Benjamin Chung

Computer science PhD student at Northeastern University. I work on types in Julia, with rocket GNC, and a bunch besides. Also @[email protected]

[Benjamin Chung @ckfinite]

@6502_ftw @queenofthed @MBielieskov With respect to the engagement geometry, getting in front of a target missile is essentially the same kinematic problem as getting below it. Where differences can arise is if the target is trying to evade with aerodynamic maneuvering, but Kh-22 was not designed to do that.

[Benjamin Chung @ckfinite]

@6502_ftw @queenofthed @MBielieskov This is well within the intercept envelope of Patriot, particularly since it's not a maneuvering target. Due to these issues, the Kh-22 is nowhere even particularly near a pacing threat for a modern SAM system. "Hard," today, would be Zircon that's doing mach 5 at 100 meters.

[Benjamin Chung @ckfinite]

@6502_ftw @queenofthed @MBielieskov Ukraine not having anything that can engage Kh-22 pre-Patriot is because of how old their missile park was (though S-300VM should have been able to). Regardless, Kh-22 is a much easier target than the maneuverable pseudo-BMs that both Patriot and SAMP/T use as reference threats.

[Benjamin Chung @ckfinite]

@6502_ftw @queenofthed @MBielieskov Kh-22, by modern standards, is not particularly fast (cruise is at mach 5 at about 27km altitude) and doesn't really maneuver in so doing. The reference threats for modern systems are things like ATACMS or Iskander that are 2x faster, coming from higher, and are maneuvering.

[Benjamin Chung @ckfinite]

@queenofthed @MBielieskov Kh-22 isn't a particularly challenging target for a modern SAM system; Kinzhal, Iskander, or even Kalibr all pose more a more challenging (faster, faster, lower altitude, respectively) threat. It was great for its era but that was ever so long ago.

[Benjamin Chung @ckfinite]

@ckdoak @ACapaccio @elonmusk If you don't mind saying, what body do you use?

[Benjamin Chung @ckfinite]

@Stormchaser_TS @rawsalerts @LiveStormChaser Is this cropped or the original framing?

[Benjamin Chung @ckfinite]

@n0x1ous @Buildzoid1 @VideoCardz No - "single seam" are better than the "double seam" that Nvidia used in their terminals (more terminal clamping force). We hadn't previously seen terminals of this design fail like this.

[Benjamin Chung @ckfinite]

@IgorWallossek Either construction would have been fine because they're modern crimp terminals. Instead, the solder-on-bus solution in the connector means that the joint is unacceptably susceptible to mechanical failure causing high resistance. It's just a nonsense of a connector.

[Benjamin Chung @ckfinite]

@IgorWallossek According to Amphenol's applications specification (https://t.co/BmADxMP90x) the connector should be built with crimps connecting either 1x16AWG wire per terminal or a combination of two wires (e.g. 1x18AWG + 1x20AWG).

[Benjamin Chung @ckfinite]

@KeysBeast1 @CommanderWolfyX @IgorWallossek @USCPSC Even if you don't have a bad solder joint solder is notorious for developing hairline fractures when subjected to mechanical loads. It's not a suitable termination methodology for this kind of cable and crimps have been de jure for decades for this reason exactly.

[Benjamin Chung @ckfinite]

@namcost1theonly @IgorWallossek Ah - 12VHPWR is not exactly Mini-Fit Jr, but Amphenol's application information (see https://t.co/BmADxMP90x) says that it does allow termination of 16AWG with crimp terminals.

[Benjamin Chung @ckfinite]

@namcost1theonly @IgorWallossek Nope. 16AWG can be terminated directly into PCIe power connectors with a crimp connection, see https://t.co/pRKzGQqCDK. This is likely what the PSU vendor's cables are doing. Nvidia's solution here is failing for all the reasons why soldered wire-to-board termination is bad.

[Benjamin Chung @ckfinite]

@LupintheI @IgorWallossek So here's the application guide for the 12VHPWR connector from Amphenol: https://t.co/BmADxMP90x. You can see that there is no solder to be found because soldering is objectively unsuited as a termination for this application! Literally all Nvidia had to do was follow this doc.

[Benjamin Chung @ckfinite]

@LupintheI @IgorWallossek Hell, you can even crimp in two wires per contact if you do it right and that with some careful design would likely let you make this adapter wherein they're splitting some of the supply pins.