
A custom NTC sensor housing is the part where the customer’s number decides the shape, and the number is usually given late and in the wrong units. So the useful thing about this page is not the housing – it is what the NTC’s own specifications mean for the metal around it, because four of them are requirements the housing has to meet and almost nobody states them.
Drawn on hydraulic presses and CNC-finished where tolerances tighten, at +/-0.01 mm on sealing faces and +/-0.05 mm on drawn profiles, with critical dimensions carrying CMM data. The tolerance scheme is right; what it is measuring is worth examining.
What the NTC’s specifications actually require of the housing
An NTC thermistor’s datasheet carries four numbers that describe the element. Each of them becomes a requirement on the housing, and three of them are requirements nobody writes on the housing’s drawing.
- The resistance at 25 degrees. This is the element’s, and it is temperature-dependent, so it is the number the customer’s calibration is built around. It tells the housing’s maker nothing directly – but it tells them that the housing must not introduce a temperature offset between the element and the process, because an offset shows up as a calibration shift.
- The beta value. How steeply the resistance changes with temperature. A high-beta element is more sensitive and more sensitive to any temperature difference between the element and the process – so a high-beta sensor makes the housing’s thermal path accuracy more important, not less. This is a real and unstated coupling.
- The dissipation constant. How much power the element can take before it self-heats. This is where the housing’s geometry enters directly: an element in a sealed cavity cannot dissipate its self-heating, so it reads higher than it should. A thermowell’s cavity has to be sized so the element’s self-heating is either negligible or accounted for.
- The time constant. The element’s thermal response. This is the number the housing’s wall and cavity make, and it is the reason a housing designed for a different NTC will not behave the same as one designed for this one.
So the most useful thing to ask at enquiry is: which NTC is going in this housing? And the second: is the customer’s calibration based on this housing? Because if it is, then the housing is part of the calibration and changing it changes the sensor’s output, which is a much bigger deal than changing a piece of metal.
The wall, and why it is the accuracy
An NTC measures by having its own temperature match the process. Everything between the process and the element’s sensing surface – the wall, the cavity, the fill, any air gap – is in the way, and each of them degrades the measurement differently.
- A thin wall in a conductive material is the best case. Aluminium’s wall would be the fastest; stainless steel is twenty times worse and a brass wall is about six times better than stainless steel. So on a fast-response NTC the housing’s material is a thermal decision first.
- A thick wall is a lag, not an offset. And a lag is worse than an offset for a control system, because it means the sensor is reporting a past temperature during any transient.
- An air gap is the worst case, because air is a very poor conductor. If the element sits in a cavity that is not filled, the cavity’s dead volume is an insulating layer between the element and the wall.
- And the cavity’s size sets the response time, not the element’s own time constant alone. A large cavity adds thermal mass that has to equilibrate before the reading is right.
So the practical specification on this part is: the wall’s thickness, the cavity’s volume, and what is in the cavity. Those three, together with the element, set the sensor’s response – and none of them is on most NTC housing drawings.
The drawn shell, and the two things the press decides
Drawn shell with the wall, the outline and the general shape made by the press. Two press-produced features decide the housing’s function and neither is a drawing dimension.
- The closed end’s thinning. The punch nose stretched the metal at the closed end, so the closed end is the thinnest part of the shell. On an NTC housing where the closed end is often the process-facing surface, that thinning lands where the measurement is. A minimum wall at the closed end is a real specification, and it is checked there.
- The corner’s work-hardening and residual stress. The drawn corner is the most heavily worked metal on the part, it has the least ductility left, and it carries the most residual stress. Two consequences: a corner that is at the cavity’s geometry will concentrate the thermal movement there, and a corner under vibration will initiate a fatigue crack first. Both are design points, and both are free to fix at the drawing stage with a radius or a relief.
Sealing faces, and the tight band
+/-0.01 mm on sealing faces with critical dimensions carrying CMM data. That tight band is correct and it should go to a small number of features.
- A sealing face: flatness and perpendicularity, as two separate measurements. A sensor that does not sit square on its face is not sealed and it is also not thermally coupled – so the same flatness is a seal requirement and an accuracy requirement.
- The element’s seat or a stepped bore: its diameter, depth and concentricity. The element’s position in the thermal path is the accuracy, and this is the feature nobody specifies on an NTC housing.
- A gland for the lead or the connector: its depth and diameter, because the seal’s compression is set by it.
- And the drawn features at the loose band – the wall, the outline, the open diameter – because those are the material’s own variation and no process control will tighten them.
Material, and the four questions
Carbon steel, 304 and 316 stainless steel, aluminium and brass, with grade and temper confirmed in writing before tooling. On an NTC housing the material is chosen by four questions and they are not all about corrosion.
- What is in contact with the process, and what is its chemistry? This is the corrosion question and it decides stainless steel versus brass versus a coated steel. 316L for anything chloride-bearing and for cleaning regimes.
- How fast does the sensor have to respond? This is the thermal question, and it is the one that gets asked last. If the response is a requirement then the wall’s material matters more than the wall’s grade, and a thermally conductive insert or an aluminium housing enters the conversation.
- Is the element soldered, crimped or bonded in? A leaded NTC is soldered, and the housing’s material and its plating determine whether the joint is reliable. A plated housing’s surface finish affects a solder joint’s wetting, and a surface that has been passivated is harder to solder to.
- And is the housing’s surface a product surface? In a food, pharmaceutical or medical application it is, and it has its own documented finish requirement and passivation record, separate from the sensor’s own specification.
The element’s seat and the seal face need opposite treatment
Plating, coating or polishing to suit the drawing, with the agreed finish travelling with the part number. On this part the finish has three requirements that are all functional.
- The sealing face, unmasked. A plated seal face does not seal.
- The element’s seat or a stepped bore, masked. A coating on a seat changes the element’s fit and its thermal contact, and a plated element seat is a different fit from a bare one.
- Any lead or terminal entry, masked. Because a plated entry changes the solder joint or the crimp’s dimension.
- And where the housing is inside a machine or a housing where the finish is not functional, it is conventional – but on an NTC housing the exterior finish is often specified for appearance and it should be stated as such so that nobody assumes it is doing a corrosion job it was not specified for.
Which NTC goes in this housing, and is the calibration based on it?
Which NTC is going in this housing, and is the customer’s calibration based on this housing? Two questions, and they are worth asking before the housing is designed rather than after. The first one decides the wall’s thickness and the cavity’s volume, because the element’s beta value and its dissipation constant and its time constant all have to be compatible with the geometry around it – and an element designed for a bare-air environment does not belong in a large unfilled cavity. The second one matters more than it appears: if the customer’s calibration curve was measured with a particular housing, then the housing is part of the sensor, and changing the housing changes the sensor’s output, and the customer’s downstream calibration has to be revisited. That is a much larger consequence than a drawing revision, and it is invisible until somebody measures it. And there is a third question worth asking because it gets missed on NTC housings specifically: how is the lead or the terminal sealed? On a small sensor housing with two wires through it, the wire entry is a seal and a thermal leak at the same time – a well-made entry is thermally insulated and sealed, and a cheap one conducts heat straight from the process into the element, which is a calibration error that will be blamed on the sensor.
Sensor Housings is the parent range for parts like this one; the overview page gathers the tooling logic, finishing options and tolerance notes that apply across the family.
Specifications
| Item | Custom NTC sensor housing |
|---|---|
| Brand | AUNOV |
| Designing | Send drawings, or send a sample to be measured – unusual walls, threads or tolerances are feasibility-checked first; the same drawing is used for every repeat order. |
| Tooling | Our toolroom covers die design, wire-cut EDM and maintenance, so a die can be re-cut against the approved drawing, and the charge is refundable against agreed volumes. |
| Samples | Sample approval is documented: dimensions, finish and material recorded for later comparison. |
| Materials | Carbon steel (SPCC / DC01), 304 / 316 stainless steel, aluminium and brass are held on site; grade and temper are confirmed in writing before tooling. |
| Surface finish | Plating, coating or polishing to suit the drawing; the agreed finish travels with the part number. |
| Processing | Deep drawing on hydraulic presses, CNC-finished where tolerances tighten; in-process checks sit between stages, not only at the end. |
| Equipment | Hydraulic presses for drawing, with CNC turning for finishing, with in-house toolmaking to keep dies and fixtures maintained. |
| Tolerance | +/-0.01 mm on sealing faces, +/-0.05 mm on drawn profiles; critical dimensions carry CMM data with the shipment. |
| Delivery terms | FOB, with DDP offered for EU and US addresses, payment by T/T, L/C at sight or D/P; repeat orders reuse the approved first article. |
Related Products
Frequently Asked Questions
What does an NTC’s datasheet actually demand of the housing?
A custom NTC sensor housing is the part where the customer’s number decides the shape, and that number is usually given late and in the wrong units. An NTC measures by having its own temperature match the process, and four numbers on the thermistor datasheet are really housing requirements. The element’s seat and the seal face need opposite treatment, and almost nobody states those four on the drawing.
Why is the wall the accuracy on an NTC housing?
An NTC reads correctly only when its own temperature equals the process, so the wall between the two is the thermal path and the accuracy. On this part the wall is also a pressure boundary if the process is pressurised and a structural member if the part is handled. Drawn on hydraulic presses and CNC-finished where tolerances tighten, the profile sits at +/-0.01 mm on sealing faces and +/-0.05 mm on drawn profiles, with critical dimensions carrying CMM data.
Which NTC is going in, and is the calibration based on this housing?
This is the most useful question to ask at enquiry, because the element, its seat, the fill and the lead exit together set the sensor’s response, and none of them is on most NTC housing drawings. On an NTC housing the material is chosen by four questions and they are not all about corrosion. Stock runs to carbon steel, 304 and 316 stainless steel, aluminium and brass, with grade and temper confirmed in writing before tooling.
How is the lead or the terminal sealed?
That third question gets missed on NTC housings specifically, yet the lead exit is a leak path and a thermal bridge in one. Plating, coating or polishing is selected to suit the drawing, with the agreed finish travelling with the part number. Answering the sealing method before tooling is far cheaper than discovering a drift after the housing exists.
Request a Quote for This Part
Send a drawing or sketch — get material advice, tooling cost and a per-piece price within 24 hours (working days).




