Latency, Jitters & Lost Steps

[Imagineering]

I've read a few posts which mention that choice of hardware may be causing lost steps.
I'm experiencing lost steps in both Y & Z using an old 486 Toshiba Laptop, and as I have a choice of Computers to use with TurboCNC I'm in a quandary as to which one to use.
I've listed below, the spec's that TurboCNC gives on startup. These spec's are almost meaningless to me and I'd like to know which of these Computers would be the best one to use?



OLD Toshiba Laptop, (preferred Computer)
Timer Latency; 11.51 Microseconds
Max is; 431.00
Min is; 11.00
10% Jitter is; 10363.0 Hz

Pentium 166 ( 2nd choice Computer)
Timer Latency; 3.87 Microseconds
Max is; 116.00
Min is; 3.00
10% Jitter is; 30833.9 Hz

Compaq Deskpro
Timer Latency; 4.01 Microseconds
Max is; 69.00
Min is; 3.00
10% Jitter is; 29775.9 Hz

Thanks in advance.

[abasir]

Either the P166 or the Compaq.

The laptop may have low voltage on the parallel port which may cause some problem with CNC controller and the step rate is limited to about 10kHz. With 4 microstep/step, 200step/rev & 10tpi leadscrew, you max rapid is only 75ipm. The other PCs can triple that theoritical maximum to about 225ipm.

[Imagineering]

Either the P166 or the Compaq.

The laptop may have low voltage on the parallel port which may cause some problem with CNC controller and the step rate is limited to about 10kHz. With 4 microstep/step, 200step/rev & 10tpi leadscrew, you max rapid is only 75ipm. The other PCs can triple that theoritical maximum to about 225ipm.

Thanks for that, I'll setup the Pentium 166 tomorrow & see what happens.
I'm actually *Gaining* steps on my Z-Axis not losing them.
I'm using DeskEngrave to write a program with the word TEST, (original huh), and it appears that the Z-Axis advances in the -Z direction between the {+Z, Rapid, -Z sequence} between letters, and then slowly creeps in the -Z direction when cutting.
BTW the lead screws are M8x1.25 pitch which is about 20 TPI so therefore much slower than your estimate above. Also I'm Full-stepping not Micro-stepping because of my fine pitch.

Can I use my original .INI file from the laptop and then adjust from there??

[ger21]

You're not gaining steps, your losing steps in the +Z direction. You should try microstepping, it may eliminate the lost steps, because the motors should run smoother. Also, the ini file should be the same. It's not computer dependant, it's machine dependant.

[rustyolddo]

Either the P166 or the Compaq.

The laptop may have low voltage on the parallel port which may cause some problem with CNC controller and the step rate is limited to about 10kHz. With 4 microstep/step, 200step/rev & 10tpi leadscrew, you max rapid is only 75ipm. The other PCs can triple that theoritical maximum to about 225ipm.

Could you explain a bit more on how to interpret the numbers?

[ger21]

10,000 steps per second from the PC. 200x4x10 = 8,000 steps per inch.
10,000 x 60 seconds / 8000 = 75ipm.


BUT, the speed that TCNC reports when it starts is not always a true indication of the speed you can achieve. I've heard of people getting double the reported speed.

[Imagineering]

You're not gaining steps, your losing steps in the +Z direction. You should try microstepping, it may eliminate the lost steps, because the motors should run smoother. Also, the ini file should be the same. It's not computer dependant, it's machine dependant.

I used the word *Gain* as what I want is;
{Z -1 to a Rapid height of Z +3, then G00 X Y to the next letter, then plunginging back down to Z-1}
But what I get is;
{Z -1 to Z +3, G00 X Y, it then goes to Z-2 through -4}.

So I thought that -1 +3 -4 is a *Gain* of 3 steps. could it be interference cross talk from one of the other Axis Controller leads perhaps??

I'm about to go and try the Pentium with the original .INI file & see what happens, just to eliminate the Laptop as a possible cause.

[Imagineering]

Back again;
The change of computer to the Pentium 166 has appeared to have solved the issue of lost steps. I ran my test program with a rapid height of 3mm & it worked OK, so I then ran it again with a rapid height of 20mm and this worked as well. I marked the Z-Axis shaft and arranged a pointer to verify this. I'll set up a dial gauge later on today to test it to a better degree of accuracy.

The next problem is 'jagged curves', specifically the letter 'S' and the number '2'.

I've created my program using DeskEngrave with 'TEST2' as the text. The letters are 20mm high Arial Font.
I'm including both, a .jpg of the problem and a copy of my G-Code.
I guess there may be 3 possible causes that I can see;
1, That I still have a Computer/CNC Machine glitch, or
2, That the program code generated by DeskEngrave doesn't interpolate curves very well, or
3, A combination of 1 & 2

Would someone please have a look at the G-Code and enlighten me on this one.
I've already had a play with the arc function under 'General Setup', but this doesn't make any difference to the jaggys, but with a setting of 0.8 the Z-Axis starts twitching.

[ger21]

It's the g-code. There are no curves in the g-code.

[Imagineering]

It's the g-code. There are no curves in the g-code.

Thanks for that, I had a play with the G02 Circular Interpolation code and engraved a perfect circle about an hour after my post, so I expected that the answer lay in the G-Code.
Is this the best that DeskEngrave gives or can the Program give a better curve than this by changing the parameters within DeskEngrave??

Having used the G02 Command, I am now a little more confused.

I did this, as outlined in the TurboCNC Docs;
G17 (plane XY specified for clarity)
G71 ; Metric units
G90 ; Absolute Coordinates
G00 X0 Y0 (get into start position)
G02 X2 Y0 I1 J0 F4
And engraved a perfect little semi circle 2mm dia

The explanation of G02 in the Doc File is;
Quote/Moves from the current point to (0,0) in an arc with its center at the current point + 1 unit in the X direction at 4 units/minute. Absolute mode assumed. The "I" parameter is associated with the "X" axis, and the "J" parameter is associated with the "Y" axis./Quote

So I changed the parameter 'I' to 10 expecting a semicircle of 20mm dia;
G17 (plane XY specified for clarity)
G71 ; Metric units
G90 ; Absolute Coordinates
G00 X0 Y0 (get into start position)
G02 X2 Y0 I10 J0 F4
And engraved a perfect *Circle* 20mm dia ????

I thought from the explanation in the docs that changing 'I' would change the Radius of the circle, but this also changed the path from 180 degrees to 360 degrees.

I know that I'm on the bottom of a steep learning curve here and any help is very appreciated.

[ger21]

The X and Y in the G2 line are the endpoints of the arc. If you don't change the endpoint, but move the center, then you'll get a different arc. You needed to change the G02 X2 to G2 X20

[rustyolddo]

Back again;
The change of computer to the Pentium 166 has appeared to have solved the issue of lost steps. I ran my test program with a rapid height of 3mm & it worked OK, so I then ran it again with a rapid height of 20mm and this worked as well. I marked the Z-Axis shaft and arranged a pointer to verify this. I'll set up a dial gauge later on today to test it to a better degree of accuracy.

The next problem is 'jagged curves', specifically the letter 'S' and the number '2'.

I've created my program using DeskEngrave with 'TEST2' as the text. The letters are 20mm high Arial Font.
I'm including both, a .jpg of the problem and a copy of my G-Code.
I guess there may be 3 possible causes that I can see;
1, That I still have a Computer/CNC Machine glitch, or
2, That the program code generated by DeskEngrave doesn't interpolate curves very well, or
3, A combination of 1 & 2

Would someone please have a look at the G-Code and enlighten me on this one.
I've already had a play with the arc function under 'General Setup', but this doesn't make any difference to the jaggys, but with a setting of 0.8 the Z-Axis starts twitching.

Go into Deskengrave, Create/Set Parameters Set your decimal places to 4. Retype and re create your text & g-code. You'll have nice smooth curves.