3 ms·

> They're very happy with their current "sell you a specialty part for $45 that costs them $2 to produce" business. How difficult is it to design these parts a

by Reelin 7y ago

> They're very happy with their current "sell you a specialty part for $45 that costs them $2 to produce" business. How difficult is it to design these parts and successfully fabricate them? Are they expensive simply because they're uncommon, or is the R&D process for them nontrivial?

extrapickles 7y ago
The main problem is the lack of scale to spread the R&D costs over. If a part takes $1m to develop, and you only sell a few thousand of them, the part will be in the hundreds of dollars regardless of how "cheap" it is to make. QA processes also add a fair amount to the final price as each part might have several hours of individual attention by a technician to make sure it operates perfectly. Even for something as simple as a strut, they might saw every other one they make in two to inspect it for defects as doing so is vastly cheaper than having something fail in launch or orbit.
madengr 7y ago
The space qualified transistors are: Hermetic ceramic and copper tungsten packages. Eutectic die attach. Hand wired bonded. X-rayed for voiding under die. Gettered to absorb hydrogen. High temperature lid soldering in helium atmosphere. Leak tested. RF-DC burned in and thermal cycled. Note the threshold voltage changes as the parts are burned in, so all these parts are aged. The PCBs are aluminum backed. Gold selectively removed from PCB traces to prevent solder embrittlement. Parts sometimes vacuum soldered to PCB (to prevent voiding). X-Rayed to look for voids. Cleaned and vacuum baked. RF-DC burn in. My stuff had the lids welded on. Vibration tested. More electrical testing and thermal cycling. Ping tested to hear for loose parts and solder balls. All this adds up the cost. Rework is a nightmare, especially after the lids are welded. All the other SMT parts are also S-level. A single tube of low-outgassing RTV (Dow 340) is $600.

by Reelin 7y ago

extrapickles 7y ago

The main problem is the lack of scale to spread the R&D costs over. If a part takes $1m to develop, and you only sell a few thousand of them, the part will be in the hundreds of dollars regardless of how "cheap" it is to make. QA processes also add a fair amount to the final price as each part might have several hours of individual attention by a technician to make sure it operates perfectly. Even for something as simple as a strut, they might saw every other one they make in two to inspect it for defects as doing so is vastly cheaper than having something fail in launch or orbit.

madengr 7y ago

The space qualified transistors are: Hermetic ceramic and copper tungsten packages. Eutectic die attach. Hand wired bonded. X-rayed for voiding under die. Gettered to absorb hydrogen. High temperature lid soldering in helium atmosphere. Leak tested. RF-DC burned in and thermal cycled. Note the threshold voltage changes as the parts are burned in, so all these parts are aged. The PCBs are aluminum backed. Gold selectively removed from PCB traces to prevent solder embrittlement. Parts sometimes vacuum soldered to PCB (to prevent voiding). X-Rayed to look for voids. Cleaned and vacuum baked. RF-DC burn in. My stuff had the lids welded on. Vibration tested. More electrical testing and thermal cycling. Ping tested to hear for loose parts and solder balls. All this adds up the cost. Rework is a nightmare, especially after the lids are welded. All the other SMT parts are also S-level. A single tube of low-outgassing RTV (Dow 340) is $600.