A profiling board is a board that goes through the oven together with your product. Thermocouples are fixed at positions you have chosen, and they record how the temperature at each position changes over time.
Five points matter most:
Measuring points: cover the oven air, a hot spot, a cold spot, and the heat-sensitive parts.
Bare board: it can just compare oven conditions; it does not stand in for the product.
Thermocouple: K type is common in SMT; fine wire responds quickly but is fragile.
Fixing method: points that will be compared with each other must be fixed the same way; a saw-tooth curve is usually a loose wire.
BGA: drill from the back of the board and bury the thermocouple at the boundary between the solder ball and the pad.
This article follows that order and marks which statements come from standards or external articles, which come from internal documents we hold (de-identified), and which are our own suggestions. The internal documents are de-identified papers from packaging and module plants; the external sources are all public web pages, linked in the text and at the end. The source list, our interests, and the limits are set out at the end: Jiahon supplies reflow profilers and thermocouple wire, but nothing here asks you to buy from any one supplier. Numbers marked “assumed” or “illustrative” are for demonstration, not measurements.
1. What a profiling board is, and how it differs from the oven’s own thermocouples
Q: What is a profiling board, and where is it used?
A: Thermocouples are fixed on a product board (or a board with the same conditions), connected to a profile recorder, and sent through the oven with the board. They record the temperature curve at the chosen positions, to confirm that each solder joint stays within the temperature range the solder paste specifies.
The IPC-7530A preview pages (Revision A, 2017; replaced by IPC-7530B in January 2025, and figure numbers below refer to Revision A) define a thermal profile as the result of attaching thermocouples to representative components on the board with high-temperature solder or copper or aluminum tape. The Wikipedia entry on thermal profiling says a thermal profile is a set of time and temperature data, usually measured in a reflow oven and judged by ramp rate, soak, time above liquidus (TAL), and peak.
The oven has thermocouples of its own, but they measure something different. An internal document, one line’s reflow oven parameter control note, divides the work like this: each zone has a thermocouple for real-time monitoring, with an alert at ±5 °C from setpoint and an alarm at ±10 °C; the first-article inspection (FAI) checks just the peak temperature, because the concern is overheating damage; ranges for each zone are not set uniformly. In other words, the oven’s thermocouples tell you whether the oven is running as set, and the profiling board tells you what the solder joints on this board actually went through.
| Item | In-oven real-time monitoring | Profiling board |
|---|---|---|
| What it measures | Zone air temperature (against setpoint) | Temperature at chosen positions (joints, components) |
| How often | Continuous | When a process is set up, a product changes, after oven repair, and at periodic audits (frequency per each plant’s rules) |
| What you read | Deviation from setpoint, and alerts | Ramp rate, soak time, TAL, peak, and temperature differences across the board |
| What it cannot answer | Whether a large component on the board reached temperature | Whether the oven is stable day to day |
The two complement each other; neither replaces the other.
Figure: what each of three measuring methods can and cannot answer (illustration, compiled from the Indium article cited below and the internal reflow oven parameter control note).
2. Can a bare board be used as the profiling board?
Q: Can I use a bare board (one with no components) as the profiling board?
A: It can be used to compare oven conditions, but it does not stand in for the product.
An Indium Corporation engineer, in an article on I-Connect007, sorts boards into three kinds:
Baseline profile: a run with a bare board, which the article says usually comes from the oven or solder paste supplier.
Product profile board: a fully assembled board, reflecting the thermal mass and laminate of each assembly; a double-sided board needs separate profiles for the top and bottom sides.
Calibration board: a standard-size copy with thick laminate (or durastone board) and no components, so its thermal mass is smaller; the article says it usually gives a hotter curve. If its curve changes, check the oven first, then re-run the product profile.
The IPC-7530B table of contents supports this indirectly: section 3.1.8 is titled “Unique Profile for Each Printed Board Assembly.” The text has to be purchased, so we can cite just the title. Internal practice is consistent with it: the BGA ball-attach oven measurement procedure (2002 edition) requires choosing the matching profiling substrate by the product type on the route card; the recorder goes into an insulated iron box, is placed in the middle of the oven belt, and is taken out with heat-resistant gloves before the data is read.
Our suggestion (not a standard requirement): for production board types, always build the curve on a fully populated profiling board. Keep a bare board just for weekly or monthly oven comparison, storing the curve of the same bare board to see whether the oven drifts. A bare board has no large components and no connectors, so it measures an easy-to-heat condition and cannot reach the positions on the board that are hard to heat.
For how a double-sided board survives its second pass and whether parts will drop, see double-sided reflow component drop. When a carrier or fixture goes through the oven, the thermal mass of the fixture itself must also be counted; see how to choose a reflow carrier.
3. How many measuring points, and where
Q: What is the minimum number of measuring points?
A: The Indium article says at least 4 thermocouples: one for the oven air (placed beyond the leading edge of the board), one at a hot spot, one at a cold spot, and one on a heat-sensitive component. For a detailed profile, at least 5 to 6.
Zarrow’s version, in the Indium blog, is simpler: at least one thermocouple at a point you think is hot and one at a cold point, so you can measure the temperature difference (delta-T) at peak. The Wikipedia thermal profiling entry says thermocouples should go where thermal mass differs, where position differs, and where problems are known, isolated from airflow from one another, and spread from dense to sparse areas.
| Measuring point | Where | Why | Source |
|---|---|---|---|
| Oven air | Beyond the leading edge of the board | A common baseline with the oven condition | Indium article (one of the minimum 4) |
| Hot spot | Bare copper pads, small passive components, usually near the board edge | Stands for “will it overheat” | Indium article; Zarrow blog |
| Cold spot | Under plastic or ceramic IC packages, leads of coils and transformers, large components usually in the middle of the board | Stands for “is it hot enough” | Indium article; Zarrow blog |
| Heat-sensitive component | Body of plastic-packaged parts | Confirms the part does not exceed its heat rating | Indium article (one of the minimum 4) |
| BGA | Centre of the array and outer row (see Section 6) | The centre runs cooler | Indium article; IPC-7530A preview figure titles |
Figure: illustration of measuring points on a profiling board (positions and counts are for demonstration; use the thermal mass distribution of your own board).
“Hot spots near the board edge, cold spots in the middle” in the table is the general statement of the sources; in practice the position is affected by oven type and by where the board sits on the conveyor. The first time you measure a board type, you can place extra thermocouples to find the real hot and cold points, then reduce to a fixed few afterwards (this is our suggestion, not a standard requirement).
Q: Should I keep a line for measuring air?
A: Yes, we suggest it.
The Indium article lists it as one of the minimum 4. In practice the benefit is a common baseline in every measurement. Jiahon’s own profiler thermocouple placement note says the same: keep one or two channels permanently for oven air or the fixture, not attached to the product, so curves from different batches can be compared.
Q: If I put several thermocouples on one board, will they read the same?
A: No.
A wireless module’s oven profile report (4 thermocouples) shows that on the same board the 4 thermocouples differed by about 3 seconds in soak time and about 1.4 °C in peak. The report’s window was a soak of 155–185 °C held for 25–55 seconds and a peak of 240–255 °C; the measured soak time was about 80–83 seconds, above the limit, and the peak was about 233–235 °C, below the lower limit. The point: with one thermocouple you see just one answer; with several you can see the temperature difference across the board.
4. Measuring the oven is not measuring the board: multi-point checks, and “setpoint is not measured”
Q: How is the oven’s own temperature distribution verified?
A: With several thermocouples at fixed positions, without a product.
An internal document, a reflow oven acceptance briefing (2012), did it this way: the oven was accepted by measuring 6 points with a profiler, and the machine capability requirement was ±5 °C. The briefing’s peak deviation, mean, and range do not agree with each other, so we cite just “6 points” and “±5 °C.” Two oven-type documents (an oven setting and control specification, and a reliability oven operating specification) use similar practice: load to the maximum loading, with 5 to 9 thermocouples spread across the centre and corners; see the limits at the end. They are not reflow ovens; we take just the practice of “multiple points plus full load.”
Q: If the oven is set to 150 °C, will the part reach 150 °C?
A: Not necessarily.
A temperature experiment on an underfill oven for a SiP module (2012, oven type) measured 11 points (5 in the chamber, 6 on chip surfaces). At a setpoint of 125 °C the chamber points read 120–133 °C, but the chip surfaces measured just about 120 °C; at a setpoint of 150 °C the chip surfaces measured just about 145 °C. The conclusion was to set production at 130 ± 1 °C and 155 ± 1 °C. This was an oven, not a reflow oven, but the principle is the same: to know the temperature of a part, measure the part.
5. K type or T type thermocouple; wire gauge and insulation
Q: How do K type and T type thermocouples differ?
A: The materials and temperature ranges differ; for SMT reflow measurement, the sources say K type is the main one.
The figures in the table below come from the Wikipedia entry on thermocouples. (The “range” row is the entry’s page text, not the usable range for reflow measurement; each tolerance class has its own temperature interval.)
| Item | K type | T type |
|---|---|---|
| Material | Chromel / Alumel (nickel-chromium / nickel-aluminium) | Copper / Constantan |
| Range | −200 to +1350 °C | −200 to +350 °C |
| Tolerance class 1 | −40 to 375 °C: ±1.5 °C | −40 to 125 °C: ±0.5 °C |
| Tolerance class 2 | −40 to 333 °C: ±2.5 °C | −40 to 133 °C: ±1.0 °C |
| The entry also notes | Nickel is magnetic, and the output deviates near about 150 °C (the Curie point) | Copper conducts heat quickly, so thermal anchoring needs particular care |
The tolerances in the table cover alone their own marked temperature intervals. For the reflow range of 150 to 260 °C, the tolerance of T type is not in the table (we did not read the text of the IEC 60584 standard; Jiahon has a separate page, IEC 60584-1 tolerance classes and the whole measurement chain), so the table does not support the inference that T type is more accurate than K type across the whole reflow range. Also, Wikipedia gives the K type upper limit as +1350 °C while the Indium article gives +1250 °C; the two sources differ by 100 °C, but reflow just reaches about 260 °C, so that part is not used.
The Indium article says K type is the usual choice in SMT, with a range of −200 to +1250 °C and an accuracy of about ±1.5 °C. The IPC-7530B table of contents lists K, J, T, and N types separately in section 7.3.1, with separate sections for wire gauge, insulation, length, and junction, and one section for calibration and testing together; the text has to be purchased, and we did not read it. To be candid: we found no public source that directly explains “why SMT uses K and not T.” The table above just describes how the two wires differ. Which one to choose depends on what type your recorder takes; the recorder and the wire must be the same type (general practice).
Q: How thick should the wire be?
A: The Indium article recommends 36 AWG (about 5 mil) for measurement; 30 AWG (about 10 mil) is also acceptable.
The SMTnet paper How to Profile a PCB also says the finer the wire the better, because readings are more accurate, but fine wire is fragile and must be handled carefully. The specifications on our product pages: AWG36 has a 0.127 mm conductor and low thermal inertia, so it responds quickly; AWG30 has a 0.255 mm conductor, is more resistant to bending, and suits fixed measuring points that are attached and removed repeatedly. For the thermocouple junction, the Indium article recommends welding it into a bead, not gluing it with solder or twisting it.
Q: How do I judge the insulation?
A: Look at the temperature limit, and at whether it is damaged.
Peaks in lead-free processes are often between 240 and 260 °C, and the temperature limit of the insulation must be above the peak; the smaller the margin, the more it must be checked before every use. PTFE (Teflon) and fibreglass insulation have different limits, so go by the datasheet of the wire you buy; when the peak is high or the number of passes is large, consider fibreglass insulation with a higher rating. In both cases, check every time whether the insulation is broken: the Indium article warns that broken insulation lets the two wires touch elsewhere and form a second junction, causing a measurement error; the Wikipedia thermocouple entry likewise says that accidental contact from damaged insulation usually makes the reading lower than the true junction temperature.
Figure: curves for a well-fixed and a loose thermocouple, compared (illustration; the curves are demonstration data, not measurements).
6. Placing a BGA measuring point, and what LTD is
Q: How is a BGA measuring point made?
A: Drill a small hole from the back of the board, without going through the solder ball or the chip, and connect the thermocouple at the boundary between the ball and the pad.
The Indium article says the hole should be small; for a small thermocouple bead, about 1.0 to 1.5 mm (the source does not say this figure is a hole diameter; from context we read it as a hole diameter). For a part with a fine ball pitch, the hole must not touch the neighbouring pads, and the diameter has to be judged separately for the part (this is our judgment, not Indium’s). The centre of a BGA array runs cooler (Indium article), so the thermocouple should be connected at the centre of the array; afterwards confirm whether the bead touches the boundary between pad and ball, which the Indium article does with 3D X-ray (a practice described in that article). Zarrow’s blog says the method is to drill a hole beneath the ball position, feed the thermocouple to that point, fix it with thermally conductive epoxy, and then fill the hole with non-conductive epoxy, so that the temperature of that joint can be measured.
The IPC-7530A preview pages list related figure titles: Figure 3-7 is the recommended thermocouple position on a BGA, and Figures 3-8 and 3-9 are examples of drilling from the bottom of the BGA to measure the inner and outer rows; the figure content and explanatory text have to be purchased. The NXP (formerly Freescale) Engineering Bulletin EB635 (April 2005) is a public example: when evaluating lead-free BGA solder joints, two thermocouples were buried in the BGA joints at the positions expected to be hot and cold on the board, and the peaks of the two curves in Figure 1 were 247 °C and 243 °C. Internal documents also describe a simple method: the profiling substrate for the BGA ball-attach process has K type wire taped with heat-resistant tape to the ball pads at the top, middle, and bottom of the substrate (2002 edition). That is a purpose-made substrate for ball attach, not a production board.
Figure: cross-section of a BGA measuring point (drawn from the Indium article’s description; not to scale).
Q: What is LTD, and why measure the inner and outer rows of a BGA?
A: The IPC-7530B table of contents lists the term “Liquidus Time Delay (LTD)”; we did not read the definition text.
In the table of contents it is in section 1.3.5; the title of Figure 3-11 in the IPC-7530A preview is “Role of Liquidus Time Delay (LTD) in Head on Pillow,” so it is related to the head-on-pillow (HoP) defect. The definition text has to be purchased, we did not read it, so we do not explain what it means. For measuring the inner and outer rows, just two points have a source: the Indium article says the array centre runs cooler, and the IPC-7530A preview has figure titles for drilling from the bottom to measure the inner and outer rows.
There is also a public SMTAI 2010 paper (the authors are from the solder paste maker Cookson Electronics; we checked the numbers against the source): the experiment dropped balls into already-molten paste to simulate the contact delay caused by component warpage; the paper does not use the term LTD and is not a measurement of LTD. The results show that with a contact delay of about 30 seconds between the ball and the molten paste, the time for the ball to collapse to 20% went from about 9 seconds to 24 seconds for paste 1 and from about 8 seconds to 16 seconds for paste 2. The cause of head-on-pillow is not just the temperature profile; it also depends on paste activity and component warpage. For causes and troubleshooting see BGA head-in-pillow troubleshooting; for how to inspect the joints, see BGA solder joint inspection.
7. Fixing the thermocouple: a short paragraph
Q: Which method should be used to fix a measuring point?
A: No single method is ideal in every case. Points that will be compared with each other should be fixed the same way; special positions such as a BGA are handled separately and noted.
Section 3.1.2 of the IPC-7530B table of contents lists five kinds: high-temperature solder, adhesive, tape, embedded, and mechanical fixing. The Indium article says: high-temperature solder (Pb95Sn5, melting point 312 °C) is a common choice; it must not melt during reflow, and ordinary solder wire cannot be used. Aluminum tape conducts heat well and goes on quickly, but the article says it does not survive many passes through the oven. Polyimide tape is not recommended, because it does not tolerate the heat well and tends to peel off the board. SMT adhesives can survive many passes. The Wikipedia thermal profiling entry adds that polyimide (Kapton) tape softens and becomes elastic above about 200 °C, so the thermocouple may lift. The SMTnet article says to use just a small amount of solder when fixing with solder.
The sources do not agree on high-temperature solder and tape. The Wikipedia thermal profiling entry says high-temperature solder is not an ideal choice (the amount differs at every position, and conductive solder may short the thermocouple), while Indium and SMTnet list it as common or preferred. On polyimide tape, Wikipedia says it is one of the common methods, while the Indium article does not recommend it. “Do not mix methods within the same measurement” comes from Jiahon’s own profiler thermocouple placement note, not from an external standard.
8. A plateau in the reflow curve: melting, or a loose thermocouple?
Q: When the curve stalls near the melting point (a plateau appears), is it a melting effect of the solder, or is the thermocouple loose?
A: Look first at the temperature at which the plateau appears, then rule out looseness. Looseness has a source; for a “melting plateau” we found no public source that discusses it directly.
The part with a source: the Wikipedia thermal profiling entry says a “jagged reading on the curve indicates a thermocouple that is not fixed well”; the Indium article lists two warning signs, one a ripple in the curve (a loose thermocouple), the other a second junction from broken insulation.
What follows is our judgment, not a standard clause: melting solder has to absorb latent heat, and with a thermocouple on the joint, the ramp rate may slow near the melting point, so a “plateau near the melting point” is not necessarily a broken wire. There are five ways to tell them apart, and the first needs no re-run. Before starting, rule out the soak stage the oven itself is set to, since that stage flattens the curve by design:
Look at the temperature of the plateau. Solder melting points have public figures: the Wikipedia solder entry gives 183 °C for Sn63Pb37 and 217 °C for the Sn-Ag-Cu ternary eutectic (SAC305’s melting range is not listed separately in the entry; go by your paste datasheet). If the plateau is at a temperature far from your solder’s melting point, it is not melting; being near the melting point is just a necessary condition, not proof (a zone boundary or a conveyor speed change can also make a plateau).
Run the same point through the oven 2 to 3 times (the fixing method must survive several passes; see Section 7). If the plateau appears at about the same temperature with a similar shape each time, melting is more likely; if the position differs each time and carries noise, a fixing problem is more likely.
Put two thermocouples on the same component, at nearby positions (for example two joints of the same part), and see whether both show it. Two parts of the same type at different places on the board see different thermal environments in the oven, so a difference in readings is not necessarily a fixing problem.
Put a thermocouple at a position on the board with no solder (an empty pad or the board surface) for comparison: a plateau that appears at points with solder is more likely melting. The fixing method at this point must be the same as at the points with solder; the thermal mass of the two places differs, so compare just whether the plateau appears, not the absolute temperature.
Shorten the sampling interval to see the detail (sampling interval and memory are in the next section); check the insulation and the bead, and if in doubt, change the wire and measure again.
Whether the measurement method itself is stable can be checked with the ideas of measurement system analysis; see repeatability and reproducibility in measurement system analysis.
9. Sampling interval and memory, and how often to replace the wire
Q: How should the recorder’s sampling interval be set?
A: First work out how long the memory can record.
The BGA profile recorder operating procedure (2003 edition, internal document) states: resolution 0.5 °C, accuracy ±1 °C (at 25 °C); the measurement rate is set by software from 0.1 second to 1 hour; the built-in memory holds up to 4,000 readings, so measurement time and sampling rate must be matched to avoid dropouts. This is an old specification, and current instruments have far more memory; here it is used just to show the calculation: recordable time = number of memory readings × sampling interval. This article uses 4,000 readings per channel as the example; for multiple channels, how the readings are divided depends on the recorder’s datasheet.
| Sampling interval | Recordable time at 4,000 readings | Is a 360-second curve (assumed) long enough? |
|---|---|---|
| 0.05 second (the old recorder cited here has no such setting; listed for demonstration) | 200 seconds | No (needs 7,200 readings) |
| 0.1 second | 400 seconds | Yes (needs 3,600 readings) |
| 0.5 second | 2,000 seconds | Yes |
| 1 second | 4,000 seconds | Yes |
360 seconds is an assumed curve length; your oven and conveyor speed differ, and so will the length. The denser the sampling, the more detail you see in the curve, but the faster the memory is used up. If memory is not enough, switch to a longer interval; do not let the curve be cut off.
Figure: sampling interval against recordable time (illustration; the memory comes from the old specification in an internal document, and the curve length is assumed).
Q: After how many uses should the wire be replaced?
A: We found no public standard that gives a number of uses.
The leads we can cite: the Indium article says aluminum tape does not survive many passes through the oven; the Wikipedia thermocouple entry says wire ages and drifts because the uniformity of the wire is lost; SMTnet says fine wire is fragile. Our suggestion (not a standard value): before each use, check whether the insulation is broken and whether the junction is loose. After a new wire is made up, fix two wires at the same point, run them through the oven, and record the difference between them; compare again at regular intervals using the same fixing method, and replace the wire if the difference has changed from the source by more than a value you set. That value must be larger than the combined range of recorder accuracy and wire tolerance (K type tolerance is in the table in Section 5; two wires that both pass, placed at the same point, can read as far apart as the sum of their tolerances on either side, so do not replace wire “at the first small difference”). Write the date and number of wire replacements into the profiling board’s history. The internal procedure (2002 edition) has a step for changing the profiling wire: tape K type wire with heat-resistant tape to the ball pads at the top, middle, and bottom of the substrate; but it gives no number of uses or interval for replacing the wire.
10. Writing the profiling board into the work standard
Q: What items should a profiling board work card include?
A: Write enough that another person, following it, can measure the same curve.
The table below is our own compilation; the fields follow the sources in the earlier sections:
| Item | What to write |
|---|---|
| Board | Part number and revision; for a double-sided board, top or bottom side; populated board or bare board |
| Measuring points | A coordinate drawing with numbers; which channel is used for air, hot spot, cold spot, heat-sensitive component, and BGA centre and outer row |
| Thermocouple | Type (K type), gauge, insulation material, length, date the wire was replaced |
| Fixing method | Material and amount; consistent among points to be compared (BGA noted separately) |
| Recorder | Sampling interval, whether an insulated box is used, which side of the board it sits on |
| Oven pass | Conveyor speed, where in the oven the board sits (for example the middle of the belt), whether the oven is fully loaded |
| Judgment | Which indicators (ramp rate, soak time, TAL, peak), the windows, and who handles anything outside them |
| When to re-measure | Product change, oven repair, paste change, board type change |
An example visible in the internal documents: one line’s first-article inspection checks just the peak, and each zone’s thermocouple gives an alert at ±5 °C and an alarm at ±10 °C, but who handles an alert and whether the line stops is not written in the document. You have to fill this gap yourself; otherwise the alert goes off and nobody knows what to do.
Further reading (external sources)
External sources (the two Indium pieces are published by a solder paste maker and have a stake):
IPC-7530B table of contents: Guidelines for Temperature Profiling for Mass Soldering Processes,
IPC-7530A ANSI preview pages,
Wikipedia: Thermal profiling,
Wikipedia: Thermocouple,
Wikipedia: Solder,
Indium: Thermal Profiling: A Practical Approach to Reflow Profiling (I-Connect007),
Indium: Best Place to Attach Thermocouples (Phil Zarrow),
NXP EB635: Lead-Free BGA Solder Joint Assembly Evaluation,
SMTnet: How to Profile a PCB.
Limits of this article
Kinds of internal data. The BGA ball-attach oven measurement procedure (2002 edition), the BGA profile recorder operating procedure (2003 edition), the oven setting and control specification (2002 edition), the reliability oven operating specification, an oven temperature experiment report for a SiP module (2012), one line’s reflow oven parameter control note, the reflow oven acceptance briefing (2012), and a wireless module’s oven profile report. All company names, personal names, equipment makers, and part numbers have been removed; just the methods and ranges of numbers remain.
External web pages were opened and read page by page. We read the content as it was at the time; the pages may be changed later.
The internal documents come from packaging and module plants, not from purpose-made SMT production boards. We take the methods. The oven-type documents (oven setting and control, reliability oven, SiP underfill oven) are not reflow ovens, as marked in the text. The oven setting and control specification requires a maximum load and 5 thermocouples (1 at the centre, 4 at the corners, 5–15 cm from the oven wall) when measuring the oven profile, held for at least 5 minutes after reaching temperature; each reliability oven is measured once a year with a 9-point recorder for temperature distribution. The oven specification’s revision record does not agree with its text (the revision record says the distance changed to at least 1 cm, while the text still says 5–15 cm); we do not take a side.
We read tidied index notes of the internal documents, and did not go back to check the source files word for word. Numbers quoted follow the notes; before going live, we suggest the fact reviewer spot-check them against the originals.
The internal documents are old. The oven measurement procedures are from 2002–2003 and the oven specification is the 2002 edition; today’s instrument specifications differ, and we have not updated item by item.
Some numbers in the internal documents are inconsistent with each other, and we have not taken a side. The reflow oven acceptance briefing’s peak deviation, mean, and range do not agree, so we cite just 6 points and ±5 °C; the oven specification’s revision record does not match its text; and why the two measurements of chamber temperature difference in the oven experiment differ so much is not explained in the document.
The text of IPC-7530 has to be purchased and we did not read it. For IPC-7530B we cite just the table of contents titles, and for IPC-7530A just the text visible in the preview; we did not read the definition of LTD.
The numbers from the SMTAI 2010 paper (collapse time at contact delays of 0 and 30 seconds) were checked against the source paper. The paper simulates the contact delay caused by component warpage; it is not a measurement of LTD and does not discuss LTD.
The two Indium pieces are published by a solder paste maker and have a stake. The I-Connect007 article page shows a placeholder date; we could not find the real publication date.
“The melting plateau is latent heat,” “the comparison at a position with no solder,” “keep a bare board just for oven comparison,” “the judgment for replacing wire,” and “the hole must not touch neighbouring pads” are our own judgments, not statements from a standard or an external article.
No search data. We checked just the Google Taiwan search box suggestions; “profiling board use,” “smt profiling board,” and “pcb profiling board” have suggestions, with no search volume figures.
Product page specifications come from our own product pages; go by the product page for wire gauge. We give no figure for the temperature limit of the insulation; go by the datasheet of the wire you buy.
Frequently asked questions
What is a profiling board, and where is it used?
Thermocouples are fixed at chosen positions on a product board and connected to a recorder that goes through the oven with the board, measuring the temperature curve at those positions to confirm the joints are within the range the solder paste specifies. The oven’s own thermocouples tell you whether the zones run as set; the two complement each other.
Can a bare board be used as the profiling board?
It can be used to compare oven conditions, but it does not stand in for the product. The Indium article divides boards into a baseline (bare board), a product profile board (fully assembled), and a calibration board; the product curve goes by the populated board, and a double-sided board needs separate top and bottom curves.
How many measuring points, and where?
The Indium article says at least 4: air beyond the leading edge of the board, a hot spot, a cold spot, and a heat-sensitive component; for a detailed curve, 5 to 6 or more. Hot spots are usually bare copper pads or small passive components; cold spots are under plastic or ceramic IC packages and at the leads of coils and transformers.
Should I keep an air line?
We suggest it. The Indium article lists it as one of the minimum 4, and the benefit is a common baseline in each measurement, which makes it easy to compare curves from different batches.
How do K type and T type thermocouples differ?
K type is chromel/alumel with a range of −200 to +1350 °C; T type is copper/constantan with a range of −200 to +350 °C. For tolerance class 1 as listed by Wikipedia, K type is ±1.5 °C from −40 to 375 °C, and T type is ±0.5 °C from −40 to 125 °C. The Indium article says K type is the usual choice in SMT; the recorder and the wire must be the same type (general practice).
What wire gauge should the thermocouple be?
The Indium article recommends 36 AWG (about 5 mil) for measurement, and 30 AWG is also acceptable. Fine wire is fragile; thicker wire is more durable but has larger thermal mass, so a reading in a fast-rising section may lag by a beat (this is a general principle, not Indium’s statement).
A plateau appears in the reflow curve. Is it melting, or a loose thermocouple?
First rule out the soak stage the oven itself is set to, then look at the temperature of the plateau: Wikipedia lists 183 °C for Sn63Pb37 and 217 °C for the Sn-Ag-Cu eutectic, and a plateau far from your solder’s melting point is not melting; being near the melting point is just a necessary condition, not proof. Then rule out looseness: Wikipedia says a jagged reading means the thermocouple is not fixed well, and the Indium article lists curve ripple and a second junction from broken insulation. Our suggestion is to run the same point through the oven 2 to 3 times: a plateau whose position is stable leans toward melting, and one that differs each time and carries noise leans toward a fixing problem; a thermocouple at a position with no solder can serve as a comparison, and a plateau that appears at points with solder leans toward melting. These are our judgments, not a standard.
How is a BGA measuring point buried?
Drill a small hole from the back of the board, without going through the solder ball or the chip, and connect the thermocouple at the boundary between the ball and the pad. The hole is small, about 1.0 to 1.5 mm for a small thermocouple bead (for a fine pitch, the hole must not touch neighbouring pads, and the diameter is judged for the part); the array centre runs cooler (Indium article), so we suggest measuring the centre and then adding the outer row. Fix with thermally conductive epoxy, then fill the hole with non-conductive epoxy.
What is LTD?
The IPC-7530B table of contents lists the term Liquidus Time Delay (LTD); the title of Figure 3-11 in the IPC-7530A preview is “Role of LTD in Head on Pillow,” so it is related to the head-on-pillow defect. The definition text has to be purchased and we did not read it, so we do not explain what it means.
How should the recorder’s sampling interval be set?
First work out recordable time = number of memory readings × sampling interval; the curve length must be shorter than it. Taking 4,000 readings per channel as an example (the memory of an old internal procedure), 0.1-second sampling records just 400 seconds; for multiple channels, how the readings are divided depends on the recorder’s datasheet.
After how many uses should the wire be replaced?
We found no public standard that gives a number of uses. We suggest checking the insulation and junction before each use; after a new wire is made up, fix two wires at the same point and record the difference, then compare again at regular intervals (using the same fixing method) and replace the wire if the difference has changed from the source by more than a value you set (the value must be larger than the combined range of recorder accuracy and wire tolerance). This is our suggestion, not a standard value.
For measuring items, Jiahon supplies the reflow oven profiler and K type thermocouple profiling wire.