Leadscrew pitch and backlash

[Festivejelly]

I finally got my build together (video incomming)

But I noticed while the X axis is pretty much spot on the Z axis is not and is quite a bit out.

I measure the pitch of the leadscrew on the Z axis and as far as I can tell its 2mm.

So here are my settings

// Main lead screw (Z) parameters.
const long SCREW_Z_DU = 20000; // 2mm lead screw in deci-microns (10^-7 of a meter)
const long MOTOR_STEPS_Z = 200;
const long SPEED_START_Z = 2 * MOTOR_STEPS_Z; // Initial speed of a motor, steps / second.
const long ACCELERATION_Z = 30 * MOTOR_STEPS_Z; // Acceleration of a motor, steps / second ^ 2.
const long SPEED_MANUAL_MOVE_Z = 6 * MOTOR_STEPS_Z; // Maximum speed of a motor during manual move, steps / second.
const bool INVERT_Z = false; // change (true/false) if the carriage moves e.g. "left" when you press "right".
const bool NEEDS_REST_Z = true; // Set to false for closed-loop drivers, true for open-loop.
const long MAX_TRAVEL_MM_Z = 300; // Lathe bed doesn't allow to travel more than this in one go, 30cm / ~1 foot
const long BACKLASH_DU_Z = 6500; // 0.65mm backlash in deci-microns (10^-7 of a meter)
const char NAME_Z = 'Z'; // Text shown on screen before axis position value, GCode axis name

Could it be that the backlash is set too high or too low?

I also noticed that the RPM jumps around a little. Only by 2 - 5 RPM, I guess this is inevitable?

All in all though it seems to work super nice, I just need to get the axis moving accurately. I do have a DRO which is pretty accurate so maybe I can use that to calibrate some how?

[kachurovskiy]

One thing to keep in mind, especially for machines that were heavily used, is that leadscrew has variable wear in different parts so it would be a natural limitation of what can be achieved with H4 in both single and bi-directional movement accuracy. With that out of the way:

1. Ensure that moving in 1 direction gives you accurate results. E.g. set a dial indicator so that it measures movement along the desired axis, move a bit to the left, zero indicator, move A mm to the left using H4, check whether the dial indicator shows A. If it's consistently different, SCREW_Z_DU likely needs to be adjusted.
2. To calculate backlash, set dial indicator so that it measures movement along the desired axis, push the carriage with your hands in one direction, zero indicator, push in the other direction, the dial indicator reading is the backlash.

DRO replaces the dial indicator nicely.

[Festivejelly]

Ill do a short video. Im going to do some adjustments someone suggestion on the Warco FB page first. I want to rule out any mechanical issues before I do anything because I think maybe the screw could be moving slightly left and right. So maybe I need to torque down the nuts and each end so that it puts a little more pressure on the thrust bearings. That said im a bit concerned about doing it on the stepper side as I dont want to over torque the plastic gear.

[Festivejelly]

Here is a video https://youtu.be/6VZD0DE_qlY?si=wypiY43MEmuZZ7Lj

i cant observe the leadscrew wobbling and it looks like 2mm to me…

[Festivejelly]

Also i put a dial indicator on the carriage against the chuck and its reading the same as the dro.

[kachurovskiy]

You might want to measure the leadscrew pitch in a more precise manner e.g. by measuring 10 thread turns and dividing the result by 10.

Seems like setting SCREW_Z_DU to 17500 should solve the issue.

[Festivejelly]

I measured over 35mm and there were 17 threads which equals 2. So im really stumped. I marked the leaderscrew and turned it a whole turn by using a spanner and the carriage moved 1.8mm So I wonder if my half nut is buggered.

[Festivejelly]

So something odd is going on. When I move small amounts it moves too little. But if I tell it to move 30mm it actually moves 32mm with SCREW_Z_DU = 18000

So I think something is up with the config?

Maybe I need to change this?

const long SPEED_START_Z = 2 * MOTOR_STEPS_Z; // Initial speed of a motor, steps / second.
const long ACCELERATION_Z = 30 * MOTOR_STEPS_Z; // Acceleration of a motor, steps / second ^ 2.

[kachurovskiy]

At first the RPM shows a few hundred over what the actual RPM is, then it settles down to the correct value after a while.

This is a sign of interference from the main motor startup showing as encoder pulses. We've seen it before in a VFD setups. Please post pictures of your encoder wiring.

So ill do some cuts and measure them.

You're likely to see problems with keeping the cutter in the thread if the motor startup produces ghost encoder pulses.

[Festivejelly]

After a while it settles down to the correct RPM. And then it becomes responsive. Its just for that first few rotations it seems to do nothing even with the lathe off. I'll send some pics of the wiring tomorrow. There were 4 wires and shielding.

VCC Red
Ground Black
Signal A Green
Signal B White
Shielding.

I twisted the shielding around ground and put that into the terminal.

I connected the signal wires to the ENCA and ENCB and I spliced a 5v wire from a 5v power supply the powers the Nano, and off that splice I run the 5v to the Stepper drivers and the encoder.

Here is a crude diagram of my wiring

[Festivejelly]

Also im not sure my motor is a VFD. I think its just a 550watt brushless motor.

[kachurovskiy]

@Festivejelly the diagram looks good. Here's an example of how I wire the encoder:

[Festivejelly]

So my wiring looks the same.

Im getting some peculiar behaviour.

If I tell it to move 10mm on the Z it moves 8.75mm. If I tell it to move 30mm it moves 28.75.... so its a constantly roughly 1,25mm error. If the pitch of the leadscrew or halfnuts was off I would expect and increasing error the longer the distance.

Also my rotary encoder doesnt seem to be working or at least its not showing movement in the controller anymore. Ive ordered a new one to be sure.

X axis still functions fine but now oddly it doesnt seem to hold anymore where as previously it did. Maybe this is because im also supplying 5v to the stepper drivers. Who knows i'll revert back and see.

Here is my config
`// Define your hardware parameters here.
const int ENCODER_STEPS_INT = 600; // 600 step spindle optical rotary encoder. Fractional values not supported.
const int ENCODER_BACKLASH = 3; // Numer of impulses encoder can issue without movement of the spindle

// Spindle rotary encoder pins. Swap values if the rotation direction is wrong.
#define ENC_A 7
#define ENC_B 15

// Main lead screw (Z) parameters.
const long SCREW_Z_DU = 20000; // 2mm lead screw in deci-microns (10^-7 of a meter)
const long MOTOR_STEPS_Z = 200;
const long SPEED_START_Z = 2 * MOTOR_STEPS_Z; // Initial speed of a motor, steps / second.
const long ACCELERATION_Z = 30 * MOTOR_STEPS_Z; // Acceleration of a motor, steps / second ^ 2.
const long SPEED_MANUAL_MOVE_Z = 6 * MOTOR_STEPS_Z; // Maximum speed of a motor during manual move, steps / second.
const bool INVERT_Z = true; // change (true/false) if the carriage moves e.g. "left" when you press "right".
const bool NEEDS_REST_Z = true; // Set to false for closed-loop drivers, true for open-loop.
const long MAX_TRAVEL_MM_Z = 300; // Lathe bed doesn't allow to travel more than this in one go, 30cm / ~1 foot
const long BACKLASH_DU_Z = 2000; // 0.65mm backlash in deci-microns (10^-7 of a meter)
const char NAME_Z = 'Z'; // Text shown on screen before axis position value, GCode axis name

// Cross-slide lead screw (X) parameters.
const long SCREW_X_DU = 10000; // 1.0mm lead screw with 3x reduction in deci-microns (10^-7) of a meter
const long MOTOR_STEPS_X = 600; // 800 steps at 3x reduction
const long SPEED_START_X = MOTOR_STEPS_X; // Initial speed of a motor, steps / second.
const long ACCELERATION_X = 10 * MOTOR_STEPS_X; // Acceleration of a motor, steps / second ^ 2.
const long SPEED_MANUAL_MOVE_X = 3 * MOTOR_STEPS_X; // Maximum speed of a motor during manual move, steps / second.
const bool INVERT_X = false; // change (true/false) if the carriage moves e.g. "left" when you press "right".
const bool NEEDS_REST_X = false; // Set to false for all kinds of drivers or X will be unlocked when not moving.
const long MAX_TRAVEL_MM_X = 100; // Cross slide doesn't allow to travel more than this in one go, 10cm
const long BACKLASH_DU_X = 1500; // 0.15mm backlash in deci-microns (10^-7 of a meter)
const char NAME_X = 'X'; // Text shown on screen before axis position value, GCode axis name

// Manual stepping with left/right/up/down buttons. Only used when step isn't default continuous (1mm or 0.1").
const long STEP_TIME_MS = 500; // Time in milliseconds it should take to make 1 manual step.
const long DELAY_BETWEEN_STEPS_MS = 80; // Time in milliseconds to wait between steps.`

Another thing ive noticed and ive taken a video of this which ill upload. If I tell the Z stepper to move left the stepper rotates the other way just a tiny bit before going left. I assumed this was backlash compensation or something but that doesnt really make sense to do if you are not changing direction. given the backlash is set as 2000 as thats what ive measured it to be maybe this is the source of that 1.25mm difference I was talking about. Seems odd to me the motor would move the opposite way before moving the correct way if im not changing direction.

[Festivejelly]

Video showing the stepper issue

https://youtu.be/QGqJmdaca8U?si=8pGdsvehPK5PJ97-

this might explain why its always the same amount out regardless of the length it moves.

[kachurovskiy]

Thanks for filming and sharing. I think the movement is due to NEEDS_REST_Z being true which means that stepper disables and re-enables between every move. It seems like enabling causes it to move. Try setting NEEDS_REST_Z to false.

[Festivejelly]

@kachurovskiy Ive read some reviews on the encoder that im using might need some pull up resistors as the signal is below 5v. Is this something you've experienced? Ive ordered a new encoder of a similar type, but I bet they all come from the same place.

[kachurovskiy]

Most encoders need pull-up resistors on signal lines, I've got them integrated into the H4 PCB already.

[Festivejelly]

@kachurovskiy setting needs_rest_z to false fixed it! Thanks for all the support. It works like a charm now and ive dialled it in to only need 0.02mm backlash! The halfnuts must be well machined!

[Festivejelly]

Measured the voltage from my 5v supply to the encoder… its dropping below 5vs. Ive ordered a dedicated supply and will use a booster if needed. Afaik its 5-24 so ill maybe set only the encoder to take 6v