Hi all,

I have been curious lately about the use of transformers in pulsed applications. Often the transformers I have seen are quite small compared to what you would need for continuous output operation.

Although the Curie temperatures of ferrites are somewhere in the region of 200 degrees Celsius and above, I have seen transformer designs that would most likely far exceed that when used in a continuous application.

I assume engineers of these transformers are utilising the fact that the transformer only operates for a short amount of time before turning off. But I have yet to see any information regarding the thermal time constants of ferrite (or otherwise) material whereby one could calculate the expected temperature rise in a given time frame. For example, if a pulse should last 10ms, and then off for 10ms, how would one select the right core size, shape, and material?

Some resistors, particularly metal strip and pulsed resistors, can dissipate significantly more than their continuous ratings if utilised for short amounts of time.

Is there any common equation or rule of thumb for calculating a cores thermal resistance given short periodic operation, and in your experience how long does it take for a ferrite material to heat up toward a steady state value?

I was also concerned of thermal shock to the core, particularly if potted; it may be that the core is intermittently dissipating multiple more times than is allowed. I have seen data sheets quote 35W for 100mT at 100kHz - but with some cores having a 9C/W thermal resistance, this probably wouldn’t fly. Other documents state the core should not exceed 7 degrees Celsius change in temperature within a short amount of time to avoid cracking due to thermal overload - but I did not notice any info regarding what is a “short time”. It could be ten seconds or 3 minutes - in pulsed applications this makes the world of difference. I know of some systems that operate multiple bursts in a couple of seconds and then turn off for ten minutes.

Feel free to sound off in the comments and let me know your thoughts.

SiC

 

I don't recall all of the transformer design formulas needed, but the key information you have to consider in selecting a core size is ampere-turns. This parameter determines the maximum magnetic lines of force the core can handle without saturation. The RMS power and surface area largely determine the maximum temperature rise. In smaller transformers, much of the heat is conducted out of the transformer into the PCB. In your example of a pulse 10 ms on and 10 ms off, the duty cycle is 50%, therefore the RMS power is 50% of the peak power (assuming square-wave waveforms).

 

I think the only way to confirm that your design is within expected operational temperatures is to conduct tests using thermocouples (measuring the actual temperature of the core).

Remember to factor in the maximum ambient temperature you expect the device to operate. The problem with ferrite material is that once over the Curie point it suffers thermal runaway.

 

I don't recall all of the transformer design formulas needed, but the key information you have to consider in selecting a core size is ampere-turns. This parameter determines the maximum magnetic lines of force the core can handle without saturation. The RMS power and surface area largely determine the maximum temperature rise. In smaller transformers, much of the heat is conducted out of the transformer into the PCB. In your example of a pulse 10 ms on and 10 ms off, the duty cycle is 50%, therefore the RMS power is 50% of the peak power (assuming square-wave waveforms).

I think the only way to confirm that your design is within expected operational temperatures is to conduct tests using thermocouples (measuring the actual temperature of the core).

Remember to factor in the maximum ambient temperature you expect the device to operate. The problem with ferrite material is that once over the Curie point it suffers thermal runaway.

I was thinking more along the lines of how many seconds would it take to heat up the core. I have read some things about thermal mass, specific heat, weight of the core halves etc but not found a concrete equation thus far where I can calculate the time. I have assured that the flux density is about half that which would saturate the core, and have calculated the power dissipation and expected steady state temperature, but I was interested in finding out how fast a ferrite would physically be able to heat to that steady-state value…

 

The TDK datasheets mention a specific heat of 800 J/Kg °K, which may be the quantity you are looking for.

 

The TDK datasheets mention a specific heat of 800 J/Kg °K, which may be the quantity you are looking for.

Yep, also spotted that, but I can’t find an equation that uses that to give me a figure for time taken to reach steady state when X amount of power is being dissipated, or change of temperate in Y amount of seconds given a power dissipation in the core.

 

Yep, also spotted that, but I can’t find an equation that uses that to give me a figure for time taken to reach steady state when X amount of power is being dissipated, or change of temperate in Y amount of seconds given a power dissipation in the core.

You get the equation from the units.

 

You get the equation from the units.

I found this equation:

ΔEt = m * c * ∆θ

• change in thermal energy (ΔEt) is measured in joules (J)
• mass (m) is measured in kilograms (kg)
• specific heat capacity (c) is measured in joules per kilogram per degree Celsius (J/kg°C)
• temperature change (∆θ) is measured in degrees Celsius (°C)

Mass is 0.28kg, ΔEt is 62W*0.025 seconds, c = 800 for ferrite.

My result is ∆θ = 0.00692 degrees.

Does this seem correct? I know ferrites take a long time to heat up, but I would be throwing 62W into the core when it can only handle around 35W. The thermal resistance of the core is 11K/W too, so the expected steady state temperature is very high (if ran continuously, which it isn't).

However, after 0.025 seconds the transformer is left to cool for a long time.

 

Yes

 

Like everything in engineering, nothing is "simple" when you really delve into it. I would recommend you obtain a copy of "Magnetic Components Design and Applications" by Steve Smith. He has an entire chapter devoted to Pulse Transformer design, and another entire chapter covering heat transfer as it pertains to magnetic components. I would describe his writing style as "extremely efficient", whereas he covers some very complex concepts with very few words and he keeps the math to a reasonable level. In other words, you have to read and comprehend almost each sentence before moving onto the next. But before too long, you suddenly realize that you actually have a good understanding of the subject. (You are ready to start your first low frequency transformer design after 9 pages.) You should be able to find a copy on AbeBooks.com.

 

Like everything in engineering, nothing is "simple" when you really delve into it. I would recommend you obtain a copy of "Magnetic Components Design and Applications" by Steve Smith. He has an entire chapter devoted to Pulse Transformer design, and another entire chapter covering heat transfer as it pertains to magnetic components. I would describe his writing style as "extremely efficient", whereas he covers some very complex concepts with very few words and he keeps the math to a reasonable level. In other words, you have to read and comprehend almost each sentence before moving onto the next. But before too long, you suddenly realize that you actually have a good understanding of the subject. (You are ready to start your first low frequency transformer design after 9 pages.) You should be able to find a copy on AbeBooks.com.

Thank you for the recommendation, my friend.