The CS website has little to no info that I can find regarding the threading module other than how to connect it.
I see no mention of PPR required/recommended/max count freq., description of operation or anything other.

I am looking for something more precise than a simple single index. I am interested in doing small close tolerance threads and I think I'm at the limit of Mach's simple threading capabilities.

I saw a vid recently where a threading cycle was started, then the spindle motor intentionally stopped, then the guy rotated the spindle by hand a few turns, restarted the spindle motor and continued on to cut (supposedly) a perfect thread. This used a 100 CPR encoder, one channel.

For those who have used this CS threading:
What PPR did you use ?
Any spindle speed limit ?

And, lets just say for instance .........

If you set up a stationary bicycle with the chain running from the pedals to a sprocket on the spindle and had your wife (God bless her) peddle the bike while doing a thread, would this thread track perfectly every pass and come out within a close tolerance ? (knowing the rpm will vary as the oxygen is expended due to the cursing)

Just curious,
Russ
 

 

I think the Enc module and the encoder inputs on the IP-A are the same and if I recall Wojtek or Andrew said they were min of 3MHz. On the Chiron I am using the Enc module for rigid tapping and my spindles encoder is set to 32,768 ppr so that is just over 2MHz at max RPM. I am hoping to get time soon to make up a belt and pulley setup for the spindle so that I can get 8,000rpm instead of my current 3,800 rpm, I will just reduce the interpolation of the drives encoder output so that I end up in the 3MHz  kind of range.

Regarding the threading, there is the issue of the delayed pull out at the end of each pass and thus an annular groove is cut and that is the only reason I have not bought an IP-A for the big lathe yet as that is not acceptable for a lot of the threads I do.
 On the small lathe (Conect) , which has the IP-S, I have done a fair amount of testing and apart from the afore mentioned problem (its a Mach issue BTW ) it threads very well. I can slow the spindle or stop the spindle or speed up the spindle during threading passes via the spindle override and it tracks beautifully in all tests I have done. I even stopped it via the override for quite a while, maybe 10 mins or more, then wound it back up and the thread continued as if nothing had happened.
On the Conect my spindle encoder is 8,000 ppr and geared 2:1 so 16,000 pulses per spindle rev.

 I had asked if Andrew could do a M76 in the CSMIO, similar to the M84 for rigid tapping and he said it could be done but time was short for him, so it was unlikely to be done as Mach4 will reportedly  solve the problem. Maybe I should ask again if there is any chance of it as looks like Mach4 Turn is not on the horizon yet.



Hood

Hi Hood,

3 MHz, That is good to know.

The pullout would not be acceptable here either.
This is one aspect of threading that Mach does satisfactorily here.

I guess it must be a daunting task to make this work properly on the software side. The "other" process I mentioned earlier basically shifts to an encoder follower / electronic gearing mode for the actual threading. I assume CS does similar. I'm guessing that there is not enough time to shift back to do the retract, as it cannot operate in both modes simultaneously.

Can't be simple, or Art probably would have gone that route initially ? ... or, fixed it later on instead of chasing the "one ppr" method.
I bet Brian could do it ... in his spare time.

 

  I had asked if Andrew could do a M76 in the CSMIO, similar to the M84 for rigid tapping

What do you mean here ?

Thanks Hood,
still looking for perfect threads,
Russ
 

Yes, basically the Z is synchronised to the spindles encoder.
The problem is there is a delay in Mach sending the plugin  the motion info for the pullout and the CSMIO is just sitting until it gets it. I found an earlier version of Mach handled it much better but there were a few things that didnt work properly, one was no spindle override. I actually have two versions of Mach on the Conect and if I really needed to do a thread on it I can just start the older version, I have not however tested it with the latest plugins so possibly there may be issues, so I would not consider that as an option without a lot of testing.

Steve had exactly the same issues when trying to use full encoder sync with the Galil and ended up just going to single pulse for lathe threading because of that.
The DSPMC or possibly the Hicon (cant remember which) does threading with the encoder and according to Ruffi it does not suffer that problem but no one has offered up any video footage of it doing the threading so I am, afraid to say, a bit sceptical that it does indeed do it. I would love to be proved wrong  as it would mean it can be done and hopefully others could find the way, but as Steve couldnt find the way..............

Regarding the M76/M84 what I mean is CS-Lab wrote a macro (and obviously internal code in the CSMIO) for rigid tapping. Instead of calling a G84 you call M84, it uses the same parameters as you would use with the G84 and it works exceptionally well. I had asked if they could do similar for G76 so that the CSMIO handled it totally instead of having to wait for info from Mach, you would just call M76 and the CSMIO would do the threading and then once completed would telll Mach it was ready for the next commands.

Hood

I understand that now Hood, thanks.
Sounds like that would stand to greatly reduce the delay, maybe not eliminate it but get it to an acceptable level.

Honest opinions cherished, (saves $)
Regards,
Russ

Here is a little tidbit from a CS Lab inquiry ........

Threading through CSMIO-ENC module is extremely accurate and reliable because CSMIO-ENC uses all encoder channels.

Spindle synchronization with axis move is very accurate and 100% repeatable.

 

CSMIO-ENC works only with incremental encoders, differential ttl5V,  index signal.

It is recommended to use encoder 10,000 pulses / rev with all edges.

They were also kind enough to reference 3 videos, all of which show excellent thread cutting and clearly demonstrating the delayed pull-out.

Please look at the videos:

https://www.youtube.com/watch?v=4GS4m5anMDU

https://www.youtube.com/watch?v=48DZoE0grLU

https://www.youtube.com/watch?v=CxL7Sv2Egno

FYI, if interested.
Regards,
Russ
 

Russ, yes have seen these vids, two I think were before I even bought the IP-S and I voiced concerns over the slow pullout.

Hood

Not sure how small your talking about Russ.

There is a real challenge on single point threading ( better to chase maybe ) for very small threads.

A good example is a 0-80. For a class 2a the lead tolerance / allowable error for a length of 0.1" is
only 0.00052". Then also consider some other tolerances like the diameter, tool radius, accuracy of setup, etc.
Depth of the thread is only 0.01083" for a truely sharp pointed V tool, but any tool has a radius, say 0.001 to
0.003 no matter how one sharpens it, thus dosen't take much to screw up that kind of thread if trying to
achieve a certain class. The pitch diameter range is only 0.0018". So setup based on the tool to be used becomes very important.

The next biggy is the CNC lathe as it must exceed the requirements. X axis repeatability must be better than the
tolerances  your trying to hold, spindle runout can come into play, same for the Z axis but it also is controlled by the index. You can check the total Z axis repeatablility by scribing a very small tick mark 1/2 to 1 thou on a trued thin disc and seeing how much it varies say for  a number of tries and the mean average will will give you a good indicator of repeatability for YOUR Z axis. Actualy you can do it with a double disc and compare the start to the finish of the thread, say 1" for a 0-80, do some math ...measure with a microscope....meticulous / time consumming but good to keep one out of the bars. A different approach is to do multi start threading as it is similar to just mentioned test.

So in summary, small and accurate requires all the stuff associated with threading to be good and there is not much room for error.  

--------------------------------
Practicaly speaking, I have cut 0-80's on the Sherline into 1/16" dia weld rod but the converted Atlas was more conistant / privided better therading and that would be expected.

---------------------------------
Electronic gearing is realy neat. Using an encoder on spindle and slaving the Z  to the spindle ( done with IMS, which i used a long time ago) one could turn the spindle by hand and could cut small fine threads. So if you wanted to spring pass a thread,just rock / trun the spindle back and forth and you could almost polish the thread and you had complete control of the x feed. The carriage just tracked in sink with spindle perfectly with the slightest movement.

----------------------------------

Art said electronic gearing would be possible, Mach code would need to be reworked, and a defined developement procedure would be rquired and that was a few years ago.

FWIW,

RICH

 

All good info Rich, thanks.

I think I can exceed the mechanical requirements (actuall am at .0003>.0004 now and there is clearly room for improvement in the next build.
The availability of extremely close tol. ground ballscrews and mounts from surplus takes care of one aspect. Small machine with short travels helps keep the $ down there too.

Super high resolution servos ... same deal, affordable and very precise. Above and beyond required (at a cost of reduced speed of course)

Linear components for ways / slides .... readily available. Preloaded, very close tol.

The motion controllers are excellent, math to 15 decimal places, interpolation to match.

I'm using Nickole Mini Systems carbide threading inserts. They come with a .002" >.004" tip rad. I have a diamond wheel and fixture set up in the surface grinder to regrind one of the sides just enough to remove that rad, then polishing a .001>.0015 rad at the tip with super fine diamond lapping compound.

Everything is peachy to this point.

It all comes down to the threading ... that's the only thing that is problematic and it's all in the sw., nothing mechanical about it.
I can have a mechanism capable of holding .00015" and it will do exactly as it is told. Just can't find anything on the sw side that will tell it exactly how to run threads .... repeatably, consistently and to a high degree of accuracy.
The CS electronic gearing of the spindle and axis is excellent provided the trigger is extremely repeatable but does not pull out properly.
Mach3 does a pretty good job provided the spindle is rock steady. Closed loop almost if not absolutely mandatory.

One of my basic threads is 48 tpi, PD +/- .0015"  which must pull out cleanly, no groove allowed so its done with Mach otherwise, I'd love to try the CS controller. At a cool grand though, I'd be a bit disgruntled if it did not hold to expectations. (the videos actually show very little to nothing useful)

56 to 60 tpi will be about the finest at this point, but would like the capability to go finer if requested to in the future. I can just barely hold the tol. on the 48 tpi right now with Mach3, and an open loop VFD.

I've a notion to make a custom tool holder/post with a mechanical trip and latch to do a precise pull-out  as required and let the CS geared threading cycle run how it wants. Nah .... too much work.

Someone, someday will fix this up to where all of the cool threading features will work right in the same sw package ..... someday.

Thanks Rich, see ya in April. (we gotta get vendor tags this time) 
Russ

Russ,
Just some more stuff .......

In the simple threading wizard the pitch for a 48 tpi thread would be rounded to  =1/48=0.021 if you input 0.0208,
but, the generated code uses whatever you input and you will see the F value in the G32 code.

So if one knew or had an idea of a tweak and wanted to adjust the feedrate some in the Gcode to account for some lead error you could do it.

...............

Pulling out cleanly would best be done with a pullout over a rev or two. Don't think instantaneous pullout is possible.
Again you can experiment some. It will realy show on say a two start thread and should  be 180 degrees out from each other with no burrs. Sort of the same idea as using a scribing on the disc at the start of a thread. You can take this to another level by adjusting the start location and playing to see how the lathe accelerates from a distance.

....................

Again in general.......threading will only be as good as the "LATHE SYSTEM" and the better you want it to be the more you need to pay since  you pay a lot for increasing accuracy.

RICH