s lower limit and below the classification temperature of sensitive components.
⑤ Cooling rateThe cooling slope after the peak2–6 °C/s (0.5–6)6 °C/s or less (the segment from peak down to liquidus)Rapid cooling helps form a fine-grained structure.

The numbers in the manufacturer’s datasheet are just for reference, and should be adjusted to the board and design. The table uses the Indium8.9HF (SAC305, Form 99211 R6) datasheet as the example; replace them with the numbers from the datasheet of the solder paste you use. The fourth column gives the conditions of the component heat-resistance classification test, not process targets, with the numbers taken from IPC/JEDEC J-STD-020F; the actual tolerance of a component is as stated in the component manufacturer’s specifications.

How to use this table: first read the 5 numbers off the profile and check each against the third column; then check the fourth column against the component manufacturer’s specifications for sensitive components (usually BGAs or large plastic-encapsulated ICs). The profile for this board counts as acceptable just when it passes on both sides. If the instrument software has a PWI (Process Window Index), it converts the measured profile into where it falls within the process window; for an explanation, see “How to choose a reflow profiler“.

If the profile looks wrong, first confirm these 3 things:

  • Inconsistent attachment methods: with five methods used on the same board, the readings differ by up to about 9 °C, more than the tolerance of the K-type thermocouple itself (±1.5 °C for Class 1) (see Section 1).

  • The thermocouples are not in representative positions: the approach in IPC-7530A is to place thermocouples on representative components, with recommended positions for large components, small components and BGAs.

  • Every measurement comes out different: first confirm the attachment method and the thermocouple wire, then check the oven’s condition, for example whether the condenser is clogged (see “Unstable reflow oven temperature and persistently high AOI false-call rates? The root cause may be a condenser that was not cleaned properly“).

Frequently asked questions

Is it enough to tape the thermocouple on, or should it be soldered?

It depends on the board. A comparative study that applied five methods on the same pad showed the readings differ by up to about 9 °C — high-lead solder read lowest, and polyimide tape read high. Tape is not recommended on high-density boards populated with components, because the tape may lift components off when the solder paste goes liquid. On high-density boards, use adhesive or high-temperature solder first.

Why use high-melting-point solder for soldering?

Because the attachment joint must not melt along with everything else in the oven. If ordinary lead-free solder paste is used for attachment, it melts together with the solder joints on the board during reflow, and the thermocouple comes loose. So an alloy with a melting point above the process peak temperature should be used (the study used high-lead, high-melting-point alloys such as Sn10/Pb88/Ag2).

Is an adhesive with high thermal conductivity better?

Not necessarily. In that study, the one that performed well was the adhesive with a thermal conductivity of 0.19 W/mK (all 20 trial runs within ±1 °C, no detachment), more stable than the 0.92 W/mK one. The purpose of the attachment material is to hold the thermocouple steadily against the measurement point while changing that point’s initial thermal behavior as little as possible, not to conduct heat to the thermocouple.

Can different attachment methods be mixed on the same board?

Not recommended. Mixing them creates false temperature differences between different points on the board, and you cannot tell whether that is the real temperature distribution or an effect of the attachment method. To compare temperatures at different positions, the attachment method must be the same.

How should the thermocouple wires be routed?

The wire itself is a heat dissipation path and carries heat away from the measurement point. Near the measurement point, the wire should run a short distance flat against the board along the isothermal direction before being led away, rather than standing up vertically through the temperature gradient. In addition, keep the lengths of the wires as equal as possible and bundle them with sleeving, and leave at least one board length of distance between the profiler body and the board.

What should be recorded with the measurement results?

At least five items: the attachment method and material lot number, the position of each thermocouple (with photos), a table matching channel numbers to positions, the carrier/fixture configuration and board orientation, and the oven settings and the actual oven temperature at the time. Of these, photos are often skipped and also often where problems arise — without photos, the next profile cannot be compared with this one.

Should I first switch to a better thermocouple, or first standardize the attachment method?

Standardize the attachment method first. The tolerance of the thermocouple itself under IEC 60584-1 is on the order of ±1.5 °C for Class 1 and ±2.5 °C for Class 2, while the difference caused by the attachment method was measured at up to about 9 °C. If the attachment method is not standardized, the improvement from upgrading the wire is eaten up by the variation of the attachment method, and standardizing costs almost nothing.

Further reading

Wire selection, tolerance and profilers are covered in these articles respectively:

Related products

Jiahon supplies all the items mentioned in this article.

Standards referenced

For the specifications cited in this article, please follow the current edition from each issuing body.

Back to top