Boxford metric lead screw fitted to imperial lathe?

The big difference between metric and imperial threads is that the ratios used are the inverse of each other which is why QCGBs don't quite work exactly in the changeover. In gearing this is achieved by taking the same set of gears and driving them in the opposite direction. The Boxford metric gearbox is the 'same' as the imperial one except the housing is modified so that effectively the input is now the output. The metric box has the 8 way selector on the left instead of the right.
If you do a bit of googling you will find images of the diagram of the gearbox Follow those up and you find the spares list. That tells you the gears.

Have a look at actual metric fixing threads, Anyway you are unlikely to need to cut more than M5, M6, M8, M10, M12.
Likewise for imperial it is only a few that get used. If you are into tractors rather than 5 in steam the actual sizes you want may be a different range but anyway it's only a handful that you need.

# Metric and imperial gearboxes are visibly different on the front face where the plungers are.

# The front face of the gearbox should have a plate saying what the leadscrew is.

# My '69 VSL is imperial gearbox, imperial leadscrew, and metric top and cross slides with metric dials. From the factory...

# Tony's lathe site should help with all the above and change wheel charts for metric-imperial conversion and vise versa.

Thank you so much, and DC31k that table is fantastic.

Just to check I'm understanding things correctly, when you say "change gear ratio", I think 1:1 would mean something like a 40 on the top gear, 80 on the middle/idler and 40 on the bottom gear? Then 7:10 would need a 35 and a 50?

I think my best bet from here, assuming I've understood things correctly, is to get a pile of gears printed for testing purposes, try a few combinations then depending on how I get on and what I decide possibly look at the leadscrew from Ian P

It's quite frustrating being thousands of miles away from my shed, but got to earn the money to buy random chunks of rust somehow!

I use a southbend 9A which is similar to the boxford . When i make metric threads using the imperial leadscrew ,i use a spreadsheet calculator that i was sent by someone in the US . Its quite good and calculates which gears you need on the gear train to use in conjunction with the gearbox. Gives all the pitch errors and enables me to cut metric threads without the need for the expensive 127/100 gear.

I would upload it but i think you can only post photos. heres a screen shot.

Edited By mark smith 20 on 27/10/2019 16:57:47

Edited By mark smith 20 on 27/10/2019 17:05:47

Edited By mark smith 20 on 27/10/2019 17:17:40

There's a very relevant response (third reply, by Bob Brown 1) in this thread;

https://www.model-engineer.co.uk/forums/postings.asp?th=98914

Cupboard,

You have the right idea, but what DC31k has omitted to say is that the driver on the spindle (after the reversing gears) will be 20T and the gearbox entry spindle needs a 56T gear in all cases.The change gears are fitted in the chain linking those two.

Sadly, and I hate to have to point it out, but there is one error in his table in the listing for 1.75 mm pitch. The change gear ratio should read 7/8, which could be 35/40 as you have interpreted the ratio.

As listed, you will get a pitch of 1.1427 mm. Otherwise, all is well

Kind regards

Brian

As Meunier and Thor replied earlier, Brian is the authority I failed to remember earlier in the thread. Though I'm hopeless, it's good to be in his safe hands!

Dave

OK, so instead of the big 80 tooth gear, I'd put a pair of gears in (at the relevent ratio) on the K shaped arms instead.

I've got a friend printing me a 24 and a 32 tooth gear (one in PLA and one in PETG as a comparison) to see what they perform like, and I'll do some experimenting from there.

If I'm taking the big gear out, I think that means I'm going to need another stud to put the additional gear on.

Sorry I'm being a bit slow on the uptake here.

Hello Cupboard,

I don't think you need another stud, but you might need a spacer at the gearbox end.

Taking an example:- A good one might be the 1.75 mm pitch that I spotted as incorrect

Using the correct values, you would set up the whole train, ignoring the reversing cluster gearing, as 20 driver, 40 driven which is also coupled, on the same stud, to a 35 driver[, that in turn drives the 56 input gear to the gearbox.

That gives a pitch of 1.7498 mm [ ie 1.75 mm] with your 3 mm pitch leadscrew. The spacer will be needed to position the 56 T wheel on the gearbox out by one wheel width to couple it to the 35 tooth wheel. If these wheels are too small to make the linkage, as I suspect could be the case, then an idler can be added at any place in the chain to 'fill' the gap. Being an idler, it plays no part in the gearing calculation any more than a belt does linking two pulleys. It will though have the effect of reversing the leadscrew rotation, readily corrected by moving the reversing cluster to the other position.

If you have the gears, the 7/8 ratio could be set up as 70/80 instead, that will bridge any spacing issues

I hope that makes things somewhat clearer

Brian

Edited By Brian Wood on 27/10/2019 20:29:17

Edited By Brian Wood on 27/10/2019 20:31:31

Hello again Cupboard,

I have been "playing" again with your dilemma and I think there is even a way round the 3 mm pitch leadscrew [without having to change it] to be able to get imperial pitches as direct reading values as if the lathe was equipped with the standard 8 TPI leadscrew.

What is needed is another set of adjustment gears interposed between the 20 and 56 gears to fool the system into thinking it has the right leadscrew

These are a compound pair, running in place of the 80 T spacer of 55 teeth mated to 52 teeth.

Setting up the chain this time is now 20 driver, 52 driven with the 55 T of the compound gear driving the 56 gear into the gearbox. As before with the metric set up we looked at earlier, the 56 gear will need a spacer so that it can mess with the 55.

Checking the maths for a few standard pitches we get for 8tpi the following

20/52 X 55/56 X 3mm pitch (leadscrew) X 2.8 (gearbox ratio for 8 tpi) gives a pitch of 3.173 mm which is 8.0048 tpi

and for 32 tpi we get

20/52 X 55/56 X 3( as before) X 0.7 (gearbox ratio for 32 tpi) gives a pitch of 0.793 mm which is 32.019 tpi

A third result for 40 tpi is

20/52 X 55/56 X 3 X 0.56 (gearbox ratio for 40 tpi) gives a pitch of 0.635 mm which is 40.024 tpi

I agree these aren't exact values but it is more readily achieved with realistic gear sizes than 120 and 127 which do give exact result. Ask your friend to print a 52 gear, the 55 is I think a standard wheel.

Time for a stiff gin now I think

Brian

Edited By Brian Wood on 28/10/2019 18:20:22

Thank you for the clarification and correction.

I am glad that it has been peer-reviewed and the bugs worked out so speedily.

I think I was a bit wooly in my terminology when I used CG for change gear ratio: it should have been more closely defined as change gear ratio modification to standard ratio (thus, when it is 1/1, there is no modification to standard/factory needed).

The incorrect 4/7 came about as I started off with 7/4 and 24tpi (i.e. gear up 1mm pitch to get 1.75mm*). That needed an extra 20t wheel added to the set (35/20). To reduce required number of change wheels, I realised you can gear down 2mm pitch but did not do well in changing the numbers. It is fortuitous that the correct 7/8 or 35/40 can come from the existing set.

Brian's calculations on getting you imperial pitches are to be highly commended.

Maybe it is bad form to recommend a different author, but Martin Cleeve's Screwcutting book is essential reading. He spends a good bit of time talking about non-standard leadscrews (mainly for speed in production work).

* And if anyone has an imperial Chipmaster, that is how you can get the missing 1.75mm on this machine, with a 63/36.

Thanks, that actually makes some sense. One rotation should move the carriage 3.175mm, with a 3mm pitch screw it actually moves it 3mm, so I need to fool it in to moving a little bit faster than it "thinks" it should.

(3/3.175)*(55/52)=0.9994

Which is near enough!

I need to find/buy/make a device to hold a pair of gears together so I can make whatever compound I need.

Edited By Cupboard on 28/10/2019 19:30:32

DC31k,

Thank you for your endorsements; as you might have guessed I like the mathematical challenge such opportunities present, particularly if the outcome is to be able to avoid having to re-equip what might be a perfectly satisfactory machine.

I too would recommend Martin Cleeve's splendid book on screwcutting, it is a bible on my bookshelf and is now stuffed with little notes and markers I have put in it over the years

I liked your neat table for the metric arrangements, it was a clever way of presenting the necessary information as a simple ratio and a method I shall remember and use if I may.

Cupboard

You have found me out! That was exactly my thinking which led me on to the adjustment gears of 55 and 52 and they came from the method Martin Cleeve uses to establish gear pairs from a ratio value. As for holding a pair of gears as a compound pair it is basically a second stud, readily available from Boxford as a standard part.

Regards to you both Brian

The corollary of using 55/52 in this instance is that it can be used to convert imperial to metric. The key here is the medium sized primes, that the above ratio reduces to, 11/13 (lower prime factors of 2,3 5,7 are readily available).

Some industrial lathes have a QCGB that provide approximate metric threads as well as imperial. For this the Raglan uses the 11/13 ratio in its gearbox, Harrison uses 19x11 and the most accurate I know of the Colchesters use 19/13 /29

Bazyle,

Now you have confused me completely! I could not follow the logic in your expression for Colchester conversions of 19/13 /29

Please enlighten me over what I feel sure is a typo somewhere.

Regards Brian

Warning - boring numbers - just skip if you are not Brian or DC.

In the Colchester QC gearbox there are gears whose prime factors include 19, 13, 29 arranged as 19/(13*29). ( I was a bit lazy in the previous post and wrote it as /13 space /29 to indicate they were both divisors.) There are other gears but just multiples of low value primes like 2,3 5,7. The 13 of course appears as 26 in practice. The use of higher value primes enables the ratios to be tweaked closer to ideal.

Ideal is 127*2/100 = 2.54
So buried in the Colchester box is effectively (19*7*36)/(13*5*29) = 2.540053 which is pretty close.
In the Harrison it is (19*11*35)/(80*36) = 2.539931 not quite so good.
Drummond tables 20*8/63 = 2.539683
Logan 2*47/37 = 2.540541
Early Myford tables (16*73)/(46*10) = 2.539130
Raglan 11*3/13 = 2.538462

In the above I have shown the main primes and some non-prime where it is the actual gear used.

This shows just how good the use of 63 turns out to be even though it is only low primes 7*9.

The use of 19 is a bit of a pain as it is in the way most of the time. It is 'factored out' on the Harrison using a 95 (5*19) in the train from spindle to gearbox input but then allowed back in to provide 19tpi. Very British.

Bazyle,

What a mine of information, thank you. And all very clever too. With the working laid out as you have done I can follow the logic now, it certainly wasn't obvious to me before.

Most helpful

Brian

I examined the Harrison QCGB in detail with the idea of making one for my Boxford. I hoped to be able to simplify it but it is a marvel of maths and engineering that is streamlined to the maximum efficiency. One day when I master CAD and 3D printing for patterns and test assemblies I still hope to go there.

Bazyle,

Interesting, but do any of these ‘approximations’ make anywhere where near an iota of difference in practice?

Totally irrelevant for anyone making both matching male and female threads, on the same machine, for a one-off model and likely less variation from the true value than adjacent turns of threads made on a worn machine?

Irrelevant over the thread of most nuts we might make with single point cutters - as the thread form will be, by far, the largest deviation from the theoretical thread specification for anyone using in-house ground cutters.

How high up in the echelons of precision engineers do we have to be, to worry about these approximations? I will never attain those levels of precision/accuracy for sure, and likely never even approach any such level,

Perhaps important for dissimilar metals and a wide temperature gradient environment (such as space vehicles, perhaps?).

Edited By not done it yet on 30/10/2019 09:21:54

NDIY - absolutely correct. It is easy to get carried away with it if you like playing with spreadsheets. Also become obsessed with getting expensive 127 gears.
Really for non QCGB lathes it is mostly possible to get near enough with the regular gears anyway and if not the 63 is the best special one to make.
For QCGB lathes it can be a little more difficult but the 55/52 mentioned above is a good idea for the Boxford as 52 is a standard as it uses even numbers in its gear range available. 55 is a special in this case I think.