Upgrade T-962 with Carbon Fiber Infrared Tubes: Feasibility and Mounting Solutions?

[Konrad1194]

Background

After researching improvements to the T-962 oven, I found a detailed explanation from [Gradient SG’s page](https://en.gradient-sg.com/t962/), which argues that carbon fiber infrared lamps are superior to traditional quartz tubes for soldering applications due to their longer wavelength. The blog post says:

The best heater for soldering are the ones that have the longest wavelength radiation. These are lamps with a carbon heater. They are very fast and the PID works great with them. In the picture below you can see the dimensions of the heating tubes for the T962. I couldn't find these so I bought the regular ones 220v 300Watt from this store: Infrared Quartz Tube Heater 25cm

The still working link for replacement tubes lead to the standard quart infrared heating tubes:

Current Situation

In the past, matching replacement carbon fiber lamps were hard to source, but I recently found affordable alternatives (10.99€ each) on AliExpress:

• [Product link (AliExpress): 4000562513957](https://de.aliexpress.com/item/4000562513957.html)

Unfortunately the mounting of the lamp is not 100% matching the M4 connection of the old lamps, but it might be possible to use these lamps and with it their advantages of energy efficient and fast heating. I am willing to tackle this topic, but I would highly appreciate your input before I start this new endeavour.

Proposed Upgrade

I’m considering retrofitting these carbon fiber lamps into my T-962. My goals are:

• Benefit from their improved energy efficiency and faster, more even heating (longer wavelength FIR),
• Achieve better reflow profiles and soldering results,
• Solve any mounting challenges (possibly with adapters or custom brackets).

Request for Feedback

Before I proceed, I would appreciate any input from others who have:

• Attempted a similar upgrade,
• Experience with custom-mounting carbon fiber lamps,
• Advice on potential pitfalls (electrical compatibility, PID tuning, mechanical stability, safety, etc.),
• Thoughts on whether the benefits are as substantial as described.

Is this a worthwhile upgrade in practice? Any modifications required for the best results?

[GitLang]

My first question having read it is why bother?. I am worried by the statement "The best heater for soldering are the ones that have the longest wavelength radiation." Where is the evidence that supports this?. I don't mean the comparative graph, I mean the absorption. Sure the best source is one that matches the wavelength of the peak absorption of the item being heated, and this will vary a lot with material.. I have said it many times before, and I remain convinced, that the main heating of the parts is the air temperature and plain old conduction. Yes, radiation must play a part, but we don't control the radiation we control the air temperature. Big industrial soldering tunnels may be a different matter, but on a small DIY oven it's the air doing the job. Maybe the carbon lamps better match the absorption of heat by the air, I don't know, but even if it does all it will give you is slightly better efficiency. The fact that the current PID values work fine tends to work against the heat absorption by the air (or even radiation absorption) being very different than the normal lamps.

[sevstels]

GitLang I am too lazy to give you a course of lectures on the basics of electrodynamics and propagation of electromagnetic waves in a closed space. It would be long and not interesting to you. So take my word for it - you are deeply wrong.

Regarding PIDs... In the firmware from Unified Engineering PID does not work as it should due to a few bugs in the code. It is in relay mode.

Konrad1194 You can ask the factory to saw off the tubes the way you want them. And put in the terminals you want. If it is the manufacturer, they will do it without any problems.

[GitLang]

Thanks, but I already had that course somewhere between my 1st Hons degree in electronics and solid state Physics, and my PhD in Physics, But it's not lectures or qualifications I am basing my replies on, just 50 years experience in industry and academia actually doing these things and fulling customers needs. Couldn't help but pick up a some knowledge on the way things work in the real world.

[sevstels]

Wonderful. Now it's my args. I have an excellent university degree in electromagnetic wave propagation. I know and understand exactly what I'm doing... And now I'm making a quantum computer at NUS at the Institute of Quantum Physics. Do you want to argue further?

[Konrad1194]

I appreciate your answers, and I will try to reply to both :)

@GitLang Thank you for your input, I understand your concerns regarding radiation, and you already mentioned that it might be different in an industrial setup. I found the following research, which at least shows that IR heating is indeed a factor in SMT soldering. This is very well explained in the thesis by Mr. Armendariz from 2021:

"Development of a Laser Solder Process for use with Material Extrusion Additive Manufacturing and Rapid Electronics Prototyping in Embedded Sensing Applications" (2021)

Citing from the thesis:

Infrared (IR) and convection soldering is the most widely used for reflow soldering (or in
combination), as they allow for mass soldering of multiple PCB’s by traversing them through IR
furnaces or convection ovens on conveyor belts, typically through heating zones for different
durations to emulate time-temperature profiles. IR soldering involves transfer of heat via
absorption of radiant energy transported by electromagnetic waves. The infrared radiation used in
this method lies between 0.72 to 1000 µm on the wavelength spectrum, with most useful ranges
lying in the short/near wave (0.7 to 3 µm) to far wave regions (3 to 8 µm) (Hutchins, 1991).
Radiation wavelength is a function of the emitter source’s temperature and the reflectance
characteristics of objects targeted, i.e., their ability to absorb and reradiate heat energy (Hutchins,
1991). The “volumetric” heat transfer takes place indirectly as the molecular structure of objects
absorb IR energy before increasing local temperatures, with organic materials (solder paste, flux,
43
substrate, etc.) absorbing most of the heat energy. As such, the PCB will get hot first, transfer heat
to the solder pads, then the solder piles, then to the leads and/or terminations on components. The
major difficulty with IR reflow is that it is not an “equilibrium” method of heating, which suggests
that the object being heated will never reach the same temperature as the heat source. Quartz lamps,
a common IR heat source, can achieve high temperatures of 1000°C and must surpass the safe
reflow temperature of 220°C to enable heat transfer throughout all constituent materials before
allowing solder paste to reflow without damaging SMD’s. Additionally, temperature differences
throughout the length of a PCB induced by uneven heating (outer areas get hotter then center areas)
of various PCB sizes and configurations makes control of time-temperature profiles difficult.
PCB manufacturers must depend on accurate pulse operation of heaters and belt speed adjustments such
that heated PCBs exit a heating zone at the correct time-temperature (Hutchins, 1991).

Another interesting source is here:
https://web.ics.purdue.edu/~gblack/RI_Section_5.pdf

and here (favoring convection, ignoring higher wavelength):
https://apr-rework.descoindustries.com/pdf/[Rework-Technology-IR-vs-Convection.pdf](https://apr-rework.descoindustries.com/pdf/Rework-Technology-IR-vs-Convection.pdf)

Temperature control via air

Regarding temperature control, I am also not very sure, as the sensor will heat up based on radiation plus convection as well as the PCB. The difference in size and geometry of the sensor and the PCB is apparent and might lead to larger errors — especially as the positioning of the sensor is crucial since radiation intensity decreases quadratically with distance.

I wonder if, with a carbon heater, it might be beneficial to place temperature sensors on a dummy PCB in a well-lit corner and have one air temperature sensor to limit the air temperature, as I am unsure if the PCB will lag behind air temperature. At the latest, to avoid overshoot, proper cooling is necessary.

Based on absorption, I am sure that plain metal pads reflect most radiation, but solder paste will heat up very well, and depending on the solder mask color, these areas are affected as well.

Maybe this idea is also wrong as in the Purdue White Paper is is written:

IR dominant systems usually range in the convection/IR ratio of 60/40 to 40/60. Convection dominant systems have ratios in the range of 70/30 to 90/10

to me this means that using air sensors I might still be able to control the heater properly?

@sevstels Thank you for your feedback. I was hoping you would respond, as it was originally your blog post that inspired this idea. I am confident the mechanical integration will work fine, but I remain somewhat skeptical about whether the current temperature sensing setup is still valid. In one case, I am intentionally heating up the air, which can be measured with thermal sensors. However, with longer wavelengths, I tend to heat objects directly, which then convect heat to the air—while I am measuring only the air temperature.

I appreciate all your input, to get into the field before blindely starting with something which is not helpful.

[sevstels]

Konrad1194 It's not clear to me, why you went back to making the furnace again? It's already lost any economic sense. The chinese solder PCBs to order, fast and cheap. About 1$ per board, following all technological norms. There's no point in fighting it, just using it.

[Konrad1194]

@sevstels it is rather a hobby project like a 3D Printer. I decided to go for the T962+ and now I have it and make the best out of it 👍 I would love to be able to do the full stack from the idea to the board at home, even if this is economical nonsense and means I loose money :D (I have all tools to just solder it by hand, but thats no challenge)

[Konrad1194]

Next to many interesting Issues, there is also more discussion on this repository in the pull request section.

Most of my question were already discussed in the following pull request, and their tests and knowledge is massive!

#132

It seems proven, that the T-962 (small version) atleast, is a radiation oven and also that the PCBs they used for their tests are heating up based on radiation more than convection. The discussion also involved a strong background on the theories for this particular oven. There also seems to be a ton of not merged pull requests. Based on these findings the cadon fiber infrared seems to be very interesting, despite the claim that 3K/s were a good enough ramping phase and it might be even dangerous as the PID is lagging, especially as the Sensor is air temperature based. However one can simply tune the PID, and make sure that we are actually close to the boards temperature.

[Konrad1194]

T962 Reflow Oven Modification Plan

Goal 1: Validate Radiation Characteristics

• Objective: Confirm that the T962 operates primarily as a radiation oven by verifying that PCB temperatures exceed ambient air temperature during reflow.

• Test Criteria:

  • 1.1) Comparing air temperature vs. PCB temperature inside the oven (in ramping phase).

• Rationale:
  Infrared (IR) radiation is the dominant heat transfer mode in the T962. Validating that the PCB reaches higher temperatures than the air confirms that IR heating dominates convection. This is critical to justify focusing on improvements in IR reflection management.

• Tools:
  4-K-type Thermocoule Board description from the Wiki

  PCB: from prefered manufacturer

BOM, e.g. from LCSC (Digikey, Mouser etc. were way more expensive, but I am waiting for customs):

LCSC Part Number	Manufacture Part Number	Manufacturer	Package	Description	Quantity	Unit Price (€)	Ext. Price (€)
C476179	MAX31850KATB+	MAXIM	TDFN-10-EP(3x4)	3V~3.7V ADC/DAC, TDFN-10 package	5	5.0168	25.08
C5188448	MX128-2.54-08P-GN01-Cu-Y-A	MAX	Through Hole, P=2.54mm	1x8P 6A Screw Terminal Block	5	0.6121	3.06
C99782	RC0603FR-074K7L	YAGEO	0603	4.7kΩ 100mW Thick Film Resistor, ±1%	100	0.001	0.10
C14663	CC0603KRX7R9BB104	YAGEO	0603	100nF 50V X7R Ceramic Capacitor, ±10%	100	0.0022	0.22
C100042	CC0603KRX7R9BB103	YAGEO	0603	10nF 50V X7R Ceramic Capacitor, ±10%	100	0.002	0.20

Thermocable Type K, AWG 30 (for rapid temp. changes and reduced lag:)

Spot welding the end --> missing documentation for spot welding device

---

Goal 2: Increase IR Reflection and Remove Wool Insulation

• Objective: Upgrade existing stainless steel internal surfaces with aluminum-coated insulation to leverage aluminum's superior IR reflectivity; remove existing glass wool insulation for cleanliness while still maintaining heat management.

• Scientific Background:

Material	Emissivity (ε)	Reflectivity (ρ = 1 - ε)	Typical Surface Condition
Polished Aluminum	0.04 – 0.09	91% – 96%	Vapor-deposited Al coating (matte to bright)
Polished Stainless Steel	0.10 – 0.20	80% – 90%	Mirror-polished surface
Unpolished Stainless Steel	0.30 – 0.60	40% – 70%	Oxidized, rough surface

• Aluminum surfaces reflect roughly 95% of IR radiation in the 0.7–14 μm wavelength range typical of reflow ovens (temperatures ~200–300 °C). In contrast, stainless steel reflects in a best-case scenario ~80–90%, dropping further if oxidized or rough.

• Test Criteria:

  • 2.1) Measure air temperature above the chamber with existing wool insulation.
  • 2.2) Measure air temperature above the chamber after removing wool.
  • 2.3) Measure air temperature above the chamber with existing wool insulation and added aluminum-coated insulation.
  • 2.4) Measure air temperature above the chamber after removing wool. and added aluminum-coated insulation
  • 2.5) Check if the PCBs temperature ramping speed is harmed in 2.4)

• Motivation:
  Leveraging a material with ~95% IR reflectivity (aluminum) over stainless steel with potentially less than 70% reflectivity improves thermal energy retention inside the oven, leading to:

  • More efficient IR heating of the PCB.
  • Reduced power consumption (less wasted energy heating the chamber walls).
  • More uniform temperature distribution.
  • Elimination of glass wool improves cleanliness, lowers contamination risk, and enables easier access for adding sensors through the ceiling.

• Tools:
  Thermal Imaging -> Risk that it is nonsense, but good first estimate, before thermocouple readings

• Reference material:
  Aluminium Isolation 1mm
  Aluminium Isolation 5mm

Goal 3: Upgrade IR Heating Elements

• Objective: Replace existing IR tubes with carbon IR tubes to improve the emission spectrum and heating efficiency.

• Scientific Context:
  Carbon IR tubes emit a stronger and broader IR spectrum peaking near 2.5 μm, matching PCB absorption better than traditional quartz tungsten tubes, which peak near 1.2 μm. This spectral match can improve:

  • Heating speed.
  • Uniformity of temperature across the PCB.

• Test Criteria:

  • 3.1) Measure and compare air temperature (ramping & static) before and after carbon tube installation.
  • 3.2) Measure and compare PCB temperature (ramping & static) before and after carbon tube installation.

• Reference material:
  Carbon Fiber Heating Tube

---

Goal 4: Enhance Accurate Temperature Sensing

• Challenge: Repeatedly gluing thermocouples to PCBs for temperature reference is impractical for ongoing testing.

• Solution:
  Install a spring-loaded K-type thermocouple sensor with a small spiky measurement tip to achieve reproducible, direct contact on a PCB via or sacrifice one well connected bare pad.

  • Typical K-type thermocouples have a temperature range up to ~1250 °C, with standard accuracy ±1.5 °C or better.
  • Spring-loaded probes maintain constant contact force (approx. 2–3 N) for stable readings.

• Potential Limitations:

  • Sensor size may shadow parts of the PCB and slightly alter local heating.
  • Sensor placement near IR tubes must avoid overheating and shading effects.
  • Sensor measures through solder mask (less effort, but slightly less accurate).

• Reference Sensor:
  Bajonettfühler - Thermoelement Typ K

---

References and Additional Notes

• Stefan-Boltzmann Law: Radiative heat exchange per unit area is proportional to emissivity and ( T^4 ) (Kelvin). Lower emissivity means higher reflectivity and less radiative energy loss.
• IR Wavelengths: Typical reflow temperatures (~250°C) correspond to peak IR emission wavelengths around 8–12 μm (Wien's displacement law), where aluminum coatings perform excellently—though this is somewhat outside the emission range of the carbon lamps.
• Material Aging: Over time, aluminum coatings may oxidize, slightly increasing emissivity but still outperform stainless steel surfaces.
• Thermal Isolation: Aluminum coatings combined with layered glass-fiber insulation synergize reflective and conductive/convective insulation to maximize heat containment.

---

The appealing structure of my plan was enhanced with support from Perplexity AI.