Big builds

File: queen of england oh yes.jpg (184KB, 634x1287px) Image search: [Google]

184KB, 634x1287px

>>1140310

I still remember when I finished my first CNC router.

>TFW precise curves are now only a G2/3 away

File: DSCN2956.jpg (849KB, 1632x1224px) Image search: [Google]

849KB, 1632x1224px

Currently working on machine #3. Just got it running the other day, still need to make the spindle encoder to deal with threading (manual-change gear boxes are fucking abhorrent). And put it in a suitable box.

Threads, tapers, and curves will now be easymode. This pleases me greatly.

>>1140327
Nice!

File: tmp_7697-IMG_20170107_2115124-578254173.jpg (1MB, 4896x3672px) Image search: [Google]

1MB, 4896x3672px

So I am a model railroader. The white gears in pic related need replaced, some have cracked radially and make noise. It is not possible to find replacements. What options are available in having some fabricated?

Looking for a place or person I can pay to have them made as well as some spares, alternatively a printer/mill/whatever that would be fine enough to make the teeth.

Right now I dont really have the ability to diy my own because it would take away from me being able to model railroad.

Any suggestions?

>>1140348
I can prolly print them for you if you can ship them. Or mill them from acrylic...

File: tmp_7697-IMG_20170107_2032026-1846038734.jpg (1MB, 4896x3672px) Image search: [Google]

1MB, 4896x3672px

>>1140354

I am definitely interested in doing this. These go on an old engine that belonged to my grandfather. I have two engines that use these gears and all four are cracked. Offgassing over 40 years caused it.

Ill try to get them pulled out for better pictures and measurements. I wanted to throw together a CAD design to print them in the future too, but I put my desktop in storage to build the train set.

Last time my local repair shop saw any was two decades ago. Please shoot me an email at [email protected].

The orange/red and green/grey are the two affected locos.

>>1140327
How are you doing threading with an encoder? I want to do this for my lathe, but I'm not sure about the programming.
Encoder counts per motor step doesn't come out to an integer and so it would have to do something to interpolate fractions of a count. Am I overthinking that?

File: 20170307_203059.jpg (3MB, 5312x2988px) Image search: [Google]

3MB, 5312x2988px

Also working on a cnc machine.

This is my 9th build, and decided to go for a traditional vertical machining center. Finishing up the welding now, then to stress relief, machine, re stress relief, re machine, then electronics.

I still got a while to go...

im almost ready to start
i have a shitload of linear stages, a shit load of cast iron, the spindle shaft , headstock bearings

now i just have to draw it

gonna make me a nice small scale mill turn cnc

>>1142080
with the quadratic factored in, do they have a lowest common denominator that is usable?

>>1142308
I'm not sure what you mean by quadratic.
With an 1800 counts/rev encoder and 200 steps/rev motor, I can express all the thread counts from 18 to 56 in terms of an integer number of counts per 25 steps, but I'm not sure if that helps me.
I can't count up encoder counts and then step 25 times, it'll make an uneven thread. I have to somehow intersperse these 25 steps evenly over the number of encoder counts.
Example, for 32 threads per inch, I need 11.52 counts per step which works out to 288 counts per 25 steps.

because encoders have 2 measurements,
cpr cycle pre rev, which is its total number of index on the encoder

then there are pulses per rev, which use the up/down of the pulses to multiply your encoder count by 4

so if your cpr is actually 1800, then your ppr would be 7200, because encoders use 4 channels a= a- b= b- then what ever index you have

so if you can work your gay imperial threads into a workable pitch with 7200 dividable units to work with instead of 1800

unless your encoder has 450 cpr and your ppm is 1800....

File: 20170310_213923.jpg (1MB, 2048x1536px) Image search: [Google]

1MB, 2048x1536px

I just restored an itty bitty vice that had been sitting in a shed for decades. A couple of days in a vinegar+salt bath, some wire brushing and a pair of layers of paint is all it needed.

>>1142080
>Encoder counts per motor step doesn't come out to an integer and so it would have to do something to interpolate fractions of a count. Am I overthinking that?

I don't even know how you got to actual steps or counts being relevant. That should be all handled in software.

In LinuxCNC (which is what I'm using), G33 is simply "move to this point on this axis at this distance per revolution". Example being:

G33 Z1 K0.1
G33 obviously invokes spindle-synchronized mode
Z1 gives endpoint of the move
K0.1 causes the tool to move 0.1 units per revolution of the spindle.


Unless you mean you're making and programming a controller, the physical quantizations of the motors/encoders don't come into play at all. If you ARE trying to program from scratch, all you need to do is just round to the nearest step. That's all you can do, really, and the resolution of any machine should be high enough that a single step should be well under the tolerance that the machine can actually maintain.

>>1143579
I am making and programming a controller, actually. I have a manual mini-lathe without a threading feed.
My encoder only has one channel, 1800 cpr, so I can get 3600 ppr. That doubles my counts/step values, making rounding a little more accurate, but I'm not sure it's accurate enough.
My worst-case value is 20 threads per inch, with 3600 ppr it works out to 14.4 counts/step.
14 counts/step works out to 19.44 threads per inch. How well will a thread of that pitch fit with a proper 20 TPI thread? Not well enough, I'd expect.

Is this just a matter of getting the ppr high enough that the rounding doesn't significantly decrease resolution?

>>1143592
>14 counts/step works out to 19.44 threads per inch. How well will a thread of that pitch fit with a proper 20 TPI thread? Not well enough, I'd expect.

Yeah, no, that wouldn't work at all.

What you need to do is work out the desired location for EACH STEP, rather than trying to use the closest integer value you get from counts/step. Then use that to figure out how many steps you need to get (approximately) to that position.

In other words:

1.) Get count from encoder.
2.) Figure out where, in decimal space (to a high accuracy; say 6 places), the tool needs to be on that count.
3.) Determine how many steps need to be taken from the current position to get to the new position.
4.) Round, if not an integer.
5.) Step.
6.) Repeat for each count of the encoder.


Using an extremely high-resolution encoder to try and brute-force your way past the incremental error you get is the wrong way to go about it. For reference, most spindle encoders on a lathe (at least, those meant only for RPM detection and threading) have a few hundred counts, max.

>>1143680
I see, I was thinking about it wrong. Thanks.

Hello. I plan to do as final project a cnc machine
( i am at electrical engineering class) nothing big, just to mill some soft woods etc. The thing is that i never did such a project (i have some experience in diy with various staff but nothing too advanced ) Anyway my question is how hard you thing building a cnc machine is from a scale 1 to 20. You needed some advanced knowledge at the porgrammable part ? thanks in advance

can you guys post your specs for shit like

>software
>hardware electrical
>hardware motors/motor drivers

as i get closer to this phase in my project, i need to get this shit in gear,
and i was wondering what everyone else is successfully using and to what ease , such as setup/modification

i am personally hoping to use linuxCNC, but i am open to what ever may support my needs better

>>1143592
can you post more details about your setup/plans? i am curious and building something similar and am interested