Choosing a spindle encoder

Hello, all,

I'm currently amassing parts for a retrofit on a 1950's Hardinge DV-59, dovetail lathe. This will be the machine's 3rd retrofit. I'm working with Jon at Pico systems and I've purchased his PWM control board and a couple of brushless amps/other accessories. But the machine needs a part that I didn't think of until just today. I need an encoder for the spindle motor so I can achieve accurate threading. The motor is a 1hp 3phase motor, likely the original motor, too. It was turning parts a few years back after I retrofitted it with Animatics smart motors and the spindle worked great. But it wasn't controllable outside of an off/on command via M3 through the former retrofit. So the motor is fine. That being said, I'm hoping to find a suitable bull-gear mounted style encoder which will merge easily with the Pico systems board.

Any suggestions out there? So far, all I've seen is the code to configure the connection to the system but not the encoder style/type that would be suitable and fairly easy to attach. Looking for hardware. The code isn't so much of a problem.

Thanks in advance,

Scot

what is the rated MAX Rpm
and what hardware are you using to red it to linuxcnc

with a 500ppm you are done at almost all parts workes fine up to 5000rpm on 7i96

I'm using Pico-system's PWM controller and hardware. The motor is 1750, belted up to 3500rpm with the original belting system. I was planning to peg the physical pulleys to their max RPM setting and running a VFD to control it. The motor is REALLY old and more than likely original to the 50's era. So attaching the encoder might prove difficult, but that's the option. I could make my own, I guess. But It would be nice to have a bolt-on version to shorten the time it takes to build.

I found one at digikey that matches my price range, with 100ppm. I was told by Jon at Pico systems that he holds rigid-tapping precision at 80ppm on his bull gear, home brew encoder. So if he can hold that with something rudimentary, I thought it may work fine with a 100ppm model. But I've only done a stepper system without encoder feedback. So this is kind of new and poses many questions.

Thanks for the reply. Any info will help.

Scot

If you build you own encoder, a lower resolution encoder will be easier to make
but if you are going to buy an encoder ( and you have a hardware encoder counter as
you should with Pico Systems hardware ) I would go with higher resolution say 500 lines
(2000 quadrature counts/turn)

Note that in general, the encoder needs to be connected to the spindle, not the motor
unless the spindle <-> motor speed ratio is 1:1

I'm going to have to think about this. The old belts are going to be replaced. So slippage is likely going to be minimized once that is done. But there will likely be slippage. And there will be an inaccuracy at the spindle if I belt the encoder to the motor shaft, even if the belts are timing belts, and if those drive belts slip. So I guess you're correct in that notion that an encoder would be best placed at the spindle. One problem is that a home-brew encoder on the spindle would be out in the open, as that spindle is out there, too, and there aren't many safe places to put it where it won't be bombarded with oil or flying parts

I'm having to reinvent the wheel, here, I guess.

If I belt the encoder to the motor shaft, there will be potential slippage on the motor belts which will alter the real world RPM on the spindle. But there isn't really a good place to put an encoder on the spindle to avoid damage for such a hi-res encoder. They're kind of fragile. Mix that with the notion of making an accurate encoder is a difficult task and it's a recipe for inaccuracy or breakage.

Making a timing belt gear for this is possible. It's just very time consuming after making the gear and mounting plates, then disassembling the collet closer to put the gear on it and reassembling it.

Old iron has it's challenges. But this machine was really accurate before I killed the old encoder. It would be a really nice thing to see it make parts. It was holding .0002" with a .0001" resolution before I fried it.

Thanks for the input. If you see something I'm missing, let me know.

Scot

Oh man! I type way too fast and I don't edit nearly enough. I meant to write CONTROLLER, not encoder!

There was never an encoder on this lathe. It was retrofitted with an upgrade technologies 2 axis system with no spindle, collet or coolant control back in 1993! Then again I retrofitted it with an easy PC control and smartmotors by Animatics after that controller died in 2004. But I fried the motors and rendered the controller useless a couple years ago, due to a bad ground between the PC and the motors. So I decided to retrofit it again, since I was holding such good tolerances with it when it last ran. It's the original motor from what I can tell which lends itself to another set of problems due to it's shape and placement. So after listening to a bit of advice, I determined the best possible and most accurate outcome will likely be to mount an encoder onto the rear of the spindle by the collet closer.

I've even looked into making my own encoder wheel to mount on the rear of the spindle by the collet closer and using an off the shelf pulse reader that you can find typically at digikey etc.

I can get 12-24 custom encoder wheels with 500ppm and an index made for about 800 dollars and turn an adapter that can attach to the spindle, but I only need one encoder and that price is twice as much as a decently high end, chassis mount encoder.

So I'm kind of stuck. I was thinking of a single pulse, but I was advised that a multi pulse encoder will be a better choice for turning in inches per revolution. I'm still unsure if that's the case, as I haven't really looked into this in depth.

The parts we're trying to make are pretty small. The smallest of which is about .040" OD. And in some cases are threaded inside with 000-120 threads, but not always.

The precision of the machine needs to be such that I can program the spindle in inches per revolution and keep myself from breaking taps when the tap nears the bottom of the hole to be tapped. I know the gcode as I've done this programming on a precision machine before. I'm just trying to make this old bird sing and hit every note.

Thanks for the reply,

Scot

andypugh wrote:

The index absolutely has to be on the spindle. But a single pulse per rev is easy to find. You might get away with having the AB phases on the motor, for short threads.

He's going to use the Pico Systems PWM controller, it does NOT support one pulse/rev spindle encoders, only quadrature with index.

Jon

Scot wrote:

The precision of the machine needs to be such that I can program the spindle in inches per revolution and keep myself from breaking taps when the tap nears the bottom of the hole to be tapped. I know the gcode as I've done this programming on a precision machine before. I'm just trying to make this old bird sing and hit every note.


Scot

No, you don't program rigid tapping like that. LinuxCNC has a dedicated tapping routine G33.1 that takes care of everything.
You do have to know how much "overrun" there will be as the spindle reverses, to allow a little clearance at the bottom of the hole.
You can actually use Halscope to capture the Z move and figure out the the exact amount of overrun on your machine. (Note, it will change depending on spindle speed and thread pitch.)

Jon