On Monday 24 September 2018 09:55:45 Les Newell wrote:
> Hi Gene,
>
> I'd take the lens off of the camera so the beam directly hits the
> sensor. The laser beam is co-linear so you don't need to focus it and
> a lens just introduces distortion. Everything else will just be an
> even colour background so you should be able to fairly easily
> distinguish between the background and the beam. Maybe put a black
> tube in front of the camera to exclude extraneous light.
> I'd also suggest dropping the supply current to the laser diode. It's
> a while since I last played with lasers but I seem to remember getting
> a cleaner dot by reducing the power.
>
> Les
I have given that some thought, consisting of extending the rear of this
thing about 15" and making a set of slip rings on the end so I can feed
a current limited 5 volts into it from the back of the spindle. It takes
2 ea 394 batteries which I suspect would not last the 2 hours or so of a
recording session, and cost $7 a pop.
Asking the missus about the real3d polaroid glasses got shot down, so
I'll have the troll thru fleabays poor search engine to find something
usable. Finding a variable cell ready made would be ideal. OTOH turning
down the current might suffice.
Right now I'm baking rattle-gun rustoleum self-etching primer onto the
mt5-5c adapter, and carving the paint back off with a mitsubishi chip
which is much sharper than the banggood version. Hopefully I can get it
hard enough it won't stick into the spindles mt5. The idea is to find
the magic taper that matches.
When I first made it, it fit well, but 2 months later I re-assembled it
to find a wobble of about 1/16" 2" out of a collet. That was good enough
to cut the brass rod for the "tap hats", but not much else. And the
spindle is still true +- .0003, which is essentially the internal
surface roughness. Plugged in the mt5 to 5c, and get 7 thou runout at
the mouth of it, and its riding with most of that at the mouth because
the small end is in-adequately tapered. So I'm painting it and diddling
the taper about half a thou per inch at a time. And I'll continue until
I get a good fit. Tempted to run a mig bead all around at both ends of
the taper but don't want to run the mig current thru the bronze spindle
bearings. Plus its thin enough to muck up the 5c fit from the heat. One
of those damned if you do, and damned if you don't things, so I'm
punting with paint, baking it with a hair drier. ;-)
> On 23/09/2018 20:13, Gene Heskett wrote:
> > Greetings all;
> >
> > I seem to have carved, polished whatever, that mt5 to 5c adaptor
> > pretty much in error, despite my last attempt to reduce the small
> > end of it to fit the spindle taper, its still at least 10 thou small
> > at the shoulder, and I am running out of shoulder as I keep pushing
> > it to try making the big end of the taper actually fit the spindle.
> >
> > But before I make a last pass to true that up, I am going to put
> > this 38 spcl laser in a TTS holder with a 3/8" collet, in the 3 jaw,
> > and see how much wobble it has as thats fairly well corrected for
> > the spindle nose bend now.
And its beam wobbles about 1/2" at 30" out, intolerable. I've not tried
the 4 jaw yet.
> > Looking at cameras, thinking of putting the camera and a rock64, and
> > possibly a small tft screen all in a nice compact qctp tool holder,
> > I'm still trying to figure out how to arrive at an image that:
> >
> > A: is ND filtered enough so as to not overload the camera, so
> > do I put the camera module inline with the beam but with an
> > nd-10 in front of it, in which case the camera has nothing to focus
> > on or put an nd-10 as a 90 degree mirror, using a 45 degree angle to
> > point the reflection at the camera looking across the bed, or
> > put a frosted glass so the camera has something to focus on, and an
> > nd-10 to keep from overloading the camera, mounted inline.
> >
> > Then we have the problem of averaging the image over several turns
> > of the spindle, and finding the centroid of the pattern that
> > results. Do we have some (probably python) code to do that?
> >
> > I'll take this latter question to the python list too.
> >
And got a link to some simple python that does that centroid thing. But
I'd like to future-proof it by making it run on python 3.5, however it
appears that numpy and cv2 are both python-2 yet. So we'll see in due
time.
> > But camera and filter mounting, I have a piece of nd-10 for a
> > dimming filter, and if dim enough, it might be possible to use the
> > camera as the integrator by taking a time exposure, that should work
> > well if the spindle turns an even integer of turns while the
> > "shutter" is open.
> >
I have a handy index signal from the spindle encoder that might serve as
the "shutter". If the camera has that sort of control ability... One
might even switch the laser on and off in synch with the index. Either
way, there's the quantizing error of the 1
