The Sorcerer's Progress (Page 38 of 62)

K caused all kinds of work for me in the Caligula! chapter. Changes still in progress.

Sometimes you get a comment like: This shouldn't be a flashback, but have it's own scene. And you agree, despite knowing it's not just a misplaced comma. *looks at everyone in the room, you know who I'm talking of*

I've spent three frotzenglarken weeks tinkering with the outline for Maurand, Day 2 and writing it bitsie by bitsier.  Now I'm typing it up and rewriting On The Fly.  And let me tell you, there's not much room for a laptop on the back of a fly.

I've got about a thousand words done.  There's about 5000 more to do (I fit about 650 words in Pitman on a letter page) and then I have to do some edits in stuff already written (pub'd and un-).

Sorry these prelims are taking so long.  I know that (most) everyone needs the points,

I'm condensing and fixing inconsistencies as I go.  This section is up to about 3 kWords.  I've got parts of it written (you've seen Pausonallie ask Merran to join her study group.)

I just hope there's enough going on to hold the readers' interest in all the conversation and small action.

~!@#$%^&*() Windoze!  I was editing a message here and needed to put the machine in sleep mode--whereupon it crashed (BSOD) losing my work-in-progress and forcing a cold boot while I'm running on battery ... and I'll have to do a full file system check later.

Anyhow ... where NJC haz bin:

I had my attention drawn to a suspended project, the W.Marshall Leach-based headamp.  In order to keep it running longer and more consistently on the batteries, I put a voltage stabilization network in.  It wasn't good enough, so I added negative feedback, which unfortunately seems also to have fed back signal ... which I didn't expect.  That reduced the gain from 20x to 2x.  20x is just barely enough.

So ... when I put it down before I was looking at ladder filters in the voltage level feedbacks.  But since I never fully internalized filter theory, I have to go back to Budak.  This is really a classic, but the growing importance of digital and FIR filters has probably pushed the book out of the mainstream.  (Note the number of offers under $150.)

I've spent the last couple of days looking at better voltage stabilization.  Unfortunately, the very simple network does not yield to easy analysis.  I'm sure someone, somewhere has a good set of formulas, but I have no idea where to look.  I've been doing algebra trying to get usable-enough formula.  I'm sure there's one to be had, but it's not coming out, and I've spent too much time.

Worse, I've got a set of experiments worth running, but they really want a test rig ... and if I have to build one I'd like to make it re-usable, and ... well. there's 20 hours of careful measuring and drilling and wiring and soldering, and of course the parts to buy.

Sigh.  And week after next, I'm going to visit my brother.

Do you see why Paddy isn't at work today?

I have gotten some stuff done, just not in the last 48 hours.  I'm going to get to work now, while trying to keep a sober head on the circuit work.  Drill guide ... angle iron to clamp the drill guide to ... how to measure from the non-parallel sides of the drill guide ... how many crosspoint pushbuttons ... how many binding posts ... aarghh!!!!

How do you know the problem is the capacitors?  What kind/value/voltage are they?

So I think I'll build a much more limited, less flexible version of the test rig.  Much cheaper and an evening instead of the week.  (But I may buy one of the tools after all.  Just a nice thing to have.)

Some years ago, one of the beer brewing megacompanies ran a series of ads based on 'exceptional' people, including the folks who wipe up sweat in NBA games.  One of their targets was the guy who was going to invent a "battery-powered battery charger."  Of course, I already had such a device in my pocket, a reserve battery pack for charging my cellphone.

This little box's disadvantages are that it charges slowly and that it requires a regulated 12V DC to charge.

That makes charging it in the car a problem.  The car has 12V, but it's 'somewhere between 10.8v and 14.8v'.  A regulated voltage is generally regulated to either 10% or sometimes to 5%.

Worse, the available voltage swings from below the desired target to above, requiring a more complex regulator circuit.  If I had a lot of time, I might look at a switching supply with a flyback design.

But I don't have lots of time.  I've had this project in back of the back burner for a long time, and I want it for my trip on Tuesday.  So I went with another option.  When the car is running, the available voltage should run from 13.2 to 13.8 volts.  That's enough to use an LDO regulator.

What's an LDO regulator?

Hmm.  Well, to start, one of the widely available and cheap integrated circuits is the three-terminal regulator.  It comes in variants with more than three terminals (pdf) and in variants whose output voltage can be adjusted.  But in the most basic form, it's simplicity itself.  You buy one for the voltage you need (you DO need a bog-standard voltage, don't you?) and the polarity (only two: positive and negative).  You hook up the 'common' terminal to your 'ground' and the 'input' to your input, and the regulated voltage appears on the output ... except it's never THAT simple.

These devices need low-esr capacitors on input, output, or both in order to work properly, and that usually means tantalum capacitors.  Okay, not hard.  And they are designed for limited currents.  Fine, I only need half an amp.  One-amp regulators are widely available.

But ... the bog-standard three-terminal regulator requires an input at least two volts above the output.  I can guarantee only a bit over one.  Oops.

Fortunately, there are 'low dropout' (LDO) regulators available.  They're just a bit harder to find.  Not much harder, but you won't find them at Radio Snack.  Fine.  I ordered one a while back, and when I couldn't find it I ordered a bunch more, and paid for second-day shipping.  Mouser promises that if the order makes it through the warehouse by 20:00 Central, they'll ship that day.  I put the order in at about 15:30 Central Time.

The order didn't reach the warehouse in time.  At about 03:00 (yes, AM!) I got an email from Mouser saying that they were sorry it was delayed, and they were shipping overnight at their expense.  They took a loss on this for a small customer--kudos to them.

Of course, that was about an hour after I found the one I had misfiled.  But it never hurts to find spares.

Refer to the PDF above.  It's a very nice device for automotive use, including protection against voltage spike that can occur when a load drops off the system, and against voltage inversions.  That also protects the output capacitor, but not the input capacitor.  I put protection on using the power mosfet trick, which will drop about another 1/10 of a volt with the device in question (see the Saturation Characteristics graph).

What it won't do is protect the input capacitor from overvoltage.  I mislaid my small stock of tantalum caps and decided to settle on Radio Snack parts.  (Made by LTE, the Budweiser of electronic parts.)  And they only have 16v tantalums.  I would have chosen 20v, at least.  (From 20v to 25v, in 1uf, 10uf, and 100uf, Mouser lists 802 products, most of them stocked, from 37 cents to nine dollars.  That's just radial, leaded, not counting axial and surface-mount.)

But 16v will do.

I wired the parts together, along with a couple of LEDs (one on each side of the box--did I mention that I drilled the box, cut a circuit board, epoxied the wired-up LEDs in place ...) that will be daylight-visible on about 4/3 of a milliamp.  (At night, the human eye gets a speckle fringe around them.)  I had the output-side connector made up on a length of wire.  (Did I mention the strain-reliefs that I drilled the holes for?)  I had the input-side 'cigarette-lighter' plug with its cut-to-length wire.

I tinned the wire ends.

Then I buzzed the wires and connectors out to make sure that what was connected should be, and what wasn't shouldn't.

And I got a short circuit in the plug.  Only when I reversed the meter, I didn't.  (Tip: Many meters reverse their polarities on continuity and resistance tests so that you don't get stray currents through semiconductor junctions.  It doesn't work as well with the wider range of semiconductor devices available today, and can fail abysmally around junction FETs, whose gate electrodes turn into very delicate diodes when their polarity is reversed.)

So either something was Very Strange or the plug contained some semiconductors.  Since I got the connections I should, it seemed very unlikely the plug had a complex device like a regulator.  (If you've gotten this far, Amy, you may be thinking Differential Diagnosis.)  But a diode to dump current on reversed input?

I carefully took the plug apart.  Turns out it was held together by its spec sticker!  Well, there is indeed a diode (hidden inside opaque heat-shrink tubing), which will blow the fuse in the plug if the reverse voltage lasts long or is very high.  Which makes the polarity protection that I put on superfluous, though better behaved.

I decided to leave the diode--maybe a poor idea--and put a legend on the plug with a paint marker.  Now that it's dried (and the epoxy smell has finally dissipated from my apartment) I'll clamp the plug shell halves together with some transparent heat-shrink, and finish assembling the box.

Then I have to test it.  Ideally, I would use a very wide range voltage supply and a variable-current load.  I have the variable supplies, but one of these days I'll have to build the load device.  (The main problem is handling heat dissipation.  Mosfets would be ideal for the semiconductor part, but they are prone to thermal runaway.  Ideally you arrange things so most of the power is dissipated in resistors rather than in the semiconductors.  More things to think about!)

AND I have to get another little demo device fixed (a battery short blew out at least one part, and I plan to replace them all).  And get laundry done tomorrow.  And adjust my sleep schedule for an early-morning drive to Raleigh.

And I have about eight things to review, courtesy of Oliver Suddin. 

Now pet the shaggy dog and get back to work!

Can it spell?

To my reviewees--I owe you right now, but I need sleep before my trip.  I'll try to get a few done while I'm away.

Okay, I see I have to get more new stuff out for you!

Just to tell you what's up here: I had my trip to visit my brother and his wife.  I was able to overlap my father's visit, so we got an evening together.  I was able to do several reviews and do a little work on my notes.

My notes have been taking a lot of my writing time.  I have dozens of small notepad notes not trasferred to rolodex-cut cards, and probably over 500 of those cards to file and organize.  Some things, like plot bits, will more or less organize themselves.  Others will be harder.

And I've not been getting that much writing time.  That little electronics project--the regulated 12V charger--has grown tentacles.  The plug is too short to fit the socket extender where it needs to go, so I need a different plug.  The LDO regulator doesn't shut off well enough on low voltage, so I need to put that in, and I decided that for safety I want the cutoff devices to break the hot side rather than the ground.

The interrupting devices are power mosfets.  Unfortunately, the positive-side ones (P-channel) that I have in stock have a very limited gate voltage--8 volts.  (NDP6020).  The max drain-source voltage is a bit low, too, so I may end up finding new ones for future work.  For the present, I can use high-resistance voltage dividers on the input, wasting a few (tens of) microamps for them.

But I need undervoltage detection.  I let myself tinker on paper with a couple of circuit topologies derived from the Schmidt trigger, but I put them aside.  I'll play with them another time.  (Among other things, they require two transistors and seven or eight resistors.  That would be a VERY cramped board.)  So I turned to a pair of similar ICs, one by ON Semi (previously Motorola) and one by Seiko.  The Seiko device operates on a tenth or less the current of the ON Semi device, but the ON Semi device has more complete documentation.

Either way I need a voltage divider, which will waste some current.  The Seiko device would waste less, but I can use the ON Semi documentation to reduce that waste, including an approximation of the device's internal behavior in the circuit theory model of the divider.  That, unfortunately, yields a messy quadratic, but I found, after a few hours of sleep, that I can simplify it greatly.  It's still a pain, but it should only cost me an hour or two to get all the terms into their proper places.

These monitor circuits have the wrong polarity for the need (ground the output when the voltage is too low) so I'll need another transistor to invert the output.  An N-channel low-power mosfet will do.  Ironically, the one I stock is the TN0620.  The output will require another voltage divider to the NDP6020's gate.

Sigh.  Such a simple little thing.  The circuit board will be VERY busy, but it will be a very solid design--up to 20 volts.  I'll also replace the input side tantalum capacitor with a 25 volt job.  (I ordered a variety when I ordered the voltage monitors.)

That stuff will be in in a day or so.  Right now I'll finish the algebra and select the resistor values, and finish up the modified perfboard layout.

Later today I have to take my car to the dealer.  The check-engine light is on.  I fear that a gas station attendent has reuined another charcoal cannister.  If that, I'd let it go except that it will mask other trouble indications.  And, of course, it might be something else.

I have to find out what Best Buy/Geek Squad will charge to swap in a new hard disk on my flaptop, transferring all the content to the new one--if I buy the new disk from them, and if I don't.  (I'd like to go to a 1TB drive, but I'll only buy drives with 5-year manufacturer's warranties, which mostly means WD Black.  BB only offers a 750GB drive, but Newegg has a 1TB WD Black at a very decent price.)

And I will reduce the backlog of notes.  (I produced about seven more last night.)

And that is why Paddy has been away from work.

I've been swamped at work. That is my only excuse. I can't wait until winter. Or for help to arrive. Or for one of my old compatriots to come back into the fold...