New guy questions related to tool holding and length

A few days ago, I started torquing my ER20 collet nuts to 39 ft/lbs per recommendations. I ensured that each tool had the proper collet size. Some of the collets now have obviously uneven gaps. From what I know, this is less than ideal. I tried redoing the trouble makers several times, no change.

I don’t know is it indicated that the individual collet or tool holder is just bad?? Or, should each tool be “custom torqued”, regardless of tool size, till the gaps are uneven, then back off 5 ft/lbs??

I’ve also been measuring tools with a granite plate and tool height gauge. when I face a part with a 1/2" endmill for example, it seems to take off too much. Ex: I tell it to take off 0.010", it actually takes off 0.012" to 0.014". Adding a couple thou to the measured length seems to fix this. I got an Edge Tech. analog tool setter to measure in the spindle, because I had heard that the TTS collars flex a little. today I did an experiment measuring the same tools using both methods. Some were the same, some were up to 0.004" different, and none were exactly the same as the one’s I “corrected” based on facing cuts.
Ex: 1/2" endmill granite plate = 3.2465" Edge Tech. = 3.2476" “corrected” = 3.2493"

Technically, I could correct square endmills based on facing cuts, other tool geometries, not so much.

I’m not sure what the root cause is. I know my axis scale and backlash are pretty dead on. Any advice on how I can measure the “functional” length of tools without experimenting, and/or suggestions for a root cause, would be greatly appreciated.


How much runout do you see on the tip of the tool? I’ve honestly never looked at the slits of collets, the only metric for collets that I’ve been concerned with is that runout of the tool is acceptable. I would not recommend torquing collets with the goal of evening out segment gaps as that can lead to tool pullout.

“I had heard that the TTS collars flex a little”

I’d be very curious where you had heard this. Yes, everything is a spring at some level, but this just isn’t a concern.

With the two different methods of measuring tools matching within .001" that sounds like the work offset is the area you need to look at instead.

Thank you,

I haven’t got around to checking run out yet. My concern over the collet gaps comes from videos put out by professional machining companies on ER collets, their holders, and the proper use of both.

So as long as the runout is good, then the evenness of the gaps doesn’t matter?

I read and watch a lot of stuff trying to teach myself, so I don’t remember exactly where I heard it, but I know it was from more than one source. They basically said that because the R8 collet pulls the TTS shank up into the spindle, the force exerted on the TTS collar is enough to flex it up to 0.0020", based on their measurements. Shouldn’t be an issue as long as it is a relatively consistent amount in a given holder, and that amount is accounted for in the tool length.

For work offset, I’m using a tormach probe. The Z portion of the probe seems to be pretty dead on. The method I used to establish trust in the Z measurements was to probe my table and set that point as Z-0. I then took two 123 blocks and two 246 blocks, set them under the probe in their various orientations and measured heights of 0", 1", 2", 3", 4", & 6" repeatedly. The Z measurement was always less than +/- 0.0005 from the nominal value of the dimensions.

Perhaps there’s something else I’m missing in the realm of Z work offsets?


1 Like


Correct, runout is paramount.

Reversing direction here, if you measure a tool, take a cut with it, and then remeasure without taking the tool out of the spindle does the tool length show the same difference that you measure in the part height? What tool are you using as well?

Thank you,

I have not tried measure, cut, measure yet, I will soon. The tools with “corrected” lengths based on facing cuts are 1/2" endmills, 1 HSS and 1 Carbide.

I redid the measurements on the HSS one today, and included a 3rd method, the 123 block/gauge block method. Here are the real numbers.

“corrected value” = 3.6463"
3" 123 block = 3.6442"
granite plate = 3.6435"
Edge tech = 3.6439"

Rechecked backlash and scale factor in Z with mitutoyo 0.0001" DTI, all dead on. Jogged a few thou in Z and used MDI to move a few thou in Z, all dead on.

At someone’s suggestion, I checked how much the nose moves when pressing the tool release button for tool change. Roughly, it loves down about 0.00025" to 0.0005". Not sure what it should be, but that sounds fine with me.

Since I had the set up, I checked the spindle nose face runout, about 0.0003" total. sounds fine to me.
Checked the spindle bore runout, about 0.0002" total. Sounds fine to me.
Tried to check the R8 to TTS collet bore run out, needle just bounced all over the place because of the slits, so no usable info.

I had a very odd hypothesis. What about something getting lost in translation between the offsets page and the physical movements of Z?

I’ve measured and checked and confirmed and reconfirmed every way I know of, that the physical movement of Z is trustworthy. 3 methods of measuring the tool, all within 0.0010" extreme spread, none of which match the “corrected” value. My thought is that there may be some “scale factor”(not the axis scale factor) that is miss-translating the tool length that I type into the offsets page.

Whatever it is, It has to be something involved with doing a facing cut, that is not present in all of the other things that I have checked and rechecked.

I don’t know why I didn’t check it before, but I went back and checked tool runout for the collets. Some were ok at 0.0004 to 0.0008, others were as bad as 0.005. The bad ones mostly correlated to the gaps being visibly off, bur some gaps that “looked” ok had tools off by 0.002 TIR. So, I did some research on fixing runout. 3 methods popped up

1: “clock” the collet position in the holder, retighten, recheck tool, repeat to find best position.
2: NYC CNC’s “tap-it” method, find high spot, tap in the tool with something soft-ish(wood, rubber, ect) to straighten out the tool in the spindle, repeat till straight.
3: “clock” the tool holder in the spindle, recheck, repeat to find best position.

Not sure if there are other methods, or if anyone has any guidance or advice, but, I’ll be playing with those to see what happens.