Using an Imperial Ainjest Threading Unit for Metric Threads

Hi all,

 

I have a background project to fit an Ainjest high speed threading unit to my lathe. Both the lathe and the Ainjest unit are imperial, so cutting imperial threads should be no problem. Just follow the instructions in the manual!

 

Within the QC gearbox on my lathe there are conversion gears for cutting metric threads. Nowhere can I find a 127 tooth gear in the system, so I am assuming that the ratio used is not exact, but is one of the very close approximations. As is normal, the advice for cutting metric threads on an imperial lathe is to leave the half nuts engaged and reverse the lathe back for the following cut. This is not possible with the Ainjest unit, as it includes an automatic knock off at the end of the thread.

 

There is a piece of equipment, called a Metradial, that can be used in conjunction with the imperial Ainjest unit to cut metric threads. I assume that this is just a fancy thread dial indicator that allows you to keep track of the starting position. The company in the UK that sells the Metradial seem to be useless, no response to emails and no 'phone number. Clearly they don't want my business, so they're not going to get it.

 

In essence the problem is how to get the lathe spindle and leadscrew back into the alignment they were originally in, for subsequent cuts. I'm not going to bore you with the number theory behind this, but in essence I think that for cutting a 1mm thread (with a 1/4" pitch leadscrew) the alignment should be exact for every 127 revolutions of the spindle.

 

But, there's a catch here. The conversion factor in the gearing is not exact, so the result will in reality be close, but not perfect. This is hinted at in the Metradial instructions, where there are correction factors to be applied if you are a bit slow in re-engaging the half nuts.

 

So, my idea is to return the lathe to exactly the point it was for the first cut. There are two things that need to be set. One the position of the tool, and second the alignment of the spindle and leadscrew.

 

Positioning the tool is easy, using a bedstop against the saddle. For aligning the spindle and leadscrew I propose fitting a rotary encoder to the spindle and counting the number the number of revolutions. If I then reverse the lathe and take the count back to zero, the spindle and leadscrew should be in the original alignment.

 

So, I have a couple of questions:

 

Has anybody used an Ainjest unit, and can give me some general pointers as to their use?

 

For cutting metric threads, will my idea work, and if not, where is the flaw in my logic?

 

Thanks,

 

Andrew

I think you blinded us all with science.

 

It seems to me that you are right, that if you can get everything back into the exact relationship that it was when you started then it should pick up OK. But it may well be that there are many other points where it is very close but not exact. Let me think about htis for a while.....

 

Incidently although I have a 127 tooth gear for the Myford, I have never tried to use it....have you seen how big one of those things is! I have never needed anything better than the approximations anyway.

 

regards

John

Hi Andrew, like your picture, but can't remember his name. However I don't know anything about the Ainjest unit, I think your theory of resetting your lathe back to zero sounds feesable. One thing I would consider, would start with the encoder on a negative, bring it up to zero and then make the cut. For each reset you could take it back to negative and bring it back up to zero before making subsequent cuts. This should make sure you are starting from the same point each time and elimenate any accumulative errors due to backlash ect. But of course you may have this already in mine.

I just put a splodge of tippex on the change gears and realign the splodges...

 

Your metric conversion gear might be a 73t.

I use a 63t because what the nut don't know won't hurt it.

Got an 8tpi leadscrew.

1/4 inch is a 4tpi leadscrew? is that right?

 Dat's a chunky one.

 

So you need just over 6 turns for an inch.

20 on the spindle = 127 MINIMUM...holy smoke

 

If you've not got a 127 in there then there must be a compound gearing arrangement to gear things up, I can use a direct drive system with an 8tpi leadscrew.

Great stuff, thanks John. It's always useful to have someone looking over one's shoulder, in case of me missing the obvious.

 

I haven't seen a 127 tooth gear, Myford or otherwise, but I understand that they are LARGE! My lathe certainly doesn't have one. I haven't yet worked out which set of gears it does use for the metric conversions.

 

A bit more playing with the maths seems to indicate that the key distance is 127 times the metric pitch of the thread to be cut. For instance, if the thread pitch to be cut is 1.25mm, then 127 times this is 158.75mm. But 158.75 divided by 6.35 (1/4" in metric speak) gives exactly 25. So, the spindle does 127 revs and the leadscrew does 25 revs. Given that 127 is a prime number this fraction cannot be simplified further. Since the numbers are both integers, then as far as the two threads are concerned, the relationship between the threads is the same as at the start.

 

Regards,

 

Andrew

The character is called Chad, in the UK at least, and was normally accompanied by a long suffering comment on the deprevations of WW2, such as 'Wot, no sugar'.

 

Nick: You are absolutely right. An encoder with one pulse per rev could lead to errors if you weren't careful. I was thinking of a 100, or even a 360, per rev encoder. Then a plus or minus one count error on the spindle position would be small in terms of degrees. It would still be pretty slow in electronic terms for the counting, even if the spindle was running at 1000rpm.

 

Ady: Correct, the leadscrew on my lathe is 4tpi. I can't find a 73 or 63 tooth gear in the system. My lathe (Harrison M300) has a few change wheels in the normal place down to a quick change gear box. The QC box has an eight way rotary selector plus three, three position, selectors. Metric threads seem to be associated with one position on one of the three way selectors. So there must be something clever in the QC box, but I haven't identified which set of gears yet.

 

Regards,

 

Andrew

The usual 8tpi arrangement is

46 ......spindle

X ........driven gear

40/20.compound gear (0.5)

73...leadscrew

 

So 46* 0.5 = 23

73/23 = 3.1739 * 8 = 25.39tpi

 

With a 4tpi leadscrew you would need an 80/20 or two 40/20 compounds

.....or more simply.... a 23t on the spindle

If you don't have a 63t OR a 73t then you must have a magic fairy dust system.

 

Let us know if you suss it out, sounds intruiging.

The EXE lathe used a dog clutch system in the spindle so you could "never pick up a thread in the wrong place".

 

http://www.stationroadsteam.com/stock%20pages/2589/pages/2589-d.htm

 

http://www.lathes.co.uk/exe/index.html

The pitch circle of the 20 DP 127 tooth gear for the Myford will of course be 6.36 inches, so with the addendum we are close enough to six and half inches, or about bread and butter plate size.I think actually a more useful setup would be to make a pair of gears in a much smaller tooth size but with the standard changewheel width and centre hole. Something tiny like 0.5 module. They would always be used together , but the rest of the train could be standard gears.

 

Anyway, back to the real issue:

 

Lets suppose that on the first pass you engage the leadscrew with the saddle in a known location. (DRO, stephens marker on lathe bed, whatever) We take a pass up the job and count the rotations to a reasonable degree of accuracy, and disengage the half nuts. Now we run the lathe backwards until the counter reads zero, and run the saddle back to the marked position. Now, if we engage the half nuts, surely everything is in the same relationship as it was when we started. How much accuracy is needed I wonder? Suppose with the quarter inch leadscrew we are cutting a 1mm pitch, the ratio will be somewhere near 6.35 to 1. (Exact if we have 127 tooth in the train.) Eg the spindle turns 6.35 turns for one turn of the leadscrew. Now, to be sure we are in the same place there are two factors...we need to get the saddle location right to a small fraction of the leadscrew pitch, so that if the leadscrew is in the right rotation position the nuts will engage in the right place. Then we need to get the leadscrew close enough to that right location. With the pitch of the screw to be cut being less than the leadscrew, the tolerance on counting the lathe rotations is not so critical, eg if we were even one turn out, it would only be 1mm of misalignment in trying to put the half nuts in. But if we were cutting a screw coarser than the leadscrew, we would have a tighter tolerance on the counting of the rotations.

 

So given that an electronic counter with a resolution of 100 counts per rotation is easy to do, I think you could use that, make sure the saddle is repositioned within less than 10% of the pitch, and everything should be perfect.

 

I guess the tippex is the up to date version of the old chalk idea. Now that everyone is using word processors how long will we be able to get Tippex? Also very useful when silver soldering.

 

regards

John

Hi John,

 

Thanks for having a look at this; I'm glad that you agree it should work! Using a reasonable count encoder it should be possible to get the spindle back within a degree or two of where it started. By using a bedstop for the saddle I guess it should be possible to get the saddle back to it's start position within a few thou. I reckon that will be more than good enough.

 

I guess my next steps are to get the Ainjest unit fitted and then learn how to use it to cut imperial threads. Then ,once I've got all the CNC milling and TIG welding out of the way for work, I can get on and try fitting an encoder to the spindle.

 

No doubt some enterprising business type will latch onto Tippex and sell it to machinists as a magic, all purpose, must have, product for the workshop at an inflated price!

 

Regards,

 

Andrew

Posted by Andrew Johnston on 05/10/2010 19:40:20:

 That's Kilroy. as in Kilroy was here

 

John S.

Hi John S, Andrews picture may have been Kilroy to some at somtime, somwhere, but when Andrew said it was Chad, I remembered that's who I remember my farther refering him to during WW2.

Kilroy, as in 'Kilroy was here', in the US, Chad in the UK. As is often the way with these things the origins are somewhat lost in time.

 

Regards,

 

Andrew

Hi Andrew