South Bend Junior Lathe Lives - Almost

Well, here we go.  It's been a while since I've posted, and I just HAD to post, since I've finally made some progress on my Jan 1930 SB Junior 9".  I bought it back in November, and it's the first foray into machining for me (after I rebuilt a 1939-1941 Craftsman Dunlap wood lathe, it was the next logical step).  I've refinished everything, cleaned up all of the surfaces, repaired the back gears using a hodgepodge of parts (try finding a decent back gear for a Junior 22YB - everything was either too big or not big enough) from a 13" and reusing the eccentric shaft.  But, it went back together.  I re-shimmed the headstock to make sure it was tight, and built a jackshaft to go between the treadmill motor from a C2000 NordicTrac, driven by a BC141.

The first time I tried to fire it up - it wouldn't turn.  I greased the zirks on the shaft pillow blocks, and tried it again.  I adjusted, thinking the belt was too tight.  Both of those helped, and the speed was nearly controllable, except when it engaged the headstock.  It barely turns the headstock.  I think I got the shims wrong on the headstock, and it's too tight.  We'll re-shim it again, and give it one more go in the near future, but I feel really close!

South Bend Lathe Motor - Finally Progress

So, I've had the South Bend Junior 9" lathe since late November.  As I've worked on getting an old lathe up and running, I've had some great experiences, from the "Eureka!" moments with calculating gear ratios for thread pitches, to finding some pretty great parts along the way.  However, the thing has yet to be put back into use.  One of the big hold-ups has been the lack of a good motor.

Don't get me wrong, I've had a good motor.  Just no way to control it.  I've had a Quadra Performance Drive 2.6 HP Peak (1.5 HP Continuous Duty) M-204432 (C3354B3352 or E62788) motor from a NordicTrack C2000 treadmill.  I also had the MC-2100 controller.  However, my pinouts must have been different from the others out there, because I could not get it to work based off of a potentiometer for speed control.

I finally gave up on all of that, and found a BC141 controller on eBay.  I ordered, it arrived today, and I've already connected everything up, and fired up the motor.  It works!  It actually works great!

My next step is to put a hinge on the jack shaft, fabricate a mount for the motor so that it can then attach to the jack shaft, and then install it.

Motion Detection with Multiple RTSP Cameras

Well, here we go again.  One of our cameras went on the fritz, so I started looking.  There were two decent cameras that had the same resolution and the same outdoor ratings (e.g. windproof, water resistance, and dust resistance), so I picked them up.  Unfortunately, they do not expose an jpeg stream, so I had to use a different Linux motion tool, available from :

Now for the REAL down-side.  I compiled it, and loaded it up with two camera streams.  It worked great, for a day or two, and then the process would vaporize. I found an obscure message in the log file that consisted of :

May  7 00:36:00 ch abrt[29014]: Saved core dump of pid 26188 (/usr/local/bin/motion) to /var/spool/abrt/ccpp-2016-05-07-00:35:55-26188 (548212736 bytes)

So, I loaded the application up, then attached to it via the debugger (gdb motion 123456) and got the following stack trace after a few days :

[New Thread 0x7fffecdfa700 (LWP 23598)]
[New Thread 0x7fffd3fff700 (LWP 23599)]
[New Thread 0x7fffd21fc700 (LWP 23600)]
Detaching after fork from child process 23603.
Detaching after fork from child process 23604.

Program received signal SIGABRT, Aborted.
[Switching to Thread 0x7ffff72a8700 (LWP 23594)]
0x00000030b5032625 in raise () from /lib64/libc.so.6
(gdb) bt
#0  0x00000030b5032625 in raise () from /lib64/libc.so.6
#1  0x00000030b5033e05 in abort () from /lib64/libc.so.6
#2  0x00000030b5070537 in __libc_message () from /lib64/libc.so.6
#3  0x00000030b5075e66 in malloc_printerr () from /lib64/libc.so.6
#4  0x00000030b50789b3 in _int_free () from /lib64/libc.so.6
#5  0x0000000000416b20 in netcam_cleanup (netcam=0x7fffa0000030, 
    init_retry_flag=0) at netcam.c:2581
#6  0x0000000000412d28 in vid_close (cnt=0x678a10) at video_common.c:555
#7  0x0000000000406f19 in motion_cleanup (cnt=0x678a10) at motion.c:991
#8  0x00000000004097b8 in motion_loop (arg=)
    at motion.c:2231
#9  0x00000030b58079d1 in start_thread () from /lib64/libpthread.so.0
#10 0x00000030b50e88fd in clone () from /lib64/libc.so.6
(gdb)

So, I then checked in the log, found the above stack trace was not related, and that that run may have had other issues going on.  My real stack trace is :

There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
and "show warranty" for details.
This GDB was configured as "x86_64-redhat-linux-gnu".
For bug reporting instructions, please see:
...
Reading symbols from /usr/local/bin/motion...done.

warning: core file may not match specified executable file.
[New Thread 4667]
Reading symbols from /lib64/ld-linux-x86-64.so.2...(no debugging symbols found)...done.
Loaded symbols for /lib64/ld-linux-x86-64.so.2
Core was generated by `/usr/bin/keyutil --command genreq --ascii --renew /etc/pki/tls/certs/cottonwood'.
Program terminated with signal 11, Segmentation fault.
#0  0x00000030b5047e2c in ?? ()
Missing separate debuginfos, use: debuginfo-install glibc-2.12-1.149.el6_6.5.x86_64
(gdb) bt
#0  0x00000030b5047e2c in ?? ()
#1  0x0000000000000023 in ?? ()
#2  0x00007fff1c7768c0 in ?? ()
#3  0x0000000000000000 in ?? ()
(gdb)

Bah!  Next, in the logs, I found :

/var/log/messages:May 22 20:29:00 ch motion: [1] [ERR] [EVT] event_image_snapshot: Could not create symbolic link [01-20160522202900-snapshot.jpg]: No such file or directory
/var/log/messages:May 22 20:29:30 ch motion: [1] [ERR] [ALL] create_path: Problem creating directory /home/media/security: Input/output error
/var/log/messages:May 22 20:29:30 ch motion: [1] [ERR] [ALL] put_picture: Can't write picture to file /home/media/security/cameras/camera-backyard//01-20160522202930-snapshot.jpg: Input/output error

I completely missed the "Input/output error" and checked the filesystem location, only to find my entire disk had failed, so it could not write videos.  Sometimes, debugging can be such a pain!  Here comes a nice recovery fee and a nice fat check going out for a NAS with mirroring!

Who is Your Friend?

Often, I'm amazed at how many people know my wife.  Considering how amazing she is, it's not baffling, just awe-inspiring regarding how many lives she's effected.  She is an amazing wife and friend.

Anyway, going back through some paperwork, I ran across the following :

It made me laugh, that memory.  I'd better tell you the story.  We were visiting family in the area, and were out traveling one of the back roads.  As we're cruising along at 60 MPH (the speed limit was 55 MPH), she nonchalantly mutters something about her old "bucket of bolts" hitting 79-80 when she was growing up.

I kept it at 60 for that phase of the drive, but on the way back, I thought, "Why not?"

We hit some speeds that would have ended up with an immediate impound and a revoked license.  I slowed down to 60 MPH, but it suddenly felt relatively slow.  I found about 75 MPH was slow, but not unbearable.  My wife (still relaxed) said, "that's a cop up there - just pull over because it's too late".  (He was driving in the opposite direction, and a mile up the road, he pulled over and turned on his lights).

We passed him, and pulled to the side to wait for him to catch up.

Turns out my wife knew him.  He gave us a warning.  It still makes me laugh.  She knows EVERYBODY!

More Math - a Lathe Countershaft

So, I already found math is very useful for calculating new change gears.  One question that has been bothering me is.... how fast does the treadmill motor turn?  It's from a NordicTrack C2000, and the motor is a Quadra M-204432.  It simply says "2.60 Peak Watts" and "1.5 Continuous Watts".  No specs on RPM.  I finally decided to determine how fast the little thing was turning, so I ran out and grabbed a few measurements.

The treadmill had a max peak of 10 Miles/Hour.  That equates to 10x5280 Feet/Hour, or 52800 Feet/Hour.  For RPM, we need a Per Minute, so we have to divide that by 60 (because everyone knows there are sixty minutes in an hour).  That gives us a distance traveled of 880 feet per minute at 10 miles per hour.  My units of measurement are in inches, so I have to multiply that by 12.  My final Max speed is 10,560 inches per minute.

I grabbed the diameter of the bar that the tread was turned by (a rod that sits on two bearings).  It came out to 6 1/8", or 6.125".  To get the maximum speed, the outside has to travel around 1724.081633 times per minute.  Incidentally, 1725 RPM is a standard motor speed.  But, we're not even to the motor yet - we're still on the treadmills rolling pin speed.  Now, the pulley ratios come in to play.

Calculating a pulley ratio is fairly simple.  Divide the diameter of the driving (or motor) pulley by the corresponding pulley on the other side.  In my case, the motor pulley is 1.125" across, and the rolling pin pulley is 3.5" across, or a ratio of 0.285714286.  Now, one is normally converting from motor speed to spindle speed for things like lathes (which is "multiply the motor speed by the ratio"), but we are going backwards.  We have to divide the speed above, so the final ratio comes out to....

... 6,037.5 RPM?  That doesn't make much sense.

But, it IS close to 6,000 RPM, and there are definitely treadmill motors out there that turn that fast.  However, I do not really want the lathe turning at 6,000 RPM, so I need to reduce the speed.  This countershaft will do two things for me - it will help me reduce the speed of the spindle (the torque will go up at the spindle), and it will let me convert from a serpentine belt to a v-belt, which is what I have on the lathe.

1725 is a good "don't kill yourself speed", but if using things like 3 jaw chucks, I'd probably say closer to 1200 RPM is a max.  I need a pulley ratio of 0.25 or lower to feel comfortable, but with the controller for speed, 0.25 is about right.  That would give me a maximum of 1500 RPM at the spindle - if I can hit the pulley ratio.  The right pulley size without a countershaft would be a 4.5" pulley.  My lathe spindle pulleys are 5" - so close enough, as it would slow me down a bit more - but I have to convert from serpentine to v-belt.

My problem is that I could only find a serpentine belt pulley that would match my motor pulley that was 2.2" across.  So, if I get a v-belt pulley that is close to the 2.2" across, that would eliminate the countershaft as the speed changer, and we'd only have to calculate motor-to-spindle - but they don't make them.  They do make 2.5" v-belt pulleys.  So, since I thought it would be close enough, I bought them, the 2.2" serpentine pulley, some CRS round bar and square tubing, a couple of pillow block bearings, and welded it onto a frame that can be put onto a hinge and bolted to the lathe table.  Before you look closely at the picture - remember, my welding was never professional - sometimes I just stink at it.

Because I have different diameters on the countershaft, we HAVE to calculate the speed of the countershaft, and then the speed of the spindle from the countershaft.  We do the work twice.

So, a 1.125" pulley to a 2.2" pulley is a ratio of 0.511363636 .  A 2.5" pulley to a 5" pulley is a ratio of exactly 0.5.  That means our final ratio on the countershaft is 0.255681818.  That ratio takes 6,000 RPM down to .... 1,534.09090909.

And a maximum of 1,534 RPM at the spindle is perfectly acceptable.  It's only spinning at 25 times per second.

Book Scanner: Mechanically Complete

After agonizing about a broken fan on the 3D printer (yes, one successful print later, the 3D printer had a third part break), I started to wonder if I'd ever be able to print a camera mount for the book "scanner".

Yesterday, I had an idea (I would call it brilliant, but it was mine, and any ideas that come out of my head must be brilliant).  Instead of printing a plastic part... why not build it instead?  My brain went immediately to wood, as it would be the easiest to "mill" or to shape.  I was worried, though, about the wood splitting.  I needed some parts with a 60 degree surface to point the cameras at the glass platten. I kept thinking "sturdy", since the two parts had to hold a camera, and I did not want the cameras falling on the glass platten and breaking the glass and/or the cameras.  (Yes, wood was not my brilliant idea.)

Then something else hit my mind.  When I worked on the old Troybilt rototiller, I needed some steel for the recoil starter's dogs that engaged the pulley.  I had picked a piece of steel bar up, about 1/8" thick.  I thought about just welding three bars together and bolting it up to the aluminum extrusion meant to hold the mounts in place.

I grabbed the torch, and started heating the bar up.  Using a pair of ducks (commonly called "tongue and groove" pliers), I bent for the first angle.  The way it turned out was even better than I could comprehend.  I simply bent the first angle to 60 degrees, and cut off a length of it - I realized I didn't need the aluminum extrusion to support the cameras - the steel was sufficient.  I immediately built a matching one for the other camera, and drilled the holes.  One hole (the one on the 60 degree surface) needed to be tapped, so I grabbed my 1/4" taps and added some threads.  I put some threaded rod in place with a nut to lock it to the surface, and then the camera could just spin onto the mount.  Hooray!

Now to figure out the CHDK firmware hacks for the Canon cameras!

There are two glass plates, called "platten" (the German word for "plates" because scanners and copies have always called those parts "plattens") both set at 30 degrees from the main surface.  There are wooden supports underneath at the same angle for the plates to press against, so that when the plates are lowered against the supports, whatever is between them will be "flat".  The glass plates are mounted to two vertical drawer slides (you can see them in the above photo), with a cable (I used a vinyl-coated wire and pulleys) to lift both ends of the plates at the same time.  When the plates are down, you trigger a camera shot, then raise the plates, turn a page, and lower the plates.  Wash, rinse, and repeat.

The cameras are mounted onto 1/8" steel bar bent to 60 degrees.  This is because 60 degrees is perpendicular to 30 degrees (required for a straight-on shot of the page).  There are two cameras, one for each page.  Each camera will be loaded with CHDK (Canon Hacker Development Kit), a "firmware" designed to allow external control of the camera, or do things the native Canon firmware cannot do (e.g. saving the pictures as "RAW" images).  I simply need the CHDK firmware to allow me to remotely trigger a picture over a USB cable, and pull the picture down to the computer without filling up the camera card or with me trying to use the SD cards.  You CAN do this without the CHDK, but I'm just lazy.

The cable had to be attached to both drawer slides - I found that a single one had the thing going skewampus, and I didn't like that.  Getting the cables the same length was a little challenge, though nothing unsurmountable.

Here is what the view from the camera looks like (though, not the camera, just right next to the camera, since I am definitely too lazy to use a picture from the camera).

So, I just need to figure out the CHDK components (which I am definitely close to doing), and pick up a Raspberry Pi to act as the computer.  The Pi will be configured with a button so that all you have to do is raise the platten, turn a page, lower the platten, and press a button.  When done, the Pi should arrange all of the images and convert them into a PDF that can be loaded onto any tablet computer for your reading pleasure!

3D Printer Project - Complete.... Sort Of

So, this week, I finished up the 3D printer.  It's a FolgerTech Prusa i3 - not a bad little printer to put together.  There were a few issues along the way :

1. The plastic parts broke when tightening - either it was weak, or I don't know my own strength
2. One motor refused to run (the X axis), forcing me to procure a new one, which had to come from FolgerTech because I couldn't find one of the same dimensions anywhere else
3. Bed leveling is nigh impossible, with the shorter springs and screws for attaching it to the acrylic frame

Aside from that, it went together fairly simply, and ran (at least once) :

I used the standard PLA test filament that came with the printer, and realized in the middle of the print that it was going to run out before it finished.  So, I grabbed the spool of ABS, and followed it through.  Not realizing I needed to increase the temperature, the joint between the two and also the ABS right after the PLA is weak, but that's my fault.  It still turned out :

So, I tried another print of the same part, using just the ABS.  I did not realize that Repetier-host (the software I am using on the Mac) will freeze hard if the screen saver kicks in.  It aborted in the middle, and the nozzle cooled while sitting right on the print, causing the plastic to curl around the nozzle and burn against it.

That part had progressed far enough along that it was still usable.  However, since that time, I failed for a day or two to get a good print.  After cleaning the nozzle again, I got one good print (a camera mount for the book scanner) :

Unfortunately, even with cleaning the nozzle, I have not been able to get a good print since then.  Very frustrating.  On top of that, one of the fan blades snapped, so the printer is now out of commission.  If I had done my math on that camera mount, it would have been great - but I needed it tilted at 60 degrees, not 30 degrees like I had printed.  Oh well!  You live and you learn!