[Donkey Hotey]

Earlier this month, I took delivery of a new TL-1. Rather than waiting for the installer to level the machine, I tried to do it myself (15" Starret machine level--0.0005/foot graduations).

If it had been done by the installers, I would have never known what I know now: the bed on my machine is twisted.

This now makes me wonder about all of the SL series machines I read about in here that supposedly can't hold tolerance. With the inclined bed of a turning center, there is no fast way to measure the twist. I have to now wonder if some of those machines are twisted as well.

• This bed has a total twist of 0.003" from one corner of the machine travel to the other (16x30).
• The geometry of this particular bed (16" wide, 10" to spindle) means that the X-position of the tool tip can vary 0.0024" from end to end.
• Because that doubles on a diameter, the positional accuracy will already add up to 0.005" error before I even start cutting.
• Even on a 10" part, there's going to be almost 0.002" of error before I start.

I was told today there really isn't a 'spec' for how much runout that the bed can have. I can't figure out where I should be centering the tailstock now (it'll change with position). I have to think about this with every part I make.

They've been very nice about checking the machine. The tech was going to align the spindle (also out of spec--0.0015 in 10") but they were going to have to remove the backsplash to do that. I didn't have time to move everything from behind the machine so we put it off for now.

Besides: I can align the spindle (ain't no big deal) but what do you align it to? If the bed is twisted, then the spindle is being aligned in the local 10" zone. It's still going to change further down the bed.

Am I splitting hairs here or is 0.003" twist too much to accept? I realize that this is a bargain-basement machine but I would expect a huge lump of cast iron like this to be flatter than 0.003".

And for those who will ask: the machine can sit on three legs and it still has 0.0025" of twist. It sat like that for a week in the hope that it would unwind. It didn't.

What would you do?

[djr76]

Call Haas and tell them to fix it. The warranty is a wonderful thing for you right now. Every time I had a new Haas installed, my HFO tech guys were usually there the same day or within a day to set it up.

[HuFlungDung]

Don't panic. I am amused by the vast numbers of people who have barely cut metal (which you might or might not be ), who think that their machines have to be perfect in order to get acceptable results. It ain't so

Holding a tolerance does not have a lot to do with the bed being twisted. You can compensate for any constant error in your program, but a true lack of holding tolerance is a random error tossed into the mix.

Now, if the linear ways were somehow loose, that could cause them to shift unpredictably from one part to the next.

I never swear by the machinist's level only. The final adjustments always come from trial cuts, and tweaking a support pad (screw) here or there. Although the machine is supposed to have been levelled or sitting flat on a table when it was machined at the factory, you'd never know if your particular casting had a little bit of spring to it which the machinist 'let go' and clamped it down anyway, and cut it. That's just the roll of the dice. Or, the casting may truly have sprung a little after aging some more.

Now if your machine is a one piece casting, and won't sag on its own, then you'll need to lag it to the floor so you can pull down the high corner.

Most of the time, you won't be turning long straight cylinders anyway, so the maximum error condition will be the diametral difference over the length of a given journal length. The widest fit of most components that ride on a shaft would be maybe 3 or 4 inches. Your lathe, as is, has a taper error of .0001/inch. Trust me, there are a zillion other reasons why that ain't going to matter. Some of those reasons are deflection of the stock, eccentric centerholes and uneven heating of the work and toolholder during the cut.

For the final setup of a lathe, I usually bore a hole about 4" deep and adjust the levelling pads to get a straight bore, on account of it being difficult to straighten a bore by handwork, although a half thousandth is not too awful in that distance.

[Donkey Hotey]

Thanks for the replies, guys. Yes, the HFO has been on top of this and of course they want to make me 'happy'. I'm just not sure what is reasonable to make me happy.

As Hu pointed out, there are a hundred other things to toss error into the problem. What bothers me is that this is one that I'm not so sure I should have to deal with. I want to be reasonable but I also don't want to be a sucker.

At some point I could end up re-machining aluminum motorcycle fork legs on this thing. They could be 15-18" long (or longer). I can't afford to have 0.003" of taper in that distance. I could compensate for it in the program but should I have to?

As for adjusting the leveling pads to get a good sample cut: this casting moves about 0.0005" when you completely unload the high corner. This casting is very stiff and there's no perceptible adjustment in it. Bolting it to the floor was suggested but is the cast iron going to like that?

The only solution for this machine seems to be figuring out which rail isn't aligned with the spindle, then shimming the linear guideway. The thinnest shims I can find are 0.0005 and 0.001.

It could get 3" shims, every 5" to bring the bed up to level (.0005, .001, .0015, .002, etc). I'm just worried about screwing up something else when I unbolt the guideways--like the parallelism between them.

[HuFlungDung]

I'd not shim the linear guideway. Right now its probably got a good nest to sit on. If you shim it, you need a continuous taper shim for the full length of it to get equal bearing.

I'd say lag the machine down. It is easy to move it that small of an amount. You may decide its not what it needed anyways, after you've put it to work, and its easy to back off a screw

[Geof]

How did you measure the twist? Are you sure it is actually twisted not just an artifact arising from the way you measured it?

Also have you put in a length of bar and taken a cut to verify whether your predicted taper actually occurs?

[Donkey Hotey]

I was very careful about how I measured.

1. I put the level on the cross-slide and parked it dead center in the middle of the machine (15,8).
2. From that location, I leveled both directions of the machine.
3. After leveling, I placed the level along the Z axis and hand jogged to each grid coordinate (measured every 5" along Z and 0,8,16 in X). For each measurement, the level was not touched or moved in any way.
4. I repeated step three with the level across the bed (X), revisiting each grid location and notating the values.

It's a nice, even, progressive twist from near left to far right.

No, I haven't cut a long bar yet to verify the taper. I was actually going to use that method to eventually align the spindle to the bed (2" dia. aluminum--maybe 10-12" extending out of the chuck).

It did fail the 'spindle bar' test shown in the machine's inspection sheet. They allow 0.0005" in 10". The inspection shows that it passed at 0.0004" or something like that. Now it's 0.0015 out in the same 10".

What bothers me the most here is that it was inspected and signed off at the factory and it's not right. It makes me wonder about the validity of their final 'inspection' process. It has me looking at future machines and what I might expect out of them.

Why bother giving me an inspection certificate if the tech has to spend a day stripping sheetmetal off of my machine and realigning it anyway? And if there isn't a published spec for bed twist, what can I expect from a larger machine? How many people don't know that their machine isn't in spec and hasn't been since the day it shipped?

For the money, I don't expect these machines to be the highest torque or the most rigid or whatever. But I do expect them to be straight and square--at least to the level they claim to be in the inspection sheet. Or to survive a 99 mile shipment from the Haas factory to my shop and still pass inspection (without the tech having to loosen the headstock and realign it).

In their brochure, they brag about doing all the bed machining in one setup to keep everything square. Why bother if they're clamping a spring into the casting before they machine it?

If I get into custom suspension work in the future and I need bores and spindles to be truly cylindrical, this shakes my confidence that a Haas can do the job--even in aluminum.

Yeah, each job can be tuned to deal with the machine but will I have that time in the future? If I buy a TL-25 it won't be because I have time on my hands to mess around with setups. It'll be because I need it to start banging out parts in a big hurry and hold some level of tolerance. With the sub-spindle and turret in the TL-25, I can't even begin to imagine all the tolerance stackup and inspection errors one of those could have.

I questioned all the claims I've read in here about SL series machines not hitting their marks. Now I see a possible reason why some seem to work and others don't.

[Geof]

.....I questioned all the claims I've read in here about SL series machines not hitting their marks. Now I see a possible reason why some seem to work and others don't.

This is a good point.

I will check my TL machines the way you describe and see how they come out.

Your comment about machining in one setup being irrelevant if they clamp a warp into the casting is quite correct. Another thing is possibly things are being done to quickly. I seem to remember being told that ductile iron castings need to be aged a bit before machining and that it is best to rough machine, age for a while then finish machine. It is likely that Haas does not follow this rigorous procedure.

[HuFlungDung]

My suggestion is to take some trial cuts. Such tests cannot even be done by a newbie or an inexperienced technician and prove anything because the tooling used has an effect. For example, I can take a brand new insert, machine a long piece, and get a bit of a barrel shape. Use a different style of insert and get another result. Then, take and grind the insert in my top secret method, and get a straight cut. So which test do you believe? What will be your final machining conditions under most circumstances? If you want to cut a straight cut with a factory honed insert, the machine will have to be tuned differently than for my top secret (custom) grind.

If you intend to do chucking style turning, moreso than tailstock supported turning, the lathe will have to be tuned differently.

If you machine dry, the toolholder will warm up and extend itself into the work. The work will grow warm ahead of the cut (if a lot of machining is done) and a few degrees of temperature variation from one end of the cut to the other will affect the cold diameter.

If the spindle gets hot compared to the rest of the machine (which is a 100% certainty), its going to grow a new shape, which may affect your tests in a certain way.

A lathe bed is not a ceramic gauge block 3 to 10 feet long. It is flexible for its span and temperature, and the casting can settle.

There is a good reason why centerless grinders have their place in industry: the long cylinders produced by a centerless grinder are difficult to produce on a demand basis on a lathe of any sort. A little bit of sanding and polishing is normal for long cylinders turned on a lathe.

I'm not saying you cannot pester the hell out of the technician Just don't expect that because he's trained from the factory, that he can coax the best out of that machine either. I can make straight cylindrical parts on demand from my assortment of manual lathes, some of which are showing signs of bed wear. A little tweak here or there can compensate for a particular job, but I really don't have to fool around like that.

CNC lathe programming on the other hand, does require a bit of tweaking if you want perfect parts. What you need the machine to do is be repeatable, because accuracy can be obtained from program tweaks, but real tolerance bands unique to your machine cannot be changed.

[Donkey Hotey]

Ahhh....HuFlungDung: the voice of reason

I understand all the points made...really. I've had pretty good luck with my 16x40 manual lathe but this has been a huge upgrade and learning experience.

Despite the intention of holding on to old tooling, I've already upgraded from my legacy KDK to a real Dorian CXA toolpost and some starter blocks (to reduce tooling movement and thermal shift). I've got a start with some inserted Valenite and Kennemetal tooling (for more consistent cutting and stronger holders).

It's a huge paradigm shift from where I was two months ago. I'm trying to get it all right. Thanks to some good advice from others on this forum, I'm getting there. I'm moving toward producing real products and want things to work.

Maybe I'll have to consider anchoring it to the floor. At least I'll gain the thermal stability of the concrete. It'll take awhile for the spindle and casting to heat up planet earth.

[duenow]

"It did fail the 'spindle bar' test shown in the machine's inspection sheet. They allow 0.0005" in 10". The inspection shows that it passed at 0.0004" or something like that. Now it's 0.0015 out in the same 10"."

So at the factory is was right & now it's not? Make them send you another "IF" you are sure. My new Mini Mill bought 07/20/04 had the hose cut to short that goes to the little oil bucket. I found out by taking the Y covers a pouring coolant in there, it just ran right out of the fitting. I bought plastic hose & fixed it myself. The other problem is if I let it set for 4 days the oiling will alarm out with no oil, they replaced the pump & tried to find the leak but never did, so I just pump up the oil before use to get a steady pressure. Haas HFO was willing to try to find the leak but after 2 days I just told them I can deal with it. It's so small that only after 4 days sitting will it do that. The machine has been great other than that & I'm sure the Dallas Texas HFO would have worked with it till they got if fixed Andy is the owner & I have know him since 1980.

[Donkey Hotey]

So at the factory is was right & now it's not? Make them send you another "IF" you are sure.

Oh, you can bet I'm sure on that one: it wasn't my test or my test bar. They mounted their precision ground, 10" test bar and their test indicator.

Okay, I've got the inspection sheet here:
Side of the test bar was supposedly -0.0003" when it finished assembly.
It's now +0.0015 (almost a 0.002" shift).

They were very nice about offering to realign it but it was going to be a huge hassle. I'm going to invest in my own 1/2" allen socket so I can get those head-casting bolts loose and do the adjustment myself.

The bigger issues to me are still:

• The bed twist: I can't adjust that and I don't think a residential garage is going to work for anchoring.
• The inspection sheet: why proudly present the owner with an inspection sheet that the actual machine doesn't meet?

[HuFlungDung]

If the bed has truly sprung (let's say from aging of the casting), it may well be that torquing the thing down to the floor will correct both issues at once. However, if you begin to fiddle with the headstock, then you'll have to undo that 'fix' if and when you fix the bed twist.

Did you chuck the test bar, and was it running absolutely perfectly true near the chuck and at 6" from the chuck? If so, that would be a miracle in and of itself. Only then can you validate the parallelism of the spindle to the bed.

If you don't want to drill holes in your garage floor, there may be an alternative. Get a one inch piece of plate to set the machine on, kind of a subplate if you will. You can make a 4 point mount on levelling screws to the floor, just gravity holding it there. Then, torque the machine down to the steel plate using appropriate shims. Now its going to be like stringing a bow, where the plate bends a little and the lathe bends a little. But, it is not a large amount that we are talking about here, so I think it would be feasible.

I still think you're swatting at gnats here. Put the machine to work and evaluate its performance and find out what the real tolerances are, then start your *****ing at the Haas rep if it really will not do what you need it to do.

[duenow]

Oh, you can bet I'm sure on that one: it wasn't my test or my test bar. They mounted their precision ground, 10" test bar and their test indicator.

Okay, I've got the inspection sheet here:
Side of the test bar was supposedly -0.0003" when it finished assembly.
It's now +0.0015 (almost a 0.002" shift).

They were very nice about offering to realign it but it was going to be a huge hassle. I'm going to invest in my own 1/2" allen socket so I can get those head-casting bolts loose and do the adjustment myself.

The bigger issues to me are still:

• The bed twist: I can't adjust that and I don't think a residential garage is going to work for anchoring.
• The inspection sheet: why proudly present the owner with an inspection sheet that the actual machine doesn't meet?

Since you just bought it tell them to get it the hell out of your shop & send you another new one! I would have done that with my mini mill but the issues were not that much. I am a screw machine & cnc bar machine & only bought the mini mill to do some second opp milling, my shop was started in 1972.
To date from 07/20/04 the mini mill has only 425 hrs on it as I don't use it much, most of my second opp mill work is done on Kurt auto air slides on bridgeport type milling machines.

[Donkey Hotey]

Again: thank you for the advice.

I'm considering something similar to your steel plate idea: an 'x' frame torsion thingy to bolt it to. I just don't know how secure I could get anchors into a 3-4" garage slab.

Yes, I'm aware that any shift in the casting will negate the head and tailstock tweaks.

And no: the bar wasn't running perfectly true. He stoned the spindle and the bar face, wiped it with the heel of his hand, then mounted and indicated it the best he could. He still got a bit of runout from the bar but he used the 9:00 position after finding the high and low spots.

The problem for me is that 'long' parts are out in the future. For right now, everything is small. After May, my warranty turns into a pumpkin.

I made the mistake of trusting that if my VF-2 was loved and cared for, I wouldn't have any problems. It didn't work out that way.

Maybe I should share why I'm so bothered. I don't want any more infant mortality problems cropping up after the warranty expires. I used my VF-2 very little while it was under warranty. At 300 'on hours' and 29 'in cut' hours, the vector drive took a big dump. It was well out of warranty by then.

I got lots of well-intentioned sympathy but little 'love' from the factory. The sympathetic people at the HFO were powerless to actually fix it and the guy at the factory was very nice but not willing to swap it. All I really needed was the internal logic board but they wouldn't sell me just the component (everything else in the vector drive was new or verified).

It looked like I was going to have to cough up $2000+ for an entire vector drive in what is basically a new machine. With tight finances, the sale of this TL-1 was truly hanging in the balance: $2000 for a vector drive was going to bust my budget. Under those cicumstances, the factory buckled (and my HFO also went above and beyond to make it right). Without this machine, I would have been forced to purchase a Vector Drive.

I don't want a repeat of that scenario. If there's ANYTHING wrong with this machine, I want it fixed or at least well documented, now.

I didn't really want to air that publicly but now you get some insight as to 'why' I'm so picky about this.
(and why I know so much about the electronics inside the Haas cabinet)

[HuFlungDung]

Yes, it is risky to buy a new machine and not use it before warranty expires I have only bought one manual lathe new in my shop, out of 12 machine tools. It was no better or worse than the others. I buy used expecting full well to have to add a bit of money to the pot to fix this or that, but it generally works out.

By all means, try the exact cut you are planning to take on a few sample pieces, ASAP while you can get help. Chances are the machine will behave differently after you've warmed it up and are taking a cut than it does cold and totally unloaded while testing with test bars.

I think you'll be using the tailstock to take this cut, so normally, if you've levelled the lathe and it seems perfect, the cut may still exhibit some error. There is no question about what to do: you adjust the tailstock to eliminate the taper, and that does not mean that the tailstock position is necessarily valid from one end of the machine to the other, but if you set it at the mean postion, you've essentially split the error in half. You won't be using the tailstock near the chuck so often, so the error in that direction is even less significant. The error on long pieces near the length limit of the lathe is going to have bow and flex problems, regardless of how dead true the bed is.

[HAILINHAAS]

what is the machines serial number?

[Donkey Hotey]

I see that you're in Valencia. Have we met? Do you work at the factory or Torrance? You ain't Sean or Mike or you'd know who it was from the story above.

SN: 3078675

[HemiRoadRunner]

I was looking at your levelling procedure and I did not see any adjustments made to the level after jogging the Z axis and seeing the twist.

If you are seeing a twist or roll along the Z axis, the machine may only be rough levelled.

Have you tried adjusting the level of the machine while jogging the Z back and forth, until the roll is adjusted out of it? I'm thinking you may have already tried this, which would explain why you now want to anchor the machine to pull the twist out of it.

If the machine is not level, the spindle and tailstock will not be in alignment either. And so, the machine will cut a taper.

[Donkey Hotey]

Have you tried adjusting the level of the machine while jogging the Z back and forth, until the roll is adjusted out of it?

Yes, it was 'adjusted' dozens of times. The final measurement I described was after many tweaks and finally settling where the center was level and the 'high' corner had just slight torque on it (not the ideal load on the casting but I think this is what you're wondering).

There is almost zero movement available in this casting for adjustment. It's very stiff. If it were right, it would be awesome. Completely unloading the 'high' corner (free spinning bolt on the foot) only takes out 0.0005-0.001" worth of twist.

Two HFO techs have taken their best shot at it. They didn't have any better luck.