4 ms·

Addressing your first paragraph, I have the feeling that the F-18 E/F Super Hornet is a real-world example of what you're talking about. A good airframe (the le

Addressing your first paragraph, I have the feeling that the F-18 E/F Super Hornet is a real-world example of what you're talking about. A good airframe (the legacy F-18) was taken as the baseline for a new aircraft that featured greatly improved systems in all areas (including electrical power generation!). Does a Super Hornet deliver more value than a legacy F-18? Absolutely, I don't think anyone familiar with the type would argue otherwise. Anecdotally, all my friends with experience on both ranges of F-18 love the Supers. I've never touched them on any kind of engineering project though. But the Super Hornet was not risk-free (not that you implied such a program would be, I feel that you gave the concept a fair shake). It encountered some problems that almost killed the program. Here: http://oai.dtic.mil/oai/oai?verb=getRecord&metadataPrefix=html&identifier=ADA387614 http://oai.dtic.mil/oai/oai?verb=getRecord&metadataPrefix=ht... Click on the link next to "PDF url" to obtain the report. This is a study of what almost wiped out the program...Trans-sonic wing roll-off, aka "wing drop". Much more readable article here: http://www.nasa.gov/centers/langley/news/factsheets/F-18.html http://www.nasa.gov/centers/langley/news/factsheets/F-18.htm... The cause was a difference in the wing-fold in the larger Super Hornet as compared to the original Hornet. You never know where risk can come from. You can, as you are no doubt well aware, try to schedule some risk out of the system, and design risk out of it, but it's always there. This is not an argument against your point -- after all, the F-18 E/F is a real world example, as I said, and a successful one -- but pointing out some of the caveats of it. Even things that seem like they should be simple can turn out to be complex. A lot like software...

nickpsecurity 11y ago
Actually, I agree the Super Hornet is an example, maybe the best, of what I propose. The results show it delivered on my promise as well. It's no surprise that they ran into problems. Matter of fact, the problem is a common one in system evolution: assumptions built into use of components in version 1 no longer apply in version 2 but it wasn't taken into account. IIRC, this was the same reason for the failed conversion that destroyed the Ariane 5 rocket. The receiving component assumed and was coded for a smaller number, resulting in an overflow. This is why I'm a fan of keeping unambiguous, mathematical specifications (eg Z) side-by-side with readable, English description side-by-side with design. The specs should embed assumptions, maybe more visibly than other things. We've seen this in software with both Eiffel/Ada's Design-by-Contract and static analysis tools with similar pre- and post-conditions. You could say typing also helps. The person using a component should be able to readily tell what its requirements were and do a manual/automatic sanity check. Personally, I'd be surprised if your CAD tools for aerospace don't have requirements or integration checking features built into them. F/18 started around the time that such tools would've both been feasible and had computers that could run them fast. Do they seriously not have ways of catching at least obvious problems with component reuse?
- engi_nerd 11y ago
  I agree with you about problems in system evolution. We both see this in our work. And about that Ariane incident, you're also right. More detail here: https://en.wikipedia.org/wiki/Cluster_%28spacecraft%29 https://en.wikipedia.org/wiki/Cluster_%28spacecraft%29. For anyone interested, I can recommend this book: http://www.amazon.com/Space-Systems-Failures-Satellites-Exploration/dp/0387215190 http://www.amazon.com/Space-Systems-Failures-Satellites-Expl... Your second paragraph sounds like a call for literate engineering, akin to Knuth's literate programming. I like this idea. About CAD tools for aerospace, it really depends on what CAD tools you use. Things like Pro/E and Solidworks explicitly allow you to put in specifications and tolerances, and also for interference checking and the like. For electrical design, there are tools that allow you to put in electrical requirements and run simulations on your design files to test against those requirements. So there are some ways of catching problems ahead of production. Now, those tools aren't available to everyone. So I'm on job #3 in this industry. Job #1 had me doing all my electrical designs in AutoCAD LT. No checking or simulations there (can't even generate a bill of materials with this either)! All we had was a library of symbols that one of the engineers had made. Job #2 had me using Cadence ORCAD, but without any of the add-ons that made it useful (could generate a BOM, but couldn't customize it, which meant that I was assigned to create an Excel macro to customize the BOM to meet our specific stock-keeping system's requirements). Job #3, I have nothing because all the vast majority of the electrical designs are done somewhere else (they use CATIA). Whenever I actually do some design work, I'm reduced to taking screenshots of relevant areas in other schematic PDFs, then stitching them together in MS Paint, and then pasting those into a Powerpoint template. I wish I were kidding. But with job #3 most of my time is spent on other things besides pushing dots, like testing and evaluating upgrades to our embedded systems (this kind of thing is one reason why I spend time here).
  - nickpsecurity 11y ago
    Wow, that's ridiculous and sounds painful! Again, I'm not a hardware guy, but I've absorbed a lot of knowledge and wisdom of it over past year trying to improve CPU designs and eliminate subversion possibilities. Studied their flows, synthesis, mask reduction, etc. Anyway, what your describing sounds behind even the free stuff. So, here's an amateur's attempt to help you out a bit with open-source alternatives: gEDA's 10+ tools to assist the process https://en.wikipedia.org/wiki/GEDA https://en.wikipedia.org/wiki/GEDA KiCad is similar w/ bill of materials and artwork https://en.wikipedia.org/wiki/KiCad https://en.wikipedia.org/wiki/KiCad QUICS for simplified usage in schematic capture & simulation https://en.wikipedia.org/wiki/Quite_Universal_Circuit_Simulator https://en.wikipedia.org/wiki/Quite_Universal_Circuit_Simula... Magic for those wanting old school (a recent SOC used it) https://en.wikipedia.org/wiki/Magic_%28software%29 https://en.wikipedia.org/wiki/Magic_%28software%29 You get used to something like gEDA or KiCAD then you can use it (even on your laptop) whenever the work tools suck. Not sure if it will help you since I understand just enough to know the tools exist and what they do but nothing further. And if you get bored with hand-made circuits, look up Qflow: from behavioral synthesis down to detail routing all in open source. Handles much of opencores.org's Verilog. Have fun with that and please tell me if any of the CAD tools are solid replacements for commercial use in your field.

nickpsecurity 11y ago

Actually, I agree the Super Hornet is an example, maybe the best, of what I propose. The results show it delivered on my promise as well. It's no surprise that they ran into problems. Matter of fact, the problem is a common one in system evolution: assumptions built into use of components in version 1 no longer apply in version 2 but it wasn't taken into account. IIRC, this was the same reason for the failed conversion that destroyed the Ariane 5 rocket. The receiving component assumed and was coded for a smaller number, resulting in an overflow. This is why I'm a fan of keeping unambiguous, mathematical specifications (eg Z) side-by-side with readable, English description side-by-side with design. The specs should embed assumptions, maybe more visibly than other things. We've seen this in software with both Eiffel/Ada's Design-by-Contract and static analysis tools with similar pre- and post-conditions. You could say typing also helps. The person using a component should be able to readily tell what its requirements were and do a manual/automatic sanity check. Personally, I'd be surprised if your CAD tools for aerospace don't have requirements or integration checking features built into them. F/18 started around the time that such tools would've both been feasible and had computers that could run them fast. Do they seriously not have ways of catching at least obvious problems with component reuse?

engi_nerd 11y ago
I agree with you about problems in system evolution. We both see this in our work. And about that Ariane incident, you're also right. More detail here: https://en.wikipedia.org/wiki/Cluster_%28spacecraft%29 https://en.wikipedia.org/wiki/Cluster_%28spacecraft%29. For anyone interested, I can recommend this book: http://www.amazon.com/Space-Systems-Failures-Satellites-Exploration/dp/0387215190 http://www.amazon.com/Space-Systems-Failures-Satellites-Expl... Your second paragraph sounds like a call for literate engineering, akin to Knuth's literate programming. I like this idea. About CAD tools for aerospace, it really depends on what CAD tools you use. Things like Pro/E and Solidworks explicitly allow you to put in specifications and tolerances, and also for interference checking and the like. For electrical design, there are tools that allow you to put in electrical requirements and run simulations on your design files to test against those requirements. So there are some ways of catching problems ahead of production. Now, those tools aren't available to everyone. So I'm on job #3 in this industry. Job #1 had me doing all my electrical designs in AutoCAD LT. No checking or simulations there (can't even generate a bill of materials with this either)! All we had was a library of symbols that one of the engineers had made. Job #2 had me using Cadence ORCAD, but without any of the add-ons that made it useful (could generate a BOM, but couldn't customize it, which meant that I was assigned to create an Excel macro to customize the BOM to meet our specific stock-keeping system's requirements). Job #3, I have nothing because all the vast majority of the electrical designs are done somewhere else (they use CATIA). Whenever I actually do some design work, I'm reduced to taking screenshots of relevant areas in other schematic PDFs, then stitching them together in MS Paint, and then pasting those into a Powerpoint template. I wish I were kidding. But with job #3 most of my time is spent on other things besides pushing dots, like testing and evaluating upgrades to our embedded systems (this kind of thing is one reason why I spend time here).
- nickpsecurity 11y ago
  Wow, that's ridiculous and sounds painful! Again, I'm not a hardware guy, but I've absorbed a lot of knowledge and wisdom of it over past year trying to improve CPU designs and eliminate subversion possibilities. Studied their flows, synthesis, mask reduction, etc. Anyway, what your describing sounds behind even the free stuff. So, here's an amateur's attempt to help you out a bit with open-source alternatives: gEDA's 10+ tools to assist the process https://en.wikipedia.org/wiki/GEDA https://en.wikipedia.org/wiki/GEDA KiCad is similar w/ bill of materials and artwork https://en.wikipedia.org/wiki/KiCad https://en.wikipedia.org/wiki/KiCad QUICS for simplified usage in schematic capture & simulation https://en.wikipedia.org/wiki/Quite_Universal_Circuit_Simulator https://en.wikipedia.org/wiki/Quite_Universal_Circuit_Simula... Magic for those wanting old school (a recent SOC used it) https://en.wikipedia.org/wiki/Magic_%28software%29 https://en.wikipedia.org/wiki/Magic_%28software%29 You get used to something like gEDA or KiCAD then you can use it (even on your laptop) whenever the work tools suck. Not sure if it will help you since I understand just enough to know the tools exist and what they do but nothing further. And if you get bored with hand-made circuits, look up Qflow: from behavioral synthesis down to detail routing all in open source. Handles much of opencores.org's Verilog. Have fun with that and please tell me if any of the CAD tools are solid replacements for commercial use in your field.