By continuing to use this site, you agree to our use of cookies. Find out more

Member postings for Andy Ash

Here is a list of all the postings Andy Ash has made in our forums. Click on a thread name to jump to the thread.

Thread: Brushless DC motor for machine tool power
18/07/2016 18:15:45
Posted by John Rudd on 18/07/2016 07:24:18:


What about horse power size? Did you go for same as existing, 400 watts?

Hi John,

The motor is 1/4 HP which is 180 Watts, considerably less than the rating of the DC motor and PSU I showed.

I just looked up the stall current on the 1/4 HP AC motor. I wanted to compare like with like.

The motor I chose looks like it comes out at 1500 Watts (2HP) in those terms, but there is no way you could use it like that continuously.

A real 1/4 HP (continuous/shaft) is more than enough to drive a small lathe or mill.

17/07/2016 22:27:24

Hi John,


I actually went for the four pole motor.

The spec sheet for the motor shows that it will happily run to 75Hz which I think is roughly around 2200rpm.

The actual motor is is characterised in the data to 4000rpm but I don't need more than 2000.

In fact I was more worried about the low speeds. I machine mostly steel and bronze, and prefer to use larger cutters where possible. I have allowed two pulley speeds. The high speed is 1:1 with the motor shaft. The low speed is 2:1 reduction.

I've been using the side of a 19mm cutter on steel this afternoon. The cut was about 3/4 of the full height of the cutter, and 20 thou per pass. It didn't even flinch. I couldn't be happier.

17/07/2016 21:22:36

I'm not sure what machine you are trying to apply this to, but I think you should save yourself some time and money.

Earlier this year I had been looking for a mill to fit a space in my shed. A bloke at work had the machine below (I don't have a picture of it at the time, but it shows the type) for sale, and he only wanted £75 for it. The problem was that he had bought it from someone who had killed the original motor/drive, and had tried (with the help of someone else) to fit a new brushless DC motor to bring the machine back to life.


The machine came to me with a 400W 48V e-bay motor a 400W 48V Mean Well SMPSU, a home made coupling and a generic Chinese speed controller. Pictured below;


There were a lot of problems with the machine, and in particular the drive. The most basic problem was that the motor gear had a different pitch to the gearbox input shaft. They didn't mesh properly and consequently the result was a terrific howling noise since the motor shaft runs at 4000rpm. Some of the gears were plastic, and so the metal ones were gradually training the plastic ones to be the right size!

The head bearings were full of the thick white paint used to make the machine look pretty. As a consequence the bearings were shot. The machine didn't look like it had got a lot of use, and I think it was probably nearly impossible to use.

At 400W you might think that this motor and power supply were man for the task, but I can assure you they are not. I put the motor above in a vice, and at full tilt I could stop the piddly little shaft in my fingers. I put a meter on it, and at stall it was taking the full 400W!!!!

The thing with most DC motors is that they describe power in electrical terms. That is to say the electricity consumed, not the Horsepower developed.

I looked at this business of BLDC motors, and I actually found one on ebay with a drive that looked good for a proper 1/4 HP. It was not a direct fitment for the machine, and it was £250. Not only is it expensive, it is also rare. If I make the effort to fit it to the machine, then I want to be able to get a replacement easily, in case I burn it out.

Given that the gearbox inside the head was plastic, and that people say these machines are noisy anyway, I decided to go for a three phase AC motor with a Z-section vee belt drive. It turns out that you can fit a pulley directly to the spindle. Quiet, efficient, simple, cheap.

I got the motor from "inverter drive supermarket" and the motor and the drive were a fraction over £100 inc VAT.

I made the steel bracket from parts that I had laser cut, and I welded them together.

I decided to go a little to town with the control box. You wouldn't have to go that far.

If you really want a decent machine drive, it doesn't have to cost the earth. Once done, it's all standard parts, so it's done forever. If anything goes wrong, you just buy new bits and drop them in.



Thread: Generating division plates from scratch
14/06/2016 13:09:51

I did think about this a little, and the basic difficulty I have is that the problem is ill-defined.

I made some of my own assumptions and it all became more interesting.

I'm assuming that I only have a lathe, and no gears. I'm thinking that the lathe is probably made of wood.

The best I could come up with was to make pulleys and use belts to do division. Obviously any fool can divide a circle into one division. If the pulley sizes are controlled you can get any division ratio for one revolution of the first pulley.

Flat pulleys are good because you can always make good belts, all the time you can still get cows!

My only problem now, is that I need a micrometer, and something must have put divisions on the thimble.

Producing a screw without gears, is a tall order too.

I don't know how they must have done it, but I guess they kept trying with the same idea we have today. Eventually things became good enough.

I think clocks came before engines, and clock-makers still make gears and escapements out of sheet with a scriber, wire saw and file.


Edited By Andy Ash on 14/06/2016 13:11:12

14/06/2016 12:57:23


Obviously only any good if you are happy about aluminium.

I recently bought pulleys from Robert. They arrived modified exactly to my request, and for a very reasonable sum.

He's really good if you need custom bores and key-slots broached.

Thread: Drawing a blank on these.......................
11/06/2016 10:47:11

It occurs to me that these could be used to check run-out at a realistic distance from a milling chuck.

Obviously you could just do that with a bit of ground stock, but these have the advantage that they have ground threads, and would be correctly mounted in a Clarkson autolock collet chuck.

Obviously you could clock a cutter near the chuck, but you would not be able to clock it at the tip where it matters. With a blank, those problems vanish.

Thread: Myford ML2 Change gear headache help needed.
27/05/2016 21:23:13

I don't think it makes a difference but I noticed something else about that machine.

There is a square boss on the back of the headstock casting. On mine the boss has a spot face and a tapped hole. The hole takes a threaded pin and is the pivot for the tumbler assembly. Given that the hole is not there, then the machine must never have had a tumbler originally.

I also wanted to qualify something I said earlier.

"The main thing of importance is that all the gears you use have the same pitch."

With hindsight this should have read;

"The main thing of importance is that all the gears you use have the same pitch, where they mesh".

Hopefully it was obvious that you can mix gear pitches in a gear train. Where the gears mesh the pitch must match or the teeth will interfere, crash and bind.

Thread: An Application to create dividing head tables
26/05/2016 00:35:52

Hi John.

MOD does what it's supposed to. And for what we are trying to do, it has more merit than most.

The trouble is that I was using it in a way that just wasn't correct for what I had wanted to achieve.

Because of that, the spreadsheet wasn't reporting all of the possibilities that it should have.

I know you have mentioned the business of floating point arithmetic quite a number of times. I don't know how to say this tactfully, but I just don't buy into it. I don't think there is anything wrong with using fractions, but I am also completely satisfied that decimals are just fine too. The floating point unit in any machine capable of running windows, can do this sum with enough precision that you could index individual atoms if your dividing head was up to it.

If you look you will see that I have explicitly set zero. I figure anything less than 1x1012 to be "zero". If there is an error less than this then it isn't worth worrying about. Bear in mind that an atomic nucleus is considered to be 1.75×1015 of a metre. This is not an uncommon issue when using floating point numbers.

The best example is comparison. In software, although you can compare two floating point numbers to see if they are equal, there is almost no point in doing so. When you think about it, that is quite reasonable. With so much precision there is almost no chance that two different quantities will be the same.

Generally the computer is less good at handling values with reoccuring decimals. This is because it only "appears" to have limitless precision. It is true to say that fractions can help humans deal with these scenarios. If you know you want the computer to do the grunt work, then I think it's decimals, all the way.

It's not actually decimals which are the problem. It's the computer. If you were to write down the whole string of a decimal, then it would be just as good as a fraction. The computer does not do this because it's designers knew better. As it happens the floating point unit is more or less a fraction machine. The mantissa and exponent are an expression in base 2 scientific notation. Nevertheless they work together almost the same way as fractional values.

As an example, floating point numbers are regularly "normalised" by the computer. This process is the same as reduction of an improper fraction. One of the main departures is that the computer does not attempt to distinguish units. It does not try to say 1 7/8 it just says 15/8. During calculation it might arrive at the quantity 30/16 which would be "normalised" to 15/8.

You have to bear in mind that I have simplified this. the sums it performs are actually in base 2. I did that example in base 10 just to make it obvious.

You might be able to choose the right fraction to deal with a reoccuring number, but the computer has no way to do that. The real observation is that I am sure that the computer is prepared to maintain more digits in it's floating point "fraction" than any human ever would be bothered to. In my opinion on precision, with floating point numbers pitted against inspired human fractional choices, you win some, you lose some. The main thing is that the computer does what it does FAST.

Don't forget that I'm not insisting that you use my spreadsheet. I fixed it because a fault was raised and I didn't want to lead anyone astray.

I think using the variable precision is an extremely useful tool. To use it well one must understand what it can and cannot give.

Thread: Myford ML2 Change gear headache help needed.
25/05/2016 23:18:29

I have an old ML4 which is very similar to that, and I'm sure that your change gears and mounting are very different. Someone has heavily modified your machine, but that is likely to be a good thing. The actual Myford gears were quite good, but the original method of mounting left a lot to be desired.

I don't know if all of your gears are metric or not. There's no reason not to use metric gears if that's what your machine has been converted to use. The main thing of importance is that all the gears you use have the same pitch.

Don't forget that idler gears can be anything. An easy way to think about this is one tooth in, one tooth out. The only way you get a change is to have two gears on one shaft. Then you certainly don't get one tooth in one tooth out.

Your lead screw is probably 8 TPI, but that might have been changed too. If it is 8TPI then the 20T and 50T gears sound right for 20TPI. Just use whatever idlers bridge the gap. It looks like you have a tumbler reverse, so you can have odd or even numbers of idlers, as you find convenient.

Edited By Andy Ash on 25/05/2016 23:18:47

Thread: An Application to create dividing head tables
25/05/2016 22:14:37

I finally got some time to check my revised spreadsheet today.

It's was set up for the Myford dividing head but now you can specify the worm by typing a new number in the box. The same is true for the hole circles.

The calculations were always right, but the filter to hilight the matches, had several problems. I'd not realised this because it had always been acceptable for the number of divisions I had needed. It has been good to put it up here though because I have found and been able to fix the problems with it.

The filtering is actually slightly different now too. I wanted exact and near matches to be hilighted in different colours. That way you can instantly see if you are about to make a compromise. You couldn't do that well enough before. Now the options which carry an error are hilighted yellow, but the exact matches come up green.

The colouring in excel works off the value in the cell, and the way I figured to handle this was to use a minus sign. In general all of the numbers on the spreadsheet are positive. In the real world, there are no negative divisions. If you see a minus then the value carries an error, and this is used to drive the hilighting.

Both the whole number of turns, and the number of holes on the circle are reported this way. Obviously the dividing plate corrects the error in the whole number of turns. You can correct any error in the number of holes in the circle as I described previously. You only need to do that if you use a number hilighted in yellow.

I found the Myford manual on the internet, and I have checked every division from 1 to 100 and the answers are all there. I think it must be right this time.


I don't know why I used the MOD function. I must have had an off day. It is very odd that it seemed to be nearly right. By analysis, I can't see how I ever expected it to be! I have now used TRUNC, but I don't think I've used it the way you have. The fact is that I don't know what you did, but that makes me think I did something different. If they are the same, then that's probably the only way you can do it.


Have fun.

Edited By Andy Ash on 25/05/2016 22:17:48

23/05/2016 23:36:30

I finally got an hour to look at what I previously did considering what Bazyle had said.

The numbers I was reporting were correct but the method of filtering was quite seriously flawed.

I think I've got it licked now. I think my spreadsheet now does what I originally thought it did, plus some new stuff.

I've no time tonight to explain and post it.

There is an update coming. ***Especially after I do some more checking!***

18/05/2016 22:54:21

I am completely confident about the principle I described before.

I seem to remember someone saying that there is an article about division coming up in the ME, and I wouldn't be surprised to find that they explain what I'm struggling to.

Perhaps you don't realise it, but you have caused me to realise there is a deficiency which could be improved. Whilst I'm in there I think I'll add the capability to do a variable worm because others have been fiddling with that. Then I know it's all covered properly. I probably will add an "exact flag" so that you can switch between exact and the current proximity, to compare the highlights. I might even be able to the exact ones a different colour.

I have to say I wasn't completely sure of the example you cited before, of the target division being a multiple of the plate. I think I tried 32 and 64 seemed to give answers that I expected, so I wasn't quite sure what the specific problem was. Irrespective I looked at what you changed and saw a problem with what I had done.

As for 15 divisions on a 15 hole circle with a 60 worm, surely that's just four whole revolutions and no holes.

60 / 4 = 15

I'll try and do an updated sheet with more searching oomph!

18/05/2016 21:10:26

Hi Bazyle,

I said I would have a look at what you had done, and comment. I still don't have much time today, but I did look.

I think I can see what you're getting at.

So as far as I can tell, there's still nothing actually wrong with my spreadsheet, but I do now understand that it might not have been generating some of the options (the ones that carry an error) that you might have expected. It's not something I had thought about but I can see now that there is an issue here.

The first thing I would say is that if you turn the proximity down to zero, if there is nothing hilighted, then there are no "properly accurate" selections. Those that are shown are accurate to the available precision of the floating point unit on the computer. I think excel uses the IEEE 754 double precision floating point value (**LINK**). If the selection is that accurate then to my mind it is "spot on". Those are the choices one should always use if possible. They have both integral and differential precision perfectly.

Then there are the choices that carry an error. Of those, the way my sheet works it will only return choices where the hole count is slightly short, never slightly over. It is fair to say that for any given selection there might be a closer (more accurate) choice which is slightly over rather than slightly under. My spreadsheet would not necessarily notice those if they existed.

I think all I would say about that, is that if you have the compromise in mind (as you possibly should), then once you have found a candidate that you like, then you will have calculated if it is accurate enough for your purposes.

I didn't have time to look at your longer function in detail. I possibly still might. Equally I think I have seen what you are getting at. I might have a go at expressing an adjustment for the flaw you have identified and see if I come up with the same answer. Also it will be interesting to see if it actually makes much difference.


Edited By Andy Ash on 18/05/2016 21:11:40

17/05/2016 22:58:15

Roger that Bazyle.

I certainly think that making the worm an adjustable parameter is a useful thing if you're exploring division possibilities. I only have the one head so it never mattered to me. It would be quite a straightforward change to make.

The spreadsheet isn't actually that complicated mathematically. What's incredible is that the computer can do so many sums, so quickly that it can answer all the separate problems in the blink of an eye.

The worm sets the number of revolutions for a complete circle. The actual rotation of the handle for one division out of however many you want can be expressed as a fraction;

Worm revs / Number of Divisions = Actual rotations of worm for one division.

If you do that sum, then you get a number, say 2.456.

The two whole revs is easy. But you also have to get the 0.456 part. It's just added on, but the holes on the plate help to define it. If a hole circle has 46 holes and you use 35 of them, this too can be expressed as a fraction.

35 / 46 = 0.760.......

That is to say using 35 of 46 holes, you can define 0.760 of a revolution. We actually wanted 0.456 of a revolution and with our 46 hole circle we can ask how many holes makes 0.456 of revolution.

0.456 * 46 = 20.976 holes

Obviously this carries an error. The closest you can actually get is 21 holes. So you're left with an error of 0.024

My spreadsheet simply asks how many holes of each hole circle do I need to use to achieve complete accuracy. On top of that is an algorithm that rounds up or down and decides if the difference (proximity) between that nearest integer is bigger or smaller than the specified limit for the whole sheet of hole circles and divisions. If it is smaller, the results are hilighted, if larger they are zeroed and not hilighted.

NOTE: Watch out. The way the columns are aligned it may not always be evident that decimal values are being returned. If you have any doubts temporarily stretch the cell width. The wider you make it, the more precision you will see.

If you set the proximity bigger, you get more matches, but you will notice that more of the results are actually decimal values. The spreadsheet might say 0 full rotations plus 22.08 holes on the 46 hole circle. Obviously you can't have 22.08 holes. You can have 22 holes. I was aiming to get 125 divisions.

125 * (22 / 46) = 59.782

Over the whole circle I come up short. How short?

60 - 59.782 = 0.217 of a revolution.

So how many holes is that?

0.217 * 46 = 10 exactly.

So if we were to divide a circle into 125 divisions using 22 holes on the 46 circle; we'd get to the end of the 125 divisions and we'd be short of 60 revolutions, by 10 holes. It's still not precise, but you can achieve integral precision. All you do is add the extra ten holes gradually throughout the division operation. If you have 125 divisions you could normally use 22 holes on the 46 circle. Then every 13th division use 23 holes.

The differential precision still has a problem, but the integral precision is now fine.

The only way you can have both, is to use a hole circle that reports an integer number (nothing after the decimal point) on my spreadsheet. These integer numbers are those which are typically only reported by the manufacturers on the table that comes with the dividing head - say Brown and Sharpe.

You can make my spreadsheet so that it only shows the integer numbers, but I find the variable precision is more useful (because I only have a couple of plates) so I do it that way. I can get at things I wouldn't otherwise be able to get.

I have to go make some zeds right now. (I hope my worked example sums were right!!! I'm sure someone will say if not.)

I'll try to analyse your worksheet functions tomorrow Bazyle, and compare them with what I did originally. I'll let you know whatever I think.

Edited By Andy Ash on 17/05/2016 23:27:44

Thread: new to myford lathes
17/05/2016 22:08:12

Hi Wayne,

I'm guessing that they have only welded the bearing to the top of the casting, so that the adjuster can still be used. If they had welded the outer to both parts of the slitted boss it would have prevented the slot from changing size.

Replacing the gib screws is always good. Make sure the locknut is good too. They'll wind themselves out if they can. Having good slots or even using socket screws can really help get the setting "just right".

The silliest little things make a lot of difference!

I don't know what to say about your headstock. The safest thing is to put it back as it was. The best thing to do is to find a way, to get a proper solution. A single taper roller bearing (if that's what was used to replace the thrust race) is distinctly non-ideal. The scoring alone, you might get away with.

Certainly if there is scoring in the region outside the actual journals then use a stone or a file, to reduce the height, so it cannot damage the bearings further. In the region of the actual journals, as little as possible change to the spindle. On the up side, because the bearings are so much softer than the steel spindle there shouldn't be too much scoring.

In the end you're just going to have to judge how far to go, based on how confident you feel, and how worried you are about making it worse. These old machines are pretty cheap, so it wouldn't be the end of the world if it was unusable afterwards. Equally if it's all you can get your hands on, or you just like the machine, maybe you'd be more cautious.

The scoring, could of course be reversed. It is possible to build up with weld, or better to have it metal sprayed. The critical thing with that would be that you would have to find someone to cylindrically grind it, back to it's nominal size. I suspect there is not enough value in the machine to make it worthwhile.

I think I said earlier. You don't need an immaculate machine to do very nice work. It might take longer, but knowing how is certainly part of the skill and the sense of satisfaction and achievement.

If you are unable to prevent the spindle being loose in the head, you will find it almost impossible to do good work. As long as it runs freely without any perceptible slack, all of the other problems are can be worked around.

Thread: An Application to create dividing head tables
12/05/2016 13:27:55

Hi John,

I'm glad you found it useful, and figured out how to customise it for your dividing head.

My current job is developing manufacturing software. Before I did that spreadsheet, I had considered actually writing an application to do the job that the spreadsheet now does. I realised I could get everything I needed out of the spreadsheet so I never bothered with the application.

I read that you wanted an application and not a spreadsheet, and I wasn't going to post. When I then read what you had said about precision, I realised I should post anyway. Obviously, with the variable precision you can compromise if you want.

I would always recommend calculating out any choice that the spreadsheet helps you to find. This is just in case the spreadsheet has an error. As far as I know it's all-right.

If you decide to compromise and use a hole circle that doesn't give an integer result, don't forget that you can distribute the error around the full circle. If you calculate how far short/over 60 revolutions (or whatever for your input and worm) you would be with a hole circle that you know carries an error, then you can drop/add those holes gradually and evenly throughout division of the full revolution.

I'll admit that it's not the easiest of things to keep track of, but you only need to use a notepad and a pencil to keep track. It's perhaps a lot better than having to find/make a plate and hole circle that you only ever use once.

If you do it that way, the errors can be surprisingly small, and you still get what you want without the "correct" plate.

Glad to be of service.

Edited By Andy Ash on 12/05/2016 13:33:32

11/05/2016 18:22:18

I have the bog standard Myford head, but I didn't have the table so I did my own spreadsheet.

With that, I know how many turns on the wormscrew there are to make the head revolve once, and I calculate decimal; whole revs and part revs separately. (Integer and fractional parts). I do that for every division from one to as many as I need. Usually around 125 divisions. You can always do more division just by extending the number of rows down.

Then I have columns which know how many holes there are in the division plate circle. In each cell for the column I calculate how many holes is the closest to the fractional part of the number of divisions for that row. Finally, that cell uses conditional formatting so that it lights up yellow if it is within a given angular tolerance of a single division for that number of divisions.

The beauty of this, is that you can adjust the tolerance as you need. If you only have certain plates, then you can relax and/or tighten the tolerance to find out if you can reach a given number of divisions accurately enough.

What I found with the standard Myford plates is that they must have been fairly clever with their choice of hole circles. If you tighten the tolerances too much, the choices all drop away together. If you loosen the tolerance then you don't get much more benefit unless you can tolerate really gross errors.

There are four Myford plates, and No.1 plate gives the most possibilities with only 7 circles. No. 2 plate gives a lot of repeats and only a few a few more (perhaps half as many again) divisions with more hole circles. By the time you get to plates 3 & 4 you pretty nearly get only one extra division per hole circle (although they're not all primes). The Myford head shipped with No.1 and 2, you could buy 3 & 4 as extra if you needed the hard to get divisions.


Obviously it's a spreadsheet so you can tweak it to your worm and plates.

Edited By Andy Ash on 11/05/2016 18:25:37

Thread: new to myford lathes
07/05/2016 16:56:19

The six jaw in a nice thing to have. I don't have one. I decided to spend the money on collets.

I'd watch out for the rams especially if they are worn. They have to be straight and round. If they are it's fine, but if they were they would probably be in hydraulic cylinders! I don't know, I'm sure you'll work it out.

The live centre is essential for actual turning operations at the tailstock end. Never trust it for setup. Dead centres are so cheap, just buy a couple and keep at least one in reserve with a really sharp point. Quite often you want to turn between centres, and you will want a dead centre in the headstock end. You can use a tap wrench as a dog, and a bolt projecting from the faceplate as a catch.

The beauty of between centres is you can make several similar items and do each operation in turn. On centres you get very few runout problems even though you are continually swapping the workpiece in and out.

A faceplate is a very useful and very simple thing. Anything you can't do in the four jaw, you can do on the faceplate. I've even had my tailstock mounted on the faceplate before. (It was broken and needed repair)


Edited By Andy Ash on 07/05/2016 16:59:13

07/05/2016 11:11:09

Only now you can run the dial gauge along the side of the bar by traversing the saddle. The bar does not need to be rotated, and it's probably better if you don't. If the stylus is not quite at centre height, it doesn't matter that much. Just keep the gauge so that the stylus moves in the horizontal plane, fairly close to centre height.

By moving the dial gauge back and forth along the bar, you will observe the misalignment in the headstock. Using, a small lead or brass weight, tap the headstock casting, low down close to it's seating plane. Gradually you will be able to bring the bar parallel with the bed of the lathe. Tighten the (loose) front bolts first. Then the main bolt under the pulley. Check that nothing moved. When you're completely happy that you have alignment, and that nothing can move, you can break the set-up down and see if you can turn parallel.

3) The final task is to set the tailstock true.

You need two centres and a long test bar. The ground bar from the headstock setup would do. You can make the test bar yourself, but you will then need a fixed three point steady. Obviously if you don't mind sacrificing your nice 1" ground bar, then the centres could be drilled into it with a hand drill. Then you don't need the steady.

If the test bar already has accurate centres, then you can just run the dial gauge along it like before. You adjust the screws on the front of the ML4 tailstock to position the offset. Adjust the screws until the dial gauge reads true. Do bear in mind that the adjustment screws and the clamp oppose, so if you tighten the adjustment screws, you'll need to loosen the clamp and vice versa.

Get an initial setting by pointing the centres at each other close to the headstock.

If you don't trust the accuracy of the centres in the test bar, then you can turn a short section of it, with the same cross slide setting at each end of the bar. When the turned diameters measure the same size, the tailstock is aligned.

Setting the tailstock on the ML4 is an iterative process. Because there are two screws to adjust the offset, then it is possible to have the tailstock barrel slanted relative to the bed. If this happens then the offset will be different depending how far out the tailstock barrel is poking.

Once you're accurately centred, you'll need to fully extend the tailstock barrel. You put the dial guage back on the cross slide, and with the saddle run the stylus along the side of the tailstock barrel. This time simply adjust one of the two screws to bring the barrel parallel to the lathe bed.

Obviously, when you do this your original accurate centre will have gone. So you need to do it again. You'll need to go around that loop until you're happy that the tailstock is true to the bed and by inference the headstock bearings.

I hope it doesn't put you off, but that's what you have to do, to bring it right.

Edited By Andy Ash on 07/05/2016 11:21:31

07/05/2016 11:10:37

O.K. So I hope you're ready for a mammoth monster post........

The method for setting up the lathe is going to vary, depending on what kit you already have.

If you're serious about using the lathe you're going to need a four jaw chuck anyway. In my opinion you're better off with a four jaw than a three jaw if you only have one chuck. That's not usually the way people think, but if you use a four jaw then you're going to need a dial gauge and magnetic base to go with it.

* Generally, once you get used to it, the four jaw chuck will be nearly as quick to use as the three jaw.

* With the dial gauge it will be much more precise than the three jaw.

* It can hold almost anything within the capacity of the lathe, which the three jaw cannot.

You're going also to need a piece of 1" diameter ground stock. A 13" piece of silver steel is ideal.

1) The first thing you're going to have to sort out is the gibs;

You can forget about the compound slide. You'll need to get he gibs for that right eventually, so you can get a good turned finish. To set up the lathe, you can remove the compound if you want. It doesn't affect anything. Gibs that matter are those on the saddle and the cross slide. The screws for the saddle are over the back, and for the cross slide they're on the right.

What you have to do is adjust and lock each screw to get a minimum of slop (twisting) throughout the whole range of the slide. You need to do this whilst not incurring excessive friction by over-tightening. On a new machine, the gibs will be easy to set. On an old machine, there will probably be some kind of compromise between the range of friction free motion, and the slop in the most used areas of the slide.

The slop will be by the headstock for the saddle, and somewhere in the middle of the cross slide. You can have the slides ground true if you want. A dovetail bed like the ML4 is the most awkward for the professional machinist to grind.

Do the lower slides first. When you do the upper slides, ensure that the lower slides are in the tightest region of their travel. That way you're only trying to eliminate the slop in one slide at a time. Start with the saddle, and when you're happy do the cross slide. If you're going to, do the compound last.

The headstock has three bolts and no dowels. There is a considerable range of movement, so the lathe can naturally turn quite a strong taper. The main bolt is the one below the pulley. It needs to be nipped, but not tight. The other two are below the front main bearing and hang down. These need to be loose.


2) Then it's the headstock adjustment;

You're not going to need power to the headstock for any of this, so make the belt drive loose.

Put the ground bar into the 4 jaw and roughly centre it by eye. There will be a whole length of bar sticking out of the chuck, but that's fine. Don't let the chuck get too tight. Put the dial gauge on the cross slide and move the saddle so that the dial gauge stylus rests on the bar within about 1/2" of the chuck. Accurately centre the bar, without moving the saddle.

Once you have the bar centred close to the chuck, you're going to find that the bar is wagging around, because there is such a length sticking out of the chuck. You're going to have to minimise that, and the only way you can do it, is to use packing between the jaws of the chuck and the ground bar. It's a process of trial and error and it will take a while, but you cannot do to good a job of it. I would recommend various different approaches depending on how accurate your chuck is.

a) If the chuck is good, then the bar will not wag much and I think you will be able to use thin hard packing, like brass or steel shim stock.

b)If the chuck is bad, then the bar will wag a lot, and you may be better off with soft packing like iced lolly sticks.

c)You might find a combination of both works.

Be methodical, get it concentric near the chuck, and then assess the runout at the end of the bar. Use a permanent marker add packing and repeat.

If this all proves too difficult you can get a test bar which is ground all over and has a no. 2 morse taper on the end. You put the test bar directly into the spindle nose and ditch the 4 jaw. This could work out quicker, but it might not work out at all. If the taper is not true in the spindle nose, then you'll never get the bar true to the bearings. With the four jaw and the packing, it will always be possible to achieve truth, but it might take forever to set up.

Don't forget that you don't have to cut anything, so if the truth is accurate but weak due to the use of soft packing, that's O.K.

Once you get the bar true to the bearings, don't put any forces into it. Treat it as precious.

Edited By Andy Ash on 07/05/2016 11:16:56

Magazine Locator

Want the latest issue of Model Engineer or Model Engineers' Workshop? Use our magazine locator links to find your nearest stockist!

Find Model Engineer & Model Engineers' Workshop

Latest Forum Posts
Support Our Partners
Eccentric July 5 2018
Eccentric Engineering
Rapid RC
Subscription Offer

Latest "For Sale" Ads
Latest "Wanted" Ads
Get In Touch!

Do you want to contact the Model Engineer and Model Engineers' Workshop team?

You can contact us by phone, mail or email about the magazines including becoming a contributor, submitting reader's letters or making queries about articles. You can also get in touch about this website, advertising or other general issues.

Click THIS LINK for full contact details.

For subscription issues please see THIS LINK.

Digital Back Issues

Social Media online

'Like' us on Facebook
Follow us on Facebook

Follow us on Twitter
 Twitter Logo

Pin us on Pinterest