Rough / Jagged Diagonal Edges

I'm wondering if anybody has encountered this problem:

I have perfectly smooth edges on X-axis or Y-axis only runs, but on diagonals, the finish is quite rough.

There is a high frequency periodic (chatter) nature to the noise which I suppose I can attribute to the single flute ("o-flute") cutter,
but what is especially disturbing is the random and fairly deep gouges.

I'm running a Saturn 4x4 with CNCRP NEMA 34 PNP electronics (ESS) as well as the CNCRP PNP GMT spindle and driver.

I'm tempted to say the mechanical system is sound because X or Y only gives superior finish.

Attached are pictures of the smooth (X only) and rough (diagonal) runs.
Keep in mind that the "smooth" picture took me 10 minutes to capture. The lines present are exaggerated by getting the lighting just right.
It is almost a perfect finish and feels like glass to the touch.
The "rough" edge will make a "buzzing" sound if you run your fingernail across it. Very pronounced.
To get an idea of the scale, the Delrin plates shown are 1/2" thick.

Some additional info:

I have the belts tightened pretty good (with my index finger I can only manage to push in the side of the belt maybe 1/8"
I have the rack fairly tight as well (only requiring 3mils of backlash compensation in Mach3)
Material is Delrin
Bit is 1/4" Amana Spiral "o-flute"
Chip loading is low at 0.0015" to 0.0025" based on varying speeds/feeds I tried
Tried RPMs from 10k to 15k, but no faster than 25ipm (I'm looking for finish quality, not speed)
DOC 0.1" to 0.2"
Tried conventional and climb milling (conventional seemed a tad better)
Also tried cutting a pocket as well as only finishing an edge
Held my shop vac nozzle near the bit the entire time
I made a part using Fusion 360 CAM (based on mostly it's defaults) and just running diagonals via Mach3 MDI tab. No difference.
Running Mach3 on a Core i7 notebook with power saving modes turned off (via Ethernet of course to the ESS).

I have 4 unique aluminum brackets (all with very long diagonals) to deliver to a customer next Tuesday. I can just sand and polish for deliveries like this - but it is getting old.

Thanks for any ideas of what to try.

Turn on 'Smoothing' in Fusion 360. You'll greatly reduce your file size and lines of code, and probably clear that right up. On Walnut about 0.850" thick, I cut on diagonal with a spiral two-flute 1/4" bit at about 1/8" depth of cut, 18k rpm, and feed is usually 150 ipm (Saturn 2x4, 3kW water cooled spindle).

David

Thanks David, I will give that a try today.

No problem, glad to assist. I guess I could have shown where it is and some settings...

Set the tolerance to something reasonable, like 0.001" and then check Smoothing -
Attachment 370700

David

Well, I gave Smoothing a try but unfortunately I still have the same problem. I wonder if maybe Smoothing really becomes advantageous for arcs - but not for straight lines?

I ran the same part on a piece of plywood. Not as noticeable there.

It really shows up on hard surfaces.

Thanks David for the tip though. I'm sure that will help my circles :-)

I'm open to any thoughts.

That's interesting because my diagonals had a similar profile but not quite as obvious. When I turned on smoothing it cured tons of issues like that. I think it comes into play on everything except straight X or Y movement, not just arcs or circles.

David

PS - I don't mind trying your file on my machine if you want to send it to me. PM your contact info to me if you want to give this a shot.

OK - I have an update on this issue.

1) I took Fusion out of the equation. I used a single G1 command on Mach3 to route a straight diagonal path (approx. 35 degrees from X).
The issue is still there.

2) I used aluminum this time (because it is easier to show the irregularity on the edge). It reflects light better.

3) I've included pictures showing the edge of a 3/8" bar of aluminum.
a) I first faced the entire edge at full depth.
b) Then raised the bit up ~0.150" and milled in deeper 0.050" to create a shelf

The idea was to check two aspects of the cut:
c) compare the top and bottom shelf routed surfaces of the step (edge of the bar) to see if the patterns had the same marks/grooves/deep cuts in the same locations (they didn't and instead were random).
Good, as that would otherwise indicate a bad rack or gear/spindle tooth pattern.
d) to look at the edges with respect to each other (from top-down) to see how inconsistent the cutting path was each time
This really shows the issue well.

4) The image with the greatest zoom really shows the wavy nature of what should be two straight lines (looking nearly top-down).
That should be 50mils separating the edges, but sometimes it narrows down to what looks to be 40mils.

Can anybody please share an image of what I should be getting for a diagonal path utilizing a Saturn with ESS and Steppers on Aluminum or Delrin? (I don't *think* wood will show this well)

I ran about 10" but it is so bad that only a few inches are typically needed to see the issue.
This manifests in my circles as well, but those are more difficult to compare to other people.

I put my oscilloscope on the 2 bipolar stepper phases to look for non-periodic pulses, but my scope memory is not deep enough to capture enough events to do a good comparison.
....And really, I can't imagine the ESS has a design issue or we'd all hear about it, so I'm thinking it's more of a router setup issue.
Unless I have a bad ESS card or Driver?


Some final thoughts:

I was wrong to assume that how I have my Saturn setup could not be potentially at fault in my first post. I thought that because my X and Y only moves were clean that it was functioning properly. Not necessarily true....

It struck me that poorly operating router mechanical linkages will not show on a pure (or nearly pure) X-only or Y-only cutting path. In those cases, if the rack doesn't move fully (or at all) when a pulse occurs (due to stiction, slack,etc), it doesn't affect the routing quality much. The next pulse or so will put enough pressure on the linkages to catch up and simply bump the spindle along the same path. So if the velocity is inconsistent with only one axis driven, it of course doesn't result in a bad (and especially never a wavy) edge. I suppose to some incredibly small degree the frequency of the small cutter marks may increase and decrease under microscopic inspection, but assuming the speeds/feeds are reasonable the finish quality will still be excellent to the naked eye.

In the case where two axis are involved (let's say a 45-degree angle for simplicity) both X and Y steppers get the same pulses (assuming equal gear pitches, steppers, etc.) If X and Y don't actually move equally and immediately (i.e. in sync) based on each driver pulse, there will be a wavy diagonal path.

The length of the completed traveled path is still essentially correct, as no pulses are actually missed (except for say one or two at the end if there was slop). It's just that the pulses don't occur in a perfectly periodic fashion throughout the entirety of the routed path.

So I'm fairly certain that is my issue. I just don't know how to figure out what I've done wrong with my build so I can fix it.


Thanks for any suggestions from the community!

I'm hoping I'm just doing something silly. Wouldn't be the first time that's for sure.

http://www.cnczone.com/forums/attach...d=370914&stc=1http://www.cnczone.com/forums/attach...d=370916&stc=1
http://www.cnczone.com/forums/attach...d=370912&stc=1

I suspect steps per distance are too coarse. Are you using micro-stepping? what are steps per mm -- I know. give it in inches if you need to.
Put a dial indicator on an axis and move it one step at a time. Each step should be the same, and MUCH small than the steps you see on the diagonal cut.
Test on X and Y axis.

I'm not in my shop right now but I want to say 2040 "steps per" rings a bell in my motor config. My units are set to inches.

Yes, good idea. I can try bumping it along one step at a time. Might be slow going, but eventually my indicator will show something (assuming it is not a strange problem that only occurs with ESS is driving both X and Y simultaneously).

As you say, maybe that will allow me to isolate to X vs Y. Or maybe I will see it on both....

I will give it a shot.

Most often, these issues are mechanical. Just not obvious at first. Breaking problem into smaller part and testing each axis will help you.
Make sure you, say do 2 steps in one direct, then 2 steps back. Should be EXACTLY at the original starting location.
Just use the tab key on Mach3 to set small increments, and use your ctrl-cursor keys to make the small steps. Quick and easy that way
Same number on display should be the same location on the dial indicator (or other crude levered indicator)

I also think it's mechanical. Specifically rack and pinion. There have been a handful of similar threads in the last few years.

cobra

just from your explanation:

1-my gut first thing would be to check servo parameters
2-check servo set screws (servo mechanics), physical connections
3- try x axis 4 pass groove , y axis 4 pass groove (multipass gives better perspective on intermittent issues)
4- mill round pocket 4 pass, same as #3
5- mill square pocket 45deg rotation to y axis 4 passes ( this will give you a test of 4 diagonals)

Question 1:

Am I the only guy running aluminum on a Saturn?

Just curious what another Saturn customer's edges look like when run at a 45-degree angle?

I need a bogie to chase in terms of finish quality.

I'm under the impression right now that most if not all the other users are only routing wood?


Update:

Just by chance I happened across just the right finish pass stepover to highlight an underlying problem.

Taking off 3 mils from the edge of an aluminum bar that was 0.200" deep, I was able to get a periodic, wavy pattern.

I could actually hear it loading, then unloading the bit....over and over.

Previously, I think the periodic gouges were masked by a bit of random issues (chatter? dunno) which was throwing me off.

I counted the number of rack teeth I passed during the pass and it was the same as the number of waves in the side of my material. Bingo.

I can change the angle and determine if X or Y is contributing more, but it is clear there is a gear mesh problem (I ran my material at 45-degrees from X).

I tried adjusting the spring tension bolt from just barely kissing to 2-1/2 turns past mating of gear and rack. It didn't really make much of a difference...maybe for backlash only.

There may be other issues, but I need to fix this one first. It is fundamental to getting decent quality.


Question 2:

Has anybody changed their Saturn to the CNC Router Parts Pro Rack and Pinion Drive and/or Rack on their Saturn? Is it any better?

It seems like the rack would take some work to retrofit due to the hole pattern being different, but the drive system doesn't seem too different.


Interesting finding:

I wrapped the barrel of a very fine point sharpie-like pen so that my 1/2" collet would tighten down on it.

I found then pen at a craft store. It is a fraction of a mm in width.

I made a 45-degree path with the tip just barely touching a piece of paper and could actually see the waves in the path.

Again, the pattern widths coincided with the rack pitch.

This proves there is no contribution due to chatter, feeds, speeds, bit quality, climb vs conventional, backlash, blah blah blah...

It is purely rack movement causing the issue.

I've cut 3/4" and 1/4" Plexiglas but no aluminum as of this time and no issues with the Plexiglas. I want to cut some aluminum but don't have a current project or even any aluminum except for small pieces of bar stock from Lowe's.

David

Thanks David. Your edges are perfectly void of any periodic waves?

Yes, unless I get the feed a touch too fast. I might see some ripples in the right light but otherwise the cuts are very smooth, especially after turning on Smoothing in Fusion 360.

David

OK, I think I finally diagnosed my problem. This appears to be velocity ripple.

I'm no R&P expert by any means. Just did some research.

Looks like nobody else has had this problem, so maybe I'm wasting my time.
I'm going to post this anyway, in case it helps somebody in the future.

Somebody with more knowledge than me please correct me here...


Velocity ripple is basically the periodic error caused by imperfect rotational to linear motion transfer.

The frequency of the waves cut in the material match the pitch of the pinion or (1inch * pi)/20 teeth, so I'm fairly confident this is my root cause.
Also:
There is no reason the ESS would cause errors at the pitch rate (it doesn't care what gears, pinions, pitch, etc. you are using).
Same for the stepper motors
The stepper pulley and mating reduction gear timing don't match the marks in my material - so I can't blame the timing belt/pulley/reduction system
I replicated the issue dragging a fine pen (i.e. no spindle turning) so it is not chatter


Getting back to R&P theory:

A pinion is able to travel along a rack with constant velocity if it's angular velocity is constant.

Many issues can impact the quality of velocity, however:

1) Wear
2) Improper mesh (caused by pitch line and pitch circle not lined up)
3) Tooth profiles not perfectly matched

For those interested, I've included a picture of what the "pitch circle" looks like on the Saturn pinion. It looks to be ~2/3 of the way down the tooth.
For perfectly smooth transmission of motion, that pitch circle must be tangential to the pitch line of the mating rack.


I can eliminate #1 because I've had this issue since day one and hopefully #2 because I've tried various depths of engagement from not touching (with large backlash) to gorilla grip (I'm sure enough to cause premature wear).
Roughly, the [rack] pitch line or [pinion] pitch circle will coincide along a tooth where the width of the tooth is 1/2 the pitch.
This is not exact for a spur style gear, as it needs some space for lubrication - hence, the backlash (I always wondered why backlash can't seem to be eliminated with single driven spur gears).
Since I don't have specs for the rack, I think it is best to simply draw a 1" circle concentric with the pinion to determine where the pitch circle hits the teeth.
Then, the same position down a pinion tooth can be applied to a rack tooth (assuming matching profiles).
Now, one simply needs x-ray vision to line up the pitch lines with the pivot arm and reduction gear in your way LoL!
(this is why I resorted to varying tension levels in small increments)

Anyway...
All I can figure is that the rack and pinion components are made by different companies and either don't use the same, matching profile or maybe I just have some bad parts.
Nate has been very helpful and I can't ask any more of him.
With his support, I at least have the Saturn running as good as it can be given the limitations of the rack system.

I only see the issue with metal parts and some plastics. Wood seems to hide the issue well.

I suppose only those of us running aluminum for long edges (at some angle) will even notice - unfortunately, I'm in that category.

So....before I tear down the whole system to retrofit with a screw-type drive (yep - I'm pretty close to that), I will be replacing the rack and pinion components
There are needle bearing based roller rack systems that have exceptionally low ripple - but they would cost thousands to upgrade the Saturn.
So I will try replacing with similar components from a well-known brand (and only one brand for both rack and pinion).
Most likely, this will require drilling new mounting holes for the rack, but that should not be too difficult.

I should add that most people that comment on parts I've made are very impressed!


http://www.cnczone.com/forums/attach...d=374430&stc=1