[migrusch]

Ok, while I am waiting for the new break-out board, I need some help from the members on this forum.....

I want to fix my problem with the PWM output from Mach3. I had this working fine and then it mysteriously stopped.
I believe that this is nothing to do with the break-out board, and here is why, when I run the A axis outputs are all fine, when I turn off A axis and use one of these pins for PWM, I get nothing (same with the original B axis pin I was using). This is with everything unplugged.

Have played with all the settings I can find in Mach3, but still no output.

Found some posts with similar issues, but none with a resolution.

Anyone have some tips?

[migrusch]

So I haven't replaced the breakout board yet, but have been playing with the machine quite a bit.
Fixed some settings in Mach3 that were making the machine run rough, especially the CV speed setting (not CV mode). Now sounds much better during curves.

Anyway, there is a part I had tried to cut but faced issues with losing steps.
Have been cutting the same file in air to confirm operation and still losing position in just Y axis no matter what setting I used.

Decided to simplify things and just perform a 2" dia circle at 16IPM.
Now I can hear and see the spot where the steps are lost.

Confirmed that it is a stall, not slipping (I can see the stepper shaft also stop)

Changed over the X and Y drives (from driver cards to motors) and the Y axis (which is now driven from the X) still has the same issue.
So it's not software or driver card etc.

Must be a mechanical stall!

Seems to happen in a similar part of the table.

I don't have any problems with rapid moves in Y only, just when performing the curve.

I started looking at the Gib adjustment. This leads to a big announcement! The base to my machine appears to be an 'RF30' not 'RF45'.... need to update the thread title!
This means only a single adjustment screw.

I was never happy with the Y axis Gib. It seems to need to be adjusted a long way out to operate. Also the screw head is loose in the slot of the gib and you can see movement of the screw (side to side) when the machine moved back and forth. I did helicoil the screw hole, hoping to sure up the screw, but to no avail. I think the main issue here is that the screw needs more engagement (needs to be set in further) but something is stopping this.....

Now that the steppers are hooked up and I don't have a hand wheel, it is hard to 'feel' the tightness.

I guess the next step is a strip down, clean and good look at the Y axis. Maybe some scraping will be in order. Guess I need some transfer dye?
Still seems strange that this only shows up when I am performing a curve? Could the X axis be rotating the table and jamming it?

[dick cnc]

Hello I just took a look back at your control console that looks like a specialty rackmount PC that's based on laptop if it is no matter how much you clean it up and take out everything running in the background. Mach 3 it is going to give you all sorts of strange problems. I know it doesn't make much sense and they say it's because of the architecture. First Mach 3 CNC router I built I tried to use a laptop. And had strange problems similar to yours but not constant. But it did seem to be in the same place. Actually I use that same laptop but with a smooth stepper on my little lathe and it works fine

[migrusch]

Hello I just took a look back at your control console that looks like a specialty rackmount PC that's based on laptop if it is no matter how much you clean it up and take out everything running in the background. Mach 3 it is going to give you all sorts of strange problems. I know it doesn't make much sense and they say it's because of the architecture. First Mach 3 CNC router I built I tried to use a laptop. And had strange problems similar to yours but not constant. But it did seem to be in the same place. Actually I use that same laptop but with a smooth stepper on my little lathe and it works fine

Hi Dick,

Actualy the PC is a regular ATX Intel PC.
The keyboard and screen are the type you find in a sever room, it just looks like a laptop.

Regardless I will be re-installing the latest version of Mach3.

The reason I don't think it is on the PC \ electrical side is that the stall occurs on the same axis, regardless of what axis the PC thinks it is diving.
Also, the more I think about it, the Gib isn't right. The rear of the screw head should sit inside the counterbore......

[dick cnc]

Sorry I just noticed what you have is Mac. This disregard my previous post I and I do not know anything about Mac or Apple

[migrusch]

Not mac... mach

Sent from my GT-I9100T using Tapatalk

[dick cnc]

Okay so it is a regular ATX pc. Everything should be okay but if you have another old one laying around might just want to plug some software into it and give it a try. I don't think it's mechanical since it Rapids okay and that's when torque is at its lowest point. The only mechanical thing would be if the ballscrew was askew on an angle and bound up closer to the mounts. But that would do it while jogging to. It was the top unit in your rackmount that threw me off when I thought it was a Mac

[migrusch]

Have to admit the Mach3 version may not be 'OK'. So will start there, then try another PC if I need to.

[Fastest1]

I have had corrupted Mach files appear from seemingly out of thin air. A reinstall usually solves it.

[migrusch]

Tore down the table to fix looseness in the Y nut mount. Fixed a few other things whie I was there.

Re installed Mach3 and things are better. ... bit still having issues with circles.

Left the x axis stepper off and can see the stall or hesitation clearly with no load.

Next step is to install the new breakout board. Then I am going t break out thr oscilloscope and trace my problem back, starting at the stepper.

Sent from my GT-I9100T using Tapatalk

[dick cnc]

What software are you using for CAD and CAM some of the low-budget software does strange things. If you want to check out a circle pocket using the wizard in Mach 3. Make sure you don't have backlash compensation turned on. If Mach 3 wizard works okay. Do the same size pocket with your cam software and compared G code

[migrusch]

Solid works for cad and hsm for cam.

Now I have simplified to just G3 commands and still having issues..

Sent from my GT-I9100T using Tapatalk

[dick cnc]

Just for the heck of it I would try a few of the wizards in Mach 3 actually very useful. In this way rule out any possible problems with your CAD/CAM

[bill south]

Migrusch;
Are you using a smoothstepper? I was having problem with my PWM controller and found that I had not set up PWM operation in my smoothstepper software. I am using a ESS and yours may be different. Just a thought.
Hope you get it going OK. Nice job on the build.
Bill

[dick cnc]

Solid works for cad and hsm for cam.

Now I have simplified to just G3 commands and still having issues..

Sent from my GT-I9100T using Tapatalk

I am not familiar with HSM but have read that a few people have had problems using Mach 3 with HSM setting up the postprocessor. Again if you use the wizards in Mach 3 you'll bypass all of this to prove it out

[migrusch]

Ok, I finally solved my problems and am much much happier with the machine operation.
Just incase any of this might help someone else I will outline what steps I went trough.


1. BREAKOUT BOARD

I had initially (a while back now) tried to use a very very cheap Chinese one off eBay. I think I paid $12 with free shipping.
This was absolute junk that had clearly never been tested. The seller sent out another one for free, but this was designed an manufactured is the same half baked manner.
Later I decided that these were poor copies of other Chinese ones and It seemed clear that the person doing the copying didn't have a very good grasp on electronics theorey.

Next I bought a 'more expensive' Chinese board from eBay. This seemed better in construction and layout, but still left me unconvinced. I mostly took issue with the power supply which didn't have any capacitors etc for stabilisation \ noise filtering. As I said in an earlier post, the USB supply from the PC seemed to be insufficient to drive the board (plus 2 additional off board 5v relays).

Finally I bought a locally (Australia) manufactured board from this company (Welcome to Ocean Controls). It's still just a simple PP BOB, but much better design, layout and construction. This one also included a charge pump circuit, which is very good safety feature as I have had issues with the spindle starting up during PC reset etc. I was very happy with this company's product and service, so recommended for those living in AU \ NZ!

At the same time as replacing the BOB, I put in a new PP cable. The one I was using came with the first super cheap board, when I cut it open I was supprised just how thin the wires were, and how ordinary the shielding was. I am now using an industrial style cable.

Result: My flood output is now rock solid and stable, but still having lost steps etc during G2\G3 moves.....


2. MACHINE SOFTWARE

I performed a fresh install of XP following all of the optimisation steps. Installed the latest version of Mach3. The results from the driver test did improve with far less interruptions to the pulse train.

The problem however was that the pulse count on the diagnostics page was all over the place. I was running a 25k kernel, but saw numbers like 1, 40, 250, not even close. Clearly these were not the actual pulse frequency numbers, other wise I wouldn't be able to rapid etc. However I was still having issues with G2\G3 moves and whilst this problem was there I couldn't rule it out. After lots of forum searching I tried some CPU 'killer' programs as these seem to have helped some people, but did nothing for me.

Finally I turned to the dark side..... Linux CNC. I have used a number of Linux distros before, but I am by no means a Linux person. However, this experience has pushed me closer to loving it.

Install was so quick and painless. Around 5 min and I was up and running! compare that to an XP install. Then I used the wizard to create a settings file (another 2 min). Now I had a working software package. Straight away it seemed clear that this system was producing more stable PP signals. Also, I liked altering all the settings from the .ini and .hal files. With a programming background this just made more sense to me that Mach3's multiple menus with radio buttons, tick boxes etc.

I am happier with Linux CNC than Mach3 and will leave it this way. However, I was still having issues during curves...


3. STEPPER DRIVER

I had been testing my issue by running MDI like the following (mm units):

F200
G1
X0 Y0
G3 X0 Y0 I0 J20

What I noticed is that the axis would appear to pause \ jump around 5mm from the end of the travel.
This was similar every run, but not exact. Also, I observed this even when the stepper was disconnected from the leadscrew.
Playing with velocity and acceleration seemed to have a minor impact, but never fixed it.

I had never had any issues whilst performing rapid jogs in any direction, or with any combo or axis.
However, I decided to perform jogs in one axis and use the feed rate override (slider bar on LinuxCNC) to sweep slowly through the operation range.
And there it was, I found the issue. Low band mechanical resonance (aprox 120-150Hz).

From here I did a bit of reading on the finer details of stepper motors (more in depth than I had previously cared about). From this it suddenly made more sense why micro stepping was important. I had always only looked at half\micro stepping as a feature for accuracy. On the stepper drivers I built, I had only selected full step as I didn't feel that the extra resolution was required, however lots of reading has shown me that in part, half stepping is a method to overcome mechanical resonance and provide smoother operation. This is not realy an issue for systems that only do rapid work and accelerate through the resonant band. however the nature of G2\G3 moves is that the feed on each axis mus vary with a cosine type form.

After I selected the half step line on each stepper drive, BANG, issue gone. much better sound too from the steppers.... more musical.

Now I am ready to build the walls to keep the chips in and start makings some stuff.

[dick cnc]

Glad to hear you got it all straightened out. Show us some videos of it cutting if you can. I would give Mach 3 another try since it really wasn't the fault of Mach 3. I was going to ask you if you were using micro-stepping I guess I should've.

[Fastest1]

Some of us are running up to 10x microstepping.

[migrusch]

I finally finished a part. This is a spanner for my spindle.

There are lots of things which still need optimisation. Mostly this is from the CAM program.
I am using HSM and am very happy, but need to set finish passes etc.

I only just got away with this part, since I ran out of super glue and had to use double sided tape. Found out this melts in contact with WD40 which is what I am running through the fog buster copy.

Next major thing is to build the walls to keep the chips contained.

Attachment 214640

Attachment 214642

[SCzEngrgGroup]

...WD40 which is what I am running through the fog buster copy.

That sounds to me like a REALLY bad idea! WD-40 has a flash point of only 120F, and breathing the vapor is not good for you. FogBusters can put a LOT of vapor into the air. If it ever ignites, you could be in serious trouble....

Regards,
Ray L.