Seems the same as a guy did on a south bend

Machining Metric Threads on the South Bend lathe part 1

Machining Metric Threads on the South Bend lathe part 2

part 2 has a useful spreadsheet in excel attachment

Well the guy who published or posted the extensive article you linked to said he found out about it from elsewhere on the net and then went ahead and bought the two gears and checked it all out and it works as advertised. So I can't see any reason it would not work.

Brian Wood's book mentions similar gear changes for various gearbox lathes but not the Atlas. So it can be done in principle. Sometimes the resulting pitch is an "approximation" and overzealous amateurs get carried away with handwringing over variations of one thou in three inches etc but generally the approximations are more than good enough for anything other than perhaps aerospace and metrology room work.

For the cost of the two gears it would be well worth it.

I have to say that having machined some HDPE just one week ago I could tell from the look of the original material and the way it was cutting that it was not the same material. The large piece of HDPE I had (Ø100) looked waxy, a bit like an uncoloured wax candle, and, when turning, the swarf came off like a piece of string being shot out of a hose. The powdery debris from cutting the gear also looked very odd. There is a possibility that what he had was more like a sintered plate of HDPE powder that was made using a heated press to form the final product. I expect this would be cheaper than injection moulding tooling and would have a shorter cycle time.

Martin C

Looks OK to me and the airgun website shows what pitches are available, identifies approximations (in red) that might be close enough for a short thread, and explains the maths:

Conventionally, Imperial gearboxes are tweaked to produce metric (or vice versa), with a 127 toothed gear. This because there are 25.4 millimetres per inch and half 254 is 127. The conversion is exact, but a 125 toothed gear may be too big to fit the banjo, hence a 63 toothed gears may be used instead. 4 x 63 is 252, close enough to 254 for short thread engagements to fit acceptably. However the error becomes progressively more serious and after about 8 to 10 threads the 63 gear bodge becomes troublesome.

Most modern lathes, including I think all Far Eastern hobby machines, do metric and imperial without a special gear pair like 52:44, 63:1 or 127:1 Instead, the change gear set provided is capable of producing the ratios needed as a mix of spot-on and close approximations. The Imperial approximations on my Metric lathe are all better than 63:1, plenty good enough for most purposes, but I would check the actual accuracy of a ratio  before trusting it to turn a long Imperial Thread. The change gears provided with the Imperial version of the same lathe are slightly different, and owners should check the actual ratio before trusting one to turn a long Metric Thread. Is the ratio spot-on or an approximation and how good is the approximation?

On the Atlas, 52:44 provides a useful selection of metric threads, but not a full set as this list shows. The shortfall may not matter!

Finally, 52:44 isn't a magic combination guaranteed to work on all Imperial lathes. It might, but it depends on the lead-screw and gearbox ratios. Someone has to do the maths!

Dave

Edited By SillyOldDuffer on 26/10/2022 10:34:32

The info in that video contains a couple of rather dumb stud gear selections. For example, 1.75mm pitch uses a 19tooth gear and 11.5tpi tumbler setting. The 19tpi gear is very weak where the keyway comes close to the base circle so a 38 tooth gear and 23TPI tumbler setting will produce the exact same pitch, be a much stronger gear and have better tooth engagement.

Similarly, 4m pitch also uses a 19tooth gear which is sure to break cutting that coarse thread. This is also completely unneccesary when a 34 tooth gear and 9TPI tumbler setting will produce the same thread and 34 is in use for 6 other pitches already.

5mm pitch is even worse, but could be improved by using a 26tooth gear and 5.5tpi setting, with a much smaller pitch error too.

Apart from that, his claimed % error is out by two decimals. He's quoting errors of (for example) 0.0005% when he should be using 0.05%