I am currently studying The Art of Electronics, third edition, by Horowitz and Hill. Appendix C.4 Resistor types says the following:

For power applications you usually use wirewound resistors, either in an air-cooled ceramic package or a conduction-cooled (“Dale-type”) metal package. High-value resistors (\$>10 \ \text{M} \Omega \$, say) are usually of metal-oxide construction (e.g., Ohmite “Mini-Mox” or “Super Mox,” or Vishay RNX-series). Film resistors are not tolerant of high peak power; for such applications use something like ceramic or carbon composition, or other styles specified for peak-power use.

In researching this, I encountered this page, which says the following:

Metal-oxide film resistors are fixed form, axial resistors. They are made of ceramic rod that is coated with a thin film of metal oxides, such as tin oxide. Metal oxide film resistors must not be confused with metal oxide varistors, made of zinc oxide or silicon carbide.

So the textbook says that, for power applications, resistors of metal-oxide construction are suitable, and that film resistors are unsuitable because they are not tolerant of high peak power. But the website seems to imply that metal-oxide resistors are (thin-)film resistors. Maybe I'm misinterpreting the authors, but the part that says "high-value resistors (\$>10 \ \text{M} \Omega \$, say) are usually of metal-oxide construction" seems to be implying that they are used for the aforementioned "power applications" in the previous sentence, no? How do I reconcile what these two sources are saying?

I would greatly appreciate it if people would please take the time to clarify this.

  • \$\begingroup\$ I don't see a contradiction. Please explain the contradiction. \$\endgroup\$
    – Andy aka
    Jun 30, 2020 at 16:24
  • \$\begingroup\$ @Andyaka It says that film resistors are not tolerant of high peak power, and that metal-oxide resistors are suitable. But the website says that Metal-oxide resistors are thin-film. So does that mean that metal-oxide resistors are suitable for power applications, or are they not suitable? Perhaps it's because this is my first exposure to this material, so the distinctions that you are familiar with are not clear to me. \$\endgroup\$ Jun 30, 2020 at 16:28
  • \$\begingroup\$ Where does it say metal oxide resistors are suitable? You say this, but no where in your quotes does it say this. Quote all relevant sections. I did skim the wiki you llinked and there are parts of it that are wrong. For example, "Metal oxide film resistors exceed the performance of metal film and carbon film for the following properties: power rating,...overload capabilities,...surges" just seems wrong. \$\endgroup\$
    – DKNguyen
    Jun 30, 2020 at 18:46
  • \$\begingroup\$ @DKNguyen Maybe I'm misinterpreting the authors, but the part that says "high-value resistors (\$>10 \ \text{M} \Omega \$, say) are usually of metal-oxide construction" seems to be implying that they are used for the aforementioned "power applications" in the previous sentence, no? \$\endgroup\$ Jun 30, 2020 at 18:49
  • \$\begingroup\$ @ThePointer I understand now that you point it out. You might want to make that clearer in your question. \$\endgroup\$
    – DKNguyen
    Jun 30, 2020 at 18:50

4 Answers 4


Those articles you are seeing are speaking from an implicit contexts, and everything is relative. There are many degrees of freedom.

For example:

  • thin film resistors are not suitable for pulse power
  • thick film resistors are more suitable for pulse power than thin film
  • but thick film is still pretty crummy so you could go even farther say film resistors (both thin and thick film) in general are not suitable for pulse power compared to other bulk element resistors

  • SMD components are not suitable for pulse power ("SMD" here might refer to chip components but there are components that are surface mount that are not chip and can handle pulse power just fine, like MELF)
  • but there are double-sided SMD components with a film on both sides for increased surface area which are more suitable for pulse power than single-sided SMD components, but still not as suitable as MELF (or axial) where the trace can be printed around the entire surface area of the component
  • axial and MELF (which are also an surface mount component, by the way) components are suitable for pulse power

  • metal oxide film is not suitable for pulse power
  • carbon is suitable for pulse power and could be carbon composition or carbon film (NOTE: I fairly certain this is not carbon thin film and that there is no such thing as carbon thin film).
  • carbon film is less suitable for pulse power than carbon composition, because it is pure carbon and must be printed and therefore cannot be a bulk material like carbon composition. But the fact it is pure carbon means it is much more controllable and stable than carbon (mixed) composition

So what does that mean? It means the following for pulse powering capability (is not quite the same as high temperature cycling stability, and nothing to do with other performance metrics such as tolerance or noise):

  • Carbon composition > carbon film > metal oxide thick film > metal oxide thin film
  • axial/MELF > double-sided SMD > single-sided SMD

But you can mix and match material with construction. So you can end up with a metal-oxide film on an axial/MELF body that might be more suitable for pulse power than carbon film on a single-sided SMD component.

Usually these articles are speaking only in a limited scope or context so will shorten their wording.

It can be difficult to tell when an article is actually referring exclusively to thin film SMD of if they actually mean all thin film when they say "thin film".

It can be difficult to tell whether an article is referring exclusively to chip components or to all surface-mounted components (such as MELF) when it says "SMD component".

The article could also be speaking from the point of view of a component manufacturer. For example, if you were making resistors it might be optimal in terms of pulse power to use carbon composition, which would be fine if that was all you cared about, but you sacrifice accuracy, noise, and cost. So you might instead choose to go with carbon film for more stability and lower tolerances and design your way around the deficiencies compared to carbon composition. Or you might decide to cut costs even more and go with metal oxide film and design your way around it the deficiencies in the material.

Or it might just be impossible to make a wirewound resistor with sufficiently high resistance so you need to go with metal oxide film, as is the case in your quote. You could go with carbon, but you might not need to and it probably costs more in terms of tooling, etc.


I'm not sure we can fully disambiguate this for you. The names are historical and don't necessarily form a clearly delineated taxonomy of resistor types.

Carbon film resistors are made from a thin film of carbon on a ceramic rod (but in a carbon film resistor series you will find that lower values are usually actually made from a metal (such as nickel) film. They are trimmed to value usually through mechanical cutting.

Thick film resistors are printed and fired, and used to be common in hybrid circuits. These days I suspect most hybrids are made with commercial resistors which are cheap and come sealed and trimmed to value. Thick film resistors can be trimmed via laser or abrasive etching. "Thick" here is still a small fraction of a mm.

Thin film resistors are made with semiconductor-like processes such as sputtering or evaporation and etching. The film thickness is sub-micron.

So-called Metal-oxide-film resistors of the axial persuasion are generally designed for power applications and use an oxide such as antimony/tin oxide over a ceramic rod. The oxides are more refractory and spread over a relatively large rod, so they can withstand large pulse power dissipation.

Chip resistors are "thin film" types and also typically use a metal or metal oxide, sometimes rather exotic types.


When they say "peak power" they are talking about things like discharging a capacitor bank, where you might go to 10 or 100 times the rated power for a very short time. Any resistor made out of a thin film doesn't have much total heat capacity in its resistive element and will fail under such circumstances. You can still use them in power applications, just don't go much over their rated power. The old carbon composition resistors have a big chunk of material to absorb the energy. Even better are carborundum resistors. We used to dump capacitor banks into them and instantly turn them red hot. Wirewounds are pretty good, but you have to find their pulsed energy ratings and respect them. We tested some ohmite 225 Watt wirewounds, which were rated for 1 kJ of energy. When you go over that you can hear cracking just after the pulse because the thermal expansion of the wire is causing the ceramic material to crack. A general rule of thumb is that if the manufacturer doesn't give you a peak power rating then don't exceed the rated power even for short pulses. Or else run your own tests, and run lots of repetitions and watch for changes in value and listen for cracking, arcs, etc.


The pulse duration may be important here.

A 1milliMeter cube of ceramic, coated with some resistive thin_fim, has a thermal time constant (determining how fast the thin_film heat can be absorbed by the ceramic core), of

  • 11,400 seconds / [ (1meter / 1milliMeter)^2 ]


  • 11,400 second / 1,000,000 == 11.4 milliSeconds thermal time constant

Pulses much faster than 11.4 milliSeconds will end with most of the HEAT still inside the thin_film resistive region.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.