[Je...] Posted September 18, 2019 Share Posted September 18, 2019 What is the difference in Spline Vs. Gauss filtering ? I'm measuring 300 mm bores with a 0.075 tolerance. 500 mm deep with about 25 circles. The cookbook states to use Gauss filtering method and I had seen some discussion on the forum about the cookbook being misleading at times . Thanks Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/ Share on other sites More sharing options...
[To...] Posted September 18, 2019 Share Posted September 18, 2019 I don't know the exact difference between the two but 99 out of 100 times, I have not noticed a difference in the result. It's been my understanding the Gauss worked best on normal geometry and it know where the data begins and ends. Where spline filtering does not know what is before the beginning and what is after the end. My 2 cents. Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/#findComment-62072 Share on other sites More sharing options...
[Cl...] Posted September 18, 2019 Share Posted September 18, 2019 Generally, GAUSS for circle/slots/cylinders, <8mm dia = 15 upr - >8mm dia = 50 upr Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/#findComment-62074 Share on other sites More sharing options...
[Je...] Posted September 18, 2019 Author Share Posted September 18, 2019 Thanks Guys Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/#findComment-62076 Share on other sites More sharing options...
[De...] Posted September 18, 2019 Share Posted September 18, 2019 The following is a pretty informative document on the history of and the relevance of different types of filtering. Fair warning, this is not light reading.. http://www.iigdt.com/1/2/Data%20Filteri ... rystek.pdf Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/#findComment-62078 Share on other sites More sharing options...
[Ch...] Posted September 18, 2019 Share Posted September 18, 2019 Gauss and Spline work almost identically for closed loops such as a circle or something. Gauss is not recommended for open ended loops, e.g. lines or curves. Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/#findComment-62080 Share on other sites More sharing options...
[Cl...] Posted September 18, 2019 Share Posted September 18, 2019 Please sign in to view this quote. Definitely not light reading, after reading part of that, I felt like a Walmart shopper. Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/#findComment-62082 Share on other sites More sharing options...
[Is...] Posted September 19, 2019 Share Posted September 19, 2019 Please sign in to view this quote. In my case, we don't use anymore spline filters in curve features. filters could modify the actual geometry; in our developments spline filters perform very bad sometimes, the folowing plot shows the difference applaying the same wavelength value. Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/#findComment-62084 Share on other sites More sharing options...
[To...] Posted September 19, 2019 Share Posted September 19, 2019 Please sign in to view this quote. What is the wavelength value? What are the step widths of the curve and the strategy? Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/#findComment-62086 Share on other sites More sharing options...
[Cl...] Posted September 19, 2019 Share Posted September 19, 2019 Please sign in to view this quote. Israel, do you use curve length / 2 = wavelength value? Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/#findComment-62088 Share on other sites More sharing options...
[No...] Posted September 19, 2019 Share Posted September 19, 2019 Please sign in to view this quote. The reason is that Gauss produces unwanted effects at the end of an open profile. Derek's document has some informative graphs comparing the effects of Gauss and spline filters on open profiles (pages 6 to 15) Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/#findComment-62090 Share on other sites More sharing options...
[Kl...] Posted September 19, 2019 Share Posted September 19, 2019 Wavelength = CurveLength/2 This will be the hardest filter, because all smaller "waves" are suppressed and only the Ground form is visible. Using shorter Lambda shows the higher "harmonics" (waves) at the contour. So Lambda depends what you looking for. Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/#findComment-62092 Share on other sites More sharing options...
[Is...] Posted September 20, 2019 Share Posted September 20, 2019 The graphic that I put is just used as visual example... for that particular case the curve length is around 28 mm and the wavelength is 8.0, in real development cases I would apply wavelength values between 0.8 and 2.0 just enough to soft the roughness. Even with those small filter values I have had bad experiences with the Spline filter.. no problem until now with Gauss. Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/#findComment-62094 Share on other sites More sharing options...
[Is...] Posted July 27, 2021 Share Posted July 27, 2021 In the following plot the curve length is around 17 mm and the wavelength value is increased gradually from 0.5 to 2.5. Spline filter generates a strange jump in the lower corner between wavelength 0.5 and 1.0, and the in the upper corner between wavelength 1.5 and 2.0. No problems with Gauss. In this case I would apply a Gauss Filter Between 0.5 and 1.5 Link to comment https://qualityforum.zeiss.com/topic/12404-filtering-question/#findComment-62096 Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in