Nils,
I wish I was closer to you, I have a few you could try. I seem to keep getting older computers when I help a friend upgrade. :)

Let me explain my frustration.

I am frustrated with me. I have reached the max of my mechanical ability. I am actually lost to some extent. When it comes to software and computers, I have some ability.

I have the greatest respect for Gerald's design and abilities to resolve problems that people have with their builds and I did not mean I was frustrated with Gerald. I also have respect for the other bulders on this forum and their abilities.

What I have done so far:

1) I have lubicated the X and Y Rails using Dupont's Teflon Dry Wax lubicant.

2) I have add some half washers to the motor mounts to make the pinon gears perpendicular to the rack. When I mean half washer here, I am talking about a washer that is cut in half. I am thinking of making some thing better.

3) I have rechecked that the Z Slide is perpendicular and the router bracked is level. When I cut flat things are good.

4) I did check the pinon gears for wear and it does not look like to me that there is a wear problem. I also checked to see if the pinon gear was bottoming out on the rack and it is not doing that.

5) How do I change the resolution? What resolution are you talking about?

6) How do I check for backlash with my geared motors?

I went back over this thread to check all the suggestions, I hope I have answered them all or have requested additional clarification.

I am still geting chatter when I cut at 70 IPM.

Before I added the washer to the gantry wheels, my cuts were better but I had chatter.

I don't know what to try next.

Thank you for your assistance.

Nils,

I am by no means an expert on the Mechmate, but I've spent a lot of my life building, troubleshooting and repairing machinery. I have followed your post and I have thought that what I would do is make some longish angled cuts in the same piece of stock with varying degrees on the angle. Perhaps you could start with a straight 'X' axis cut and then make subsequently steeper cuts in 10 degree increments until you reach a straight 'Y' axis cut. This would allow you to examine, compare and contrast the quality of the cuts under differing angles using a known straight vector.

You also may want to try this twice. Once climb milling and once conventional milling.

Just a thought.

Regards,

Jeff

Jeff,

You beat me to it. I was also going to suggest the angled cuts. However, I would like to add an explanation as to why Nils should try this technique. There are a lot of us that think this is a resolution problem. Nils, since you seem to have a regular pattern, when you change the angle of the cut, depending on whether the angle is shallower or steeper, the regular pattern should lengthen or shorten. In my opinion, this will provide additional evidence that it is a resolution problem.

You asked about resolution. A short answer is your gear box ratio. You said you know about computers. Imagine a couple of curves. One curve is analog and the other is digital. The analog curve is perfectly smooth but the digital curve is composed of a series of short lines and is sort of choppy like the picture Gerald drew. If you increase your resolution by changing your gearbox from 3.6:1 to 7.2:1 or to 14.4:1, you will have a smoother curve. One of the cheapest ways to change the ratio is to use a smaller pinion gear.

Oh yeah. You can check the backlash in your gear box with a dial test indicator. Google it and I'll bet you find pictures of the set up. Harbor Freight has cheap dial indicators.

Nils, when I looked at your .dxf file of the diagonal in question, it appeared to be a pure 45 degree line......is that correct?

Everyone else, are any of you having issues with 45 degree diagonals? I assume not.

One of the factors that can make such a stark difference between similar machines is the performance of the PC. But Nils does not have a spare PC to test with.

However, you could try something else...........

A 45 degree diagonal means the pulse rates of X and Y are identical. For a test, program your Mach to cut a straight line at 50 IPM in the X direction. But, disconnect your y-gecko's inputs from the PMDX and connect it in parallel to one of the x-gecko's inputs. The machine will run at 70IPM on the diagonal, but Mach3 will not have the duty of synchronising the two axes.

For experimental purposes, you can eliminate backlash by pushing against your y-car & gantry while it is cutting.

Gerald,

Thanks for the suggestion.

Just so I don't screw things up. I want to disconnect the three Y wires at the PMDX and connect them to the same connectors that the X wires are connected.

I am working on trying to find a PC that I can borrow.

Best to "park" the y-car away from the end of the gantry before swopping wires - you don't know for sure which way it will go...

You only need to move the "step" and "dir" wires onto the wires going to a x-gecko next to it (I assume they are G203V's). Leave "the" com where it is.

Good thing I asked.

Could there be a problem with the ribbon cable that runs from the bottom of the Controller Box to the top where the PMDX is located? The ribbon cable is taped to the side of the box.

Gerald,

I could not use the cables that run to the Gecko. They are too short so I attached a small wire and ran that to the X. It did not like it at all.

The X with the new input did not like it at all.

I also recalculated the steps for motor tuning.

30 pinons : 1.5 X Pi = 4.7124

2000 X 7.2 = 14,400

14,400 / 4.7124 = 3055.7677. The old setting was 3057. some thing

I have verified that the ribbon cable is not a problem. I had a regular printer cable so I bypassed the ribbon cable the problem is still there.

I will have to put a file together to try Doug changing diagonal cuts.

I am still thinking of people that might have a computer I can borrow to do a test.

Hey Nils,

From a trouble shooting point of view, frustration is the worst kind of mind set.

I can't count the number of times when I'm trying to solve a problem (with the MM also) and I'm pissed. I do more damage than good. The answer to the problem is staring you in the face and you can't see it for the blinders.

I'm always amazed when I put it away for the evening and return the next morning (or whenever) how easy the problem is to see.

Anyways, don't forget to take a break and clear the head. Now, if you have a project and customer is chomping at the bit, that's different.

You'll figure it out. :)

Nils, I am going to back away from this thread. I find it very frustrating to try and work with you. Example: In post #49 I ask you to test something. 20 minutes later you are onto a different course. An hour later you tell me that you have discarded my test, and that you are onto yet another course.

Helping people to build and run MechMates is not my profession, it is my hobby. Right now that hobby is too demanding and I am pushing the heavy aspects of it onto the backburner.

Gerald,

I am sorry you feel that way. I have ADHD. So that is the way I operate. I jump from one thing to another. It frustrating for me also.

I did not reject your test. I did try it but it appeared one x motor ran one way and the other ran the opposite direction. That was after I attached the Y Dir and Step into the X with the X still connected.

I was concerned that I would damage some thing if I continued to try. So I reported back and in the mean time I tried some thing else. Maybe I did not set the test up correctly.

Thank you for your time.

Nils, tome precauciones antes de actuar con la electrónica.
Las recomendaciones son buenas.
Sigo este hilo con atención.
Me interesan mucho sus pruebas y los resultados obtenidos. Gracias.

The mechanics of your machine seems perfect.
You can put one fine ballpen in the router site and trace some diagonal and circular lines
in different places of the table for see the accuracy.

Good idea Ros.

Nils, if you move the car into the middle of the table with the Z axis all the way up, you shouldn't damage anything during Gerald's recommended test. Just don't cut anything at first. This step is just to confirm that things are working correctly before you try cutting.

So the first step is to move the car into the middle of your table, place the Z axis all the way up, then connect the wires like Gerald told you too. When you use the keys on your keyboard to move the X Axis in any direction, you shouldn't hit a stop for a few seconds. A few seconds will be enough time to react and remove your finger from the button. With the wires connected per Gerald's instructions, the cutting bit should move on a diagonal when you press the X Axis button.

In your last post, you wrote that the motors seemed to turn in opposite directions. Yep. The X axis motors have to since they face in opposite directions on the gantry. That's why you want to put the car in the middle of the table. Like Gerald said, you don't know which way it will move. (You could figure it out but putting the car in the middle of the table is much simpler.)

Okay, back to the test. Once you have confirmed that your X and Y axes are moving together, program your Mechmate to cut a straight line on the X Axis. It will make a diagonal cut. If the cut is smooth, this would seem to indicate that changing the computer might fix your problem. If you get the same pattern as before, this would indicate that changing the resolution might help.

I agree with Greg J too. It seems the more pissed off I get about a problem, the harder it is to solve. Relax and have a barley sandwich like Greg. :)

Doug,

Doug thanks for helping.

This AM, I went back out and tried again. I thought that maybe I had a bad connection. So I reconnected things again according to the what I understand.

What I understand is this: Take the Dir and Step wires from the Y Gecko and connect them to one of the X Gecko Dir and Step. I left the X wires there that are already connected. So I had the X and the Y Dir and Step wires in the same X Gecko.

I was not clear. Things don't move too well. It appears that the one of the gantry with only X is moving in the positive X direction and the other end of the gantry is moving in the negative X direction. They are in conflect. It doesn't make sense to me but it does not work the way it should. Because there is a conflict, I don't keep it running. I stop it.

"connect it in parallel to one of the x-gecko's inputs" I guess I don't understand what this means.

I think I am making the connection incorrectly.

Send me you phone number by PM. May be you can walk me though it.

Thanks

Nils,
Actually you need to move the connections for the Y motor at the pmdx. You want to put it in paralell with one of the x motors pmdx outputs. This way, you are driving the y motor with the same signal as one of your x motors. The connections at the geckos remains untouched. You will have one pmdx output open and one x motor pmdx ouput coneected to one x gecko and one y gecko. The other x pmdx output remains as you had it. Hope this helps. Also see my pm. :)

Heath,

Thanks, that is what I was going to do the first time but I misread Gerald's and thought I needed to do it at the Gecko's. Thanks

I don't know if I'll be much help Nils. This might be a case of the blind leading the blind. I'm no whiz on the electrical stuff but I'm happy to try to assist.

Sorry I didn't answer you before now but I've been working on a research paper all day long.

Thanks for adding your post too Heath.

Take your time Nils. Unless you've got the money to buy a new computer or a Smoothstepper right now, there is no point in rushing through this. You'll just make a mistake or overlook something.

What happened when you changed the wires and made the diagonal cut? Was it smoother?

I looked at a stack of objects I cut out the other day and they had a diagonal cut on them. All of them had a slight pattern. I'd never noticed it before because I always sand them down.

Nils

Sorry to but in especially since I'm not any sort of authority on any of this but it just seams to me that a couple of likely things have been skipped here. The first response to this question was that it is likely a mechanical issue, am I correct???

It would also seam to me that you have inspected the visible mechanics and particularly the bits you built yourself well therefore (by process of elimination) if it's going to be a mechanical issue it's going to be somewhere you haven't looked or can't see my first port of call here would be your gear boxes. Have you tried the test that Gerald suggested in post #47, simple fast effective but you didn't give any feedback (maybe you've done this). Also in post #16 Gerald suggested that this types of thing is accentuated with small bits (I'm assuming lower pressure against the gantry and therefore the motors), Have you re-cut at larger size to allow use of a larger bit and seen the effect?

On the PC thing (and I'm probably not telling you anything new here) all large financial institutions and large corporates rotate their computer every three years, this means that about now they are tossing the last of their P4's often these are sold to small auction houses for next to nothing and then auctioned of (either on the web or in local sales) I managed to pick one of these up recently for $20 in good nick even had a reasonable quality (though old obviously) graphics card in it, so i wouldn't just overlook the option or purchasing one.

Anyway long story short, 90% of what I have said is probably wrong but that's fine the bigger issues is that there seams to be big holes in the tests you have tried (or the feedback supplied to the group). This likely what is fustrating Gerald, to solve these things you will need to be methodical and with all these ideas flying around it's important to cover off what has and has not been tried. Perhaps now would be a good time to go back over the thread summarise the suggestions and detail which have been tried the results, which have not then you can cover off the issues that have not been addressed.

Sorry for the essay.:)

Thanks to Heath, I was able to run Gerald's suggested test cut in post #46. I actually ran it twice but the first time I forgot to set the speed at 50 IPM.

I created a 3" horizontial line, I used a new .5" O Flute End Mill and the material was solid surface .5" thick and the speed was 50 IPM. The 45 degree cut turned out to be 4" in length. I made two cuts.

Both cuts have chatter but they are closer together and what is even stranger to me is that 2" towards the end of the cut it is smoother though you can see fine chatter lines.

Since I have chatter that rules out the computer?

Since I have made changes to the Z Slide, added the quick change plate and I modified the Gantry V Wheels to sit better and I had chatter before I did all this but not as bad, it would seems to me that I have a set up problem or a mechanical problem of some kind.

I am using 30 teeth pinon gears and others are using 30 teeth but are not experiencing chatter. What is different with my machine?

What do you think my next steps should be to resolve my issue.

I also want to thank all those that have been posting ideas to help me resolve this problem.

Thanks

Great input Jed.

Yes, Nils. It seems to me that the fact that you still have chatter would indicate that the computer is fine.

I still think it is a resolution problem. The chatter you keep writing about appears to have a pattern. It isn't random. If you had a loose mechanical connection between two pieces, you might get a random pattern. Or if your wheels weren't seated properly on the rails, you might get a random pattern. If you look back at the line Gerald drew, your pattern looks something like that. In my opinion, a smaller pinion might help somewhat and a Smoothstepper would be the best route.

I think the reason that your chatter got worse after you made those mechanical changes is because you tightened things up. Even though you have "chatter" I bet your cuts are slightly more accurate now than they were. The slop in your machine probably obscured the pattern and it simply appeared to be smoother. Just my guess.

I'm not sure that "no one else has chatter problems." Like me, maybe they hadn't noticed before because it didn't matter to them based on what they are using the machine to cut. In other words, if someone was using their Mechmate to cut out parts for a pocketwatch, the chatter you have in your machine would be a problem. However, if that individual was cutting out cabinet parts with it, the chatter wouldn't be a problem.

Hmm. I just re-read this whole thread. I think what is being stated is:

- Changing the cut speed has no effect on the chatter pattern.
- The chatter pattern got closer together when you ran the post #46 test.

If that is true, I think you just ruled _in_ the computer. To confirm this without another computer, I propose the following:

Go back to the normal gecko wiring.

Change your computer setup to use 1/2 the current gear ratio settings. Since you are at 3056, go to 1528. This should cause your machine to cut everything at 1/2 size and 1/2 speed, so make your test cut twice as long if need be to see clearly.

If the chatter stays the same, it's mechanical. If it changes (gets closer together, perhaps by 1/2), then you have computer troubles, or gcode troubles.
Note that if it's mechanical, it may be that the smaller pinion will get closer to your expectation. If it's computer, smoothstepper might get closer.

Speaking of gcode troubles, for the simple diagonal test case, can you confirm that it has only 8 (or 4) G1 lines, one for each side? If that gcode file is longer than about 20-30 lines, it points towards something funny.

Also, in any future photos, would you include a tape measure? It may help inspire thoughts in some of our visual thinkers.

I know what the problem is, I have had the same issue and resolved it,
Nills if you will put your ADD a side for a few minutes we can get you up and running with greater accuracy,
what is going on is three fold, since you have checked all the physical attributes and are sure every thing is tight, it comes to backlash in the geared motors, I was running the 7.2's and had about 5 to 8 thousands backlash in the gear heads, which is fine in wood no good in aluminum, you need to change the drives to belt drives you can see my posts about this in the Driving Mechanisms, 4 to 1 transmissions for sale? also the larger tooth (30) pinion gears magnify the problem you need to switch the pinions to 20 tooth gears, I would suggest that when going to belt drives you go as big of reduction as possible ( that is why I made the 4 to 1's) the bigger the reduction the smaller the steps the smaller the chatter marks same goes for the pinions.
the rest of the problem lies in the tool head, I believe that you are running the Milwaukee router for this application, I have the same one, you need to get rid of the run out in the tool head, to do this you will need to by new bearings for the router and replace them it is quite simple to do there are threads here an how to do this, but the key is to replace them with better bearings than you currently use, I am on my third set of bearings with about 100 hrs of time on the machine, the bearings that come in the router lasted maybe 8 hours or so, if you are running the factory bearings they are for sure junk by now... I spent more on the bearings than I did buying the router, but by doing this I cut the run out from 8-10 thousands down to 1.5 thousands if you do this you will eliminate most of the chatter and your machine will run quieter with less effort than ever before the same goes for the belt drives once you go belt you wont go back they are far superior, check out the posts good luck//chopper

Brad,

I will try the test later Friday. I did look at the G code and there were 8 G1 moves. 4 for Z and 4 for X.

Chopper,

I am not ready to move to belt drives.

I do believe that runout is a problem. I checked the runout again just the other day. Without the router running, I have 4 thousands and I am sure that it is higher with the motor running. I have sent Chopper a PM requesting the part # for the bearings he is using.

I am sure that the bearings that Milwaukee uses are fine for hand routing. I called them earlier this year and asked them what they considered factory runout to be and they had no idea or they didn't want to share that bad number.

I sent Nils a PM with the info he requested,
but others may not know this that the bearings that come from the factory are not rated for the speed that the router will turn, I believe that the bearings in the Milwaukee were rated for 12,000RPM at best and to accomplish this they add play to the bearings to keep the heat down because the ball bearings are not machined to high tolerances,APDEC 3, by replacing the bearings to a higher APDEC rating the tighter the tolerances get which = less run out,
I am currently using APDEC 7 bearings, I have thought about and actually designed a collet system for my router to eliminate the run out, I am confident that this system will work but I have come to the conclusion that it wouldn't be cost effective, since the price of spindles has come down (Chinese)
to a reasonable price,
the single best improvement I have made to my machine is the belt drives,
the second would be the bearings
//chopper

I don't want to sound argumentative but how could run out in the spindle be responsible for creating a pattern? I don't mean chatter, I mean a uniform pattern like Nils has.

Doug,

Think harmonic resonance.

I cant honestly answer this with precise information, all I know is it works,
I think most of the issue is in back lash, but I also believe that both problems magnify each other,
it is not a matter of argument Doug all opinions are welcome, hell I am no deciding factor or guru, I am just trying to share what I have learned with my cnc you can use it or not just like Nils can decide to use the info or not, all I know is since I switched to belt drives and changed the bearings, my cut quality has improved drastically, the bearings were changed since the photos were taken in the 4:1 transmission posts so the cuts are even better than the ones shown
//chopper

I have had a lot of you ask about the bearings here is the info I have;
the router that I am using is a 5625-29 Milwaukee
the lower bearing is a New Hampshire Ball Bearing
pn: R-104SS KEA7P58 (this is the good one )
www.nhbb.com
tel. 603-924-4100 this is the info off the box,
the original upper bearing is an NSK 608DW this is directly stamped it the bearing
with this info you can cross reference them..
hope this helps,
Charlie
AKA chopper

Doug and Jed

I will try to take a picture of the results on post #63. I am responding, your post came at the same time or close to when I posted my results.

I think runout whether that is caused by the collet and / or the worn router bearings is a factor in the quality of cut. Since I am not a mechanical guy, I don't know if it can been seen as uniform pattern like what Gerald high lighted. One of the reason spindles are used over routers is that they have high level of tolerances in the bearings and collets along with the motors are designed for long running applications. Though routers work, they are not designed to be used in a CNC application and they don't have the tolerances. Thank goodness they do work because most of us can not afford the cost of a spindle.

It is possible that I am asking the Router to provide a quality of cut that it is not capable of giving. I think the MM is capable, more than capable. Since I don't have a background or experience in CNC, I want to be sure I have my machine set up correctly and that it is cutting well. That is why when I found the chatter and it seems to have gottern worse after all my modifications, I posted what I found and asked for help.

At this point, I do not believe the computer is a part of the problem.

I'm not buying the resonance argument when the runout is so small and the pattern looks like Nils' does. Sorry.

I had significant chatter with the original non-geared 1:1 motors on my Shopbot. When I installed a 3:1 belt-drive transmission, most of the chatter was eliminated, but, depending on the material, and the type of cut, there was still some chatter. Finally, I replaced the original motors with 7.2:1 geared motors (Oriental Motors ASM98 Alpha motors). I have about the same amount of chatter with the geared motors as I had with the belt-drive transmission.

Both the belt-drive transmission and the geared motors reduced the amount of chatter, but neither eliminated it.

The harder the material (plastic, solid surface material, aluminum), the more evident the chatter. Even MDF shows significantly more chatter than baltic-birch plywood.

(My machine has a Colombo spindle, so run-out is almost non-existent. Chatter "test" were run using a full sheet of MDF being held in place by a 4-zone two-vacuum (FEIN vacuums) system. None of the cuts went entirely through the material, so material movement was not a factor.)

Richards,
I do not know what is wrong with your machine or what is different about it than mine.
but I went from the 7.2:1 geared motors to the belt drives (mine are 4:1)
I cut primarily aluminum, and the chatter was almost eliminated I had significant backlash with the 7.2's there is no way that the 7.2's are as good as the belt drives at least in my case, maybe it is because your drives are made out of aluminum,plastic, mdf, or some other inferior material other than steel which will flex, or maybe your tolerances in making those drives are to loose I don't know but something is wrong, I am down to 1.5 thousands run out on my router (Milwaukee)
and it appears that is about the amount of chatter I have on my parts,
when I switch out the router for a spindle I believe that I will be close to 0
on chatter marks, currently I have a second machine set up and running the same drives He cuts mdf and is currently having 0 chatter marks while doing so with a porter cable router
I have zero chatter marks in plastic and mdf ( I do not cut much mdf)
the only material that has issue right now is the aluminum and that is my primary material, if I were you and I had a Colombo spindle and some one with a Milwaukee router was getting better or equal cuts I would be looking for what is wrong
//chopper

Chopper, I think you missed the point that Mike has a ShopBot, not an MM, despite the fact that he's one of the most valuable contributors to the forum. So he has some inherent machine differences that likely explain the difference in your experiences.

Brad,
I do not get your reply?
if Mikes machine is not the same then why are we comparing to it?
He may be one of the most valuable contributors to this site but that doesn't mean I cannot disagree with what he says, in fact I am not disagreeing I am just stating my findings with my experiences,
and comparing them to what he (Mike) has stated, now if his machine is so different then explain to me how his findings are relevant?
it seems to me you missed the point....
and some how are upset that I have come to a different conclusion,
these are not personal attacks, digs, or anything else, no disrespect intended or implied, towards you or Mike or anyone else, I came here freely to learn and to share what I have learned that is all.
//chopper

Chopper,

I know that you are trying to sell belt-drive transmissions. I have no argument with that, but lets look at the inherent weakness with any belt-drive transmission. In the fifteen different models that I designed, manufactured and installed on my Shopbot, the weak point was belt flex. A stepper motor can only have a certain amount of "overhang" or pressure/force on the motor's shaft before problems arise. Browning recommends at least 1/4" of flex in the belt when moderate thumb pressure is applied to the belt. That's the amount that I used with the 150-tooth belts that I used. The second weak spot is the hinge bolt. It has to be loose enough to allow vertical movement of the belt-drive transmission, otherwise you would have to have a perfectly "flat" rack.

Between the flex in the belt and the movement allowed by the hinge bolt, any flex in a 1/2" thick aluminum mounting plate would not be measurable.

If you look at the various photos that are posted on this forum that show chatter, you'll see the same pattern and the same severity that I had pre-belt-drive / pre-gearbox. The chatter was still visible after I added belt-drives and after I removed the belt-drives and added the gearbox equipped motors, although the depth of the chatter marks was greatly reduced.

All of the belt-drive transmissions that I tried gave the same results. There was no difference between transmissions that I prototyped using MDF, plywood, 1/2" plastic or 1/2" aluminum. That tells me that the frame is not the weak link in a belt-drive transmission.

I'm sure that you're aware that the $1,700 Alpha motors use a different gearbox than the $257 PK296A2A-SG7.2 motors, although my PK296 motors have remarkably little back-lash, much less than the tech support person at Oriental Motor told me to expect.

I freely admit that the MechMate is a much more robust machine than my PRT-Alpha Shopbot. I wish that I had one, but I only have room for one machine in my shop and I already have a Shopbot.

A CNC machine is made up of a lot of different "systems", i.e., frame, motors, router/spindle, hold-down method, etc. Any one "system", i.e. transmission/gearbox, cannot compensate completely for weaknesses in the other "systems".

Mike,
I make these statements as an observation, I put the belt drives up for sale to help others who could not or cannot make them for their selves,
I am offended that you would suggest that I am pushing this issue to sell drives,
that is not the case, if the belt drives were inferior I would say so, in fact I here by ask Gerald to remove my post from the market section, I will no longer offer them for sale, if you want to build them you are on your own....
I cannot believe that you all get so pissy when someone questions you or has a different conclusion than you, sorry I tried to offer help to someone who asked for it....
I may not be a guru of cnc, but I have been around the block a few times and I know what I know, I will put my drives where my mouth is I will put them up against the oriental 7.2's and they will and do out perform them day in day out
I have tested and proved this over and over ...on two separate machine
//chopper