I made a spur gear setup that has a high number of digits when you work out the pitch circle ratios.
When you check the properties of the mechanism session, the ratio is truncated to only 5 digits.
This causes a mismatch pretty quickly in a simulation. I have tried every way possible to define these including generic gears with user defined connections. User defined input will take a relation value with the high number of digits but the output doesn't change. It still calculated with 5 digits.
Anyone else run into this, and have any of you reported this? Is there a solution with some sort of precision setting?
Solved! Go to Solution.
Uhg!
I received a clarification e-mail from the PTC engineer responding to this request.
I am sorry to inform that rations are only stored at 6 digits.
Not only 6 digits beyond the decimal point, but a total of 6 digits.
A sample provided value: 123456789012 is stored as 1.23457e+011 (the mantissa of the value?)
Anyway, if you are dealing with friction or toothless gears or even bearings, be prepared for truncation.
Any cylindrical perimeter value will be truncated since the diameter will be calculated with Pi.
Even a fixed perimeter will be truncated because the diameter will be a value with significant number of digits.
I thought we were well past the days of single precision values?
I post this so people are aware of the limitations for long running mechanism analysis.
You could end up with cumulative error.
Please vote for this idea if you can to have PTC review this situation. Double precision for mechanism gear ratios
And another edit: PTC is looking into this with SPR 2202259
EDIT:
So I figured out my 1st problem due to an assumption... the ratio is obviously 2.4 even. ...10T and 24T mesh.
I had finally figured out exactly how to make these gears (actually 2 ways) but the question still remains... what if I have a very odd ratio (belt drive or friction gears)... does the software consider the Nth digit or is it truncated at 5. I have posed the question to the Support people.
If anyone is interested in this gear-style, let me know and I'll post a document.
Here is the video. After changing the ratio's to the actual diameters (DIAM1 and DIAM2), I re-run the analysis yielding the same results. Notice that the gear dialog does maintain a lot more digits than the information dialog reflects.
These are the gear definitions: Each has the same diameter "teeth" and the pitch diameter is determined by the angle. These should be "perfectly meshed" gear sets where nice round ratios are rarely achieved (re: 1:1)
I unlisted the video on YouTube since that was my error.
I received a reply from PTC R&D and they assure me that ratios are calculated to 11 digits.
I did follow up with a question regarding the entry values... where an entry of 2/3 may become .66667 or is in fact processed as .66666666667.
You might enjoy the Fellows Gear videos on Youtube. They were generated (pun) to show the geometry behind involute gear generation. What I found nice about them is the quality of the demonstration.
In searching for them I just found this: www dot bugman123 dot com/Gears/ Just Wow!
Nice find, David.
I still don't get the magic solution in modeling gears. Fortunately there are other people that do.
Uhg!
I received a clarification e-mail from the PTC engineer responding to this request.
I am sorry to inform that rations are only stored at 6 digits.
Not only 6 digits beyond the decimal point, but a total of 6 digits.
A sample provided value: 123456789012 is stored as 1.23457e+011 (the mantissa of the value?)
Anyway, if you are dealing with friction or toothless gears or even bearings, be prepared for truncation.
Any cylindrical perimeter value will be truncated since the diameter will be calculated with Pi.
Even a fixed perimeter will be truncated because the diameter will be a value with significant number of digits.
I thought we were well past the days of single precision values?
I post this so people are aware of the limitations for long running mechanism analysis.
You could end up with cumulative error.
Please vote for this idea if you can to have PTC review this situation. Double precision for mechanism gear ratios
And another edit: PTC is looking into this with SPR 2202259