Accuracy and the Influencing Factors

<< Click to Display Table of Contents >>

Navigation:  Sensor Platforms > WaveRIDER® NL 2 >

Accuracy and the Influencing Factors

There are two fundamental measurements the WaveRIDER® NL 2 measures: temperature and time. The accuracy, resolution and range the WaveRIDER® NL 2 is capable of is stated in the Specification section of the manual. The purpose of this paper is to show how these specifications affect the solder machine parameters presented on the WaveRIDER® NL 2 printout.

 

Time Specifications

Measurement time by the WaveRIDER® NL 2 is influenced more by the log rate rather then the overall accuracy of the timer. The WaveRIDER® NL 2 takes data at a log rate of 0.1sec. Thus every reading of time has a tolerance of ±0.1 second. The net effect on time dependent variables is best illustrated on the graphs below.

 

Dwell Times

The % effect that ±0.1 seconds has on the measured dwell times decreases as the dwell time increases. To determine the maximum error of dwell time measurements, locate the measured dwell time on the X axis of the graph and run vertically to the curve. Now run horizontally to the left and find the % error on the Y axis. Clearly the maximum % error that 0.1 seconds has on the measured value increases dramatically as your dwell times approach 0.1 seconds (100% error!), but practical dwell times run from 2 to 5 seconds or 5% to 2% error respectively.

 

Conveyor Speed

The % effect that ±0.1 seconds has on the measured conveyor speed increases with the actual conveyor speed. To determine the maximum error of conveyor speed measurements, locate the measured speed (ft/min) on the X axis of the graph and run vertically to the curve. Now run horizontally to the left and find the % error on the Y axis. This is the maximum % error of actual conveyor speed you could expect to see. Typical errors run at about 0.75% at 4ft/min.

Important Information

Time measurement can be altered if there is one or more open T/C(s) on the coupon.

Contact Lengths

Contact length is the product of conveyor speed and dwell time. Thus the % errors of both must be added to determine the total maximum % error for contact length. To determine the maximum error of contact length measurements, locate the measured dwell time on the X axis of the graph and run vertically to the curve that is closest to the measure of conveyor speed. Now run horizontally to the left and find the % error on the Y axis. As you can see from this graph, conveyor speed has little effect on error as the dwell times get much below 2 seconds. Typical errors range from 4% to 6% of the actual dwell time depending on the conveyor speed and the dwell time.

 

Temperature Specifications

Absolute temperature accuracy is dependent on four basic factors:

1) The accuracy of the thermocouple sensor (T/C)

2) The accuracy of the Cold Junction Compensation sensor (CJC)

3) The accuracy of the overall Analog to Digital system (A/D)

4) The conformity to the International Practical Temperature Scale (IPTS)

Other factors could be mentioned here but they are either too small to be concerned with or are “lumped” in with those mentioned above.

 

The maximum possible error for a given temperature reading is the sum of the errors listed above. For the WaveRIDER® NL 2 this maximum would be:

 

T/C ± 4°F + CJC ± 4.5°F + A/D ± 0.3°F + IPTS ± 0.1°F = ±8.9°F

 

This may seem extreme but it is the theoretical maximum error (see Specification in the WaveRIDER® NL 2 manual). However, we calibrate each WaveRIDER® NL 2 to ±1.8°F (1°C) at 212°F (100°C) and 1832°F (1000°C). With conformity to IPTS at 0.1°F and A/D error at 0.3°F, the practical maximum error would be (when using a traceable voltage standards as the calibration source):

 

± 1.8°F + IPTS ± 0.1°F + A/D ±  0.3°F  = ±2.2°F

 

This leaves only the T/C error to deal with. The WaveRIDER® NL 2 uses “Standard Limits of Error” T/Cs which have ±4°F tolerance. Practical use of these standard T/Cs shows that these tolerance limits are rarely approached. Tighter tolerance T/Cs can be used at extreme cost, with a tolerance no better then ±2°F. This small gain in accuracy is not worth the extra cost, considering the temperature range being measured by the WaveRIDER® NL 2 (<500°F typically).

waverider-0072

waverider-0074

waverider-0076

Solder Wave Variation

Solder wave variation happens. It is by nature a dynamic process caused by the pumping of a liquid. Through various channels, weirs, rudders, and nozzles, all designed to balance flow across the surface of the wave, liquid solder makes its way to the surface and finally breaks one way or the other depending on the wave shape. Lacking the presence of a printed circuit board, this wave looks even and more or less  smooth on typical wave solder machines. However, some ripples and eddies are visible as the flow of solder moves toward the spill point. These ripples are everywhere and only some of the bigger ones are pointed out in the picture below.

waverider-0078

Important Information

Typical Ripples and Currents that appear on otherwise smooth looking waves.

Currents running throughout the wave take on a whole new manifestation when they run into the underside of a printed circuit board. They will cause the crest of the wave to vary, much like ocean waves as run up a sandy beach. An ocean wave will never rush up the sand and stop in a perfectly straight line even though the beach may be very flat. Thus the crest of the solder wave, as it flows against the underside of a printed circuit board, will be very dynamic. You can view this using a tempered glass plate, such as the one called “Lev Cheek” by Hexacon Electric, as it runs slowly across the solder wave just like you would solder a printed circuit board. A typical solder wave using this glass plate is pictured below.

waverider-0079

Important Information

Typical wave crest pattern as viewed through a tempered glass plate.

The crest of the wave is highlighted to make is more visible. As you can see, the edge is very rough and is constantly changing shape as it progress across the board. It is also noticed that the sides of the wave tend to be narrower then the middle.  Based on this snap-shot, one would expect this wave to read parallel with sensors “A” and “C” reading lower contact times then “B.” This high in the middle observation is typical in many waves and is not because the glass bent down in the middle. It is most likely due to increased pumping action near the center of the wave, which may also be causing the increased ripple action near the center as seen in the first wave illustration. Also, with this typical wave dynamics one should expect successive measurements to have some variation. In the above illustration, at 4 ft/min conveyor speed, one can expect up to 0.8 seconds variation from sensor to sensor.

 

The real issue in making exact measurements from your solder wave is to determine what your solder process can tolerate, and not so much to make the solder processes perfect. Clearly some parameters are more important such as conveyor speed, solder temperature, and preheat slopes, but to expect a pumped liquid as dense as solder to behave with perfect geometry is asking a bit much from the machine manufacture. If your solder process is not producing good results, a measurement tool will quickly point out the problem. But the goal of measuring a good process over and over is to obtain data which can be used to determine how much variation is acceptable and help predict when a good process is going bad.