I went through similar thoughts, not so analytically, recently. The school has a Gallien-Krueger bass amp, all aluminum. The heat-sink forms the top of the unit. The handle screws were threaded into this extrusion. The beast has many years of rough service. One handle screw musta pulled-out, because I found a wood-screw replacing it. (Cleverly blunted so the tip did not skewer the PCB under the heatsink!) The other screw was loose and tightening it didn't feel good. My gut said that #10 screws in ~~0.2" of soft alloy mighta been OK the first decade, but was too close to failure for "infinite duty" (defined as "until I retire from here").
My bodge was to drill for Tee-nuts. Effectively a steel insert with huge bearing on the underside of the extrusion. I know these handle screws give reasonable service in T-nutted wood.
This handled heatsink attaches to the box with a dozen amazingly small screws into C-channels in the extrusion. This looks like a weak link but all were in good shape.
> I'd say definitely more. ...comparison to the depth of a nut seems logical, but you are tapping AL, and a bolt is not AL.
True, but this is not an Engineered Joint either. Where did the proportions come from? Not from structural analysis.
Pardon me for simplifying, and using archaic units.
A 60deg thread can develop the full strength of the -area- of the base of the threads. If both components are equal, it strips on the bolt. Often the bolt is special and the block is bulk-stuff, weaker, it strips on the block. For cheese-steel in aluminum bar, we expect the AL to strip.
3mm is like 1/8 inch. The circumference is then 3/8". Say the threaded depth is reduced to 3mm or 1/8". The area is 3/64 or 1/20 square inch.
If the bulk were 20,000 pound per square inch stuff, it would take a 1,000 pound pull. We might have 10KPSI aluminum, so 500 pounds is needed to pull the bolt out of the Aluminum. 230KG outside the US.
Interestingly, 500lb is also the tensile strength of a 1/8" rod made of 32KPSI steel.
So with one 3mm screw in 3mm depth of aluminum, Peter could not pull the module out with both hands. And if the module is 5 pounds and we apply a 99 Gee drop-test, the module would not pull out (assuming on-center drop).
Anyway, 500 pound force on a 8mm square bar a foot long would put a permanent bend in most steel. Assuming the bar is 10KPSI material, as we assumed for the screw-hole, if the bar is more than an inch long the bar will fail first.
Assuming 1 pound module and four screws, we have a 20:1 factor of safety even on the drop-test. At full 8mm depth, maybe 50:1 FoS. Assuming Peter won't brutalize his modules and this is not a touring company, the threading can be exceptionally sloppy (oversize pilot hole) and still have ample margin.
The 8mm comes from Rob's estimate/experience of what he needs for the Span, and the 3mm happened because smaller screws aren't cheaper, just more awkward.
We could micro-optimize everything like a bridge, which is used by unsupervised strangers and is a major dent in the local budget. Then for stiffness in a 16" rack we might assume 1" or 1.5" beams. Since abuse can be any direction we favor round or square. Strength is a lesser issue so they may be hollow, indeed the wall thickness might be controlled by screw depth demands. Four 1mm screws 3mm deep or 1.5mm 2mm deep would resist Peter's pull. But who wants to find a dropped 1mm screw? And they are not cheaper, perhaps more costly.
> hand-tapping flanges the first few threads a bit because of the tool's "wiggle" in your hand.
A good point, and a reason not to micro-design too far. I'd actually tried to re-tap that G-K's extrusion, but without a proper setup I could see that I was making a cone not a cylinder.
BTW: the standard plastic replacement guitar-amp strap handle sold in the US now comes with Metric screws, not #10 as it used to be.