# Power dissipation of resistor

I am using a couple of resistors in my circuit as heating elements. However, the input voltage is in a wide range, and the maximum voltage exceeds the power rating of these resistors. I am thinking of pulsing the resistors when the voltage is in the higher range to lower the average dissipated power.

My question is, at what frequency should I pulse these resistors?

• Without any details of the resistor, you can use any frequency you want as long as all parameters are within the resistor limits and no parameter is exceeded. Commented Dec 20, 2022 at 5:26
• To be specific, you will need the thermal response or pulse rating of the parts to determine that. What resistors are they (MFG/PN, data sheet)? Commented Dec 20, 2022 at 6:28
• The datasheet will tell for sure but in general, if your pulse is significantly shorter than the thermal time constant of your resistor, your method is sound. Commented Dec 20, 2022 at 8:21
• the datasheet can be found here: bourns.com/docs/Product-Datasheets/PWR263S-35.pdf
– HV16
Commented Dec 20, 2022 at 12:48
• Please edit that link into your question. From the datasheet: "Short Term Overload (2x Pr for R < 2 Ω, 1.6 x Pr for R ≥ 2 Ω, V < 1.5 x Operating Voltage" You need to stay within that limit. Commented Dec 20, 2022 at 14:17

I don't have enough rep to comment so here's an answer:

1. The pulse time must be significantly faster than your resistors thermal time constant if you are slightly over the resistors power rating.
2. If you are largely over the resistors power rating pulsing doesn't do much (you will destroy the resistor anyway)

But why make it complicated with pulses? Can't you just have two resistors with the same power rating in parallel? Or get a resistor with a higher power rating?

• what's the power rating we are talking about?
• What do you want to heat and how? Fixed temperature, heat as much as possible? ...
• How large is the input voltage range? AC or DC?

Edit: I think I understand your question now. Given that the resistor is rated for 35W and you want to heat with no more than 30W the pulsing is supposed to limit the total heat and not to protect the resistor from overheating. Therefore, the frequency is bound by thermodynamics from below (how fast do you need to switch in order to stay in the acceptable temperature range?) and by electrical characteristics from above (what's the maximal frequency you want on your PCB?).

If you have a huge range of acceptable frequencies, which you most likely have, I would take one that's achievable with components your company already has in stock.

If you want to maintain a certain temperature I would implement a controller instead of fixed frequency switching. I.e. design a circuit s.t. it heats while Y < Temperature < X.

• here is the datasheet of the part I wanted to use: bourns.com/docs/Product-Datasheets/PWR263S-35.pdf
– HV16
Commented Dec 20, 2022 at 12:50
• the input ranges from 40Vdc to 100Vdc. At 100Vdc the power will be 30W. The ambient temp this is intended to work at is -30 celsius. I already have a few in parallel, I cannot have more because I don’t have enough room on the PCB.
– HV16
Commented Dec 20, 2022 at 12:55
• see the edited answer above. Commented Dec 20, 2022 at 13:26
• I actually am worrying about overheating the resistor itself. For a thermal resistance of 3.9W/Celsius, 30W of power results in a temperature rise of 117 Celsius from -30, which end up being 87 Celsius. If you look at the derating curve, the rated power at this temp is almost half of 35W which is around 17.5W
– HV16
Commented Dec 20, 2022 at 17:59
• you can lower the total thermal resistance by e.g. adding a heatsink on top of the chip Commented Dec 22, 2022 at 14:51

For heating only, you can use any PWM frequency between 100 Hz and 10 kHz.

Less than 100 Hz, the dynamic effects may allow the resistor to overheat within a pulse.

More than 10 kHz, a poorly designed or poorly laid-out electronic switch may start heating due to switching losses which would rob some of the expected heating out of the resistor.