Attaching an occasional use encoder to a lathe spindle.

[tonyfoale]

There are times when I need to fit a rotary encoder to my lathe for precise measuring. Accurately measuring the profile of camshafts is one example use. I originally used a 3 post mount, but as you can see after the drive belt conversion (described in a previous post on this forum) and now a large 50 tooth timing wheel that is not possible. The solution, which was inspired by a friend, was to support the encoder with the spindle plug and then it is only necessary to prevent rotation. The torque is absolutely minimal so I used a piece of 2 mm plastic sheet which is rigid in the rotation direction but flexible in the axial direction, thus preventing any misalignment from loading the bearings in the encoder.
It only takes 30 seconds to mount it when needed.

Click for full size images

This is my original encoder mount. A piece of rubber tube connected the pip on the spindle plug to the encoder shaft.



Here is the new plug made in Delrin. The fit is so good that the O-rings are not necessary but seem like a good idea. All machining including the hole for the encoder spindle was done in a single setting to ensure concentricity. The encoder spindle is a push fit in the plug and no other fixation was needed.



Everything in place. The plastic sheet is wide and hence rigid in the direction to locate the rotary motion of the encoder, but thin and hence flexible in the axial direction which allows for misalignment tolerances.

BTW. The eagle eyed may have noticed that there is oil in the sight glass on the old photo but there is no oil in today's photos. A while back I changed from oil to grease for bearing lubrication. The bearings run much cooler now, due to the lack of oil churning.

[metric_taper]

Tony, I have a 1976 Jet 1024 lathe. It was made in Taiwan. The head stock is similar casting as well the sight glass is the same. I found when I pulled the spindle many years back, that there was only a single hole going from the oil reservoir to the bearing (the hole level with the normal oil level). The sight glass is a front plug for this reservoir. This hole was straight back to the roller bearing. So the sight glass could show oil at the lower line level, but the bearing could be dry. So I modified both the front and rear bearing reservoirs, by drilling a 1/8" hole from the rear bottom of that reservoir to the bottom of where the bearing cup is. So now when I run the lathe, the oil is pumped out of the reservoir and circulates with the bearing (which was unintended). When running the sight glass shows the level has dropped to nothing on the sight glass. Your lathe probably did not have the design or manufacturing error, so the bearings were working as an oil pump.

I know you reworked your whole spindle arrangement on this lathe, and did away with the back gear. I think you filled the head stock with concrete to increase mass and reduce vibrations.

In your last two photos of this post, you do have a photo-optic shaft encoder. Is this added one more resolution?

[DIYer]

Thanks tonyfoale! We've added your Rotary Encoder Mount to our Lathe Accessories category,
as well as to your builder page: tonyfoale's Homemade Tools. Your receipt:

Rotary Encoder Mount by tonyfoale

tags: lathe, spindle, mount

[tonyfoale]

Tony, I have a 1976 Jet 1024 lathe. It was made in Taiwan. The head stock is similar casting as well the sight glass is the same. I found when I pulled the spindle many years back, that there was only a single hole going from the oil reservoir to the bearing (the hole level with the normal oil level). The sight glass is a front plug for this reservoir. This hole was straight back to the roller bearing. So the sight glass could show oil at the lower line level, but the bearing could be dry. So I modified both the front and rear bearing reservoirs, by drilling a 1/8" hole from the rear bottom of that reservoir to the bottom of where the bearing cup is. So now when I run the lathe, the oil is pumped out of the reservoir and circulates with the bearing (which was unintended). When running the sight glass shows the level has dropped to nothing on the sight glass. Your lathe probably did not have the design or manufacturing error, so the bearings were working as an oil pump.

Yes, a totally crazy and incompetent design. I also drilled a hole as you describe, some years before I changed to grease. I think that we have both made posts about this in the past.

In your last two photos of this post, you do have a photo-optic shaft encoder. Is this added one more resolution?

This is for screw cutting with my electronic conversion. That 50 tooth wheel was only made and fitted about a week ago. Originally when I did the electronic conversion I used the single pulse/revolution RPM signal to drive a velocity based screw cutting method. The Mach3 CNC software uses the same method. Basically, you measure the spindle speed and set the carriage speed to set the thread pitch. The concept is flawed because it is incapable of following spindle speed variations within a single revolution. In practice this was not a problem for my lathe, it has a 5hp motor and a heavy chuck, the inertia of which reduced speed variation during a revolution, but you still have the effects of pulley eccentricity etc. A smaller lathe would not be so accommodating.

Although the single pulse worked well enough I just did not like it and a while back I added a 10 pulse wheel for speed sensing but kept the original single pulse wheel for indexing. So that updated the spindle velocity 10 times/revolution. To be honest I did not notice any difference in the thread quality, but I felt happier. However, the screwcutting algorithms were still velocity based, whereas a thread is a position based concept, a pitch based on a rotation. Real CNC machines use position based algorithms which need rotary encoders of fine resolution. I did consider fitting a rotary encoder but to keep the spindle hole clear it would have had to be belt driven which I did not want.

I thought that a 100 tooth wheel might be enough to produce good threads using a position based algorithm. Without any carriage inertia, general structural flex and carriage motor delay that would mean that every single pitch would theoretically be made up of 100 steps of 0.01 mm for a 1mm pitch thread. We do have inertia and other smoothing influences so the actual results would be better and I designed some software velocity based smoothing which I thought would make it close to perfect. It was a simple software change to test the effectiveness of the software smoothing with the 10 pulse wheel. It worked well but further reflection led me to consider that a 50 tooth wheel would be better than 100 and that is what you see in the photos. It was fitted a few days back and I feel happier knowing that I have a position based system with velocity smoothing, even though the thread quality seems the same as the single pulse velocity based system. For a low inertia mini-lathe this is a much better system than the single pulse Mach3 style solution.

Click images for full size pix.
This was the previous 10 tooth wheel on the left. The single index pulse sensor is mounted inside the head. The multi-tooth wheels would have been also but it means pulling the spindle to do that. hence the external fitting.

[tonyfoale]

FWIW here are links to previous posts about my lathe conversion to NOT CNC.

Ball screw and electronic lathe conversion
Improved lathe controller.
More on my lathe conversion
Final?????? update to the lathe electronic control

Some other lathe mods

Improving a lathe spindle head.
Screw-on chuck mounting to bayonet.
Lathe cross slide conversion to a ball screw