100W Aluminum Power Metal Shell Case Wirewound Resistor ±5% 0.1R 0.5R 1R 2R 6R 8R 10R 20R 100R 1K ohm Aluminium Shell Resistor – Dummy load project (WHL #56)
Something very basic for today: Resistors. Yaay.
I was tasked with troubleshooting a Blu-Ray player that cut the sound a few minutes into movies repeatedly. Sounds like a thermal issue to me, and after the local electronics shop grabbed 35 bucks for an “inspection”, yielding a “motherboard issue” which makes this Samsung heap of sh.. a total loss, it was given to me for a last look into it, before it was tossed. Of course the owner would never buy a Samsung player again, as this high-end model played Blu-Rays just fine, but does refuse to do so with double-layer DVDs, which his previous Samsung DVD player was absolutely capable of. Samsung confirmed this is correct behaviour. Well yeah, different story.
Turns out that thing runs just fine, even in 3D mode and with “Power Bass” enabled. To cite the manual:
Power Bass: Increases the volume level of bass tones and provides thudding sound effects.
Thanks, Samsung. The world was in dire need of Power Bass.
Maybe the shop accidentally cleared something that was blocking the fan, as they didn’t even blow the thing clean with compressed air, nor did they touch anything wiring related. For my tests I only got one of the speakers of the “1000W” 5.1 system, so maybe the dude is deaf and that only happens with a full set of speakers on full blast. As I do not own enough speakers in general, none of them having the special shitty connector that has built in, and none of them once again are of the 3Ω type that Samsung uses (wtf?), I had to improvise.
Just look at those poor connectors, LAN jack nearby for size comparison. 165W to 170W specced output power at 3Ω each, so for a DC load (and let’s face it: For RF voodoo guys, audio basically is DC), that’s about 7.5A at 22.5V. No way these are jacks are rated for 7.5A continuous current. Furthermore, at 3Ω design impedance, they’ll never use the voltages this pin pitch would allow for.
(and don’t get me started on the Java part!)
So as usual, I started googling around, and 50W resistors are about 3-4€ a piece in Germany, while 100W ones are more like 6€+, meaning one could just use two 50W (half/double resistance) for the same money and be more flexible. As I recently got a couple of METERS of large-ish heat sink for scrap value, I am flexible. What do our friends in the People’s Republic charge for it?
About one Euro per 100W, including shipping.
I’ll have some of those, please.
So I got a pack of ten 4Ω (standard audio impedance) resistors, “100W”, at 5.40€ plus 4.50€ shipping.
They are somewhat larger than 50W resistors, but not of the typical 100W size. For once, they use the same case size as regular 50W types (15-16mm width, a little more rectangular on the top), but with 60mm instead of 50mm case length.
They also have two mounting holes on one side, while 50W have one on opposing sides each, and regular 100W have four, two on each mounting flap. For a flat base, this shouldn’t be a problem, it’s just a matter of contact area to the heat sink.
So I trust them in being more capable than my known 50W resistors, but I doubt they are 100W. I’ve never seen a 75W resistor in person, but that’s probably close to their rating. Some more googling required…
Arcol, one of the more available brands for high wattage resistors, has a nice graph in their HS series datasheet:
Given their 75W, 100W and 150W types use the exact same shell in different length settings, one can extrapolate the required length increase for power rating gains (surface area, T⁴ black body woo-woo, etc). For this very case of this very manufacturer, it’s about 80% to 85% additional length per 100% increase of wattage. For a crude estimate, my 50W@50mm resistor would yield about +25% for the 10mm or 20% increase in length. So that’s 60ish watts, add in a couple of watts for the larger mounting area, and this could be a 65W to 70W resistor in European rating terms. 2/3 of what’s advertised by the Wan Hung Lo factory. Healthy numbers, still a bargain.
Now, back to building the dummy load. Grab your heat sink:
Be delighted that there are already ten almost-evenly spaced M3 holes in there And get to work. I had to mount them close to an edge, as the other mounting hole would have required to put the second mounting point through a fin of the heat sink. Doesn’t really do all that much in terms of heat dissipation on such a huge unit. It’s 75×13.5×2.5cm³ (3mm base thickness) at about 2.1kg, and it’s one of the smaller ones I got…
The Fischer SK 463 family would be similar in performance to this, maybe a tad better due to a thicker base. It’s rated 1.0K/W at 15cm length, so there’s that.
First test run was done with the HP 6644A from Project #14, a 3.5A/60V lab power supply. By pure coincidence, this dummy load almost maxes out this unit, which I never did before. Great, let’s get it on:
One resistor connected, 3.58A at roughly 14V, P~50W, temp peak 65°C, avg 36°C: (this is all steady-state images after like half an hour of operation)
Three resistors in series, 3.58A at ~44V, P~150W, temp peak 85°C, avg 60°C:
And full load at four resistors in series, 3.58A at 57.3V, P=205W, temp peak 98°C, avg 72°C:
3.585A at 57.291V by the way makes this a 15.98Ω load including wiring, which is fantastic for four roughly 4Ω resistors with unknown temperature coefficients at “50%” advertised or 75% real world load.
Interestingly both eyeballing the average temperature as well as calculating a rectangular box average with the crappy FLIR software yields a pretty “linear” temperature rise with dissipated power, which I did not expect due to the highly nonlinear temperature term in the Stefan-Boltzmann law (two dudes, not one). Turns out dissipation by radiation at those “low” temperatures is not the major player, with convection even winning out by an order of magnitude when forced, i.e. when using a decent fan. It also does not change that drastically as T⁴ from 18°C to 72°C would suggest, as this obviously needs to be computed in absolute temperature, hence an offset of 273K.
Quick pi-equals-three calculation: The heat sink envelope is 750mm x 135mm x 25mm, which creates an effective surface area of ~2500cm² or quarter of a meter squared – total area with fins is closer to 6500cm². Emissivity of an anodized aluminium heat sink could be 0.85. Radiated emissions for this scenario would be:
18°C: 87W (baseline – the heat sink radiates 87W to the environment, and gets 87W back at room temperature)
36°C: 110W (Δ23W)
48°C: 127W (Δ40W)
60°C: 148W (Δ61W)
72°C: 171W (Δ84W)
Add in the resistor body (35cm² each) at peak temperature and ignore the leads and any thermal gradients to not overcomplicate things:
18°C: 4.8W (baseline)
36°C/65°C: 6.8W (Δ2W)
48°C/79°C: 8.8W (Δ4W)
60°C/85°C: 10.4W (Δ6W)
98°C: 12.8W (Δ8W)
Sum that up, and the radiated emissions would be:
25W @50W in
44W @100W in
67W @150W in
92W @205W in
So radiation accounts for about half of the entire heat dissipation, the other half being mainly carried out by convection and some losses via wiring and the contact points of the heat sink to the ground and the rack.
I’d be satisfied with that conclusion, as I know calculating this other half is pretty darn hard, finite element analysis level hard – contrary to black body radiation which basically is accounted for with assumptions like massless isotropic spherical cows in a perfect vacuum. Turns out…there’s a free online analyzer available. Just a demo of a paid product, limited to three power sources (and a HTML-level cap on heat sink size), but abso-fucking-lutely usable for a second look into that calculation. Try it yourself over at heatsinkcalculator.com!
And now watch this:
Power distribution also lines up:
50W: 23W radiation (matching my first estimate, -2 of my second one), 27W convection
100W: 43W radiation (+3 / -1), 57W convection
150W: 63W radiation (+2 / -4), 87W convection
(no 200W simulation due to demo constraints)
The absolute basic approximation with the bare heat sink underestimates radiation, and the correction with the peak temperature resistor addon overshoots, both are less than 5% out – fantastic.
So in conclusion, those Wan Hung Lo resistors make for a great dummy load, way better than needed for such an application. The dummy load works absolutely fine and has a thermal performance of up to Δ54K per 205W or 0.26K/W in my highest possible scenario, even with the non-ideal four hot spots instead of an evenly distributed power source. While having a 2kg / 72°C body of aluminium around might not be everybody’s cup of tea, there’s still the option for a fan, very likely making this into a kilowatt+ capable unit at reasonable temperatures.
The Samsung player did not drop sound throughout a two hour movie at 100% sound level, but I bloody murdered two small harvested TV speakers when putting them in parallel to one of the speaker channels with just a 15Ω resistor as protection. Anyone who does this with real speakers likely will be deaf at the end of the movie…
Bonus content #1: The HP6644A fan doing its job, toasting the wallpaper behind the rack
Bonus content #2: A totally useless IR video showing the cooldown of the entire dummy load – without a temperature scale embedded or searchable temperature data recorded in metadata. Thanks FLIR…what a waste of space.
For the record: This was done by a simple MKV container speedup from the original 8.56333fps to 480fps (56x) via mkvtoolnix, then noise filtered and reencoded to 60fps for YouTube via Handbrake.