[ChrisT88]

Hi,

I have a G0704 that I just finished converting to CNC and after testing the backlash found its far from ideal at 0.02"+ with the gibs slightly loose. If I remove all slop in the ways it turns into 0.04"+ so I plan to do a few things. I need a granite surface to work the gibs into shape(may lap ect later), purchase double ballnuts and maybe try repacking them with oversized balls and if that doesnt do it go for some C3 screws.

I may also add gearing for my steppers vs direct drive as they don't sound too smooth at low feed rates and all of my cuts will be made at 5-15 IPM.

After looking everywhere I can think of I only found single 1605 nuts. I would prefer to purchase from the US but I realize this may be impossible.

[ChrisT88]

Finally found a seller on aliexpress but the price is fairly steep at $95 shipped for 3. I really only need two I guess as X & Y are the two big ones but I cant help but think I could just go strait for some name brand C3/C5 screws for around 300 and sell these though the Y is an odd configuration being rear mounted Y.

[CNCMAN172]

Chris,
That is actually pretty cheap for three ball nuts. A Japanese NSK ball nut for example might go for $200 just for one.
Russ

[SCzEngrgGroup]

If you think you can buy a ballscrew from one source, and a ballnut from another source, and have them play nice together, you are very likely setting yourself up for disappointment. There is NO standardization of the screw and nut thread profiles across manufacturers - they all use their own dimensions and profiles, and getting two to work together at all, much less work well, is a total crap-shoot. If you have too much backlash, re-ball the nuts, with larger balls, DON'T replace them. THAT is the only correct way to reduce backlash.

Regards,
Ray L.

[dracozny]

I purchased mine through Chai aka [email protected] this is the same vendor as seen on Ebay. For some reason the double-nut Single Flange setup he doesn't list on ebay at all. Ask him for a quote and submit a drawing for how you want the ball screw ends machined. This was what i got from him back in 2014:

1pc 2005-532.8mm overall length with 1pc DOUBEL BALLNUT AND end machined as your drawing(2pcs locknuts as gifts)
45 usd
1pc 2005-1013.7mm overall length with 1pc DOUBEL BALLNUT AND end machined as your drawing(2pcs Locknuts as gifts)
53 usd
1pc 2505-750.6mm overall length with 1pc DOUBEL BALLNUT AND end machined as your drawing(2pcs Locknuts as gigts)
59 usd
Sub-total
157 usd
DHL shipping
69 usd
Total
226 usd

[ChrisT88]

Chris,
That is actually pretty cheap for three ball nuts. A Japanese NSK ball nut for example might go for $200 just for one.
Russ

Hi Russ, I considered this but there are plenty of 14-20mm Japanese C1-C5 screws from surplus sources that shouldn't run me more than $300-350 for the X and Y which includes a pre-turned end with -32 locking nut and high quality angular conact bearings. I Would have to cut the other end to length, grind the threads down then turn it to my bearing ID but it would be superior to the screws I have now.
I purchased this G0704 as a new-used unconverted CNC package otherwise I would have went with arizonavideo's low backlash kit or DIY. Luckily I am only into it for just over 1k so I dont mind upgrading a few things.

If you think you can buy a ballscrew from one source, and a ballnut from another source, and have them play nice together, you are very likely setting yourself up for disappointment. There is NO standardization of the screw and nut thread profiles across manufacturers - they all use their own dimensions and profiles, and getting two to work together at all, much less work well, is a total crap-shoot. If you have too much backlash, re-ball the nuts, with larger balls, DON'T replace them. THAT is the only correct way to reduce backlash.

Regards,
Ray L.

Thanks Ray. I definitely considered this and hoped to get some feedback on my options. I just heard back from Chai and it seems there are two sizes of 15.5mm and 15.98mm which mine appear to be 15.48-15.5mm(difficult to measure on machine) and he quoted $98 for 3 DHL to US. I am really surprised nobody has attempted this and I guess that means they truely are binned from the factory or everyone fears that they are, lol. I have also considered buying a second standard nut for $10 and attempting to tension two against each other like a lot of people used to do with the Roton screws though I would probably want to upgrade my steppers due to the additional friction (thinking 3 phase leadshine hybrids).

After seeing what Arizonavideo achieved by reballling his PM-25's screws I definitely had plans to let him or myself try to achieve similar with double nuts but I am now thinking some real ballscrews are probably the way to go. I found some 14mm C1 screws that would be interesting to try though I'm a little worried about going too small and 20mm may not fit my X axis without having to remove too much cast iron from the saddle. 16mm surplus are a bit harder to nail down in the 5mm pitch but they do popup.

I guess my choices are guinea pig double nut or spend the extra $200 and go to ground screws. I would be happy with 0.0005" backlash over a small distance but having some insane precision would be welcome considering the lack of precision this machine has in the ways ect.

[ChrisT88]

I purchased mine through Chai aka [email protected] this is the same vendor as seen on Ebay. For some reason the double-nut Single Flange setup he doesn't list on ebay at all. Ask him for a quote and submit a drawing for how you want the ball screw ends machined. This was what i got from him back in 2014:

I think the prices have gone up since then as I was quoted $98 shipped from Chai for 3 double nuts by themselves. I ended up buying two NSK 1605 C3Z 850mm ballscrews for $198 shipped. I was really considering a 420mm C1 grade screw for the Y axis but I really wanted to maintain the bearings at each end which requires at least 450-470mm.

Next purchase is a pitch gauge as these didnt come with any lock nuts and I am not sure of the exact pitch but it looks fine.

[arizonavideo]

A 14mm ground screw is fine for the X and Y. The end work will have to fit.

As far as backlash in a axes you have three main sources. The ball nut, support bearing and the way/slide stiction. One thing most people dont realize is the sliding surfaces on the China mills are rarely perfectly flat and square. My Y on my IH mill has a ground screw with really nice matched pair of AC bearings and a really tight nut but I still get backlash. Why?

Because the slide rocks before it moves by about .001".

The Z on most the mills will do the same thing. They will rock before they move. The standard RF-45 are the worst, The 704 and PM-25 have a fairly long Z slide so they work better but still have the same effect.

From what I have seen a screw with zero lash at lets say 15 pounds of force will still have some lash as a system on the mill.

To get under .0008" is quite hard. Flex from the coupler, end plates, slide rock , stick slip and a little from the end bearings all add up.

I'm not saying a ground screw is bad but a mill with C7 screws that are tight and end bearings with good preload and slides that are flat and move well you wont notice any real difference compared to the .002" of flex you get from tool hop, chatter when milling.

On the double nuts.

They have two preload systems. The most used is just a shim to separate the two nuts. This makes the preload. Most have no preload and have lash. You can make a shim but its a real PIA. Sometimes you can go to a larger ball size on one or both nuts but as you might think that is no easier then re packing a single nut.

You can ever so slightly bend the shim to act like a spring but you have to be really careful because the are cast and crack.

I often replace the shim with a new spring loaded system but it takes a lot of work.

[Chris D]

Chris,

Have you actually determined that there is backlash at the ball nut? I assume what you have are rolled ballscrews from China. We all know their quality varies greatly, but I have these screws on a machine I just built. Interestingly, the nuts really didn't have much backlash - I measured about .0002" which is very good. Where I had a LOT of "backlash" was slop in the fixed bearing. This bearing block is supposed to be a matched set of angular contact bearings, matched set being the key word. When the bearing set isn't matched, you have to shim either the inner or outer bearing races to take up the slop. I believe it was the inner races I shimmed and the amount of shim was around .020".

The machine has been run-in a lot during the last few weeks and I checked the actual backlash of each axis and they are both still at .0002".

I would highly recommend you study your components before trying to do things like double-nuts etc. You may be spending a lot of time and money on the nut, only to find out you need $2.00 in shim stock to pre-load the bearing block.

Chris D

[ChrisT88]

Thanks Dave. I have definitely factored in the poor ways/saddle on the Grizzly which are much worse than the Precision Matthews 45 I have experience with but plan to work the Grizzly into shape even if that means cutting the dovetails off for some linear rails though thats a last ditch effort if I am unsatisfied with my attempts at finishing what Grizzly started. I look at going with precision ballscrews as a good base to help me eliminate them as a variable since as you pointed out there are countless ways for backlash to build up and even going DFU + reball is no guarantee the screws I get will achieve the numbers I'm rebuilding for.

Chris,

Have you actually determined that there is backlash at the ball nut? I assume what you have are rolled ballscrews from China. We all know their quality varies greatly, but I have these screws on a machine I just built. Interestingly, the nuts really didn't have much backlash - I measured about .0002" which is very good. Where I had a LOT of "backlash" was slop in the fixed bearing. This bearing block is supposed to be a matched set of angular contact bearings, matched set being the key word. When the bearing set isn't matched, you have to shim either the inner or outer bearing races to take up the slop. I believe it was the inner races I shimmed and the amount of shim was around .020".

The machine has been run-in a lot during the last few weeks and I checked the actual backlash of each axis and they are both still at .0002".

I would highly recommend you study your components before trying to do things like double-nuts etc. You may be spending a lot of time and money on the nut, only to find out you need $2.00 in shim stock to pre-load the bearing block.

Chris D

Hi Chris, My setup uses radial and thrust bearings with each end of the shafts constrained allowing for me to carefully adjust tension via nuts. I first assumed my issues were in the coupler, bearings, clamping of the ball nuts(Grizzly uses some crude methods to secure the ballscrew nuts) and/or the ways themselves which I measured with the gibs at many tensions. Its now fairly obvious my ballscrew turns about 20-40 degrees freely back and forth inside the nut before engaging. With the gibs at what I would call very loose and the ballscrew bearings preloaded quite a bit it translates into around 0.0024" of backlash which is about the standard for these ballscrews.

I think Dave mentioned most arrive with 0.002-0.003" initially and that the better sets can be rebuilt and tweaked to achieve 0.0005 if all goes perfectly. How yours are achieving the precision of a professional VMC and numbers Tormach doesnt even achieve with ground screws and double ball nuts I cannot understand. Maybe you ended up with high quality rolled C5 screws somehow?

Btw, I also talked to the seller again about my NSKs and he said there will be locking nuts and shims included so all thats left is to find some nice angular contact bearings. I plan to redesign the mounts to accommodate the threaded end of the screws as I cannot duplicate the high quality threads they already have. If I can get at least 40 or 50 bucks for my ballscrews it will have been cheaper to go with C3 ballscrews than pursue optimizing the Chinese screws.

[arizonavideo]

I'm not achieving ground ball screw precision by packing the ball nut fairly tight on a rolled screw. We are making a trade off of speed, life, backlash.

The rolled screw are really quite good with most allowing you to have positive preload on the balls for the full screw length. The preload may vary but you don't have just plan slop.

The thing here is if you have no preload but zero lash you will have a few thow of lash under load. The balls are really more like springs if you look at the amount of flex.

A ground screw will have a much greater contact area because all the groves will have contact with most the balls. A rolled screw has more error so less balls will contact, so for the same amount of preload a ground screw will give less backlash. So a good ground screw can have vary low drag which is needed for high speed use and to give low backlash. For a rolled screw to give the same lash, if it can, it will need to have the nut packed tighter. This will cause over heating of the balls at 500 IPM and shorter life. Lucky for us we only go 150 IPM.

So most of the rolled screws I send out have zero lash under no load, and around .001" under light load. They wont last 15 years of 8 hour per day use and you can't run them at 500 IPM all day long.

For weekend use of making a nice smooth part with a good finish for 5 years or so they are perfect.

Is it better to grind off the dove tails and go to high preload rails and C3 ground screws? 100% yes but you really have a different machine and at a different price point.

If you need that, which some people do, then go for it.

I was looking at CAMBAMs web sit the other day and this guy I have watched for years make watches. He really is amazing and he built it mill a few years ago and for a small mill it works really well.

Watch!..

For this ground screws help.

Whats funny is his spindle.

Stay cool

Dave