Please help - missing steps ... Mach3

[JAZZCNC]

What voltage do you have coimng out of the PP.? This problem sounds very much like PP is low power probably 3V type. This will cause the problem you are having because there isn't enough voltage difference between HIGH/LOW.

Also one more thing that can cause this and in some way related to the above is if your Step pulse is on the wrong edge of the pulse. ie: On the down slope(falling edge) when the controller is pulsing on the up slope (rising edge). This affectively puts the timing out and you'll loose one step with every direction change.
To test this for each Axis just write some G-code which moves the axis Back n forth small amounts but 1000's of times very fast.
ie:
G0 x10
x0
x10
x0

Copy and paste into a file to make several 100 lines. Make a Mark before starting and run the file. It should return and if it doesn't then you are on the wrong edge of the pulse.

Some drives will let you select the rising or falling edge, Some won't. To Force or change the edge just toggle the Active Low in Motor outputs. This will reverse the motor direction. So goto Homing and Limits and Click Reversed for that axis.

Try and let us know.?

Or better still do your self and machine a favor Dump the PP and get your self a external motion controller.!! . . . You won't regret it.

[cropwell]

My understanding of the operation of the drivers is that the direction pulse should overlap the step pulse at both ends, so whether direction is set one way or the other, High Low, 0v 5v whatever it should be in a settled state before the step pulse, then whether it is a positive going pulse or negative one, both the leading and trailing edges are within the direction state. I would imagine that if different protocols are used for step and direction then results would be unpredictable. As would happen when the timing of pulses is in error, then direction would not be properly set. - Hence missing steps.
JAZZ - correct me if I am being Naive

[JAZZCNC]

It's more to do with the Controller issuing the step pulse on the rising edge and the drive set to process it on the falling edge there becomes a timing issue so when a direction change occurs the step is in front of the Dir and a step or MS is lost on each change of direction.

Having the wires on the drive wrong way around will also cause this. So Step + being connected to Step- on BOB. Also a BOB with slow Opto's could cause this but seen has the OP tried direct to drives I reckon not.!

[Dragonfly]

Emil,
is this happening only to the Y axis?
And I didn't see what drivers you are using.

P.S. Saw the drivers in the first post.

[cropwell]

It's more to do with the Controller issuing the step pulse on the rising edge and the drive set to process it on the falling edge there becomes a timing issue so when a direction change occurs the step is in front of the Dir and a step or MS is lost on each change of direction.

I think that is what I said i.e direction not settled before step.

[JAZZCNC]

I think that is what I said i.e direction not settled before step.

I didn't read it that way at first but reading again then yes same thing really.!! Time is shortened and next signal comes along before gets chance to complete last thing so something gets dropped at each direction change.! . . . . . I've eaten since then so my brains fired up again. . Lol

[emilvv]

Hello,

THANK YOU for all of You.

Problem is not resolved ... :(

Yesterday i tried with LinuxCNC live DVD . 2.7.

It is easy to try but the problem is the same !

I moved only one ( Y ) axe because I want to fix the problem.

I moved only in one direction .... from Y = 0 to Y = 600 ....

When I moved Y ... after 200 mm ...300 mm I am receiving that motor knock ...

In Linuxcnc - the same situation ... I tried many configurations .... for STEP pulses, driver setup ( I mean microstepping modes ... )

A little bit better on slow speeds ..... and 400 steps per / rev .... but that is not good idea ..

I have to check LPT voltage .. Also will try to drive the drivers by arduino .. to change speed and to prove driver quality ...

Any ideas are welcome ....

Thank you in advance,

Emil

[Clive S]

Well first of all you seem to be jumping all over the place and not concentrating on trying to sort the problem out in a logical way. It is no use just trying random changes. So take it one step at a time giving clear and concise feedback.
.
Some good pictures of the machine would help, are you using ball screws or screwed rod etc? You have already been asked to check the pp voltage out put. ie is it 3.3V or 5V but you have not answered. This might just be a simple pp and bob problem.
.
Sorry to be a bit harsh. ..Clive

[emilvv]

Hi,

Problem solved. :) Thank you for all your help.

The PC LPT port generate strange pulses .... from 4000 ns to 300 000 ns.... ( based on linuxcnc tests .. )

Why I not saw the problem ..

I tested all the electronics with my another PC ... ( also with the same problem .... but a little bit better !!! ), I used another PC with CNC machine for manu years, but just for some small cuts .. with perfect shape ... )

So, I suppose my first PC is OK .... with both PC s .. the problem exists .... so the problem is in the electronics ....

But when I tested it with NE555 generator .. motors are fine !!! So ... I became almost with no idea what to do ..

I grounded all, used 74HCT245 ... 6N137 ... to change the logic voltage levels .... and finally on the last started the linuxcnc
for N-th time and saw strange situation ....

After start i have 4000 ns step period ... and after a while ... 1 second ... it increased to 300 000 ns.....
after reset statistics button .... the same situation ...

Mach3 doesn't see the wrong signals .... message is "System excellent" !!! :(

I didn't investigate to check signals by scope in time period ... I have no time for that ...

So, please fine attached first cut picture - 4 mm carbide tool in wood ..... Sorry I am not woodmaker ... it is just a test

Have a nice day to all of you,

Emil

[komatias]

Please do not be offended by this:

You are wasting your time the parallel port! You should seriously consider buying a motion control board like we have all suggested previously.