Backlash , Any suggestions ?

**komatias** · 08-10-2015

Have a look in your code to see the start and finish positions of your circles. It seems that they may be different.

As for backlash, yes ballscrews do have backlash if they are not preloaded, you will also get it from the bearings. Try cutting a square, it is usually the best way to see if you do have backlash. Also use a flat endmill not a ball. A very quick way to test is to measure the actual movements you are doing to verify them.

If you really want to test the machine have a go at the NAS Test also known as the Circle Diamond Square:

**philgiggins** · 08-10-2015

Ill try the NAS test tomorrow evening when i get a chance and post the results. Theres definitely something amiss. I need to eliminate possibilities one at a time !

Philip

**howser37** · 09-10-2015

To me it just looks like the tool geometry is wrong, as Komatias said are you using a ballnose ?. If so and the software thinks it is an end mill, then you will be left with material like you have in the pictures on the end points of the circles. You can change to an endmill or you can simply tell the machine to overrun the toolpath by a few mm.

If those steps were caused by backlash you would see signs of it on the top and bottom of the circles on the direction change, but they look fine, so it looks to me like you just have the wrong settings in.

The reason the steps arent in a straight line is becasue i assume each vector is starting in opposite directions and that is just the bow wave effect pushing material infront of the bit, so you get that alternating step effect, if you ran the same program with all the startpoints and directions the same on the vectors you would still have the steps but they would all be in a straight line, either overrun the bit or use an endmill, and even then i would overrun a few mm.

**JAZZCNC** · 09-10-2015

Originally Posted by howser37

To me it just looks like the tool geometry is wrong,

+1 on this. If you had backlash issues that V carved lettering would show it straight away.!

**magicniner** · 09-10-2015

There's a kerning fault with the lettering, but the cutting looks good.
You almost always need to manually kern in cad as very few software packages do the job well.

**philgiggins** · 11-10-2015

Hi all,

Thanks to everyone who has replied.
I have been doing some experimenting and have identified the problem. I originally had some backlash issues with the gearbox that drives the 2 pinions, it turned out it was a stretched belt ( my own fault from over tightening ) After i sorted that out i was still getting slight lines on profile cuts ie. 3 passes- slight step on each, and as it seemed to be occuring at a point where the axis changed direction i didnt look elsewhere for the problem. I was cutting on Fri and i got this massive step (see picture) which made me realise the problem was with lost steps and not backlash.
This is where i am looking for some advice ... i contacted the company i bought motors, drivers and gearboxes off. The drivers are leadshine Dm856 and the motors are 4.5nm. They are running off a 70v power supply. I suspected at the time of purchase the motors where slightly undersized but i was assured they (the company i purchased from) ran a similar sized machine with a similar weight gantry from these motors and drivers. I am using an ethernet smoothstepper as my controller. The company reckon mach3 is possibly the problem (they use Usbcnc) But so many people use mach 3 with no problems ,i have doubt that this is the case.
As you can probably see from my pictures i have built this on a budget (no hiwins on X etc.) But it runs very smooth and is very very sturdy. My point is ,this machine will be in use most days as part off my business and i need it to be reliable. Am i as well just biting the bullet and purchasing bigger motors and drivers now , or are these correctly sized ?
Sorry this is so long and rambling :)

Thanks Philip

**philgiggins** · 11-10-2015

Tool geometry was incorrect on that toolpath, quite a mistake on my part. I had issues earlier that with steps as i have said before and that evening i thought ill try some circles it should be a good test and when they didnt join i just assumed .... and then posted it ... thats what i get for rushing it!
Thanks Philip

To me it just looks like the tool geometry is wrong, as Komatias said are you using a ballnose ?. If so and the software thinks it is an end mill, then you will be left with material like you have in the pictures on the end points of the circles. You can change to an endmill or you can simply tell the machine to overrun the toolpath by a few mm.

If those steps were caused by backlash you would see signs of it on the top and bottom of the circles on the direction change, but they look fine, so it looks to me like you just have the wrong settings in.

The reason the steps arent in a straight line is becasue i assume each vector is starting in opposite directions and that is just the bow wave effect pushing material infront of the bit, so you get that alternating step effect, if you ran the same program with all the startpoints and directions the same on the vectors you would still have the steps but they would all be in a straight line, either overrun the bit or use an endmill, and even then i would overrun a few mm.[/QUOTE]

**JAZZCNC** · 11-10-2015

Nothing to do with Mach3 so forget that.!! . . . . They are correct thou that those motors are more than good enough when run at that voltage and with those drives.
It's a common mistake to blame motors and go for larger motors only to find out they weren't the problem and now your performance is less.!!

If really is loosing steps(I don't think it is.!!) This will be down to setup with motors being over tuned or mechanical issue causing binding.
My own feeling is that some coupler or pulley is slipping.? Missed steps tend to be one of two states.? Incremental showing slowly over the course of job in tiny amounts. Or Random with no ryhm or reason.?
Incremental often comes from wrongly setup motors/drives. Random from Electrical noise or poor wiring.

Give us some info on the tuning for each axis. Also Info on the Pitch of pinions or screws etc.

**JAZZCNC** · 11-10-2015

There is one thing you can try that does happen and will cause a missed step on each directional change.
This happens when the Step Pulse is sent out on the wrong pulse edge to what the drive is set to expect. Resulting in the drive sees the Step pulse at the wrong time and one step or micro step is dropped on every direction change due to the fact step pulse comes before the direction pulse so it gets dropped.

This is easy tested by knocking some G-code to move each axis back n forth 1000's times then returning back to start. Make a mark before running and see if it returns to same mark after run the test.

If it doesn't then you have the Pulse edge set wrong. Easy fixed by changing the Active Lo setting in Motor tuning. ( Or some drives have a setting for this) This will then make your motors turn the wrong way so you'll have to go into Homing and limits and use the Reverse axis option.
People Often make the mistake of thinking swapping the Dir active low is for changing axis direction when it's not. It does but it's not correct way.

**philgiggins** · 11-10-2015

Hi Jazz,
The first thing i did was go around and check every relevant nut bolt and coupler ! They all seemed to be fine. But no harm to have another look.
Rack and Pinion are 20' pitch, pinion moves 94.2mm per revolution. Geared down by 5:1. Ball screw on Y is a 2010 its ran from a belt 1:1 (2 no 34-5M-15mm (TL) Timing Pulley) and on Z its a direct drive 2505,

I was having some resonance issues at slow speeds when i first started cutting so i upped the microstep setting on the driver and it helped a lot. I upped it to 4000. When i talked to company they suggested putting it back down to 800. I will try this on Monday.

Im pretty sure its not a binding issue. I can jog around no hassle as 10/12 m/min. Although maybe that could change under load. I need to eliminate it 1 thing at a time.

Acceleration is set to 100 ( this is low i think ) Max speed is set to 5m/min.
If i cut below 2.5m/min i have no issue its when i go above this the problem starts....

Thanks Philip