Feeds, speeds, and other topics for the home machinst

[David]

I soldered a 5-pin DIN plug to the tool length setter tonight and it seems to work!

It took some head scratching to decide how I should wire it in but the test I described above seems to have been good enough and no smoke escaped from the control cabinet. I was very careful to determine which pins on the socket were the ones I needed... the Tormach manual has a diagram of the socket 'from the solder side' ie inside the cabinet. What I didn't see until later was it had a little table right below the diagram with the wire colours. The documentation is pretty good with the Tormach stuff.

I plugged it in and pressed it down and saw the indicator on the screen flash so figured all was well. But I was having trouble getting consistent trip events and was wondering why the indicator was on all the time rather than off and just flashing on. It turns out I need to tell the Tormach it is passive probe rather than a tool length offset probe. Something to do with NO and NC switches. I thought my test above showed this to be a NO switch which is the same as the Tormach one but I don't care enough to figure it out now it works.

I didn't like Tormach's idea of entering the height of the probe so setting Z=0 to the top of the table. That just seemed like a sure fire way for me to plunge a tool into the table or hard down into the tool setter.

My original thought (before reading the Tormach manual) was to set the top of the tool setter as Z=0 by touching the spindle nose to it with tool 0 selected and making that Z=0. At that point I went looking on YouTube to see if this might work and found confirmation that it does if you trick PathPilot into accepting 0 for the ETS height using floating point rounding errors :)

I was getting fairly consistent length readings on my 4mm spot drill. The lengths are a bit different to what I get using the surface plate and height gauge but that's dependent on the initial tool 0 reading on the ETS. As long as all the tools are measured relative to the that tool 0, ie where the spindle nose tripped the switch, I don't think their absolute length matters.

So now I need to make a mounting plate for it and one day I might even unsolder the plug, put some heatshrink on the wires, and put the plug body on, which I forgot to do until I'd soldered the wires in... a bit like tail nuts on copper pipe.

It's probably not that accurate but I think it is more accurate than the height gauge I have, and it won't eat batteries when I forget to take them out. Looking forward to setting a bunch of tools for job and seeing how it turns out.

[David]

On cnczone I mentioned I had to tell PathPilot the ETS was a passive probe as opposed to an ETS, and that mine didn't have the LED usually included on the base. Someone replied that I probably had the same one as them which was just a two switches with no electronics in the body.

It seems mine has a sensitive NC switch for the tool probe, and I have no idea what the other two wires do. Nothing happens when I check continuity on them while pressing the ETS right down, which is where other ones have an over-travel switch.

So the 'manual' that came with mine is complete fiction and I think I've bought a switch in a fancy housing for over $100 AUD. That's ok if I get good enough measurements from it, but it certainly isn't the thing described on eBay or the leaflet that came with it. Shocked!

I'm going to unsolder the two wires I assumed were powering the internals. One of them is connected to the 12v pin on the Tormach and the other to 0v and I don't want to risk a short-circuit between them.

There are what I assume to be 3 tiny screws at the top and nothing else I can see to get into it. I'm going to leave it alone and just use it for a while.

[David]

Someone on cnczone set me straight on custom thread defns. The file I wanted was the one in ~/gcode/thread_data/ not the one in ~/v2.2.1/. I edited the correct file and my new threads showed up. There is still a bug where two blank entries show at the top of the list but they do go away when you scroll to the bottom.

[David]

I took the RapidTurn apart today to file the base flat where it had burred up and fit the new sensor.

Put it back together, dialled it straight and put the motor on. I found out what put the big dent in the RapidTurn motor cooling fins - its power plug.

The plug has 4 pins with a threaded shroud that pulls it into the female plug and I couldn't make it screw on. Tried the mill motor, it was fine. Checked the threads, looked ok. But it didn't turn freely anymore and as I was screwing it on I noticed a pronounced wiggle in the male half of the plug so that was clobbered by the motor and bent. I'll have to replace it. The motor must have fallen on the plug and given it a mighty clonk.

But I forced it on as far as I dared too try the sensor and try and get some safety valve bodies made.

Reset my tools etc, tried to run a simple OD turning program and no go. The sensor ran but the head/table wouldn't move once it was onto a cutting path. Turns out the new sensor isn't sending a signal that the control cabinet can pick up.

I was too angry to take it all apart and start debugging tonight. I'll have to do it tomorrow while I'm on weekend tech support duty, if I can find a quiet time. I was too stupid to take a photo of the cable run of the old sensor and how the 'strain relief' cable ties were put on in relation to all the joins near the sensor. I did my best but I would not be surprised if I've buggered one of their solder joints by bending it the wrong way or too much.

Me and Tormachs don't get on. The first one delivered was broken. As soon as I started to feel comfortable with the mill and put the RapidTurn on I've had nothing but trouble with it too. I'm cursed, or it is. I really want to sell the lot of it and pack it in right now but with my job going to sh*t as well I need to keep working on this and see if I can make something of the mongrel machine. It would free up about 20% of the workshop though.

Feeling pretty damned frustrated right now.

[Roger]

Those are annoying problems, but there doesn't seem to be anything that can't be put right with a little patience. Try to look at the machine as a project in itself, I'm sure you'll soon have it working as it should.

[David]

Good news, it's working again.

1. Plugged old sensor in and held it near spindle - it triggered. So the plug on the control cabinet ok, and the tool length setter didn't blow anything up, and the setting I changed to make it work does not impact RapidTurn.
2. Took RapidTurn apart, tried the new sensor near the spindle. It had to be really close to trigger. I thought I had it as close to the steel 'flag' on the spindle as was safe, apparently I was wrong.
3. Put it back together and used a folded piece of paper as a feeler gauge. Took it back over to the mill and tried it by just turning the spindle by hand. Seemed to trigger.
4. Put it back on the table, put the motor on etc. I tested the sensor at every stage to ensure it still triggered.

I did a few quick test and it seems to be able to read it at speeds above 1200 rpm now. Only tried it to 2000 rpm because I have the belt in low speed posn and don't want to fart about with it - I want to use it.

The motor connection plug still needs to be replaced.

I'm not happy with how the sensor cable has to be bent to get it all in place. It's quite stiff with what seems to be a couple of joins at the sensor end and this end need two pretty severe bends in it to get it to fit. It's a problem waiting to happen and I'm sure the solder joints will fracture soon enough.

[David]

While waiting for problems at work, which happily didn't arrive until after I signed out, I spent the day making safety valves. I made 4 bodies and 6 caps which are a small top hat shape with a thread and a hole through them. This won't be a surprise to anyone else but the actual machining time was trivial compared to setting up/tearing down the RapidTurn, opening/closing collets, moving stock around, changing tools, etc. So it took hours, most of which was not actually cutting.

It took about an hour to dial the RapidTurn to be square on the table and checking/setting all the tool offsets to be sure they were correct. This should not have been necessary, but was.

I didn't get to drilling the 1.6mm holes in the base (I forgot), and I only tapped the stem of one body. That's the most time consuming operation and it locks the safety valve stem (column?) to the fixture it screws into and I don't have a good way of releasing them yet.

That was basically 9am - 5pm with a break for lunch and a few tweaks to the model and programming!

Two things I learned about the aluminium fixture I use for drilling 'boring' them on the mill:

1. I found a way to get good enough repeatability in X and Y by chance. The plan was that I'd milled one end of the fixture square and that was meant to line up with the ends of the vise jaws. I didn't think far enough ahead to realise this put the stem over one of the slides of the vise and I'd probably drill down into it if the stem sank too far into the fixture. So I have to position the fixture such that the stem is in the gap between the vise slides, where the jaw screw is. I did this and was sliding it across to try and get it over the deepest part of the gap when the thread at the bottom of the stem clunked into the side of the slide. The nice thing is that this uses the stem itself for positioning the X axis and the Y seems close enough.

2. I was so pleased with the above I forgot a tiny variance in the diameter of the stem taper makes a big difference to it's vertical position in the fixture. So the bottom of my bores vary by about 1.5mm! Given you're adjusting by screwing down on a spring this would be fine except the springs I have are already too long and I couldn't fit the caps on the valves after I'd set them last weekend.

I was advised to just grind a bit off the springs. I can also gain about 1mm by making the plungers that hold the ball down from 8mm stainless rather than the 6.35mm I have now and I won't need the brass spring seats I made for the first valves. Additionally I cut 1mm off the threaded part of the caps.

I may run the boring op again on the ones I made today, resetting Z every time. I'll see how they go on air before I bother, it may not be necessary.

It occurred to me I should have left the RapidTurn on the mill table and mounted the vise further to the right, so I wouldn't have to set it up again. Maybe next time. I wouldn't want to leave it there permanently but I should have done so until the batch is done.

[David]

So those safety valves. No good. The thread is undersized so I've spent the last few days trying to figure out how to turn one that works. Yes, a die does it in less than a minute, but you know how it is. I paid for the rotten machine so I have to make it do the work!

I think I've finally turned one that's ok - 3/8x32 ME. I turn the major dia as close to nominal as I can, then I use a 55deg insert to cut the thread to a programmed 8.4mm minor dia. No idea what it actually is though. But the last thread I did I cut it to 8.525 figuring that would be right given ME threads are 0.5mm deep and the nut didn't quite go on. I figured I'd go down in 0.025 increments until the nut went on and it only took one extra go. It all sounds pretty obvious, I know, but I made 8 dud valve bodies not doing it the simplest and most obvious way. Bloody thread tables that give you all sorts of weird numbers on them.

Yesterday I cut down to 8.34 before the nut would go on, but that seemed excessive. I also didn't know for sure the insert was a 55deg one and not 60deg, helpfully it's not marked on the insert itself.

The thread doesn't look as good as a die cut one due to the sharp crests, but as long as I don't supply it undersize the owner can decide if they want to clean it up. The insert doesn't round the crests, that part of the insert never gets anywhere near the work on a 0.5mm deep thread.

I've now had a fair amount of practice setting tool offsets etc using the offset table so that's something as long as I don't forget it next year. I should write my method down! I managed to change my tool 3 to a larger theading insert size (the only one I knew was 55deg) and didn't crash or otherwise stuff anything up.

[jasonb]

If you are using a partial form insert which it sound like you are then the tip radius will be smaller than that for 32tpi so you will need to increase depth accordingly or invest in a full form insert that will not only cut to the right depth but also round the crests. Best way to measure it would be with thread wires then you know the flanks are correct regardless of crests and valleys.

If you are selling then then can't you could just set it rotating and run a needle file along the crests to round them if you don't want the expense of a full form insert.

[David]

Would there even be an insert available suitable for ME threads? In metric and Whitworth threads I assume the thread depth varies with the nominal diameter of the thread. How do full form inserts cope with that - is there a different one for every thread?

I don't want to cut the threads down at all. I was told they were under dia and given these are for safety valves I don't want to risk making them small. I wouldn't care much otherwise. If the purchaser wants a nicer thread form they can use a die that they no doubt have. If I do it, I might have the die too tightly closed!

I would love to supply a perfectly formed thread but I don't see how I can do better than an approximation that isn't undersized.

[vulcanbomber]

A 32 tpi full form insert will cut that form at any diameter.

ME threads use whitworth form 55degree inserts.

[David]

I am using ER16AG55 inserts at the moment, and have ordered ER11A55 (with SMX35 at the end of the name - no idea what that means) for use with the little toolholder that comes with the RapidTurn in future - are they both suitable?

Still confused how a full form insert can handle different thread depths when it comes to the outside crests but I'm not likely to ever own one so that's not really a problem.

EDIT: As far as I can see 11ER32W or 16ER32W are the codes for full form 55deg 32TPI threading inserts. Is this correct?

[David]

I just spent a long time transforming my test program from the last few days to the real model and tried a test cut. Straight away the thread dia was 0.05mm (0.002") undersize.

This was puzzling given I've been turning dias to within +- 0.025 since yesterday when taking test cuts myself.

I did some tests typing G0 X... commands into the command prompt for dia and skimming the tool over the stock with a 3% jog speed.

It turns out the error happens when the mill head (so lathe toolpost) starts lower than the dia and moves "up" to it which I never do in my test cuts but F360 does all the time. Each time I move the head low then move it back up to where I want and take a cut, it's 0.05mm under dia. If I start the head high and move it down to where I want it's within 0.005mm. I did two tests for each situation and they were consistent. Not a huge sample size, I admit.

This makes no sense to me. The head is constantly moving up and down so won't any lost motion just accumulate over time and get worse as a program runs? My Tormach has stepper motors with no feedback so the controller can't take account of this lost motion - it doesn't have backlash compensation either.

I'm going to increase the thread major dia by 0.05mm as I can't think what else to do! But I'm stumped as to how you can run any sort of real part and expect it to come out accurately given all the movements for anything non trivial. As it is I think I'll have to reset X as well as Z for every run.

[jasonb]

Yes those are the right codes for 32tpi inserts. one will cut different diameters but not different pitches as thread depth is the same be it 3/16 x 32 or 3/4" x 32. Also worth looking up what depth of cuts the maker suggest, looking at one they do it in several decreasing depth passes which is much like you would if doing it manually on the lathe so will let any spring out of the tool.

Sounds like slight backlash in the Z axis, probably something in your system to compensate for it, Mach3 has the option to do so. Easy enough to check if you put a dti against the spindle and jog down and then back up, if needle does not move for 1st or second up jog then that is your backlash.

As I said thread wires are really the only way to measure the thread you have cut accurately if using partial form, if using full form then that should be closer when measured over the crests provided the whole of the insert has been cutting.

[uuu]

If you're getting lost steps, then it may be excessive friction in the head support, or the acceleration that the controller is expecting is beyond the power of the motor. You can address the first with a bit of maintenance. Are your acceleration settings adjustable (I'm using Mach3, where these are easy to set)?

Wilf

[David]

PathPilot is Tormach's own controller so it should have the correct settings for the motor.

As for backlash compensation... no. I looked through their Python files once and saw they defined variables for them but they aren't used as far as I can tell and there is no place in the UI you can set the values. Although I was wrong about single stepping and the button was right there on the front page so who knows?

I've just googled it and someone says given it's based upon LinuxCNC it should be possible to do it with LinuxCNC config, even if PathPilot doesn't allow you to set it.

But I haven't even done the X/Y backlash measurements or gib adjustment procedures so you can see I'm not really on top of these things. I did measure the Z lost motion with a DTI once, and there is lost motion, but beyond that I can't recall what it was.

[David]

I found a good feature in PathPilot today. It has conversational programming and you can append conversational operations onto the end of existing files so you can build up a more complicated program from simple steps.

I'd never bothered to do this because I didn't think you'd be able to edit the resulting program except as a normal text file / big g-code program. I didn't even know you could edit a single operation file in the conversational interface.

However I pressed the 'conv edit' button for the first time today and it uses all the comments it puts in to fill in the conversational fields. For files made from more than one operation it even gives a list of them and you can reorder, edit, insert new ones, remove them, etc. I was impressed.

Some parts and operations are much more easily done with the conversational interface than faffing about with F360. Simple stuff like facing, turning down to a diameter, chamfering, threading, parting off, can be done this way fairly trivially and avoid the back and forth of post processing and copying g-code from the PC (even thought that's really easy with a SAMBA shared drive).

You wouldn't do it for anything you wanted to simulate or that was much more complex than the sequence I gave above. But I can't get threading to work from F360 so I was using it to do everything else and PathPilot conversational to just do the thread. Meaning I had 3 programs to run including the parting off. I didn't want to try and jam the threading code into the middle of F360's output in case things went bad.

F360 has means to put your own code in, and linuxcnc can make calls to external files, but I've not got either of those work yet.

[David]

Guess which profile here suffered from the cutter re-cutting chips and getting clogged up.

This was a brand new cutter. 2mm dia, 5000rpm, 45mm/min feed, 0.8mm depth-of-cut. This is why I cannot leave my automated machine unattended! Luckily I was in the room at the time and noticed the change in the noise being made. The spindle was already getting warm by the time I got there. I let it finish the cut figuring it would be easier to edit the g-code to start on the next bit. I got away with it and the cutter didn't break.



I was able to stop it, pry the jammed chips out of the flutes, edit the g-code, and continue.

[jasonb]

Can you not set up a constant air blast or turn the coolant on, the other guy I know cuts just about everything on his Tormek with flood coolant to clear the chips more than actually cool things. I'd rather clear the chips than compromise the actual cutting.

It's the only thing that stops me leaving the machine unattended on certain materials/cuts but I'm in the process of setting something up that will also add a small amount of lubricant to the air if needed for materials like aluminium. At the moment I'm just using a small blow gun as and when needed

[David]

Yes, I'm using an air gun too with very low pressure.

I actually bought one of those cheapo misters over a year ago. Just not got around to installing it because I do so little work with the machine it never seems worth it, and any workshop time would be better spent on something else. But it seems I should get on with that project.