Just put the finishing touches on my converted mini-mill. Using a Hobbycnc 4AUPC board and TCNC 4.01. Mill is set up with 5TPI ball screws, controller is 1/8 step.

I've got a simple stick font engraving file that I'm using to test. It engraves a single line of text .100' high. When I run the file the, the text doesn't end up being on a straight line it kinda rides a wave. Some letters like G, O & S are compressed or smaller than the other letters like L, F, I, etc. At first, I thought it was the backlash comp. I went and set that up on all 3 axis using a DTI to tweak it in on all 3 axiz. It didn't make a difference.

I'm running 4.01 on a Pentium laptop, booting from a DOS floppy. I disabled all the power management in BIOS. Made sure the 'Precision' was set to 0 to eliminate rounding. Reverse axis delay is 100ms, ARC factor is 1.

I'm at a loss as to what the problem is. I know the code is good as I've used the exact same code to do the work on a Bridgeport. I'm not running anywhere near the limits on accel, start & max vel and at these small movements, I doubt the max speeds would come into play.

Any suggestions?

Tried with a desktop PC? We have several similar threads in the past... the voltage from the parallel port may be too low.

Tried with a desktop PC? We have several similar threads in the past... the voltage from the parallel port may be too low.

Thanks. I checked it tonite, port voltage is 3.3V on a H, too low for a valid HI state. I recall seeing some kind of adapter or port board mentioned somewhere that would correct the problem. Hopefully I can find that reference again.

Tried a desktop PC tonite. It made no difference. Going to go witha new copy of Turbocnc and try reposting the code. Maybe a rounding error??

Did you check the voltage on the desktop port? Sound like lost steps. What size motors? any noticeable binding?

Did you check the voltage on the desktop port? Sound like lost steps. What size motors? any noticeable binding?

I did check the voltage on the port, with the laptop it was 3.3V so I went with a desktop computer. Same results. I've got Vexta PH296 NEMA 34 motors. I checked the gibs, Im using 5/8" ball screws, I can turn the motor couplers with my fingers and move the X & Y axis.

Tonite I put a 2" DI on the X-axis and I've slowed the accel, start vel and max vel way down. 75 Start, 750 Accel, 2500 Max. That seemed to have made the difference. I measured the backlash to be .0035" and set it to that. Using a 2" DI, I can go <alt-g> from 0-1" and it will move 1.000". If I go 0-.5", it really goes to 0.48" if I go 0-2.0" it really goes to 2.002". So I'm short .002" on a .5 " move and .002" over on a 2" move. Is this normal error? Any way to correct for it? When I go to 0" from whereever I'm at, it does return to 0" each time on the DI.

Previous settings were 1000 Start, 3000 Accel, 9000 Max, this was causing missed steps.

What voltage arre you running those motors at? What the rated current of those motors? I looked at the torque charts, and depending on which model you have, the torque drops to less than 100 oz-in by 200-300rpm. What is the model? 296-02BA?

not sure if this is touching base with your prob or not i had a sim prob i was using a star as pattern a couple ot the legs of it were wavy shaped now for the answer ........i had a lead screw that wasn't straight :( replaced the lead screw for the x axis and prob dissappered not sure if it your prob just an idea to check

I did check the voltage on the port, with the laptop it was 3.3V so I went with a desktop computer. Same results. I've got Vexta PH296 NEMA 34 motors. I checked the gibs, Im using 5/8" ball screws, I can turn the motor couplers with my fingers and move the X & Y axis.

Tonite I put a 2" DI on the X-axis and I've slowed the accel, start vel and max vel way down. 75 Start, 750 Accel, 2500 Max. That seemed to have made the difference. I measured the backlash to be .0035" and set it to that. Using a 2" DI, I can go <alt-g> from 0-1" and it will move 1.000". If I go 0-.5", it really goes to 0.48" if I go 0-2.0" it really goes to 2.002". So I'm short .002" on a .5 " move and .002" over on a 2" move. Is this normal error? Any way to correct for it? When I go to 0" from whereever I'm at, it does return to 0" each time on the DI.

Previous settings were 1000 Start, 3000 Accel, 9000 Max, this was causing missed steps.

What voltage are you using with what steppers? Performance seems awful low.
Dave Rigotti
HobbyCNC.com

"I measured the backlash to be .0035" and set it to that. Using a 2" DI, I can go <alt-g> from 0-1" and it will move 1.000". If I go 0-.5", it really goes to 0.48" if I go 0-2.0" it really goes to 2.002". So I'm short .002" on a .5 " move and .002" over on a 2" "


are you talking on the the indicator or on the program??? if you program is showing right and your indicator is off ....... this shows more of a prob of the bent screw

What voltage are you using with what steppers? Performance seems awful low.
Dave Rigotti
HobbyCNC.com

Dave: Using a 28V P/S. 15A

ger21: PH296-02 steppers that are 5.5V 1.25A. They are rated at 174oz holding. If you have a link to the torque curves, let me know.

Strat. The error shows in the DI not what the software indicates. Software is functiong fine. I won't rule out bent lead screws or screw error but they were new Roton stock and I turned the ends in a Hardinge collet lathe. Roton hand fitted the ballnuts. Using motor oil for lube, not any heavy grease. I have a teflon light white grease I'll apply when I'm done. Oh, I made new gib strips from Si/Al bronze.

When I"m checking the backlash, I just set the jog increment to .0005 and jog left & right & tweak the software until there is ever so slight movement in the needle. (.0001" or less) I'll stick a DTI on it and fine tune it better than that when I've worked the other bugs out.

I did have my feed rate in the test program set to F10. It seems that the letters like 'S', 'G' that had alot of interpolation were the ones getting compressed. That error had to be missed steps. Haven't run the program since I've been playing with reduced speeds. Does the accel rate get taken into account on every move? I didn't have any success with the G28 Home command and it may be because the Max speed was too high and the Start & Accel don't get taken into account on a G28 command?

Thanks for all the help folks!

I misread your pevious post and thought you had PK series motors. I think those motors just don't have enough power, and need to use a much higher voltage to use the power they do have. I couldn't find a torque curve for those.

Max speed is used for rapid moves only. Unless you try to cut at an F faster than the Max speed. Yes, every move accelrates and decelerates. You might want to try a much lower start speed, say, 25, and a then slowly increase the acceleration until you notice missed steps. You may be able to increase the max speed as well.

I think the bottom line, though, is you need more voltage and / or bigger motors. More voltage alone may not be enough. Perhaps Dave can give you more info.

Update:

Spent the afternoon playing with the setup. The problem was indeed missed steps. I lowered the rates and it did the job albiet realllllly slow. Guess I'll need larger motors to run higher rates.

Still have three other problems:

Using a DI, when I jog using the Alt-G command, from 0-2" I don't get 2" of travel, I'm about .004" shy. All I can figure at this point is since the ball screws were fitted for minimal backlash, that the thread pitch has changed ever so slightly. I suspect I could correct this in software? Suggestions?

Secondly, how is the best way to measure & adjust backlash? Been trying to jog +\- .0005" and adjust the backlash so that I get .0005" of movement. Don't seem to be having much success.

In the jog mode, I can jog the axis by using the discrete mode but the continous mode doesn't work, it just kinda grunts.


Lastly, the G28 home function won't work. I've got my N/O home switches wired to 10,12,13 & Gnd. Set the ports to active low, "Home is Limit" enabled. When I enter the G28 command, it sounds like the controller trys to move the motors, they kinda grunt but nothing happens. Also, when I'm running a program, it randomly stops due to a "limit error" yet no limit switch is activated. Using the port monitor, I can see the switches change states when manually activated. I may post this as a seperate subject in a new post. Thanks.

Using a DI, when I jog using the Alt-G command, from 0-2" I don't get 2" of travel, I'm about .004" shy. All I can figure at this point is since the ball screws were fitted for minimal backlash, that the thread pitch has changed ever so slightly. I suspect I could correct this in software? Suggestions?

Secondly, how is the best way to measure & adjust backlash? Been trying to jog +\- .0005" and adjust the backlash so that I get .0005" of movement. Don't seem to be having much success.



If you're not using ground ballscrews, they can vary slightly as you move along them. Try the same 0-2" test at different parts of the screw, and see if the results are different in different places. This might tell you if it's screw error.

Do you mean you're adjusting backlash on the nut itself, when you're trying to get .0005? Or using software backlash comp?

If you're not using ground ballscrews, they can vary slightly as you move along them. Try the same 0-2" test at different parts of the screw, and see if the results are different in different places. This might tell you if it's screw error.

Do you mean you're adjusting backlash on the nut itself, when you're trying to get .0005? Or using software backlash comp?

Software backlash comp. No room for a second 5/8 ballnut to adj. mechanical backlash.

I don't think backlash comp works during jogging. And even if you MDI a single axis move, it probably doesn't work there either. I don't know the specifics, but I believe it only comes into play when an axis changes direction.

I forgot to mention, regarding uneven screw lead. Mach3 now has screw mapping, which let's you compensate for uneven screw lead. Not sure how complicated it is to set up, but it will (should, anyway) help to make machines more accurate. Provided you can run Mach3. :)

I don't think backlash comp works during jogging. And even if you MDI a single axis move, it probably doesn't work there either. I don't know the specifics, but I believe it only comes into play when an axis changes direction.

I forgot to mention, regarding uneven screw lead. Mach3 now has screw mapping, which let's you compensate for uneven screw lead. Not sure how complicated it is to set up, but it will (should, anyway) help to make machines more accurate. Provided you can run Mach3. :)

Backlash comp does work during jogging.

Going to back figure and see if I can correct for the screw but I need to get the backlash comp correct first so It's not introducing error. Mach3 would be nice but I'm using an old laptop and that's all I have room for right now, Windoze is very slow on the old laptop. Many new laptops don't have parallel ports, you have to use a USB port replicator which is a pita in itself.

So if you make one jog move, then the backlash comp doesn't do anything, but if you then jog in the other direction, it then compensates? Just trying to understand how it works.

If you're screw is off by the same amount in different locations, you should be able to adjust out the error by playing with the step size.

So if you make one jog move, then the backlash comp doesn't do anything, but if you then jog in the other direction, it then compensates? Just trying to understand how it works.

If you're screw is off by the same amount in different locations, you should be able to adjust out the error by playing with the step size.


Correct. If I keep jogging to -X, it simply moves, if I jog +X, I'll get a backlash comp movement+jog then it will continue to jog until the next direction change.