Medkit Preview

Still waiting on the final selector knob, and the Mysterious Part that does all the healing and stuff (Shapeways order sent in just before the holidays).

But outside of those, and the associated wiring (because the Mysterious Part just had to have some LEDs put inside it!) the Medkit is done.

I'll take some nice pictures soon.  Until then, these will have to do;

 

I didn't take a lot of work-in progress shots either, sorry.  Anyhow...boring text stuff (but not very much of that either) below the fold:

Answering Questions

Sometimes I look at my blog sites and I see interesting search terms that led someone to here.  Sometimes I wish they'd stay and write me a note; because it is something I actually know the answer to.

Here's a random selection from this past week:


qlab how many tracks:  As many tracks as your computer can handle.  Actually, I've had more trouble over running several long tracks (ten minutes each) at the same time, as at some point QLab runs out of RAM and has to stream more data from the hard disk.  A friend of mine has had some breakup when running 10+ stereo sound files simultaneously on an old Mac Powerbook.

nanokey broken:  The thing that keeps happening to mine is the keys getting pulled off.  They can be put back, if you are nimble (and don't loose the little contacter boot/cup thing).  But it is enough trouble so lately I've just done without a few keys.

shield arduino ULN2803:  I've not seen this specifically.  I made one myself, of course.  Plus I made a one-board minimal Arduino + ULN2803 using one of Adafruit's darling perma-proto boards.  There are quite a few driver shields out there, tho.  Enough so you can almost certainly get the equivalent performance.  If not better (as in, many driver shields are arranged around latching registers so they give you more controlled outputs on fewer I/O pins).

oaklahoma little wonder:  I built this prop long ago.  The description in the play is straight-forward; it is a frontier Viewmaster with a hidden switchblade.  Basically, a portable version of the peepshow machines popular at the time; you pointed it towards a light and looked into one end, and saw either a slide, or through various cunning mirrors and similar, more than one slide, of a pretty girl in various stages of undress.  In the specific device used in the play, if you know which catch to press, a knife comes out -- according to the script, the knife pops out the bottom and you then grab the tube from the rube and stab him with it.  The version I made was slightly more efficient; a spring-loaded rubber dagger came out the viewing end!

attiny button led pwm:  Um, sure.  All of the above.  Trick to note; the tinys have only one or two internal clocks that can be used for hardware PWM.  These don't line up well with the expectations of the Arduino software so you are better off writing in C -- and be prepared to read up on register flags!  You can also do software PWM on any pin; I reprogrammed a Blink-M this way, and it fades okay without too bad a flicker even though I didn't use very efficient code.

Atompunk USB:  now THAT is a nice idea!  I've got a couple of flash drives sitting around -- maybe my next prop should be a nice atompunk mod for one of them.  Dare I find a really small gear motor and put a fan on it?

Bah

The magnets don't work.  Next project, I'll plan how I'm going to fasten down the lid before I finish the sculpts!  So hot glue is going to have to do.

The dress nuts I ordered don't fit my buttons.  Which didn't look right anyhow.  So I filed down the shafts of the buttons until the new nuts would fit over them, pried off the buttons, clipped down the stems and painted the stubs black.  Then epoxied them into the holes (as they are now way too big.)  The decorative nuts were superglued over that.  Inevitably, at the last part of this last stage one of the buttons stuck, and in my frustrated efforts to free it I dinged up the new paint job.

Was also working on a very slight weathering.  Mixed up an acrylic wash, but it pooled and looked spotty and didn't seem to want to stay in the panel lines and around the letters.  Also mixed up a lighter color and dry-brushed some edges.  And it was all very messy and annoying and I have a headache now and at the end of it came back with a fresh coat of base color to knock most of that down because it didn't look very good.  Now I'm putting down coats of Krylon Crystal-Clear to seal it and make a sturdy finish.  Which is gloss, not the matt I'd prefer, but I'm not running out and buying yet one more can of paint at this point!

At least the water slide decals seem to be working now.  Given my experiences with dinted paint, the fragility of the decal, and the fact that there are buttons glued into their holes instead of threaded with proper nuts, the boxes are going to be more fragile than I'd like.  Not exactly the kind of thing you can toss around.  So one more thing to learn how to do better.

The desk is still covered in electronics scrap.  I just need to throw some minimal effects into the Medkit and I can clean up the desk/clear the decks for new projects.  Like the microphone repairs I'm sure to need before the next show opens.  Or the robot I may be building with my niece!




Next entry should be pictures of the finished props.



Oh:  just for laughs, here's an approximate list of the different "threats" the randomizer will pull up when the CBR Kit is set for "Simulated Threat" behavior. 

SARIN, CHLOLERA, TULARMIA, TYPHUS, SMALLPOX, NOVICHOK, BUFOTOXIN, RICIN, EBOLA, BOLULIN, MARBURG, ARSINE, PHOSGENE, CHLORINE, BROMINE, CYANIDE, ANTHRAX, SPECTROX, C DEIMOS, IOCANE

 The list is not meant to be complete or even representative; it is mostly a list of reasonable size of simulated threats that look okay on the display.  Seven-segment, 8-digit display, remember.  So "X, M, N, V, W, T, K" are all very hard to work out, "R, U, N, B, Q" are all lower-case, and "I, O, S" all look like numbers.  So "SARIN" displays more like "5Ar1n" and "NOVICHOK" looks something like "n0u1CH0H."

As the client and I have discussed, if war holds off for another decade the Morrow Project upgrades to LCDs or character displays.  And if they made it to within the current decade, it would be full-color OLED screens!



To add argh! to argh! as soon as I soldered the speaker down, clipped on a brand-new fresh 9V battery, and hot-glued the top of the case down, a software error showed up.  I still don't know exactly what is going wrong but after several hours of work I traced it down to one program line and commented it out.  So, now, when the simulated radiological threat is in the penultimate stage -- just before the Rad Alert triggers -- only the last two digits of the displayed number continue to change randomly.  Instead of the last three digits, as I had originally programmed.

I have suspicions as to what is going on, but it looks very much like a problem of interaction.  A problem where the engineering gets real instead of done cargo-cult style, that is.  I'm actually surprised I didn't run into trouble before, what with PWM'ing and playing tones at the same time, plus what are probably jolts of back EMF coming off the speaker.  It looks very much as if at some point, a buffer over-runs.  Well...it seems to work now!

Another complete oddity.  While the problem persisted, over the entire simulated radiological threat display sequence, the VFD would not display all the segments of an "R."  Every other letter or number was normal.  Not "R."  Which sounds even more like somewhere, a buffer is running over and something is leaking into program space where it isn't supposed to be.

Well, the Medkit is simpler.  I'm actually looking at three different functions (one power switch and two "easter eggs") but there is no micro.  Everything is hard-wired.  In fact, one of the easter eggs won't even share a power supply.

I think.  I will test if the VFD will fire up on a 3v supply but... 

Sticky!

What's even more dangerously sticky than fresh superglue?  A supermagnet dipped in superglue!

I've been fastening the tops down on the Morrow Project boxes.  By attaching small neodymium magnets to the cases and bits of stainless steel to the lids.  With cyanoacrylate glue.  As I was attaching one, it flipped around on the tweezers, smearing them with fresh sticks-to-everything glue.  As I carefully wedged it free with wooden tools, it flew across the desk to grab on to the permanent magnet of the installed speaker.  In trying to remove it, I broke the tip of an X-acto knife and the broken tip struck me in the eyelid! 

NOT safe.  This is the sort of thing that should really be done with goggles at least, plus gloves, good light and ventilation, and plenty of sleep too.



Today was also the day to finish off the code.  And work with the water-slide decals.  So... my printer wouldn't make a clean job of it.  Neither of the copy shops I tried had anything that could work with inkjet paper.  My friend's printer wouldn't feed the thick paper.  Finally we got a set printed.  The first smeared into a mess despite a coat of Krylon Crystal-Clear.  The second came apart (and also showed signs of smearing) despite a coat of Blair Spray Fix.  The third I hit with three heavy coats of the Krylon and that finally worked.

Here's the artwork:


These are water-slide decals on thin sheets of polished tin (4 cm by 2 cm square). The white parts (like the letters) is exposed metal.  So far it looks pretty good.


At some point in the next few weeks, besides taking pictures of the final props, I want to write one combined post that lists the parts, gives the schematic, provides links to the 3d models, etc.  Basically, everything that I can provide electronically that would allow someone else to make one of these.

Stop Casting Porosity!

Volpin makes it look so easy.

The difference is between being able to stumble through all the steps, and being able to do them with confidence and control.  I made a better box mold and a much better slush-cast for the Medkit body, and the lid fit first try.  But there is a bit of alignment still needed and I did have to Dremel out the lid a bit to fit the electronics.  Which means I'm far below the level where I would be offering a kit to other people!  Heck -- my panel lines are still barely acceptable.

 Next project I have to mold I want to try a jacket mold.  I'm tired of going through so many bucks worth of silicone, and the mold boxes I've been doing are barely good enough to keep the molds from folding over, turning my boxes into parallelograms. 

I'm finally seeing the end of this.  Still waiting on custom-printed knobs from Shapeways, have another couple lines of code to make the capacitance-sensor triggered "injector" look right, and of course solder some minimal components into the Medkit.  Oh, and print out the data plates and so some touch-up paint.

Which sorta of segues into an attempt at a schematic for the CBR kit:

I can't vouch for the accuracy of this.  I didn't exactly draw up a schematic before I started soldering, nor did I document that well during the project.  Mostly I opened up the appropriate datasheets and made the appropriate connections on the spot.

In the upper left is a basic 7805-based regulator.  I left the power switch off the diagram.  Below it is the 6-pin programming header that works with my in-system programmer.  Left of center is a minimal Arduino, essentially; 22 kHz resonator, power and ground.  There's actually a .1 uf filter cap across the power leads I didn't bother to draw in.  And some people hold RESET down to ground with a 1 meg resistor but this didn't seem to be necessary.

The important part is the Supertex HV5812, which I found at Mouser; a 20-channel high-voltage driver designed for use with vacuum fluorescent displays.  Like the Soviet-made, old-stock ILC1-9/8 eight-digit, seven-segment display I am using.  I found those on eBay and they were shipped from the Ukraine (cheap, too!)

Above the HV5812 is a Recom DC-DC converter also picked up at Mouser.  It works, but just barely; 24 volts is low for the VFD, which really wants 30-60V.  It also wants a voltage-controlled AC filament voltage, but a small resistor works well enough for this application.  To the right of the Recom a power Darlington (the venerable TIP120) is operating as a switch to (over) drive the poor speaker with nasty square waves.  No attempt at a DAC or waveform generator of any kind here.

Two of the AVR pins are being used for a capacitance sensor using the Arduino library (many AVRs will do capacitance sensing natively as well, which is a more efficient way to go if you have lots of pins and want a fast reaction time).  The indicator lights (20 ma LEDs) are all running off PWM channels, and the two control inputs (button and rotary switch) thus are on the otherwise unused analog inputs.  The rotary switch is using a resistor ladder, thus allowing a single pin to detect a larger number of different switch states.

As circuits go, its actually pretty simple.  Or, rather, no part is complex; it is just a melange of different simple circuits.  The HV5812, for instance; high voltage is supplied to Vpp and the regulated 5v to Vdd, but the four control lines (clock, blanking, strobe and data) are just fed any four arbitrary Arduino digital I/O lines.  Wire up which ever is convenient and then assign them the right pin numbers in code.  The outputs are the same; I pretty much wired them one-to-one to whichever VFD pin was nearest, and then solved the assignments in the lookup table.

Beloved Denton

There's a thread at the Replica Props Forum about literary props.  Not props that are literature -- they call those "paper props" -- but props from works of literature that have not (yet) been turned into movies or television shows.

The one I've been thinking about on and off for a while is the pistol carried by James H. Schmitz's character "Trigger" Argee.  (Not actually my favorite character of his, but Telzey Amberdon would be more likely to carry a set of law books.  In a "reader," which is a very nice 1940's prediction of an eBook with Internet connection, but still...)

Trigger's weapon is interesting both as a high-end collector's item; a very expensive, classy but ruthlessly effective weapon suitable for someone who is a crack shot and makes a living on the frontier, and as the signature item of a strong female character from the Golden Age; as much a signature weapon for Trigger as the twin DeLameter's were for Kimball Kinnison.

That said, there are maybe ten words of unique description spread out among the various short stories and the one novel-length adventure, "Legacy."  Some sort of weapon appears in the cover art for the Baen re-issues (so far I've not seen an Argosy or other magazine cover from the time of original publication that features Trigger, and the only other artwork I've seen for "Legacy" also fails to show her weapon), but those aren't as much help as they might be.  That rather makes the design up for grabs.

For that matter, DeLameters or the Valerian Space-Axe have only barely been depicted -- and not named as far as I know -- in a Lensman anime that bore little resemblance to "Doc" Smith's epic space opera otherwise.  Or, of course, the Lens.  And it hardly matters what the latter may have been depicted as before; you walk into a science-fiction convention with a polychromatic jewel set into a wristband, you hardly need to add a plain gray leather outfit for the locals to "get it."



Anyhow, I've been thinking for a while of making some SF weapon that isn't from any particular movie or game.  And I've also wanted to go in a different direction than the brutal, heavy, highly-weathered aesthetic these days (and not in the some-what overused steam or atom-punk directions, either.)  More in something that has a boutique aesthetic; fine details and rich materials like expensive watches, cigarette lighters, etc.  So technical details, yes, but something that also looks expensive and well-cared for.  Not like a mass-manufactured gun that's been through a war.

Plus of course integrate materials, fabrication methods, and internals with the design process.  So instead of trying to cram LEDs into the shell, design the shape around the necessary batteries or circuit boards or whatever.  And leverage CAM, 3d printing, photo-etching; every time-saver that modern technology has opened up.



But it is hardly like I need a new props project!  For all the above ambition, I'm struggling to finish the seemingly much simpler Morrow Project boxes.

My first pour of the Medkit faceplate was a failure.  The pour holes were too small. 

This is my second two-part mold.  I molded the faceplate face up, first.  Then, since the original model is solid balsa, I flipped the mold and built up a thickness with clay, attempting a technique I'd read about to achieve a uniform thickness (you roll out cords, then lay them edge-to-edge to fill the space).

I was in a hurry to make the second pour and completely forgot to add mold release!  Fortunately, I'd put down a layer before starting the clay and it was enough to keep it from turning into a solid mass of silicone.

The second pour was more successful.  Bubbles mar several of the letters (to achieve raised letters on the model I attached styrene letters sold by Plastrut).  The backside of the hood needs to be shaped to fit the vacuum fluorescent display (although I'll probably be using the dead one, for simplicity, and back-lighting with a couple of LEDs.)  And all that sanding and patching and shaping seems to have altered the shape of the cut-out enough where it isn't a good match for the box anymore.  I'm not sure if I'm going to try to repair that with sheet styrene (or a lot of sanding on the box, once that is cast).

Ah, well.  A little more clean-up, and I'll be ready to mold up the box.  I still haven't decided if I'm going to try a two-piece mold and attempt thin-shell fibreglas, or just do a box mold and see if my hand is better at slush-casting now.

I've been flipping back and forth between adding details to the battery compartments on both boxes, and prepping the new models for molding.  Just yesterday I splurged on a Squadron panel scriber, which turns out to be much faster with a smoother result than scraping panel lines out with X-acto knife.  Except that the bulk of the Medkit body is the soft green plastic of an Army first-aid kit, and it doesn't take details cleanly, not at all.  Been a lot of go-around with Bondo spot putty and Rustoleum filler-primer on that one!

With luck I'll be able to pour silicone today for the last mold for this prop.  It will be a relief to clean up and put away at least some of the tools and supplies I have spread all over the apartment!



I keep thinking "this is the day" I can finally fasten the lid down on the CBR.  But as of this moment, I still have to touch up the paint job, re-glue the components that go inside the lower body, solder them to the mainboard, add a couple more lines of code to decrement "doses left" and play a buzzer/change the "injectable" status LED when all the doses are used up, and swap out the Atmega168 for a 328 because I'm down to the last 40 bytes of program memory!  Oh, yes -- and fasten down the lid, which as of this moment is probably magnets inside, and some dummy screws as decoration.  But I might epoxy some nuts and make the screws practical instead.

O.R.(W)T.F.

I had a short gig to record a chorus (and reinforce them as necessary).  At the very last moment (aka while we were loading in, a half hour before the house was to open) we changed from Clavinova to a Baby Grand, and moved the guest star 'cello quartet to the center of the horseshoe.  Oh, and wrapped the horseshoe around at the ends because the singers couldn't hear each other (even before the 'cellos fired up!)

So the resulting stage plot looked like this:

There's actually less space between singers and instrumentalists than I've indicated.  There is also the physical layout of the stage, which further limited my options for placement.

So my primary mics for trying to get singers up over accompanists were three condensers set at chest height and crammed in between singers and instrumentalists.  My primary recording device was a stereo set on the tallest mic stand I had available.

Unfortunately the nice mic bar we used to own had gone missing.  So I taped the pair of Oktava's to one of the home-made brackets for my mini monitor speakers, in an arrangement I promptly christened the "ORWTF."   (It was actually more like a narrow A-B setup, given the limitations of my improvised rig.)  An A-B stereo pair depends entirely on timing differences for localization of sound sources; as contrasted with XY which is purely level-based (plus some tonal differentiation).  The biggest problem with A-B is that it doesn't collapse well to mono; you get phase cancellation.

On the second day I added a spot mic inside the quartet; again improvising with the black electrician's tape and a table-top stand to make an XY pair sitting on the floor.  Had I the gear, I would have used either a crossed pair of figure-8's, or an MS pair.   The intention here was just to give me a spot mic with some localization to increase my available options when I go to mix down the multis.

The piano was on no-stick, with the front third folded back to expose the hammers.  If it had been alone in a room I would use a large diaphragm condenser four to six feet above the lid.  Instead I stuck a mic a mere sixteen inches from the tiny slice of exposed strings.

The only good luck I got was that on the first performance, the singers managed to end up grouped around my large condenser -- which I'd put in an omni pattern for that location.  This was pure chance and not repeated, however.  I've mentioned to the conductor that if they continue to work with instrumentalists, they either need to rethink where they use pianissimo, or start including awareness of microphones in their stage choreography.

Oh, yes.  And all this listing of brands -- two Behringer B-5's, two CAD CM217's (actually, one, as the other went bad and I taped a Karma silver bullet on to it with yet more black electrician's tape) -- sounds like I was making careful choices.  I wasn't, really.  This is what I had available.  But I will add that I've used both the Behringer's and the Oktava's before in similar roles, and the PG-81 has become my go-to mic for piano (although I recently heard a very nice result from an MXL on a grand on short-stick.)




So far the recordings sound decent.  I record in Audacity for stability and options.  The first step is hand-editing to take care of loud clapping, loud random noises and equipment noises that may be on the tracks -- Audacity is handier for this kind of point-edit.  Then I normalize and export as individual mono tracks.

Sure, you can tame some stuff with compression.  But one-of-a-kind events (like someone knocking into a mic stand) are better treated locally, not globally.  And most compressors don't kick on instantly; they leave the first transient.  In the case of a pop -- or a hand clap -- the compressor may pull down the average of the peak, but it will still leave the clipping-level peak intact.

In CueBase, I pan the tracks and chop and trim down to individual songs, and (often) re-normalize the individual slices.  Then on a per-song basis I adjust pan, EQ, level, and effects.  For some situations I'll need to draw automation curves as well (usually level, as when I have to ride a too-soft solo.)  This particular concert looks easier than most; they didn't have much in the way of solos and they didn't move into so many different positions (not like last year, when they arranged themselves around the audience for one number and entered singing on another).

I'm hoping, in fact, to basically use my AB pair, unsweetened, and bleed in just a little of the spot mics where necessary.  I listened to the four "solo" numbers by the 'cello quartet, and although it is a good concert hall sound on the ORWTF pair, the closer position of the XY pair at their feet is a much more exciting sound.  Almost certainly, my final mixdown for them will be a combination of both.

The most complex thing I'll probably have to do is some very small time offsets -- since the spot mics are hearing the sounds up to 20 milliseconds before the area mics.