pretty good rant

Where's Jim "design, then verify" Thompson when you need him...

But I can understand the sentiment. The problem with many problems in electronics design is that the requirements are underdetermined - too many variables, not enough constraints. It can be impossible to sit down and write out a set of equations describing some circuit and have the equations tell you by inspection if some particular solution is actually any good or not.

Sort of like Genearl Relativity - it's fairly straightforward (for a physicist at least) in many cases to set down what the distribution of matter and energy is in some region of spacetime and solve the field equations somehow for the metric...in theory there's no reason why you can't run the equations in reverse to determine what distribution you need to construct the metric for a "warp bubble", but it's a lot harder exercise and then the equations tell you you need exotic currently impossible stuff like negative energy.

Right. Optimum performance isn't always defined. We can fiddle the sim until we like what we see, then write the specs.

That's been my approach, I shoot my arrow (as best I can) and then write the spec's. This often bites you as every prototype has a golden part...

George H.

te:

e:

at he claims to know he seems to have got from the denialist song-sheet.

The point about the stuff posted over his signature was that it was the reg ular denialist clap-trap. John Larkin posts essentially same story at all t oo regular intervals when he picks up one of the bits of denialist propagan da that the denialist propaganda machine has planted in the Murdoch media.

I can understand why you don't waste your time reading that stuff, so you w on't have realised quite how obviously Will Harper has sold out, but I've r ead enough for it to be blindingly obvious to me.

sees a real threat to their cash flow, and have bought both politicians an d retired scientists. Most of the money gets paid to right-wing think tanks , originally set up by the tobacco industry to spread doubt about the damag ing effects of smoking on health. They've subsequently been exploited by ot her special-interest groups.

s, ambition, greed, and what seems to be a hereditary human need for a righ teous cause. What better cause than saving the planet? Especially if one ca n get ample, secure funding at the same time? Huge amounts of money are ava ilable from governments and wealthy foundations for climate institutes and for climate-related research."

eement, > lively debate.

prevailing orthodoxy. The denialist lobby cherry-picks the occasional scept ical paper, and keeps on publicising it long after the literature has ident ified its defects.

s agree

ical

d actually looked at.

, Roy Spenser and John Christy. None of them has been "squashed" in the nam e of political unity.

He did come up with a plausible and testable anti-anthropogenic global warm ing hypothesis - or at least and idea that would predict that cloud cover w ould increase to limit temperature rise, Sadly. it was tested, and wasn't r ight.

rom doing good work, but there's at least one occasion when they took some prodding to do the work that made their data less useful to the denialist l obby. They still got serious jobs and have kept them for many years.

get the same kind of numbers for the Relativity Theory, Quantum Theory and the Atomic hypothesis - around 1900 Ernst Mach still didn't believe that a toms were anything more than a useful modelling tool.

so more than 95% acceptance of any idea implies that the population being t ested has had some of the nuts filtered out.

There is no debate on climate science. There is a well-funded propaganda ef fort to make the general population more dubious about the science than the y should be, and it works on people who don't understand the science, or ha ven't bothered to read much about the science.

English-language science journalists don't know much about science, and som e are dumb enough to think that there's a debate going on. Dutch science jo urnalists have at least undergraduate degrees in the sciences they report o n, and they aren't fooled.

It's in much the same state. We've got a very good idea of what's going on, and sometimes we know enough to make reliable predictions. Try to predict the chemical properties of a synthetic element - something like technecium which doesn't last long enough to exist in the natural world - or the biol ogical properties of a synthetic protein and you've got a prediction proble m, not as bad as that faced by climate science, with a whole world to squee ze inside a computer, but bad enough to notice.

Very sensibly.

But it's the intuition (or experience, perhaps) that constrains the design at the start. You don't just randomly hook pins together.

Sure, but they'll be back!

That damned DDT ban!

My dad was stationed on an LST in the Pacific. They evaluated him carefully, thought about it, and made him a typist.

Mo's dad was a radio operator, but when things got tough on Guadalcanal, he had to fight. Killed some J***.

My father had just been called up for rifle duty (had a critical need deferment) and was just about to drive the family back to MI.UP to their parent's place on VJ day. ...and here I am.

Well, then there is AlGore.

Yeah, my FIL was on a carrier. His job was trombone player. He was so handy that his GQ assignment was his bunk.

Depends what you're doing. In my business, it's usually possible to do a bit of theory and find out what the fundamental limits are, i.e. how good the gizmo _could_ be. Then one can figure out how to get there.

Without that guidance, it's difficult to know when to stop--one so easily winds up settling for performance 20 dB or more off the pace. (I see that a lot in my consulting, both business and pro-bono.)

Cheers

Phil Hobbs

The interesting situation is where some specs trade off against others, or something trades against cost. That's not as easy a call as optimizing one thing. Exploring sets of possible specs, some just feel better than others.

One other problem we have is quantifying guaranteed specs. That's a statistical carp shoot [1], especially temperature, since we do not generally temperature cycle production units to make sure they meet specs.

Well, of course everything always trades off against cost, once you start getting close to the optimum. (One often sees designs that are,

Keep pumping slugs into that barrel--either you'll hit the fish or all the water will drain out. ;)

Cheers

Phil Hobbs

Sure I try and do that. Well two examples come to mind. The first is noise from a cap mult. that is on the output of a few watt power supply. I built several and all had noise near 1 nV/rt Hz. So I wrote the spec as 2 nV/ . Well the next batch of transistors were worse, and 2-3 nV is now more typical. (It's not really a big deal, and no one has called me up asking about it.)

The other has to do with Helmholtz coils. Here it's easy to do the calculation for the field inhomogeneity (minimum linewidth of my signal), and the first batch of coils were close. But it's harder than you want to know to make precision coils. We've got most of the bugs out. But I still have a set of little brass shims that I use to tweak up a less than stellar coil set.

George H.

I agree about the Helmholtz coils. For the cap multiplier, a transistor with a lower R_bb' ought to fix it. BJTs actually follow their noise models, unlike FETs (especially MOSFETs).

Cheers

Phil Hobbs

No, not at all. It just means that most climate scientists can't find any importnt gaps in the existing coverage, and are looking forward to developments that don't contradict the currently accepted picture. You can probably get 99% agreement that today is Thursday (in my time zone, at least), and that doesn't mean that dissent has been squashed, either.

R-bb, do they list that on the spec sheets or do I have to pull it off a graph. I'm using a 2N4401 (and 4403 for the other polarity.) Is there something better, (Did I mention it needs to be in a to-92 pac.) Like I said it's not a big deal, I've got a stash of the "good" transistors, and if I find one that is particularly bad I switch in a good one.

The big problem is actually the voltage regulator upstream of it. This was designed while I was developing my noise "chops", so there are some bits I would choose to do differently today.

George H.

raph.

g better,

Win and Paul list a number of suoer low noise, medium-power BJTs in their l ow noise chapter. (I'm riding in the car just now so I can't easily look it up.)

Cheers

Phil Hobbs