Comment by contingencies | Hacker News Reader

Comment by contingencies | Hacker News Reader

Parent Full thread

So as neither an EE nor an ME (nor a qualified SE!) I've been in charge of some fairly complex mechatronic systems development for the last near-decade. IMHO the whole harness thing seems to be on the way out, excepting very specific situations where every last bit of weight really matters. Why? First, cost. Not having point to point means you aren't an off the shelf item, which means costs 10x immediately. Second, communications friction. While numerous providers can create arbitrarily complex harnesses as a commercial service, they generally disagree on everything from the manner by which cable qualities are specified to the sequence by which a termination should be made to what a length means with respect to a radius, etc. especially when crossing national boundaries. Third, complexity. Shared buses are harder to manage from a project evolution standpoint, and make it much harder to isolate and debug misbehaving nodes. Now I can see why all of this is worth it on a spacecraft, where weight matters, or in a car, where volume is assumed, custom connections are probably needed owing to a high vibration environment and every communicating component is probably either relatively well tested (auto grade transceivers) or relatively simplistic (window or door controller) or both, but for the rest of us where a smaller volume is assumed it seems to me harnesses are falling out of favor. Their chief benefit is weight reduction and nominal material cost reduction but the accrued costs in real terms outweigh these savings for the majority of projects, IMHO. Does this echo your own experience?

Space is very slow to change, and traditional harnessing is going anywhere for the time being.

I cannot speak to the newer systems on the market like Starlink, their economies of scale are closer to car manufacture, so for harnessing, they likely do it similar to that industry. Form boards and very repeatable methods for producing a harness that has been well designed into the chassis of the spacecraft to be installed at a specific point. But much of those techniques are relatively unchanged from what I do, just the timing and overall design is more optimised.

Most of the harnessing I have done is very bespoke, and happens at many stages throughout the integration as things are installing, finalised, and what-not.

Thanks for taking the time to reply. I'm going interpret that as a soft yes with reservation since you seem to be extremely knowledgeable in this specialist area and additionally state that you are doing 'very bespoke' work within that. Interesting stuff, great to hear from someone with such knowledge. I suppose Starlink/SpaceX people are NDA'd to the point they can't comment, maybe we'll hear from them in future. FWIW internally we've achieved autonomous cable production ex termination (length, strip and cut including measured splice-points - rarely used), and custom bus bars which I guess are a nominal harness in a sense. Anything more complex we need pre-terminated still gets sent out.

I would also add that the 'New Space' scene is going to be much more willing to accept autonomous manufacturing methods. People like ESA are much much more resistant to change, and will take a lot of convincing that it is as good as the slow manual method.

Honestly, as things are looking, I am likely to be leaving the space sector in the near future, so I am not too worried about protecting my job or future on that front. The more accessible space is the better, even if that means people like me become less common.

Reply on news.ycombinator.com