Larkin, Here's mine...

If you vote for the state machine they'll even pay you. ...many times.

Hmm, I wouldn't value a diploma very high :-) When I need to interview people part of the interview is taking a test. I've come across 'engineers' which could only engineer on (diploma) paper.

Good one!

--
Failure does not prove something is impossible, failure simply 
indicates you are not using the right tools... 
nico@nctdevpuntnl (punt=.) 
--------------------------------------------------------------

Even assholes can be fun.

--
Well, yes. 

Even _you_ can be amusing.

the better. The

And, unfortunately, things like deadlines and children that make it much harder to spend enough time doing the exercises to get really good at it.

I agree in general, though I'm far less cynical about the value of course work than seems to be general here. I see that I've been lucky in that essentially every course I ever took has been directly useful at some point in my career, even my orbital mechanics and galactic dynamics classes. (They used a bunch of really useful sleazy approximations.)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 USA 
+1 845 480 2058 

hobbs at electrooptical dot net 
http://electrooptical.net

A spreadsheet is a wonderful tool for composing works of fiction.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation 
Picosecond-resolution Digital Delay and Pulse generators 
Custom timing and laser controllers 
Photonics and fiberoptic TTL data links 
VME  analog, thermocouple, LVDT, synchro, tachometer 
Multichannel arbitrary waveform generators

Indeed. MIT only gave me the _tools_, not the knowledge and experience... you learn to design circuits on-the-job.

Back in the days when I had to hire people I used a simple

3-transistor bias circuit as a "test". In all my years, only ONE person attacked the problem properly... all the rest were laughable.

I also posted it here... no one got it exactly right. I'll try and find it, if not, I'll redraw it... it's pretty simple-minded.

...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| Phoenix, Arizona  85048    Skype: Contacts Only  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

the better. The

I've always wondered how people compute multiple rocket burns to move a satellite in orbit, or correct a Mars probe.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation 
Picosecond-resolution Digital Delay and Pulse generators 
Custom timing and laser controllers 
Photonics and fiberoptic TTL data links 
VME  analog, thermocouple, LVDT, synchro, tachometer 
Multichannel arbitrary waveform generators

the better. The

Interesting question. With difficulty, apparently. The optimization problem for continuous (eg. ion) thrusters is seriously challenging, some folks have been working on it for years. Two part burns tend to be used with conventional thrusters because they're not impossibly complex to calculate. Or so I'm told.

the better. The

There are classical analytic solutions, e.g. the Hohmann transfer and the bielliptic transfer, that minimize the required delta-V.

I don't know how NASA, ESA, and those folks do it, but if I had to reinvent the celestial sphere, I'd probably start with the osculating orbit, which is the heliocentric or geocentric ellipse that 'kisses' the true perturbed orbit at the moment of the burn ('osculare' is Latin for 'to kiss'). From there, I'd compute the burn required for the Hohmann transfer, and then take that as a starting guess for an iterative improvement scheme using the best available numerical model for the relevant parts of the solar system. (That was probably how it was done circa 1975, at a guess.)

Bielliptic transfers sometimes require less delta-V than Hohmann transfers, but can take a lot longer to execute--ISTR that the optimal bielliptic transfer from Earth to Mercury winds up going out past Jupiter in the process. You save a lot of delta-V by doing all your burns at perihelion, because that's where a given change in momentum gives you the biggest change in kinetic energy.

Being able to execute multiple-flyby slingshot orbits as accurately as they usually do is pretty amazing.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 USA 
+1 845 480 2058 

hobbs at electrooptical dot net 
http://electrooptical.net

Should have added: the other reason for the long transfer ellipses is that adjusting the angular momentum of the orbit is cheapest at aphelion, the further out the better. So you want to do orbital energy adjustments at perihelion and angular momentum adjustments at aphelion.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 USA 
+1 845 480 2058 

hobbs at electrooptical dot net 
http://electrooptical.net

much the better. The

I'd expect that there is some amount of simulation with successive approximation involved.

If you have a uP with multiply but no divide, you can do a nice integer square root by bitwise successive approximation.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation 
Picosecond-resolution Digital Delay and Pulse generators 
Custom timing and laser controllers 
Photonics and fiberoptic TTL data links 
VME  analog, thermocouple, LVDT, synchro, tachometer 
Multichannel arbitrary waveform generators

much the better. The

approximation

Awhile back there was a guy who built an approximate N-body code that ran in Nlog(N) time. As you'd expect, it did vaguely the FFT thing, i.e. it ran the nearest neighbours one by one, then lumped them into a single centre-of-mass to calculate the resulting perturbations further away. I forget his name, but it was an interesting approach--it works pretty well in a lot of cases, because it's second-order accurate. Tidal effects are completely wrong in general, though, because those are third-order.

N-body orbits are usually (maybe almost always) chaotic, so since accuracy is going to decline badly with time anyway, it's a win for lots of purposes to be able to get a general idea say 100 times farther into the future, which is what the Nlog(N) algorithm gives you.

I saw the integer square root algorithm in Knuth iirc, but I've completely forgotten the details. (Even ARM Cortex M3s have hardware divide now!)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 USA 
+1 845 480 2058 

hobbs at electrooptical dot net 
http://electrooptical.net

It's not a theory. It was the reality of the late 1960's, when the real motivation for getting an education was closely tied to getting a student deferment and avoiding a one way ticket to Viet Nam. There were far too many college age students that wanted to get into the low cost state subsidized colleges. In order to thin the ranks, the colleges literally turned the education process into an obstacle course. For example, the courses and curriculum necessary to graduate with an engineering degree were impossible to complete in 4 years. It took me 3 colleges and 6 years to graduate.

That's fair. Collaborative efforts were deemed little better than cheating at the time. Today, group efforts are encouraged as part of learning how to work on a "team". Times have changed.

better.  The

It might be possible to sell such a generality to trusting managers. I never had the credibility, so it was necessary to justify my assertions with hard numbers, calculated estimates, and sane projections. I put what I learned to good use with a summer job working for an insurance actuary, where I participated in such exercises as putting price tags on missing body parts. The lessons learned in this class, which I still consider to be the most important things I learned in college, were also applied when I temporarily took over my fathers lingerie business. What I managed to learn in Engineering Economics was very superficial, but provided the basic concepts necessary to do further study, and to make the all important financial mistakes that are part of the learning process.

Incidentally, your statement of selling below cost is rather amusing. During the dot com boom of the 1990's, I heard "Yes, it's a net loss, but we'll make it up in volume". At first, I thought it was a joke, but then realized that the manager offering that line actually believed it.

That's not what I meant by "why". The problem is what motivates people, what inspires people to go beyond 8am to 5pm, what makes buyers purchase one product, and not another, etc. Little things, like color, weight, shape, rounded corners versus sharp corners, and even smell, have a huge effect on product design. While most of this fits better into "industrial design", with a small company, the designer tends to do everything. If you fail to understand and appreciate the value of understanding what motivates people, you're going to be making quite a few marketing mistakes. If you don't understand what motivates engineers, techs, and managers, you're going to step on a few toes trying to make things happen.

I was never a graduate student, so I wouldn't know. Mostly, I have an incurable desire to test the limits of things. How fast can it go? How big a hole can it dig? What happens if I go over the red line? It's all part of "Learn By Destroying" where one learns far more by breaking and then fixing a device, than by studying it in a book.

Not possible. In order to make more money for the textbook monopolies of the 1960's, the required text books were rotated every few years. The source and supply of used current text books was very limited. Somehow, I have difficulties believing that basic concepts go obsolete every few years, but that's the way the textbook game was played.

We had Shaum's to help decode the textbooks. The fraternities shared the class notes. Sharing note was commonly done in the 1960's, but was considered borderline cheating at the time. I probably should have borrowed class notes, because I also had various part time jobs in order to pay for skool, but I didn't.

Good plan, but it didn't fit well with my study style. I had a fairly good short term memory. By trial and error, I determined that an all night cram session, which usually meant speed reading the books, was the most effective for me. Not the best way to learn anything, but good enough for surviving the ordeal process.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

A variation of my employment exam schematic...

...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| Phoenix, Arizona  85048    Skype: Contacts Only  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

oximation

Sounds a bit like the Fast Multipole algorithm, which the AIP Computers in Science journal included in its year 2000 issue within the list of "top ten algorithms of the century".

Regards, Mikko

approximation

That was probably it--the name sounds familiar. Cute idea, anyway.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 USA 
+1 845 480 2058 

hobbs at electrooptical dot net 
http://electrooptical.net

I find it a bit hard to believe that almost all the interviewees were really that clueless.

Are you sure it wasn't something a little less perfectly symmetrical? Maybe you asked for Rout, or beta wasn't quite infinite, or something like that? Or maybe they got psyched out, thinking it couldn't be quite that easy? Or you moved the goal posts by saying that the array was fast enough that it oscillated? Or pulled the fire alarm?

Inquiring minds want to know. ;)

(My brother's first year physics final had a question like that, worth something like a quarter of the credit: "A 200 lb man pushed against a slippery rock, with his body inclined at 30 degrees from vertical, but was unable to move it. How much work was done on the rock?")

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 USA 
+1 845 480 2058 

hobbs at electrooptical dot net 
http://electrooptical.net

I compute out Vdd/2, with no temp dependence.

Numbering BJTs on the left, top to bottom, as Q1 and Q2; on the right as Q3 and Q4 (bottom pair are Q2 and Q4, top pair are Q1 and Q3):

1. Vdd - R*Ic1 - Vbe1 - R*Ic2 - Vbe2 = 0

But Ic1=Ic2=Ie1=Ie2 (beta is infinite.) Also, since the bottom pair of Q2 and Q4 are a perfect mirror, then Ic3=Ic4=Ic2=Ic1.... nice as this means all the Vbe's are equal to each other. So:

1. Vdd - R*I - Vbe - R*I - Vbe = 0
2. Vdd - 2*Vbe = 2*R*I
3. I = (Vdd - 2*Vbe) / (2*R)

But the shared base node of Q1 and Q3 must be:

1. Vb = Vdd - I*R

And Vout, then:

1. Vout = Vdd - I*R - Vbe

So,

1. Vout = Vdd - [(Vdd - 2*Vbe) / (2*R)]*R - Vbe
2. Vout = Vdd - [(Vdd - 2*Vbe) / (2)] - Vbe
3. Vout = Vdd/2 + Vbe - Vbe
4. Vout = Vdd/2

Which is pretty nice. Is does depend upon T, but all the BJTs use the same Is and are ideal. And the Vbe's cancel, anyway.

So I think (10) is it. Hobbyist, thought. So I'm interested.

Jon

The original was slightly less "symmetrical", but of third order effect.

And you're not the typical interviewee ;-)

But, at the same time, you imply a level of understanding that I don't see on these newsgroups as a majority talent... observe my simple bootstrap math example, and the puzzled responses.

That's a good one... I'll have to come up with a circuit analog and see who I can entrap >:-} ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| Phoenix, Arizona  85048    Skype: Contacts Only  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.