Silicon Valley Redneck

Pondering the Stockman frightener

Yes, everyone's linking to it: David Stockman says we're doomed.

Which, I'm inclined to agree, we are.  To my knowledge, financial markets have been severely, and increasingly, out of order at least since sometime in the 1990s (I used to think the dot-com boom had been masking the problem, but now realize that it was a symptom).  While there's innovation trying to happen at low levels (bitcoin, kickstarter, and so on), the government and the large financial institutions are by now completely painted into a corner.

And yet... as I scan through Stockman's piece, I see some arguments that are just out of whack.

This explosion of borrowing was the stepchild of the floating-money contraption deposited in the Nixon White House by Milton Friedman, the supposed hero of free-market economics who in fact sowed the seed for a never-ending expansion of the money supply. ...

Under his successor, the lapsed hero Alan Greenspan, the Fed dropped Friedman’s penurious rules for monetary expansion, keeping interest rates too low for too long...

I'm not familiar with the details of the policy, but I think I see gold-bug logic behind this.  In a well-functioning market economy, "a never-ending expansion of the money supply" is a necessary thing.  Wealth is constantly being created, so surely the money supply should expand just as fast as necessary to represent the wealth, no?  There's no need to keep the price of gold constant, so long as the price of bread doesn't get too far out of line (as, in point of fact, it has, as the expansion of the money supply has far outpaced the creation of the wealth that it should represent).

Oh, and:

That Mr. Greenspan’s loose monetary policies didn’t set off inflation was only because domestic prices for goods and labor were crushed by the huge flow of imports from the factories of Asia.

I rather think that's how trade is supposed to work.

Most of the rest of the piece seems to make sense, though.  And, his tone, with regard to corruption in high places?  When you can't tell an old-school conservative from a Communist bomb-thrower, then maybe... just maybe... there's something seriously wrong with the system.

While I'm here, I'll toss in a note and a prediction.

Note: a much-needed increase in interest rates (other than certain real-world rates, such as credit-card debt) Ain't Gonna Happen.  For one thing, it would lead to a wave of defaults on those ill-conceived adjustable-rate home loans; for another, it would make the eternal refinancing of the national debt suddenly more expensive.

Prediction: within the next two years, as states (possibly led by California) run out of options for paying their bills and those of their pension plans, the Ben-Bernank will start another program... QE5, or QE6, or QB VII, or whatever... whereby it will buy vast quantities of long-term state bonds, thereby debasing the currency still further while kicking the can down the road by another several years.

Busy, busy...

The recent lack of posting is on account of my having rather a lot of Work currently at the front of the cue, plus not having so much as started on my taxes yet, plus productivity being impaired this week by a backache which gets worse if, among other things, I sit at my desk.

I think my body is telling me to blow off work (and taxes) and go for a quick 10-mile hike.

Oh, well: I got out for a bit under 6 miles this morning, actually, and if I manage to get a fast start tomorrow morning, perhaps I'll get ambitious and do the 8-mile loop, with the proper hills in it.

Actually, on one of the high-priority tasks I'm not yet holding up the parade - the hardware fabrication schedule has slipped by another couple of weeks, and I only have a few days' worth of useful stuff to do on that project before I have the hardware in hand - but this is a case of prioritizing on the basis of billing cycles, and the end of the month is nigh.

And now, time to go straighten out my back for a while, before returning to TMFAAC: The Massive Firmware Adaptation And Cleanup.

Update: and when I lie down to give my back a rest, I'm likely to be trampled by a kitten stampede.  For the Dave Barry fans out there: wouldn't "Kitten Stampede" be a good name for a Ukrainian bubblegum-punk band?

Stateful op amp? WTF???

Once again, I find it necessary to delve into the wonderful world of analog signals, in an effort to cope with the fact that simple copper wires are the most complex component of a system (as well as being the most expensive).

In this case, I need a unity-gain buffer for a signal.  Well, one for each of several signals.

For the lower(-ish) frequencies, I've selected the LMV116, on the basis of running on the voltage that's handy, having a decent common-mode range and 45 MHz bandwidth, and matching the footprint of the part I used for the higher frequencies.

In initial testing on my bench, it seemed to get the job done.  Installed in the system this morning, it didn't.

So, back to my own workbench with the little dinky boardlets.

If I administer a 25 MHz square wave, I should get, in general terms, a 25 MHz sine wave out, right?  The harmonics don't make it through, but the output should be good enough to feed to a frequency counter?

Hm.  No output.

Try 5 MHz.  Get little glitchies, but no actual waveform.

Try 1 MHz.  Works fine; output looks like a squarish wave of correct amplitude.

Crank up the frequency by MHzes.  Output looks good up to 4 MHz, then goes away, leaving only the little glitchies.

Crank the frequency back down.  Output doesn't come back until 2 MHz.

The input-to-output delay of this op amp (in a basic unity-gain buffer connection, remember) turns out to be frequency-dependent, and also frequency-history-dependent.  Crank up the frequency too far, and the delay increases dramatically, not returning to normal until the frequency has been reduced below where the effect started...?  Frequency-domain hysteresis?  Is that even a thing?

The datasheet doesn't characterize input-to-output delay, but in normal operation it looks like right around nothing (at 50ns/div), then when the input frequency gets out of the comfort zone, the delay suddenly jumps to about 75ns.

Oh, well... time to knock off for the day.  I have various options for dealing with this, probably on Monday.

Update: inspiration struck whilst I was out for a Sunday-morning poodle drag.  That 75ns input-to-output delay?  'Twas on the rising edge only.  The falling edge was handled promptly.  So... um...

Yeah.  Input coupling was a little too capacitive, and the falling-edge negative overshoot must have been going beyond the -0.3V that the LMV116 can handle.  This, presumably, was giving the chip a wedgie, from which it needed a couple of hundred nanoseconds to recover, after which time it would recognize that the input was high and resume tracking.

So: remove the capacitor from the input circuit, and it behaves as it should, to include turning a 25 MHz square wave into a pretty fair sine wave, at much reduced amplitude (because it's running into the slew-rate limit).  If I reconfigure the input circuit as a divide-by-three or so, the slew rate should cease to be a factor, and it should have better immunity to strange input signals (which it may well encounter in the real application).

I'm not sure where the hysteresis came in, but the mystery of the vanishing output is solved.

The Mouse Problem strikes again

What with using (a) Linux and (b) CAD applications, I put rather more wear & tear on the middle mouse button than does the average PC user.

And thus it came to pass, yesterday evening, that the middle-clicky on my not-all-that-old mouse ceased to operate unless I pressed the button both very firmly and just so.

So, after walkies and other errands this morning, it was time to swing by Fry's and pick up a replacement mouse.  Basic: Logitech or perhaps Microsoft, wired, and capable of speaking PS/2 because that's how things are set up here currently.

Well!

Fry's is in the middle of a great rearrangement... and, among other things, the double-height shelf units in the computer accessories aisles are all now single-height, which would seem to point to Less Inventory Anticipated.  Also, everything's getting moved around.

I found the mice.  Actually, I found three categories of mice: Mac, Super-Expensive Gaming, and Wireless.  Nothing with a cord, a PS/2 adapter, and a sane price, unless it was hidden in the Mac department.

Grrrr.

Oh, well.  As things stand at the moment, the mouse doesn't really need to go through the KVM switch, as the only two active computers in here are the workstation and the server, and I really only need to connect keyboard and monitor to the server, on those occasions when I need to talk to it directly.  So, I get a $20 Logitech wireless mouse, plug the USB micro-dongle into the workstation, and all's well for the moment.

The scene at my desk this morning

Top Hat lies atop a pile of important papers, making them rather less than fully accessible.

This is problematic, as I need to get at them to make progress on a serious rush job (Monday evening, the priority suddenly went from "sometime this year" to "the end customer wants it Tuesday, and right.")

So, she'll have to move momentarily.

The fuzzballs are all dusty this morning, after I rubbed diatomaceous earth into their fur in an attempt to rid them of fleas.  It remains to be seen how well that'll work.

Update: it gets worse.  Now they're awake.  No kittens on the keyboard!  I'm trying to program in C, not FUZBOL!

Update 2: and stop chasing my cursor!  Ya got any idea how much trouble it is to clean squished cursors off an LCD?

Update 3: the old diatomaceous-earth-in-the-fur trick may actually be working... 'cause the little black things in it are not moving.

Polycultularity

Whilst out and about this morning, I passed the residence of a German immigrant and observed someone preparing a whole small pig for cooking.

Must be for an old-fashioned St. Patrick's Day luau.

Li'l Huckleberry's gettin' a mite too frisky

Once upon a time, not that many weeks ago, Huckleberry was a tiny, timid, wobbly, sad-faced thing, cautiously exploring a small area of his new environment.

Now, he's scampering everywhere.

And my stacks of papers are undergoing spallation due to relativistic kitten impacts.

Parsing

There's this billboard I see along 101.

Picture of somebody, and the text: "I CANCER BEAT THIS."

I keep thinking it needs a coupla commas: "I, cancer, beat this."

Because that's the only way it comes out grammatical.

Yes, we all know that, but...

Surprise!  Most of the horrid security holes on Windows PCs are in aftermarket software, not stuff that Microsoft ships.

Well, duh!  It should be immediately obvious that, e.g., Flash Player and Acrobat have fundamental problems, given that (on Windows) they need extra privileges and a reboot after installation, and why should applications that do what they nominally do need any of those things?

But!  It's still Microsoft's fault, because Windows has fundamental structural flaws, always has, and always will (unless MS decides to break compatibility entirely and have a totally different API for Windows X).

Item: the official Windows API is incomplete, and there are backdoor hacks for getting important stuff done.  While there are now supported functions for using serial ports, for example (when I first had occasion to use a serial port under Windows, many years ago, it was pretty much necessary to talk to the hardware directly), there is still no documented way of determining what ports are available... and yet the device manager and the MODE command both know how to do this.

Item: it's possible to run antivirus software on Windows.  Antivirus software hacks into your system in exactly the same way that malware does.  If Windows didn't have basic structural insecurities, antivirus software would be not just unnecessary but impossible.

And, yes: even with proper system-level security, there would be the potential for application-level malware to mess with user files, so long as there are scriptable apps with access to the filesystem... but they bloody well shouldn't be able to install boot viruses unless you're running everything as root.

Speaking of soldering irons...

...the use thereof, and other basic skills...

Got a project in the works that involves a smallish plastic structural component.

There's a bit of unexpected time lag in arrival of 3D printed parts, in addition to which the samples we just received for the customer end of the assembly fail to match the drawing in some rather significant ways, so the parts on order wouldn't be right for the near-term test anyway.

And so: out to the garage, where some of my old hobby tools live; dig up some suitable plastic stock; fire up the air compressor and phase converter, then bandsaw, lathe, mill, and lathe again.  Voila!  Tolerances are kinda sloppy 'cause I was being impatient, but it'll do for initial testing.

Took less'n an hour, start to finish.

Came back indoors, and found two kittens on my office chair and the third in front of my monitor.  Guess they didn't like the noises coming from the garage.

Dude, where's my pet cow?

Or whatever that book is that Sam Vimes reads to his kid at bedtime.

Sunday evening, being the eve of Monday when the vet is working, I set the Momcat trap again.

Monday evening, being just moments ago, I heard something making a minor commotion out on the patio, with definite galvanized-wire notes to it.  Oho!

Out to investigate - the trap is sprung!  Lift cloth cover, aim flashlight... yes, it's occupied... by something big and gray and hairy... and wearing a bandit mask.

That's not my cow!

The story so far:

We caught Momcat in the old trap, but she escaped.

Then we caught her again, in the old trap, but she escaped into my car and proceeded to stink it up for a day before the vet techs had a look and declined to try to remove her; we ended up leaving the car doors open in my driveway to let her escape to her familiar hunting grounds.

We caught Cow-Print Cat in the new trap.

We caught Southmoon's Dad in the new trap.

Now we caught a Big Ol' Rackety Coon in the new trap.

This leaves the Giant Rat of Sumatra, and a Two-Tone Kitty that we haven't caught yet.

...Plus any foxes, coyotes, badgers, wolves, and midget circus bears that may be lurking about the neighborhood.

You are engineer? Where is soldering iron?

My occasional colleague Boris says his first boss greeted him thus.

The wisdom of that question is evergreen.

That prototype I got Friday was hand-built by a manufacturing technician.  Did a lovely job on the SOICs and QFPs, and quite satisfactory on the general passives.

But...

It's apparent that this was a soldering-iron job, not a paste-and-reflow.  A rather critical part in a DFN package turned out to have two vital pads not soldered at all (though there was some attempt at getting solder on the central pad), and a crystal that was installed 90° out of phase (hey, those packages are almost square) also turned out to have no solder under one of the pads, when I removed it to re-install in the correct orientation.

As you may gather from the preceding paragraph, I've been busy doing SMT rework on a Sunday afternoon.  Bouncing it back to the technician (and coaching him through the process) would take at least as much of my time, plus a bunch of his, and would add to the schedule slip.

So, yes: it's important for an engineer to know how to drive a soldering iron.

And: the hardware is now working, at least as far as powering up and running some test firmware meant for a different board.  The core is working; I still need to test a couple of simple interfaces... and write the firmware for this board.

There, I fixed it!

Got prototype #1 of a new gadget today.

The client's technician hadn't been sure which way 'round the main connector (a Harwin J-Tek vertical header) went... and had flipped a coin.  It was, indeed, backward.

Option 1: desolder the connector and try to re-install it.  Ugh.

Option 2: make up a backward wiring harness and paint both of the connectors orange to avoid confusion.  Better, but still ugh.

Option 3: uh, wait.  Polarization is by a single notch in the housing.  Cut a roughly-matching slot on the other side... yeah, the mating connector will now fit either way... then cover the inside of the original slot with a bit of sticky tape, mix up some 5-minute epoxy plus some graphite for rigidity and color, and plaster over the original slot.

Hey presto!  A couple of simple mods to the connector housing, and it's installed the right way around!

The epoxy should be about set by now.  Back to the lab to see how it's doing, and to make up a minimal wiring harness for a smoke test.  (Then more wires for the firmware-installation test, and so on.)

Aieeee!

I was eating lunch at my desk.

I told Huckleberry.  Don't get on my desk now, I told him.  Not a good time for that, I said.

And I kept putting him back on the floor.

And then, suddenly, he leaped almost-gracefully onto the slippery pile of paperwork atop my desk, and the unfinished plate of lunch atop that.

Comically realizing that he'd landed with a front paw in something squishy, he beat a hasty retreat.

Knocking over my glass, which he'd somehow managed to miss in the initial leap.

Milk everywhere!

Gah.

A mere meanwhile away: that mission-critical project I mentioned a couple of days ago?  There was another complication that the remote worker encountered in bringing up our part of it, which looked suspiciously like we weren't getting output from a COTS calibrator.  This is a piece of equipment intended for calibrating life-critical products, bought from one of the big-name test-equipment manufacturers, and it had a calibration certificate and everything.  And... remote guy shipped it back to the head office, the boss took it to a site where there was a working one for comparison... and It Doesn't Work.  Outputs are dead.  But it has a certificate of calibration!

And now, time to finish cleaning up after the Kitten Apocalypse and head off to a meeting.

Update: that calibrator?  Haven't got the full details yet, but apparently it had the wrong wall wart... so the UI appeared to be working, and there were no error indications... but there was no output.  Doesn't anyone know how to play this game?

One year on...

Been a year (less a couple hours) since I saw the news that Lex had died.

The wheels of official investigation grinding slowly, we still don't know what happened, beyond the NTSB's brief preliminary report and some scattered informal accounts giving additional details which may be spurious.

And meanwhile, I've utterly failed to follow through on either of the inspirations that struck within a few days.

Inspiration #2: come up with an awesome technical solution to the problem of landing blind in a low-budget plane at a field that doesn't have all the right equipment.  I actually did come up with what seems an entirely plausible concept; mass-produced, it would cost a few hundred bucks (plus several times that for installation), and I figure it would improve an experienced carrier pilot's odds in that situation from basically zilch to better than 80%.  Trouble is: it would improve Joe Cessna's odds to maybe 20%... and probably be blamed for encouraging him to get into that situation in the fist place.  So, not likely to be a viable product.

Inspiration #1: become more Lex-like.  Not in the sense of flying jets, or embarking on a military career (you wouldn't believe how many waivers I'd need, and I don't think they even have one for members of the risk-averse community), but....

See, here's the thing about Lex.  Apart from the interesting stories, he was a rare sensible voice in a mad world.  And: he had a sort of Personal Decency Field that was reflected in his comments section.  It's easy to keep a civil comments section when you have few commenters, or you allow only like-minded voices.  But Lex had an active comments section, and while it was dominated by military types, random civilians, including lefties, could participate civilly.  (As it turns out, the most persistently irritating contrary voice was actually a conservative in liberal guise, there to challenge prevailing assumptions.  Sounds suspiciously like a training tactic, but it also helped keep things less unfriendly to those who actually held views outside the comment-section mainstream.)

Another thing I notice is that, despite an active career in which he interacted with a great many people and held positions of authority over some of them, it seems that he never made any personal enemies.  De mortuis nil nisi bonum and all that, but I haven't heard anything negative about Lex in the past year, nor while he was alive either.  A possible clue: a rare post on something he regretted.  Once, as a young officer under a lot of pressure, he'd snapped at an enlisted man.  A momentary breach of professionalism and courtesy: this was his big nagging regret.

While it's true that I haven't accumulated much in the way of personal enemies, I've mainly achieved this by never being in a position of authority, not having a lot to do with people and their issues, and generally running away and hiding when things get personal or political.  This is the kind of stuff you can get away with when you're an engineer... but, if I'm to be an entrepreneur, I really need to develop the sort of people skills Lex had.

Inspiration #0, also not followed up on: do the analysis to support #1 in detail, and compose an epic post thereupon.  (Sort of the antithesis of Cosmo Lavish's obsession with the outward trappings of Vetinari-ness.)

Inappropriate poll answer

Current Slashdot poll: What Is Your Favorite Polearm?

All the answers on offer are blades on sticks, of one sort or another.

No one seems to have mentioned the obvious answer: the Radom pistol.

What?  It's a Pole arm, innit?

Mission-critical, we've heard of it

Currently involved in a long-running, er, Classic Foxtrot, involving a device that Must Not Fail, Ever.  (Not our design.  We're working on something that connects to it.)

One of our associates, who's working remotely, was having difficulty communicating with the device's main board.

Turns out he could get it working by pressing the reset button.

Apparently, if prime power comes up a little too slowly - and the rise time in question is not what I'd call slow - the main processor fails to wake up, and it doesn't have a watchdog timer that forces recovery from this situation.

Many years ago, my occasional colleague Boris and I standardized on a particular POR/watchdog chip because it, and no other (at the time) would generate repeated reset pulses as long as the processor failed to come up and feed the watchdog within (IIRC) 1 second.

Some better watchdog chips have come out since then... and perhaps some worse ones.

But here we have a safety-related product that doesn't even have a proper power-on reset, never mind a does-not-fail watchdog.

Yikes?

The things you notice when you're paying attention

Whilst waiting for my laundry to do its thing yesterday, I watched the Nova program "Rise of the Drones".

Not that I was paying that much attention, mind you, but:

Hobby guy points to GPS antenna, and says that what sends the signal to the satellites.  OK, so maybe a lot of hobby guys think that's how it works, but didn't they have some technical consultant to suggest it should be edited out?

Interesting way of lighting the faces of interviewees (more than one, but not all of them).  Not the standard light-from-below sinister effect, but more from the side... and still giving the sinister shadows.

Video clips (I noticed this in more than one place) from Predators at, what, 25K feet?  Video shows splodey.  Audio has simultaneous boom.  Uh... Predators got an audio feed, from way up there?  And, let's see, speed of sound, call it 1100 ft/sec, so, is that 25K feet above sea level?  Call it 20K feet AGL, so... shouldn't there be about an 18 second delay?  Or has someone been, like, adding sound to the video clip to make it seem right to the typical PBS-viewing* ignoramus?

This is just stuff I noticed on casual viewing and remember a day and a half later.

*Remember!  If you're watching PBS, you're still watching TV.

Helpful hints on feline launderment

Curiously, when I wash the kittens I disregard essentially all the advice given in this informative video.

Next time, I'll have to get Joy to act as videographer so I can prove I actually did it.  (Helpful hint: start when the subjects are very young.)

Their way is funnier, though.

Jumping on the breakfast-photo bandwagon

Eggs & sausage, at home.

Huckleberry and Southmoon were being very persistent this morning.

Also, all three of the kittens have taken to emitting a strange whirring-cooing sort of sound, pioneered a couple of weeks ago by Huckleberry, who at the time hadn't even heard of a tribble.  Now that they've all seen the show, I guess they think tribble noises are cool.  (Next, they'll be claiming fezzes are cool.)