We have been discussing how to conduct temperature uniformity surveys (TUSs) and system accuracy tests (SATs) to comply with specifications such as AMS 2750D (Pyrometry), and we want to continue to expand that discussion. Let’s learn more.

Fig. 1. Single-chamber vacuum furnace temperature-survey rack in position (Photograph courtesy of Solar Atmospheres Inc.)


Fig. 2. Typical temperature uniformity survey chart (Photograph courtesy of Super Systems Inc.)

Tips for Conducting a Successful TUS and SAT - Vacuum Furnaces

Here is some advice gathered from years of doing surveys and from a number of industry experts on what you should watch out for when conducting temperature uniformity surveys in vacuum. Specifically:

1. The advertised “working zone” must be the same size as any racks or setup used (Fig. 1). In the field, it is not uncommon to find survey racks or thermocouples placed well within the actual working zone of the furnace. Nadcap auditors, for example, often measure the rack and compare it against the advertised working zone. If the advertised working zone is bigger than the rack, it will be a major noncompliance issue.

2. Make sure the thermocouples (used) are correct to the certification of calibration. Correction factors can be as much as 2.2°C (4°F) or 0.75% of the reading (whichever is greater), and if the wrong correction factors are used, the survey can be highly inaccurate.

3. Make sure the location of the sensors is correct to the record. If performing a TUS, be sure adjustments made to bring the working zone into a better uniformity and the sensor locations are correctly logged. It is not uncommon to find people adjusting locations incorrectly, making it worse and wasting time.

4. When crunching the numbers afterwards, make sure the approach to temperature is included in the “high” reading calculation for the test sensors and any other recording sensor(s). A sensor could overshoot a little, and if you do not analyze the approach and include it in the report, you are not certifying the furnace accurately.

5. Make sure the thermocouple (TC) wires are not bent sharply or stretched too tightly when routing them to their respective locations. The wires themselves go through phases of expansion and contraction. If too tight, they can become shorted during the process. A shorted reading will most certainly skew the results or cause a survey to be aborted.

6. Make sure to include the known deviation of the portable TUS recorder into the correction of the readings. An instrument may be off as much as 1°F or 0.1% of the reading (whichever is greater). This could add a deviation of 1.1°C (2°F) at 1095°C (2000°F). If not compensated for, this can seriously affect the results of the TUS, and it could be a major finding in a Nadcap audit.

7. Do not “step” the ramp for a TUS (to prevent overshoot). It is never a good idea to attempt this in production either. For example, do not ramp 16.5°C (30°F) a minute to 510°C (950°F), then 5.5°C (10°F) a minute into the 540°C (1000°F) setpoint of the TUS. It is imperative that production ramp rates are used for the TUS (Fig. 2). An auditor knows this can be a bad habit and will typically look for this. Again, this is a major finding should it be noted in a Nadcap audit.

Fig. 3. Typical jack panel (Photograph courtesy of Ipsen USA)

8. Always measure the insertion of the controlling thermocouple before each survey. It must always be the same, which is critical to setpoint versus uniformity distribution. If a thermocouple is replaced and inserted to a different depth, the uniformity from setpoint may change.

9. Thermocouple wire feed-through assemblies and jack panels are an insidious source of EMF error. If using a thermocouple jack panel, make sure the panel is absolutely clean between the legs as even the slightest film could produce errors in the readings (Fig 3). Always keep dummy plugs in unused jack locations. The wires from the chamber feed-through to the back of the jack panel are also a potential source of error. Fiber insulation attracts and holds contamination and water vapor that can bridge the gap between the wires, causing a slight EMF error. It is important to keep in mind that even on the more forgiving type-K thermocouple, the difference between 300°C (572°F) and 301°C (574°F) is a mere 0.041 millivolts. It doesn’t take much contamination to cause an error in the difference between the hot and cold junction of a thermoelectric circuit.

10. With a calibrated run-up box (thermocouple generator source), inject a signal into the jack panel to simulate a given temperature through every channel to ensure correct readings back to the official recording system. On vacuum systems, temperatures in the ranges of 93°C (200°F), 538°C (1000°F), 800°C (1475°F) and 1205°C (2200°F) are not uncommon input signals.

11. On average, TC jack-panel feed-through plates should be replaced every year as the female connecters become impregnated with process debris and hot-zone trace elements. Coupled with this, the springs become annealed causing loose connections and subsequent errs in the readings.

12. Be absolutely certain that the instrument used to calibrate the portable TUS recorder meets the accuracy requirements of a “secondary standard,” which is ±0.3°F or ±0.05%, whichever is greater. Many common field-test instruments do NOT meet this accuracy requirement. If you outsource the calibration, you should always have a current copy of the calibration from the field test instrument (used) and make sure it meets this requirement. In addition, the calibration should have been performed as the instrument is used. For example, if the field-test instrument is used as a thermocouple simulation source (sending the temperature signal to the device being calibrated), the appropriate thermocouple types should be checked while the instrument is in the source mode. Having a calibration certificate that shows how accurate an instrument is in read mode does not always qualify it in the source mode. Communicating these kinds of requirements to your supplier is very important to getting the service you require to pass your Nadcap audit without findings.

13. First, check the element resistance to ground. For example, one manufacturer’s equipment is designed for 87-110 ohms prior to attempting a temperature survey. Do not attempt a survey if the resistance is bad on any particular zone. The readings should be fairly balanced prior to attempting a survey.

14. Ensure load TCs are new and preferably (not required) from the same lot of source wire for ease of tracking deviation factors.

15. Make sure that your load TCs maintain correction factors that are less than the survey spread allowance. For example, surveying for a spread of ±5.5°C (10°F) and utilizing load TCs with correction factors of ±6°C (11°F) or higher is not desirable.

16. Extrapolating setpoint temperatures on load TCs in steps greater than 120°C (250°F) are not allowed. Load TCs for vacuum furnaces tested to AMS 2750D are supplied with certification sheets typically from 540°C (1000°F) up to 1205°C (2200°F) in 120°C (250°F) steps. Hence no extrapolation required.

17. Before starting a TUS survey, check that the furnace has recently passed its reference probe check.

18. Check all element connections to be sure that they are tight and show no signs of arcing.

19. Remove all lower-temperature materials and alloys from previously processed runs prior to conducting surveys.

Fig. 4. TUS rack with metallic-sheathed thermocouples (Photograph courtesy of Vac Aero International Inc.)

20. Where possible, use metal-sheathed load TCs (not cloth versions) (Fig. 4). Nadcap may ban the use of cloth TC in vacuum environments due to the vaporization of the actual TC materials coupled with impregnation of process (vaporized trace elements, which all lead to downward temperature readings).

21. To prevent transient noise from reaching the control system, use load TCs that are ungrounded.

22. Always utilize a temperature-survey fixture with appropriately sized heat sinks unless otherwise specified by a particular customer-driven internal specification.

23. Always center the temperature-survey fixture front to back and left to right for consistencies in readings and repeatability in future surveys.

24. Design the load-survey fixture and heat the furnace at a ramp rate and vacuum level that best emulates your process requirements.

25. After heating to the first test temperature, allow the furnace ample time to reach steady-state power consumption prior to entertaining any adjustments. For most vacuum furnaces this will be approximately one hour at the first setpoint.

26. Observe the temperature spread throughout the load at a given setpoint before making adjusts (e.g., manual trimming, PID values and offsets). Remember that the best adjustment strategy may depend on instrument type or manufacturer.

27. If available, adjust zones in order to balance the load front to back and/or top to bottom again depending on particular product line and manufacturer. Adjustment must be made slowly at the lower temperatures and one zone at a time. The cause-and-effect relationship from these adjustments is not realized for at least 15 minutes after the adjustments are made. Patience is a virtue that manifests itself when running temperature surveys.

28. All adjustments on an older system need to be made and locked at the lower temperature. Newer systems fall under slightly different guidelines.

29. One should never be too concerned about deviation from setpoint when making the preliminary trimming adjustments. Remember that trimming yields uniformity while control TC placement rectifies load-to-setpoint anomalies.

30. Once trimmed at say 540°C (1000°F), the low TC reading should be at least 540°C (1000°F) while the high TC reading is still within the required uniformity range. The goal is to have the mean within specification but on the high side of the lowest setpoint, as this ratio will drift downward during incremental upward temperature steps through the survey.

31. Prior to moving away from the 540°C (1000°F) adjustment permissible range, the mean of the load is now adjusted to reduce deviation from actual setpoint by adjusting the control TC either inward or outward depending on what direction the survey fixture needs to be adjusted to be compliant to control.

32. Each progressive setpoint will take less time to soak out because the radiation levels are increasing over the black-heat area of the 540°C (1000°F) range.

33. Be sure in-house and AMS 2750D documents for sampling rates, time at temperature and stabilization periods are compatible.

Final Thoughts

The requirements of AMS 2750D for temperature uniformity testing are an excellent place to start when developing a statistical capability study for any type of heat-treating furnace or oven. These types of surveys give the heat treater a greater level of confidence that the equipment will be able to perform to the required temperature tolerances. When working to the requirements of AMS 2750D, proper planning avoids failing a survey. If a survey fails, all jobs that were run in this furnace to the AMS 2750D requirement would have to be reviewed for affect and disposition. In a job-shop environment especially, this can be a daunting task. Using statistical analysis techniques to monitor the temperature uniformity can eliminate such discrepancies. IH