CONTINUED: involute gear generator preview !

quirxi · Post by **quirxi** » Sat Nov 02, 2013 10:52 pm

It seems that the board here has problems with threads that are too long.
Since i got problems posting at the old thread here:
viewtopic.php?f=10&t=3704&start=120
I will continue our discussion in this thread here if nobody minds?

@jmhrvy:

Turns out that while the Sept 7th code correctly calculated the circular tooth tickness and mirror angle, it wasn't compensating for the angular distance traveled by the involute from the base circle to the pitch circle. With this small change, the CNC machine is now cutting gears that both mesh & roll well.

Cool ! Thanks for testing the code in reality and sharing your improvments ! I knew there must be some problem with it - there always is when u dont test the stuff

. I hope its ok when I add your improvments in an update of the old code ?
If you dont mind and if u have time enough it would be great if you could post some pictures of the 'good' and the 'bad' gears you cutted out - would be interesting to see the difference ...

@shoogen:

I think that creating an involute gear by cutting simulation is not the right solution.

Probably there are more ways to skin a cat. There might be adavantages/disadvantages to each of the two construction methods. I think both of them provide a correct involute curve, you can provide clearance and shift profile with both methods when applied correctly. Maybe its easier using the cutting method to generate a trochoid root curve than to calculate - but i am not sure about this ? As far as i know it will also be easier to cut arbitrary (non-involute) tooth-forms with the cutting method than to calculate these. On the other hand the cutting process seems to be very cpu intensive.
Maybe it would be good to make kind of a script that takes a (half) tooth as parameter and in return constructs a gear from it like jonasthomas pointed out in some prior post ? That would make it possible to test and verify different constructions methods fast and easily ...

But instead of modifying the rack to cut a clearance in the gear, the uncut should be designed to other goals.

Sorry, this sentence i dont understand ? ^

The tooth base should be designed to minimize the notch effect.

Can u explain what the notch-effect exactly is and how it could be minimized ?

Post by **shoogen** » Sat Nov 02, 2013 11:56 pm

I guess you are one of the lucky persons who never had to do calculations according to DIN 3990 in an exam.

The crucial factor regarding the stability of tooth base is the local stress. Having a rough surface increases the stress.
making the radius bigger would reduce the stress, but removing more material from the tooth base increases the stress.
http://en.wikipedia.org/wiki/Stress_concentration

quirxi · Post by **quirxi** » Sun Nov 03, 2013 12:13 am

shoogen wrote:I guess you are one of the lucky persons who never had to do calculations according to DIN 3990 in an exam.
The crucial factor regarding the stability of tooth base is the local stress. Having a rough surface increases the stress.
making the radius bigger would reduce the stress, but removing more material from the tooth base increases the stress.
http://en.wikipedia.org/wiki/Stress_concentration

Guess you are right - I never ever have made an exam that was even slightly related to DIN 3990 or gears in general

...
But how exactly do you think the tooth base should be designed in order to reduce stress ?

Post by **shoogen** » Sun Nov 03, 2013 1:31 am

the cycloid would be enough, IMHO. But we should not coarsen the surface with the cutting simulation. The effects might be neglect able in some cases but not in general. And if we use the simulation and the try something to make curve smooth afterwards we need to make sure the gears still meshes where there should and provide enough clearance where there shouldn't.
I don't think we should introduce advanced modifications yet. This is not my area of expertise. What we need is basic tools, that reflect the state of the art 50 years ago. Profile shift and helical gears are things that we need.
I just want to remind that cutting is not the only way to make a gear.

NormandC · Post by **NormandC** » Sun Nov 03, 2013 3:38 am

quirxi wrote:It seems that the board here has problems with threads that are too long.

I don't think so, my "Show your FreeCAD project here" topic is almost 3 times longer than the involute gear topic. I'm sure there are a few other topics longer too.

I think the problem is the forum has serious problems right now. This afternoon, and a few nights this week, I had timeouts and I would have lost a couple of replies if I hadn't copied them to a text file prior to hitting the Preview button (I've been burned in the past so I'm cautious now). At one point I had to wait for a few hours to get access back to the forum and paste my saved reply.

Let's leave this new topic as it is, and wait for the forum to get better. Maybe later we can merge this topic back to the original one. I already posted a link to this discussion in the original topic.

BTW you had created this topic twice, no doubt because of the forum troubles. I deleted the other one which didn't have any replies.

quirxi · Post by **quirxi** » Sun Nov 03, 2013 7:16 am

This afternoon, and a few nights this week, I had timeouts and I would have lost a couple of replies if I hadn't copied them to a text file prior to hitting the Preview button (I've been burned in the past so I'm cautious now). At one point I had to wait for a few hours to get access back to the forum and paste my saved reply.

I had the same problems as you describe here and lost a few posts. Therefore i pm'ed jriegel - who seems to be siteadmin - and he gave me the advice to reopen a new thread, what i did. But the first time i tried to create a new thread the problem occured again. I hope they can solve the problem here at sourceforge soon - would be a pitty to loose some good contributions because of that

.

quirxi · Post by **quirxi** » Sun Nov 03, 2013 7:34 am

shoogen wrote:the cycloid would be enough, IMHO. But we should not coarsen the surface with the cutting simulation. The effects might be neglect able in some cases but not in general. And if we use the simulation and the try something to make curve smooth afterwards we need to make sure the gears still meshes where there should and provide enough clearance where there shouldn't.

As far as i know there wont be any difference between the simulation and the calculation method concerning the roughness of the surface. For example: you can *calculate* 100 coordinates along the involute in order to fit the curve in (and maybe smooth it afterwards), and you can 'simulate' the cutting process so that you get exactly 100 coordinates along the involute. IMHO the result of the simulation method should be the same as with the calculation method what the smoothness of the curves is concerned. Or am I missing something ?

I don't think we should introduce advanced modifications yet. This is not my area of expertise. What we need is basic tools, that reflect the state of the art 50 years ago. Profile shift and helical gears are things that we need.

I agree ^

NormandC · Post by **NormandC** » Sun Nov 03, 2013 8:18 am

quirxi wrote: I hope they can solve the problem here at sourceforge soon

The sad truth is, expect nothing from Sourceforge. Last year, they announced that web apps such as MediaWiki and phpBB would not be supported anymore. They want their "customers" to migrate to their own developed platform, but AFAIK even with some early promises they have not made a migration path easy.

Image (jpeg) attachment on the forum has been broken for at least 18 months, and they will never fix it. As it is, the current admins (jriegel, wmayer and yorikvanhavre) have limited privileges over the forum functionality.

This is why the wiki was moved to an external server earlier this year. Sooner or later, the forum and the bug tracker (Mantis) will have to follow.

I don't know if this is related to SF's disinterest toward unsupported web apps, but you can't find the FreeCAD forum through a web search anymore.

quirxi · Post by **quirxi** » Sun Nov 03, 2013 11:15 am

Probably it would be better just to migrate to another place, even when it is a lot of work. But normally things dont get better after 3 years

jonasthomas · Post by **jonasthomas** » Thu Nov 07, 2013 4:41 am

Re: CONTINUED: involute gear generator preview !

Postby jmhrvy » Wed Nov 06, 2013 11:31 pm

quirxi wrote:by quirxi » Sat Nov 02, 2013 10:52 pm ]If you dont mind and if u have time enough it would be great if you could post some pictures of the 'good' and the 'bad' gears you cutted out - would be interesting to see the difference ...

Try this link http://www.youtube.com/watch?v=YUcWnFlfag0. It should take you to a short video, showing gears cut with both the original code [as posted on pg11], and the modified code [pg13, of the original thread]

I hope jbhrvy winds up making it to the new forum..
First.. He winds up cutting gears on a machine tool (of his own making??)... top it off he's making a stirling rhombic drive... very very cool