Does spending over an hour vainly pushing my Norton up and down after a rebuild (good spark, petrol ok) then finding the pushrods neatly wrapped up in an oily rag count?
Brilliant! I do that sort of thing.
I remember changing the oil in a van I used to own and poured 6 pints of fresh oil in having forgotten to refit the drain plug. The mess!
When I was young and silly (I'm older now, but still silly), I drilled bolt holes down through the rear floor of my car, to install anchor plates for seat belts.
I drilled straight through the hydraulic brake pipe .....
So in trying to improve my chances in an accident, I had converted my vehicle into a brakeless death trap ...
Last edited by 603DX; 22nd Feb 2012 at 10:42.
Reason: "install" has two l's, dammit, I'm British!
>And in addition, it must be said that a truly competent engineer seldom trusts computer outputs completely. It is an instinctive reaction to do several simplified checks manually, on the validity of results. This attitude has been known to detect gross input errors, incorrect computer modelling, wrong material properties, and so on. So it is simply a convenience that the "jukebox" does everything in the blink of an eye, but don't cut the safety factors because of that - raise them rather than rely on the output unquestioningly, if anything!!<
So right - if that is done.
As was said in an international standards meeting:
'Simulation is like masturbation. The more of it you do, the more you prefer it to the real thing'.
I do find it amazing that simulation results are believed implicitly even when you can see they don't make sense, and, it seems, young engineers brought up on computer based design systems either can't. or haven't been taught, to do a manual sanity check on the results they get.
In one case where the calculations were being done as to separation needed between two radio systems on the same frequency to avoid interference, the results were quoted in decimal points down to 100 microns - for things kilometres apart. And the young man didn't see anything wrong in that because 'that's what the computer says'.
Another chain drive problem is stretch, which apparently is well known on the full and half size 'spam cans' - although there's only half size spam cans left with it.
The sleeve valves weren't that much of a success, as was the fire brick lined boiler.
FWIW there's quite a good book on the subject. I think it was too much too quickly rather than a goof. Lack of development was probably the biggest problem. Compare it with some of the early experiments in dieselisation.
I do find it amazing that simulation results are believed implicitly even when you can see they don't make sense, and, it seems, young engineers brought up on computer based design systems either can't. or haven't been taught, to do a manual sanity check on the results they get.
Clearly, we are on the same wavelength on this topic, radeng. For the early part of my 43 years as a professional engineer, I designed almost everything using a slide rule, which provided adequate accuracy. Multiple-figure logarithms were applied if fine tolerances were called for, such as processing topographic surveying data, and a few other things. And the habit of double-checking became ingrained.
So when fast computing came in, I made full use of it, but viewed it with a jaundiced eye, and continued with my checking practices. In due course I became one of the approved Supervising Engineers for the ICE in the firm of consultants I was with, responsible for on-the-job training of several younger engineers. I made a point of stressing the importance of checks and balances in design, to these youngsters.
As a result, in retirement I can sleep soundly at night, in the knowledge that I have done what I could to verify my own designs over the years, and also to pass on my attitude of questioning everything and accepting nothing at face value without checking.
Mr Grunt and Mr Mouse. Snap! Moggie split windscreen - with non-sidevalve bored out 950 fitted, to replace the standard 803 sidevalve, stage 3 head, twin strombergs etc........(Later to receive a 1275, but that's another story)... Whole re-build took ages, and cost real money - (I had none).
No matter how much time, money, effort and so on one puts in to a car engine, they do need a drain plug. I had forgotten that..................
I do find it amazing that simulation results are believed implicitly even when you can see they don't make sense, and, it seems, young engineers brought up on computer based design systems either can't. or haven't been taught, to do a manual sanity check on the results they get.
You may wish to repost that on the climate change thread.
one wonders if over engineering used to be much more the norm, and these days, with computer simulations, safety factors are less.
Ectually, one would argue that what once was thought of as "over-engineering" in fact was admission to not really understanding the problem well enough to define a "right" answer, so then adding guesswork padding based on superstition, intuition, and kids 'tuition was a form of obfuscation intended to create at least an appearance of doing the right thing.... when, in fact, time, information, resources and will to exactly get it right were insufficiently available.
Quote:
'Simulation is like masturbation. The more of it you do, the more you prefer it to the real thing'.
If done properly, one may not be able to tell the difference. Vive la simulation., eh?
Reality is that a lot of things built today - with values into the trillions of whatever annually - couldn't and wouldn't be built without the full benefits of simulation.
Essentially everything that is built into semiconductor LSI is highly dependent on simulation -- at the functional process level, at the timing and logic level, at the electron level, at the mask level, the fab process level, timing analysis, design validation. With the right bank numbers and a few friends, one can now go in some few weeks from a sketch on a napkin to a few dozen fully conceived, designed, working, tested prototype chips -- each one containing perhaps a few hundred million distinct working parts in a unique custom architecture - arriving before lunch in a fedex envelope. Not too long ago, the same level of effort might have taken five or ten or twenty thousand person-years effort to get an equivalent result. Simulation - at many levels - is what makes the difference.... and it works like gangbusters in the LSI domain with 100.0000% accuracy, consistent with the driving data.
Quote:
... and a nother thing...
Even better than simulation is REALITY. Now that we have a better bag of tricks, systems, machines, buildings, (and soon, maybe, living things) can be designed and built using best practices, and also instrumented up the proverbial ... so that the correctness of materials and design assumptions can be validated and measured throughout the design life of the thing, whether it be a bridge, a Hardon Collider of some sort, or a charcoal brazziere-thingy.
"I do find it amazing that simulation results are believed implicitly.."
When working on the box-girder bridge checks one team in our office was urged (old pals act?) to try out a new analytical program which was supposed to be completely compatible with what we were using.
One engineer had run a simulation using existing data and asked me to confirm what the units for end rotation were, degrees or radians?
I spent a bit of time looking at the output and could not make sense of it either, until I realised that the bridge deck ends were pointing downwards at just a bawhair off the vertical and the middle had apparently deflected by about 1.5 km.
Turned out that one program was working in newton millimetres and the other in kilonewton metres, a difference of 1,000,000 which the new program designers had not noticed.
Last edited by A A Gruntpuddock; 22nd Feb 2012 at 20:16.
For complex digital solutions in IC design, there's no alternative - Mrs Radeng tours the world delivering training courses on using her employers software to do digital design.
However, back in about 1982, a direct conversion pager using 150 bipolar transistors was designed in 7 weeks by an engineer using no more CAD than an HP41. It took him 5 weeks to lay out by hand on blue grid, 3 weeks to digitise and 4 weeks to process and assmbel and it worked first time.
Fast forward 15 years and another device of similar complexity used around £3million of computer and software, 3 engineers, took 10 weeks in design, 8 weeks in layout, was 20 % smaller, took 6 weeks to process and assemble and didn't work first time.....
A new satellite was launched a few years ago which had a spot beam for New Zealand to broadcast PayTV.Other transponders were for Australia only and some covered both countries.During in orbit testing it was discovered that the NZ spot beam was configured for the wrong polarity which meant that many thousands of people's sat dishes had to have their LNB's changed over time.The initial fix while the LNB's were swapped out was to migrate the transmissions to a transponder which did have the correct polarity but a lower signal strength.The satellite was specced in Australia and built in the US.A lot of the world uses H (horizontal) and V (vertical) but the US uses X (horizontal) and Y (vertical).Why it wasn't discovered during the final ground tests is unknown.The customer project director had also resigned a month before launch.All kept very quiet.
Last edited by TWT; 22nd Feb 2012 at 23:06.
Reason: typos
Clearly, we are on the same wavelength on this topic, radeng. For the early part of my 43 years as a professional engineer, I designed almost everything using a slide rule, which provided adequate accuracy. Multiple-figure logarithms were applied if fine tolerances were called for, such as processing topographic surveying data, and a few other things. And the habit of double-checking became ingrained.
As a part-time teacher of engineering subjects, I regularly see results which neither need nor merit more than three significant digits, given to much greater precision.
One of the subjects I supervise includes students taking river gaugings, ie measuring the flow rate. We always get the students to make an estimate before they do the measurement, and have an informal competition among the students and teachers. We teachers can almost always get within 10%. A gauging, correctly done, should be within 5%.
The big risk is good engineering but not enough of it! In days of yore extra material was allowed for things that it was impractical to optimize. Now we engineer to take into account all (and only) the things that we can predict. When the unpredicted comes along, the whole thing collapses like a pillar of s**t in the sun. The big buzz in the North American Electrical Engineering world is the "Smart Grid". The idea is that the old grid, increasingly unmaintained, overloaded, and subject to the destabilizing vagaries of wind power etc. will continue to serve because it's now managed by computers. When the time comes it's going to collapse like the pillar mentioned above!
LATCC was brilliantly situated at West Drayton. As originally conceived with Linesman/Mediator (itsself deserving of a nomination) it would have been at the heart of the UK's Air Defence. The Russians were obviously considered incapable of finding Heathrow then flying a couple of miles north in order to bomb the place. An attempt was made to hide it in the middle of a council housing estate though. Entry to the replacement at Swanwick is made by a road at the rear of the building. I heard that there were issues after construction was commenced and the planned appproach from the front was cancelled.
