Thread Pitch Error.

[richardandtracyParticipant @richardandtracy]

Is there a general rule of thumb for what error will work between two thread pitches in the size of a normal nut?

Over the weekend I didn't have access to my normal calculation program, yet needed to get a 0.75mm pitch on my Warco. This is not one of the standard ratios quoted by Warco (giving 0.7 & 0.8 mm pitch, and the half-way gears were not supplied). The closest I could come up with by trial & error was 0.747723mm (0.3% error), which worked. When I got access to my usual program again, I found I could have had 0.750031 had I been able to guess the correct wheels.

But it made me wonder, is there a general rule of thumb for the error that will work with a normal fit between a thread & normal size nut? I assume if the nut is double normal length the acceptable pitch error will halve & vice versa. So, does any one have such a rule?

Regards,

Richard.

[richardandtracyParticipant @richardandtracy]

General Rule of Thumb

[John Stevenson 1Participant @johnstevenson1]

How accurate is your Warco leadscrew ?

[richardandtracyParticipant @richardandtracy]

The leadscrew 'good enough' for general engineering use. The precision is not something I've verified. I expect its precision to be a bit degraded near the headstock due to wear, though by visual inspection I can find no signs of wear.

I'm looking for a rule of thumb that can be generally used, but with a suitable level of caution, and the caution increasing with machines of lower quality whether it be due to intrinsic quality or age & wear. For model making use, I'd rather expect the absolute value of the acceptable error to be smaller, but as a percentage, it's likely to be similar for similar quality work.

Regards,

Richard.

[Gordon WParticipant @gordonw]

I guess it would depend on the tolerances of the mating parts. Some bolts and nuts I have could use different diameters , never mind thread pitch.

[HopperParticipant @hopper]

Dunno about Warco leadscrews but I have an old Myfrod inspection sheet that calls out .001" per foot for leadscrew pitch tolerance. IE, 1 in 12,000. Pretty tight really.

Re allowable pitch error, Martin Cleeve's seminal book "Screwcutting in the Lathe" includes a list of imperial change gears for turning metric threads on an 8tpi leadscrew. The largest error he lists is 1 in 380, which is .263 per cent.In another chart of approximations he lists the grossest error as 1 in 255 which is .392 per cent. He seems to consider these quite acceptable for professional quality work.

No mention though of what he considers the limit on such error. As others posted above, it depends entirely on how tight you machine the clearances on the threads. Basically, use what ratio you have and when you get to full thread depth, move the tool sideways with teh topslide to widen the V groove until the nut fits. Being that the length of the thread in a nut is small, I imagine a pretty large error could be accommodated. EG, 1 in 380 error is less than three thou per inch. A nut might be, say, half inch long, so that's 1.5 thou error over the length of the nut. So two thou of extra side clearance between the threads will accommodate it. You are doing pretty well if you are machining threads to that close a tolerance anyway, with a hand-ground tool bit on a worn lathe of hobby-grade quality with a bit of slack int he top and cross slide ways and feedscrews etc etc.

[Clive FosterParticipant @clivefoster55965]

Deceptively simple question with a huge bone in it! Couldn't find any useful references on t'net (but plenty of tempting rabbit holes).

From a practical standpoint what is needed is the maximum pitch error that still lets the two parts be screwed together. But pitch errors are cumulative so it depends on how long the thread engagement is. Something I exploit when trying to figure out if a coarse thread is imperial or a very close metric size.

Needing to sort out some mixed imperial and metric feed screws and nuts for Boxfords and Bridgeports I did a spreadsheet tabulating the length of thread for 1/2 a pitch error for all the closely related coarse pitch sizes. In both units. Separating sheep from goats then becomes a simple ruler measurement. No way will you split 5 TPI and 5 mm pitch with a standard thread gauge. The interesting corollary is how much wear in the feed nut is needed to let the wrong breed of nut screw on the screw. Not a lot for a Bridgeport or Boxford. Certainly within the more backlash than I'd like but still quite useable range!

Getting back to Richard's starter for ten any useful rule of thumb must allow great enough engagement lengths for a full strength joint. Easiest practical definition of a full strength threaded joint is one where over torquing shears the male fastener before the threads strip or deform beyond safe re-use. General rule of thumb for that is 1 to 1 1/2 D of good thread depth where D is the diameter of the fastener. So a thread pitch error small enough that 2D of thread screws on OK seems reasonable.

If slightly mismatched thread pitches are to screw together some axial play will be needed in the threads. Thread fit classes are defined in terms of maximum and minimum pitch diameter. Staying within any given tolerance the smallest permitted male thread in the largest permitted female thread will have most axial play. Conversely the largest permitted male thread in the smallest female thread will have least axial play. Largest male and smallest female are the Maximum Metal Condition (MMC) for each component.

If that is acceptable its fairly straightforward trigonometry to figure out how much axial play there is is 2 D of MMC fasteners to any specified tolerance and hence compute a permitted pitch error. Do that for a reasonable sample of fastener sizes and round out to a rule of thumb. Spreadsheet time.

Clive.

[richardandtracyParticipant @richardandtracy]

Working through the theoretical Warco ratios for my lathe, they all hover around the 0.04% error, though 1.25 & 2.5mm pitch are 0.3% error. One is obviously a misprint in the manual, being 12.55% off for 3mm pitch. Someone typed in 56 instead of 63 for a gear wheel when compiling the manual. When using the corrected value, it's again 0.04% error.

Regards,

Richard.

[BazyleParticipant @bazyle]

There will be a difference between fit and strength. If you have clearance on peaks and troughs and shave off a bit extra by applying a bit of longitudinal offset you could make it fit over a longer length. This wouldn't be doing the strength any favours.
A perfect nut bolt pair will contact all along the thread. If one is wrong pitch and they are infinitely hard contact will just be at one point on one flank along the thread until the error takes up the clearance and the opposite flank hits and it jams. If the say the bolt is elastic when put under strain it can stretch putting more surface into contact. However the bolt thread might fail because the force is not spread over enough length of contact area.

So quality needed depends on the type of material and how much load you expect it to take. It should be possible to calculate the maximum load relative to a particular error that allows bending of the thread to distribute the load without reaching the yield point.

[Neil WyattModerator @neilwyatt]

A 1% error will be 5% by the time you have ten threads (take the middle thread as perfect, so the two ends are +/-5 threads out)

5% is nothing compared to typical tolerances for ordinary fixings. For a 1mm pitch thread that will be a 0.05mm error or 2 thou out at each end of a 10-thread engagement. Most nuts use far less than ten threads of engagement and you don't need that many for studs either.

Conclusion, for fixings 1% is well within the required accuracy.

Neil

[RichardNParticipant @richardn]

Does it also depend on the length of the threaded portion to mate? If I just need a thin nut (1/2 thickness lock nut might be around 0.3x thread diameter) tolerances could be far greater than the thread within a T-nut that has a threaded portion perhaps 3x diameter?

Edit- see Neil's point above- beat me to it.

Edited By RichardN on 06/03/2017 13:55:18

[Brian WoodParticipant @brianwood45127]

Richard and Tracy,

​I have done a little work in trying to quantify this particular element in thread fitting and made a mathematical model to analyse it. The results came down on working values of 3% for fine pitch threads [BA sizes for example] and 1.5% for coarser fastener type threads. These values were confirmed in practical tests.

​In another aspect of that work, like Hopper, I tested the pitch accuracy of my Myford ML7R which gave an accumulative positive error [ ie greater than target pitch] of 0.0018 inches over the full leadscrew length of 25 inches. By comparison, leadscrew pitch error on commercial lathes has been found to be 0.0001 inches per inch, or 1 in 10,000. Proper tool room lathes are about 1/10 of that

If you do the maths, the Myford result is actually better than the commercial lathe, but that said it isn't being used in production conditions.

Regards Brian

[richardandtracyParticipant @richardandtracy]

Thanks Brian. It does put bounds on how bothered you should be if the gear ratio isn't exactly right, which is what I was aiming to find out.

I think I shall adopt the following rule of thumb:

• Less than 1% error, don't care.
• 1% to 1.5% Assess machine's accuracy prior to cut
• More than 1.5%, try to avoid
• Halve error for doubled engagement.

Regards,

Richard.

[Brian WoodParticipant @brianwood45127]

Hello Richard,

​The error values I quoted already allow for + and – where they might be + on the screw and – on the nut, whichever was the sloppiest case, so you are being over cautious. It quite surprised me how little that aspect really matters on short engagements into nuts for example.

Regards Brian

[JonParticipant @jon]

Good points above but what you will find the finer the pitch the less room for error.

I do a lot of metric 0.75 internal and external zero aiming for tolerance in ti and aluminiums – no warning even with oil before test fitting each it galls and of course scrap job on the two parts. I only use full form profiled decent tips, on such an M30X .75 pitch x 17mm long theres detectable play if taking 0.01mm off radius all relative to depth of thread pitch with full engagement.

On the other hand that 0.01mm quite acceptable for 1mm pitch you can barely detect the play. However a 3% pitch error no where near going in unless a short thread.

Another point to make is amount spring passes dependant upon actual tool used ie cheap far east imports may require 5 passes to get rid of te flex where quality branded may mean 1 pass.

[duncan webster 1Participant @duncanwebster1]

If you're making nuts and bolts there could be a (very small) argument for making the bolt slightly finer than the nut. Why? because when the tighten it up, the bolt stretches and the nut compresses, so the pitches could come back in line. You'd have to be good at sums to work out the required pitch error to achieve uniform pressure on the flanks along the length of the nut, and I'll wager the required 'error' will be within acceptable tolerances anyway.

[Brian WoodParticipant @brianwood45127]

Jon,

​There will always be exceptions to any generalisation and Richard was looking for a 'Rule of Thumb' which I tried to provide.

​The model analyses threads of 65% engagement which would be adequate for most work and is intended to apply to normal steels and threads in brass and cast iron for example.

I imagine working with a reactive material like titanium rather more care has to be taken and perhaps even the use of specialised lubricants to minimise the risks of pick up and galling. Such examples might be best regarded as special cases, maybe based on a starting point and suitably tweaked for that condition, but they will be outside the general envelope.

Duncan.

​Surely the nut will only tighten up if the load on it comes from the final threads in the nut to give the condition of compression you put forward? It will not be a uniform effect either throughout the nut as the degree of compression will vary turn by turn according to the fit between the two components.

The model was only a simple case and it took as an assumption, just to be able to calculate anything, that such effects could be ignored.

Regards Brian

​

[richardandtracyParticipant @richardandtracy]

Duncan, I'm considering normal usage on earth. Had it been a space craft use where every fraction of a gram saving on launch costs by getting dead even thread loading, paring away the nut so it flexes nicely to the thread and use of exotic materials etc, then I think your idea is worth considering. No, I'd go further, it would be downright foolish not to consider and implement it.

As it is, at home, I work in

• brass (grade unknown),
• 'steel' (En1a leaded, En8, En16, En24, En54a and occasionally that ubiquitous grade 'Gash' )
• 'ally' (LM4/6, 2017A-T6, 7075-T6511 or 6082-T6),
• silver ('fine' & 'sterling', cleaning up items sintered from 'Precious metal clay' )
• bronze (sintered components from metal clay)
• and plastic (Acrylic, many acrylic alloys of varying colours, stiffnesses & compositions, Acetal. PTFE, Nylon, Tufnol, Polyurethane resins & Polyester).

Just normal 'stuff' to normal 'stuff' tolerances. Nothing to push the boundaries of any technology.

The item I made over the weekend that prompted this thread was a lighter weight cap finial for a Chinese fountain pen. It was made from brass, and I wondered if making it out of aluminium would be better. The pen is a £15 Chinese copy of a £350-500 Parker Duofold Centennial. It is a really good, re-engineered, copy of the Duofold. The only flaws in the Chinese pen are the brass finials at the end of the cap & barrel. I have made 500 odd polyester finials to go at the end of the barrel for people around the world, but was trying a cap finial for the first time. It's rather more highly loaded, so I don't think a polyester moulded one will be adequate.

That is my nearly worn out test pen, showing the effect of severe usage on a 6 year old pen. The old Kaigelu 316 finial is loose, & the aluminium one I made is fitted to the cap. The recess in the aluminium finial can take a sintered bronze metal clay logo if I get around to it.

Regards,

Richard.

Edited to get rid of involuntary emoticons

Edited By richardandtracy on 08/03/2017 11:21:15

[Neil WyattModerator @neilwyatt]

Posted by Brian Wood on 07/03/2017 10:10:00:

Richard and Tracy,

​I have done a little work in trying to quantify this particular element in thread fitting and made a mathematical model to analyse it. The results came down on working values of 3% for fine pitch threads [BA sizes for example] and 1.5% for coarser fastener type threads. These values were confirmed in practical tests.

​In another aspect of that work, like Hopper, I tested the pitch accuracy of my Myford ML7R which gave an accumulative positive error [ ie greater than target pitch] of 0.0018 inches over the full leadscrew length of 25 inches. By comparison, leadscrew pitch error on commercial lathes has been found to be 0.0001 inches per inch, or 1 in 10,000. Proper tool room lathes are about 1/10 of that

If you do the maths, the Myford result is actually better than the commercial lathe, but that said it isn't being used in production conditions.

I hear there's a definitive new book on screwcutting coming out soon

Neil

[mark costello 1Participant @markcostello1]

I don't know if anyone else has noticed this effect of screw cutting but when fitting a nut to a screw on the lathe I get the thread to a good fit. Then when turning the nut around it is very slightly tighter. Wonder if it is pitch error creeping in?

[Brian WoodParticipant @brianwood45127]

Mark,

​Not an effect I have observed but I can offer an explanation.

​If your threading tool is angled slightly so that one flank is cutting at 29.5 degrees say, the other should be at 30.5 degrees for a properly formed 60 degree tool, I submit that would have just such an effect. I can't see any mechanism for pitch error in such a case.

Standing by for shot and shell!

​Brian

[duncan webster 1Participant @duncanwebster1]

Posted by Brian Wood on 08/03/2017 09:58:50:

Duncan.

​Surely the nut will only tighten up if the load on it comes from the final threads in the nut to give the condition of compression you put forward? It will not be a uniform effect either throughout the nut as the degree of compression will vary turn by turn according to the fit between the two components.

The model was only a simple case and it took as an assumption, just to be able to calculate anything, that such effects could be ignored.

Regards Brian

​

I did say it was a very small argument. Perhaps I should have included a smiley. There was a very good article in ME or MEW recently about forces in nuts and bolts, well worth a read. In a long career in engineering I've never come across anyone trying this, but I have seen nuts with an annular groove from abutment face to almost the top face so that the nut threads are in tension. Only used in very critical applications as they must be expensive to make, and they were pretty big.

[Brian WoodParticipant @brianwood45127]

Hello Duncan,

​Argument accepted of course! The article you mention must be in ME, I haven't seen it in MEW

​The special nuts you describe sound interesting but I haven't got a picture in my mind of the geometry; which would be the top face as you describe it?

Regards
​Brian

[duncan webster 1Participant @duncanwebster1]

It's a rubbish sketch, the thread is mean to be parallel. You can see that the threaded part of the nut is in tension, and why they would be so expensive

Edited By duncan webster on 08/03/2017 19:14:42

[Brian WoodParticipant @brianwood45127]

Hello Duncan,

​Thank you for the sketch, it explains what I couldn't imagine with crystal clarity.

​I'm going 'off air' for a day or two so I'll let the design mull away in the background, it is certainly one I have never come across. In the meantime, what application were these used on?

Regards
Brian

[Author]

Posts