Photodiode amplifier noise

I had ONE course in transistors... Germanium... how to bias without thermal run-away ;-)

You didn't have a course in passive circuit theory and analysis?

Like AGW ?:-)

I'm not sure about your last statement. I tend to think in a language called "circuits", so I'm really good at debugging... even by remote. Ask the Stheno guys ;-)

...Jim Thompson

it.

It is, but include cooks as well. Had salmon at Sudwerks yesterday, exquisite. But I had a very similar salmon dish in other restaurants and it was bland. Good cooks must know how much a tad, a dash and a smidgen is yet very few really know. Just like there are engineers who slap the biggest hinking inductor they can fit onto an EMC problem, and later many units go bzzzap ... *POOF* upon disconnecting power.

it.

Did you learn all those sound effects the hard way ?:-)

...Jim Thompson

I had part of a class in RLC circuits that I could have taught, but that's all. The rest I learned from poring over the ARRL Handbook, and (especially) databooks and app notes. Plus blowing up stuff, and not chasing girls till I got to be 18 or so. ;)

The main thing that helped was the math, and lots of practice at home. I got put in charge of all the TX side timing and frequency control on the world's first direct-broadcast satellite system, two months after getting my astronomy bachelor's degree (June 1981). (It was the Spacetel system from AEL Microtel--They really needed people, and I got insanely lucky.) I'd never built a PLL in my life, but it all eventually worked fine. That was where I really learned to use a lot of algebra when designing circuits.

I did have a really good undergraduate class on astronomical instruments (Astronomy 421, taught by the estimable Professor Gordon Walker), which is partly where I got my interest in electro-optics. Gordon is a great guy.

I don't think that it's impossible to be good at both design and troubleshooting, though debugging is partly t'one and partly t'other. (Michelangelo was a famous fortress expert, f'rinstance.)

Cheers

Phil Hobbs

No, there ain't much to write home about. Clare has some really big depletion mode devices but as usual there are hardly any specs useful to the analog signal guy because these things are meant to switch power.

Japan offers some nice JFETs in very small geometries such as the

You could carve out a niche catering to the fiber optics scientists, an area where extreme profit margins are customary. But it depends where you live. California is currently too biz-unfriendly for me to start anything where I'd need employees or where sales are involved. If you live in NY state it could be the same.

all

it.

it's

it,

Mine where mostly from my teenage days and of the tougher kind, with plaster falling out of the ceiling and so on. But nothing compared to the three-alarm a chemist friend had. Ended up in the ER. He said the lab had no door anymore and it was just one big black hole.

I could always multi-task ;-)

Math _is_ very important. I _did_ start out my career doing a complete Algebraic analysis of an OpAmp. Now-a-days I can usually "see" those solutions, rather than having to crank out each and every one, though occasionally I have a complex multivariable situation that brings me back to pencil-and-paper.... like the dB linear gain control for a sonar, compensated over temperature, power supply voltage, and process corners... posted here a few years ago to taunt those claiming they understood... I couldn't get a single taker who could analyze it ;-)

...Jim Thompson

all

it.

it's

it,

Are you accusing Joerg of being a *POOF*ter?

all

into it.

it's

it,

Sno-o-o-o-ort ;-)

...Jim Thompson

That thought has come up in conversation. Stay tuned.

Cheers

Phil Hobbs

Interesting! BTW, it would be good if your web site states a few more details about exactly what kinds of projects you feel competent to take on, experiences and so on. If a client needs a consultant in the optics area I would know where you'd be a good match but I found that a lot of people come to my web site out of coincidence, and then call. No referrals or anything. About half of those "inverse cold calls" result in actual consulting. Some of them really interesting, the usual analog stuff but totally new markets for me.

When you open a real brick-and-mortar kind of business which you almost have to if you produce anything, watch out for overregulation, over-fining and stuff like that. Even if it's just you and a couple of techs some day they might hit you because one bathroom isn't 100% compliant with some law or the lettering on some poster was too small. Best to do that in a state that's very biz-friendly and also won't tax your biz into oblivion.

Heck, I've written a couple of million lines of code, including three RTOSs, and I've never taken a programming course! I think that helps, actually.

The basic EE courses, circuit analysis, Signals and Systems, Fields, Electrical Machinery, the physics courses and such, were tremendously helpful to me. Self-taught designers can miss this stuff, and that can be limiting.

I suppose that circuit designs are also "out there" somewhere in some zillion-dimension solution space, waiting to be found. But, as you say, there's no hard "right" or "best" circuit. I've though a lot about the mental process of designing circuits, and who is good at it and who isn't, and I absolutely don't understand it, except to generalize that goofy, funny people are good at it and fatheads aren't. And it helps to start young. And you can do it in your sleep.

Troubleshooting is usually a consequence of sloppy design. Usually, after however many hours or days or debugging the hardware, the problem turns out to have been in plain sight, and could have been found and fixed by checking the design, before the board was etched.

John

BF862 is a pretty good part for a class of signals, like particle beam detectors in accelerators, or small piezos, where the source is capacitive. I have one paper that calculates that the best s/n happens when the capacitance of the fet is equal to the capacitance of the source, sort of a capacitive maximum-power-transfer thing.

I've been meaning to look into varactors as low-noise LF amps; they were popular for a while before fets got good. I have no idea what the noise mechanisms are.

Distributed amps have interesting noise behavior, when you can use them.

John

I think one could make a nice living finding scientific instruments that are operating far from their theoretical limits, and finding ways to improve them. It wouldn't really be all that hard. You'd want a field that's fairly competitive, so you could shop the ideas around; that has better sales dynamics. The biggest issue in selling ideas like this is usually the competition from the inside engineers, who will often start to sandbag you to management the minute you finish your presentation. With a little practice, you can usually tell early-on when you're going to lose that battle.

John

IME it's actually related to trying to do everything without external help. Many of my consulting efforts results in a small amount of billed hours, followed by time-consuming and expensive re-design on the part of the client after which it (normally) works fine. It's not the parts cost, I try to do everything jelly-bean style. It's the "miscellanea" such as $5-10k for a layout plus board spin, another $5-10k for a new round at the EMC lab, $20k or so for other agency cert, $100k++ in missed business because they couldn't deliver product on time. All this totally avoidable had they called me in before the layout, latest at the design review stage where we could have stepped on the brakes and re-do some stuff.

There is, but in the analog world camouflaging is not easy at all. Other than grinding off part numbers and potting the whole chebang there isn't much more to obfuscate. The big kahuna of the particular marketplace will surely notice when turf is invaded and try their utmost to find out how yours works.

I've looked at a $1000 device from because an engineer came into the lab with it already open and asked "How does this work? Could we do it that way?" All P/Ns were filed off but within a few minutes I knew how it worked and what chips they had most likely used. It was cheap alright, probably no more than $20 worth of parts in there (ain't that a nice fat profit margin?) but I told him that we could do it for a couple of bucks ...

I found the only method is to entice an atmosphere of "together we can improve this". Even if in essence you are doing it all by yourself, never say so. Don't take the glory, only take the check :-)

Some stuff is easier to visualize than other stuff, but it's true that you can stare at perfectly functional hardware for quite awhile before realizing it. The Footprints sensor had some seriously weird normal behaviour that took me a long time to understand--partly because I was trying to debug it using buggy firmware (written by a summer student) talking to buggy software (written by pros). One of the weirdest parts was the nonlinear capacitance of the LEDs, which was hard to probe because the whole net was just a nonlinear cap in series with a linear one.

I'm with you on the importance of the math. My two favourite courses in grad school were an applied math course on asymptotic analysis and advanced calculus of variations, and Bracewell's course on Fourier transforms, which was called EE but was entirely math. The math prof had an EE bachelor's degree, so he taught it the same sort of way.

(BTW the applied math course's textbook was Bender & Orszag, "Advanced mathematical methods for scientists and engineers", and is the best math book I have ever seen, bar none. Have a squint at it if you get the chance.)

Cheers,

Phil Hobbs "A good sleazy generalization is worth a dozen exact special cases."

Thanks. So far it's a partly-baked idea, but I'm doing some work on it in the lulls of my current expert witness gig. The idea is to take some of the problems that bedevil complicated optical setups and just nail them to the floor, so it's like they were never there. It's a collaboration with a colleague who has first-rate design and manufacturing resources, so I can stick to what I'm good at.

At the moment I'm a sole proprietor with a DBA as ElectroOptical Innovations. My Dad was a first-rate business man, but that seems to have skipped a generation--I haven't balanced my cheque book in 25 years. (On the other hand, he had Grade 11, so he and I really weren't in competition in any area. He was a great guy.)

You're right about the web site needing an overhaul--most of the folks who have contacted me have been readers of my book whom I've helped informally in the past. One of my daughter's friends is a freelance graphic designer and Web human, so I'll probably ask her to redesign it.

My current policy is that the first hour or two is free--lots of people have helped me over the years, after all, and besides it's interesting, saves a lot of wasted effort, and makes it easier for people to call.

Cheers

Phil Hobbs

Good. Then he probably has a brick-and-mortar business place already and knows how to keep bureaucrats from messing things up too badly.

Yes, a degree can't hurt but it is not an absolute requirement for business success. In his days things were easier though, less bureaucracy. You hung out a shingle and the business was founded ;-)

Make sure she doesn't put fluff in there ...

That's very normal in consulting. I often spend 1/2 to 1 hour on the phone with a new client and that's non-billed time. Sometimes they want a coarse estimate what it would take in terms of billed hours to get to an xyz approval, and for that I need to spend another hour or so perusing schematics, Gerbers and such. So yeah, 1-2h is about right. Always been that way, even back in the early 90's.

BTW, take a look at HP optical spectrum analyzers if you need one. I was surprised how little money clients had spent at auction for those. Far less than what us analog guys have to shell out for RF versions. Same for optical benches and stuff (but you need several burly guys to haul it in).