Where to host photos for display here (Page 1 of 2)

My first attempt to use photobucket has been ... disappointing.  Tried with two browsers, got the same misfunction from both.  Web pages load and take you to the bottom of the page.  The upload page doesn't do uploads.  Drag-and-drop causes the browser to load the image for a page rather than having it captured by programming in the page.  Select mode is unresponsive.

Disappointed.  May try again, since I've seen another TNBW user successful.

Okay, it works.  Photobucket can't deal with more than one tab or window open at once.   AAAArrrgghhh!  Why in the name of the Great Pumpkin can't people work within the web page programming model instead of running widdershins with the right foot and deasil with the left--all while walking backwards on their hands?  It makes it -very- hard to meet them face to face.

Three multivibrators.  I described some of it in the Sorcerer's Progress thread.

First is an astable, using a diode trick to be sure it doesn't lock up on startup.  (Here and in other places I use Schottky barrier diodes because the larger voltage drop of standard PN diodes would kill the circuit's operation.)  It's asymmetric, periods of about 300 msec and 1300 msec.  (Flash-flash, flash-flash, flash-flash.)  Characteristically, turn-on is much sharper than turn-off, so turn-on gets through the coupling caps into the second multivibrator, a monostable.  Each turn-off pulse from the first mv results in a pulse of about 7.3 msec from the second mv.  That pulse removes the pull-down that holds the third mv off.  All three are on a 3v power rail, but the unregulated voltage, from four batteries, will range from about 6.4 volts down to about 3.6 volts.

The third mv is an astable producing narrow pulses, whose width varies with the unregulated voltage (used for one side of the timing).    At 6.4 volts, the pulses are about 65 or 70 u-sec wide, at 3.7 they are about 135 or 140 u-sec wide.  They drive a mosfet that gates current from the unregulated line (with about 2000 low-esr microfarads to deal with resistance in the batteries and the mosfet that turns the whole thing on) through an Osram Platinum Dragon LED rated at 700 mA.  In fact, the pulses can break 5 Amps at the higher voltages.  (The mosfet drive has a capacitor sized to ensure that if for some reason the circuit stays on it won't fry the LED.)

The pulse spacing is about 630 to 640 u-sec.  It's hard to keep the wide side constant because the recharge of the timing circuit has to be accomplished rapidly and when the recharge period varies by 2x it's hard to get to about the same point on the curve.  That same recharge also slows the rise time of the output voltage when the transistors turn off.  The rise time problem is dealt with using the diodes (1n914) and recharge resistors separate from the collector resistors.  Even this isn't enough unless you make one of the recharge resistors very small--small enough that with the transistor on that side on it would draw more charge in the off period than the LED draws in the on period, so there's another recharge resister, quite small, parallel with the first one--only it's gated by a mosfet switched from the other side.  (All the mosfets are 'logic level'.)

This arrangement, without the specified heat sink on the Platinum Dragon, will flash for over 48 hours on four very cheap carbon-zinc AAA cells.  On the four alkaline C cells, I plan for it, I'm guessing it's good for well over 800 hours of flashing.  But it's only supposed to be allowed to flash for a few minutes at a time.  If the rest of the box works as planned, those four C cells should last about 6 years.  (My standby draw budget is 135 u-A.  I don't think I'll quite make it, but 160 u-A is quite reachable.)

Not sure if you've mentioned this already, if you had, I missed it.  But where do you intend to use this?

I think I know those Laws well enough.  Which one do you accuse me of violating?

Go read about a circuit design called the Joule thief.  Note that it works with short, bright flashes.  With the battery depleted it can run at over a megacycle.

Edit: Better yet, go build one!  For greater efficiency put a capacitor across the resistor (Circuit B, here).

Edit: Oh, I forgot to mention the Schottky flywheel diode and the tantalum snubber capacitor around the LED.

But explosions are much more dramatic! 

njc, in medieval times you would've been a sorcerer for defying thermodynamic laws (all of them) like that! 

Any brand of battery can leak, unfortunately.  There are some things that might make it harder, but I don't know that any manufacturer does them.  A deep labyrinth at the seal with the can would make it harder for the reaction to consume its way past the seal.  I'm sure someone has thought of it; it's got to be harder to manufacture.

Oh, and it's mostly because I don't like worrying when the busy signal goes on for hours.

And where do you go accusing me of thermodynamic crimes?

It looks like a cross-stitch pattern.  I remember looking at a CT of the face once (sagittal and coronal views) and seeing the shroud of Turin.  When I pointed it out to the radiologist, he admitted I was right, but had long stopped seeing anything but facial bones. 

A

It looks like a headache!    And just as complicated as a CT of anything would seem to me.  That's why I'm not a doctor!

Easier for you, maybe.  Have I mentioned that I'm a patterner?  Reading CT scans is a patterning skill, so I've gotten REALLY good at it. I outread my attendings.  I call radiologists and say, "Um, did you guys see this little ol' thing and doesn't it look really bad?"  My shining moment was when I called an outside group and got them to rewrite their interpretation. 

On the other foot, medicine is full of logical habit-formers.  Patterners don't thrive in medicine.  They don't learn fast enough for the logical thinkers preference, so they get chased away.  I had a lot of years where I wasn't one of the anchors in my department.  Now, I'm the one who teaches everybody rather than making them do the 'see one, do one, teach one," philosophy that used to be so common around me.

(Snicker) One of my co-workers asked what book I learned from, so he could study it.  I told him to look at a thousand scans and that I used the same part of my brain to read them as I used to sort socks.  He was so disgusted... :-)

Anyhow, here's the flasher operating:

The narrow pulses dropping are the LED being turned on; that's the voltage between the mosfet and the big LED.
I'm driving a testbed (not the battery-powered test article) on only about 3v, so the on pulses are quite wide.

The other trace is one side of the sequence timing multivibrator.

Let's look at it more slowly:

Here we see one side of the sequence mv, with the two sets of pulses.  Because this is a digital scope, some of the pulses disappear in the sampling process.  Note the slow rise and sharp fall.  The other, much longer, side of that mv has an equally sharp fall.

Here's the battery-powered test article, running right now on the bench supplies.

I was able to catch it the LED flash by using a 1/10 second shutter time, but the camera can't show the real effect of the flash.  (I'm setting this test article up to work out the revised line voltage detector, visible just in front of the big IC.  The two green LEDs, with their resistors and mosfet switches, will be used to read the line state off the detector.

Here's the whole 'lab' in its present form:

The works you see on the test article, along with some other stuff, will go on a circuit board about 3+1/2" x 1+7/8".  It will go at one end of a box 4+1/2" x 7+1/2" x 1+3/8".  Most of the box will contain the battery box and the input filter, which will also provide some protection from a high voltage transient on the line.  It will have loops on the back for hanging from a wall.

But somewhere I have the diagram for a 'mobile' of parts that is actually a functioning AM radio.  Diode detector, not superhet, and I think I'd have to make up my own loopstick for the rf tuning tank.  And today's transistors are so good they'd probably overdrive the circuit--drive it into clipping.  But that can be fixed.