Saturn 2 4x4 - Nema 23

[DDgitfiddle]

The machine being discussed the gantry was removed to get inside a building. No marking or scribing was done before hand to make it easy to get back together correctly. My machine is pefect and I use it every day. My problem is I can’t seem to get enough free time to use it!

I had the same self-inflected pain on my 4x2...I had to remove the gantry to get it into my shop. Getting it squared up was a process, but not impossible. I didn't square it up the same way recommended above, but in retrospect, I wish I had, because it would have been a good bit easier. I also had an additional problem...my gantry was not secured properly during shipping, and it slid forward and bent one of my hard stop plates. As a result, one of my stops is definitely NOT even with the other, so I knew I couldn't rely on that.

It is still a "kit" machine. It takes effort and some fine tuning to get it up and running, but when you are done, it's a pretty nice machine for the price...maybe even for twice the price. You won't regret it.

[PLJack]

I can get behind the idea of keeping both sensors, and adjusting them so that both trigger when the gantry is homed and square as long as the home signal only triggers off one sensor to control both motors. The other sensor will not impact operation but will be there as a sanity check if there is any concern over the squareness. It could also possibly serve as a backup over-travel sensor if your controller will allow for two of those.
-Robert

Bingo! That is my plan.

Jack.

[PLJack]

It is still a "kit" machine. It takes effort and some fine tuning to get it up and running, but when you are done, it's a pretty nice machine for the price...maybe even for twice the price. You won't regret it.

Is there such thing as a double bingo?

[PLJack]

Well. That was a long road.
Took me all night to work out a procedure. I had to restart a few times as you can see.




I now have the gantry squared to ~0.5mm over a 1120mm diagonal.
To start I removed the stop plates and grounded the tap holes. I'm doing this to every one I come across since the second gantry mounting. They seem to protrude a bit.
Don't see the "O" face. you can't unsee it.



Reinstalled and started squaring. Here is the procedure I came up with after learning a few lessons the hard way.



XML attached if you want to modify it on https://www.draw.io/

When finished this was the difference between the two stop plates. The other plate has the end of the screw flush with the back of the stop plate.



I'm calling the squaring good for now. I will address it again another time. It's pretty quick once you have a plan.
Moving on to shimming the gantry.

Thank you all very much for the help.

Jack.

[GME]

I have one reason that I can think of to NOT use two sensors for homing on a Saturn machine. There is a always little variation in the repeatability of these sensors, and it is difficult to adjust them to trigger at EXACTLY the same time. This variability can lead to differences in when the two motors stop during homing. I firmly believe that the Saturn gantries are extremely stiff as long as they are appropriately tightened to the bearing carriages. With this stiffness, the pinions will load up if they try to push one end of the gantry further than the other. This "preload" will stay in the system unless the motor slips (e.g. loses a step or two). This can have a number of possible effects:
-Robert

Thank you, Robert. I am left wondering how much variability there is in sensor repeatability. That will drive how I intend to proceed. I am not particularly concerned about getting the 2 sensors to trigger at exactly the same time, unless the variability gets in the way. My intended approach was to have the gantry loose, or removed, and home the mounts in the conventional way. As has been pointed out previously, it doesn't necessarily matter whether the mounts are perfectly parallel to one another (as measured from side to side). In fact, they are not likely to be, if only moving one side to mechanically square. That also means that if one sensor block is a tad in front of or behind the other on the Saturn 2, it won't matter. Absent an issue with repeatability", the two should trigger at EXACTLY the same instant.

So, the question is repeatability and what that means in quantifiable terms. If it amounts to .010" then it's a no-brainer. I would consider other options. However, if were are talking about .001" or ".002 or less, it could work. I did some very quick research and found the found the following, which, on its face, sounds encouraging:

Today's quality inductive proximity sensors can have trigger points that are repeatable to 0.0001 in. To obtain such precision, though, the detectable object must be moved the reset distance away from the sensor after each time the sensor is triggered.
https://www.electronicdesign.com/com...oximity-sensor

Given the nature of CNC work, moving "the reset distance" shouldn't be a problem.

I haven't found other sources for repeatability stats yet, but I'll keep looking. Do you have more specific information/references?

Thank you,

Gary

[difalkner]

By using a single sensor, both motors are commanded to stop at the same time, avoiding this extra tension in the system.

-Robert

I agree, Robert, and that's one reason I used only one. That gantry is incredibly stiff. The other reason I'm only using one is that Nate said one was sufficient. I just don't see the steppers pulling it square by using two sensors without causing other issues in other areas.

It's great that y'all have dived into this squaring issue to this level. I hope Nate is monitoring and taking notes.

David

[GME]

I agree, Robert, and that's one reason I used only one. That gantry is incredibly stiff. The other reason I'm only using one is that Nate said one was sufficient. I just don't see the steppers pulling it square by using two sensors without causing other issues in other areas.

It's great that y'all have dived into this squaring issue to this level. I hope Nate is monitoring and taking notes.

David

David, I think there is a misunderstanding going on and on and on . . . . I do not want, nor do I advocate for using 2 sensors to square the gantry. I couldn't agree more that it would be a very bad idea and for all the reasons already discussed. I haven't seen anyone who disagrees. I don't understand why it is being treated like a point of controversy.

Here are the sets I propose for using 2 sensors:

1. Loosen, or remove the gantry altogether
2. Home the gantry mounts
3. Immobilize the mounts on each side at the homing backoff positions, since the software automatically backs off a few thousands and stops
4. Once the mounts are immobilized, turn off the motors.
5. With the mounts immobilized, square the gantry (however many sequences it takes to get it square)
5. Remove the immobilization blocks.
6. Start the machine, home (to both sensors), and operate as usual.

In theory, at least, the gantry should then home at the exactly the same place, at exactly the same time, and without any sort of squaring taking place. The initial setup starts at the sensors, and ends at precisely the same spot. The only wildcard in the process (at least so far) is variability in sensor repeatability. Robert said it's there, and suggested it would cause problems. I then found one source addressing it, which said good quality inductive sensors should be repeatable to within .0001" (not a typo, it's 1/10,000"), which sounds darned good. If true, both motors would start to approach the sensors at exactly the same instant, and would stop together at essentially the exact same instant. No squaring or racking involved. I haven't yet had a chance to do more exhaustive research on inductive sensor repeatability, but plan to.

I'm not sure why it keeps happening, but just about every time the subject of using 2 sensors comes up, it gets turned into a gantry squaring issue. It isn't and never was. However, no amount of explanation to the contrary seems to matter. It keeps circling back to squaring. I'm not angry, but I am getting really frustrated at my apparent inability to more clearly express myself.

[Ntl]

I agree, Robert, and that's one reason I used only one. That gantry is incredibly stiff. The other reason I'm only using one is that Nate said one was sufficient. I just don't see the steppers pulling it square by using two sensors without causing other issues in other areas.

It's great that y'all have dived into this squaring issue to this level. I hope Nate is monitoring and taking notes.

David

Ok good that’s what I thought, I tried moving one side of the gantry on mine because on one side it didn’t hit the bump stop and we’re talking like a 1/16” and with all my force I couldn’t keep the gap closed. If a motor was trying to do that I seriously doubt it could hold it. I’m not a weak human lol. I’m positive that the pinion would just skip gears unless the backlash screw was tightened all the way and then I cannot imagine it would be good for the motor to try and hold it if it could to begin with. Also the blocks with the bumpstops aren’t even close to level at least on my machine so don’t use them as a reference for squaring . It’s either because the holes aren’t tapped straight or the end caps aren’t welded on level. Obviously unless everything is machined after construction and there is no paint in between assembled parts nothing will be perfect. I just hope that the machine is good enough for the non precision work I do.

[DDgitfiddle]

David, I think there is a misunderstanding going on and on and on . . . . I do not want, nor do I advocate for using 2 sensors to square the gantry. I couldn't agree more that it would be a very bad idea and for all the reasons already discussed. I haven't seen anyone who disagrees. I don't understand why it is being treated like a point of controversy....

Part of the misunderstanding may be on my part. The Saturn (1) system has the home sensors mounted to the main frame so that the carriage approaches the tip of the sensor axially. Originally it would trigger off of the grease fitting, so I moved mine so that they trigger off of the aluminum bearing carriage. When things go wrong, the sensor gets smashed. (I think David had that unfortunate experience). Either way, these sensors are meant to be triggered by sweeping a flag across the face of the sensor. They are more repeatable in that configuration, also. I THINK the Saturn 2 has changed the mounting so that the sensors ride with the carriage and trigger on a fixed flag, so that is a good change.

The method you are suggesting should work OK and would minimize the tension on the RnP, but you MAY find that you don't have enough adjustment range in the gantry holes to get the gantry square. That will depend on how much offset you end up with between the carriages. If you run out of adjustment range, you would just need to adjust the flags (or sensor locations) to give you a little more space. If I am understanding your method correctly, you will be squaring the gantry by adjusting the gantry relative to the two bearing carriages rather than moving one carriage fore-aft relative to the other carriage. If that is correct, then your range of adjustment will be limited by the size of the gantry bolt holes relative to the size of the bolts and also the accuracy of those holes relative to each other. With the same hole sizes, in theory you can get more range of gantry adjustment by allowing the bearing carriages to move relative to each other (for initial machine setup only, while mechanically squaring the gantry). Or maybe you were thinking about adjusting the sensor flag location to do this during initial setup? That could work also and would give you pretty good control over squaring

There is also still a SLIGHT error in when the motors stop. It is the 0.0001" sensor repeatability plus the very small error in the pullback positions for the motors, plus any small differences in controller signal timing. These are probably less than 0.002" in total. (estimate). There could still be a little residual tension, but it should be pretty minor.

[difalkner]

It's probably not so much of a misunderstanding as it is my not taking the time to really articulate what I'm thinking, Gary. Skim, respond, move on... not always the best approach when we're left to typing and getting a particular message across. Much easier in person where inflection, body language, facial expressions and the like really help! But yes, to your point, while very interesting and informative this has become a bit of a circular discussion at times.

David

[PLJack]

Ok, Shimming complete. Keep in mind these are not adjustment shims. Just pressure shims.

The right side took the worst of it.




the left just needed shims in the front.




And here is a 60mm square I cut after everything was shimmed and tight.



Top: 59.94mm
Left: 59.89mm
Bottom: 59.88mm
Right: 59.86mm

Might need a tweak to the steps per but she is definitely square!!!
Getting a good reading on wood with the sharp end of a caliper is a little touchy as well.
Calling this done.

Now I can do one more facing and cut the spoil board square.
BTW:
I was shocked when I trammed the spindle after all that gantry action. It was better than before! Especially in the -+Y direction. I'll tram again after the next facing.
If it needs adjustment in the Y direction I'm shimming the spindle!!!! Enough with the gantry.


Thanks.

[difalkner]

Looks good, Jack. I would switch to Maple or Poplar or some other close grained wood rather than Red Oak for your small square test. Now cut one as large as possible on your machine and measure that. When I cut small squares, rectangles, and circles it looked pretty close just as yours does. But when I cut the large squares and rectangles I found out just how far off it was. Now when I cut the smaller ones they're about dead perfect.

David

[GME]

Part of the misunderstanding may be on my part. The Saturn (1) system has the home sensors mounted to the main frame so that the carriage approaches the tip of the sensor axially. Originally it would trigger off of the grease fitting, so I moved mine so that they trigger off of the aluminum bearing carriage. When things go wrong, the sensor gets smashed. (I think David had that unfortunate experience). Either way, these sensors are meant to be triggered by sweeping a flag across the face of the sensor. They are more repeatable in that configuration, also. I THINK the Saturn 2 has changed the mounting so that the sensors ride with the carriage and trigger on a fixed flag, so that is a good change.

The method you are suggesting should work OK and would minimize the tension on the RnP, but you MAY find that you don't have enough adjustment range in the gantry holes to get the gantry square. That will depend on how much offset you end up with between the carriages. If you run out of adjustment range, you would just need to adjust the flags (or sensor locations) to give you a little more space. If I am understanding your method correctly, you will be squaring the gantry by adjusting the gantry relative to the two bearing carriages rather than moving one carriage fore-aft relative to the other carriage. If that is correct, then your range of adjustment will be limited by the size of the gantry bolt holes relative to the size of the bolts and also the accuracy of those holes relative to each other. With the same hole sizes, in theory you can get more range of gantry adjustment by allowing the bearing carriages to move relative to each other (for initial machine setup only, while mechanically squaring the gantry). Or maybe you were thinking about adjusting the sensor flag location to do this during initial setup? That could work also and would give you pretty good control over squaring

There is also still a SLIGHT error in when the motors stop. It is the 0.0001" sensor repeatability plus the very small error in the pullback positions for the motors, plus any small differences in controller signal timing. These are probably less than 0.002" in total. (estimate). There could still be a little residual tension, but it should be pretty minor.

Thank you, Robert. You accurately described what I propose. You also correctly observed that what may work theoretically may not work so well in practice. I had the same concern, but hadn't gotten to the point of expressing it. It all comes down to whether the gantry mounting holes will permit enough adjustment, if the sensor flags aren't positioned in sufficient relative alignment. If the sensor flags are not aligned well, the choices are (1) enlarging the mounting holes to allow more adjustability, or (2) moving the sensor flags. I'm not too excited about either choice, and what I see in practice may take me in a different direction. I won't really know any more, until my machine arrives and I test it.

I'm guessing any error occasioned by "the pullback positions for the motors" would be the same, regardless of whether homing was done with one sensor or two. I'm also guessing that it could also be the same with controller signal timing. I'm just guessing. If my guesses are incorrect, I it would be great if you could help me understand why.

FWIW, I asked Nate about repeatability for the sensors he sells. He confirms that they are repeatable to ? .01mm (?0.000394?"). He also says the same spec applies to the cheaper Chinese sensors. Apparently, the main functional difference between the two is longevity. The Chinese sensors only last 2-3 years. Some folks might be interested when making a buying decision. For some, the price difference vs longevity could be an important consideration.

I am tentatively planning to document my build, so folks can evaluate the option I've proposed. Nothing like moving beyond the theoretical to seeing pictures of the practical.

Thank you, again,

Gary

[GME]

Ok, Shimming complete. Keep in mind these are not adjustment shims. Just pressure shims.

The right side took the worst of it.




the left just needed shims in the front.




And here is a 60mm square I cut after everything was shimmed and tight.



Top: 59.94mm
Left: 59.89mm
Bottom: 59.88mm
Right: 59.86mm

Might need a tweak to the steps per but she is definitely square!!!
Getting a good reading on wood with the sharp end of a caliper is a little touchy as well.
Calling this done.

Now I can do one more facing and cut the spoil board square.
BTW:
I was shocked when I trammed the spindle after all that gantry action. It was better than before! Especially in the -+Y direction. I'll tram again after the next facing.
If it needs adjustment in the Y direction I'm shimming the spindle!!!! Enough with the gantry.


Thanks.

Nice job, Jack. While the closeups are wonderful, it's hard to get a sense of how much adjustment was actually required. In the photos, the gaps look huge. I'm sure they aren't nearly as bad as they look. Can you give a rough idea of how much you shimmed in the worst area. That gives some perspective.

Looks like you got it dialed in pretty close. Within about .0012" one way, and .0024" the other (some of us still have trouble thinking in metric).

Gary

[GME]

Looks good, Jack. I would switch to Maple or Poplar or some other close grained wood rather than Red Oak for your small square test. Now cut one as large as possible on your machine and measure that.
David

+1

[PLJack]

Nice job, Jack. While the closeups are wonderful, it's hard to get a sense of how much adjustment was actually required. In the photos, the gaps look huge. I'm sure they aren't nearly as bad as they look. Can you give a rough idea of how much you shimmed in the worst area. That gives some perspective.Looks like you got it dialed in pretty close. Within about .0012" one way, and .0024" the other (some of us still have trouble thinking in metric).
Gary

About 32 thou per shim. Highest is three shims at 2.5mm (0.098").

Jack.

[GME]

About 32 thou per shim. Highest is three shims at 2.5mm (0.098").

Jack.

Yikes! More than 3/32".

At least you were able to accommodate the situation, and that's a good thing. I admire your determination.

Gary

[PLJack]

Yikes! More than 3/32".
At least you were able to accommodate the situation, and that's a good thing. I admire your determination.
Gary

Thanks Gary. I am trying.
Is what it is.

[PLJack]

Now cut one as large as possible on your machine and measure that. When I cut small squares, rectangles, and circles it looked pretty close just as yours does. But when I cut the large squares and rectangles I found out just how far off it was. Now when I cut the smaller ones they're about dead perfect.
David

I just did a preliminary with a .7mm pencil led in the spindle. The 700mm square that it drew was:
Top: 702
Right: 700
Bottom: 703
Left: 700.

Aside from some steps per calibration on X it looks like I am out 1mm over 700mm.
I'll do a few more to see what is what.

Thanks.

[DDgitfiddle]

I'm guessing any error occasioned by "the pullback positions for the motors" would be the same, regardless of whether homing was done with one sensor or two. I'm also guessing that it could also be the same with controller signal timing. I'm just guessing. If my guesses are incorrect, I it would be great if you could help me understand why.

I don't really think that the pullback distance and signal timing will add much error. I estimated 0.002" total to be on the safe side. I also don't think you will hurt anything by giving this a try. Worst thing that happens is it doesn't work and you have to try a different method. Very, very low chance of damaging anything.

I am tentatively planning to document my build, so folks can evaluate the option I've proposed. Nothing like moving beyond the theoretical to seeing pictures of the practical.

Looking forward to it. I love seeing the different decisions people make on there machines. I've seen alot of pretty cool ideas on this site and alot of creative solutions to various challenges. I've learned alot and I'm happy to try to return the favor.

-Robert