Resistor temperature rise and heat dissipation

[iroc86]

This is sort of a follow-on question to a post from last month where I described a circuit to rectify 120 VAC to 48 VDC for dampening a fluctuating input to a relay coil. I've been experimenting with the circuit and have some questions about the temperature rise and heat dissipation of the dropper resistor.

In the schematic below, R1 drops about 60 Vrms and has an average power dissipation of 1 W as calculated by LTspice. The voltage waveform across R1 is half-wave rectified AC. I can confirm these numbers within about 5% on the bench.



In my first design, I used a 2 W metal film resistor (Vishay PR02) and the surface temperature reached about 100 degrees C steady state. This is within spec and seems to agree with the datasheet's hot-spot temperature rise at 1 W dissipated power. The ambient temperature of my lab is 22 degrees C, so 22 + 75 = 97 degrees C; close enough.



100 degrees C is a little hot for my liking, so I thought I would try a higher-wattage resistor. I replaced the 2 W metal film with a 5 W wirewound (TE EP "S" Series - "Small Size"). To my surprise, the steady-state resistor temperature was only about 5-8 degrees cooler. The physical package of the 5 W resistor is nearly five times larger than the 2 W (by volume), not to mention its power rating is 150% higher.

The TE datasheet doesn't have any charts for temperature rise, but I wouldn't have expected such similar performance between two seemingly different resistors. In both tests, the ambient temperature was around 20-22 degrees C. I'm measuring the temperature with a FLIR thermal camera.

Now, I did find a temperature rise chart for another 5 W general-purpose wirewound resistor (Vishay AC Series), which suggests that a temperature rise of ~75 degrees at 1 W is appropriate. I don't know how closely this correlates to the TE resistor, but it's probably close since the construction seems similar (ceramic/cement) with similar power ratings.



As a final "controlled" test, I dropped 60 VDC across each resistor (~1 W) from a bench power supply and observed similar temperatures within a few degrees. So, I think everything checks out.

What would cause both the 2 W and 5 W resistors to reach about the same temperature when there's such a difference in power rating and physical size? Could it be the composition (metal film vs. wirewound)?

Short of using a much larger resistor (7 or 10 W), or putting a few in parallel, I'm not sure how else I might reduce the surface temperature. (I know that this isn't the most efficient circuit, of course, so let's not comment on that ).

[Kim Christensen]

Why not just use a capacitive dropper circuit instead?

[WattsThat]

In the PRxx datasheets, look for the difference in temp rise between 15 and 20mm of lead length, detailed on pages 5 and 6.

Conclusion: Conduction beats convection. The result is similar device temperature rises in spite of the differences in device area.

[T3sl4co1l]

Right, for the same dissipation rate (per area), the bigger part MUST have the higher rating for the same max temp, or a lower temp temp rise at the same power.

That leaves:
- Temperature uniformity: if the ceramic body is less conductive (e.g. steatite vs. alumina?), the ends and leads won't get as hot as the center of the body (unfortunately, neither product discloses the material)
- Emissivity: difference in enamel IR emission (but, radiation should be a small part of the total at this temperature)
- Lead material: the EP5WS claims solid copper wire, while the PR02 can use a selection (which exact part number did you use?)

Yeah, a capacitive dropper is practical; you'll need to add another diode to conserve charge.  Keep the series resistor, just use a bit less of it -- limits turn-on surge.  This doesn't answer the underlying question of course, which is interesting (I would also expect bigger resistors to run cooler!), but alas we may not have the information to solve it.

A destructive test to confirm body material, might be: break it open and do a Mohs test.  Steatite I think should be fairly soft, I mean still hard like a stoneware, but not like porcelain; porcelain or glass is a possibility and should be harder (Mohs 6-6.5?), and smoother in appearance.  I'm not sure what other ceramics are used; cordierite and mullite are possibilities, having similar hardness.  And all of these I think are similar in conductivity (modest to low).  Alumina is 9 by definition, though may test lower due to porosity (which I would think is true of any fired body), but is very conductive in comparison.

Tim

[iroc86]

I'm always interested in learning best practices and new approaches, so thanks for sharing the capacitive dropper circuit. I've never used one before, but I've seen them. I guess as long as that dropper cap doesn't fail closed, all is well.

WattsThat and Tim, thanks for the insight. The 5 W resistor takes longer to reach its steady-state temperature due to the extra thermal mass, but I guess I would have expected the heat to also dissipate more readily into the ambient surroundings (i.e., run at a significantly cooler temperature). I'm using copper-leaded resistors in both cases. The leads are about 0.75 mm diameter for the PR02 and EP5 models.

I also noticed that the 2 W and 5 W resistors are rated differently for maximum ambient temperature. The PR02 datasheet states 200 degC (250 degC max film temperature) and its derating curve matches up accordingly. The EP5, interestingly, identifies 155 degC as maximum but this only correlates to a ~55% derating, not 0% as one might expect. This curve derates to 0% at 275 degC, implying that *that* is the maximum ambient temperature. (Of course, I don't think you can really deduce temperature rise from these derating charts, but maybe the difference in temperature range has something to do with the thermal efficiency between the two types of resistors and how they're rated.)

One last tidbit: the EP series is available in "normal size" and "small size" packages. I'm using the smaller version which measures 6.5 x 17.5 mm, whereas the larger version measures 8.5 x 25 mm. I wonder if it'd run cooler because it's significantly larger, of if it'd eventually settle around the same temperature in steady-state operation.