News:

Welcome to the Tormek Community. If you previously registered for the discussion board but had not made any posts, your membership may have been purged. Secure your membership in this community by joining in the conversations.

www.tormek.com

Main Menu

Grinding software upgrade is coming

Started by wootz, October 15, 2019, 08:42:11 AM

Previous topic - Next topic

wootz

Quote from: smurfs on October 23, 2019, 05:33:50 PM
Hi Vadim,

Could the angle issue simply be due to the reduction in the jig projection length during grinding, as illustrated in the schema diagram in your initial post? If it is would it not be better to recheck the projection after each grinding step in your sharpening procedure, and if necessary return to the initial projection length using the adjustable stop? Perhaps such a tweak is all that is needed to ensure the target edge angle is always maintained.

Or may be I've missed something and it is not as straightforward as I think  :-\

Andrew

I wish it were that simple.
What we've been doing for the last 2 weeks - grouped knives by blade thickness behind the edge: under 1mm, near 2mm and near 3mm, and sharpened each group from 10 to 20 degrees at 1-2 degree step, on two T8 and two T7 machines, in the standard position and edge-trailing using our Frontal Vertical Base.
The laser protractor numbers tell us that the ground angle we get depends on 4 factors: the angle, the thickness behind the edge, whether the blade is double-bevel or single-bevel, and whether the grinding is done edge-leading or edge-trailing.

Our formulas now factor in the difference between the double-bevel and single-bevel blade, and the blade thickness behind the edge. It is not the same as the thickness at the spine, so has nothing to do with modifying the knife jig to center the blades better. You can have a blade 5 mm thick at the spine but thin behind the edge. But of course, you still have to center the blade in the jig as practical.

The new computer algorithm in our software not only has formulas that factor in the blade thickness, but also follows the experimental numbers we've obtained for each edge angle from 10 to 20, ground edge-leading vs edge-trailing, to grind the correct target angle no matter what.
We've made a big change from a pure theoretical algorithm to the one fine-tuned by 100 of experimental sharpenings. This makes the match to the honing angle perfect. At the end of the day, all the man-hours we've put into this improvement are to better the deburring and sharpness.

wootz

A word on home-made FVBs.
4 months ago we started CNC machining the FVB base, and have tolerances tight.
If your self-made FVB does not have the base of the same thickness as ours or legs in the wrong position, you will be getting less accurate results, and all the man-hours we've put into this software improvement will be wasted as you will not be getting better deburring and sharpness with it.
I'll be stopping making custom applets for home-made FVBs because of that, because formulas by themselves will never match the algorithm fine-tuned to our particular FVB make.

van

Quote from: wootz on October 23, 2019, 07:29:23 PM
Quote from: smurfs on October 23, 2019, 05:33:50 PM
Hi Vadim,

Could the angle issue simply be due to the reduction in the jig projection length during grinding, as illustrated in the schema diagram in your initial post? If it is would it not be better to recheck the projection after each grinding step in your sharpening procedure, and if necessary return to the initial projection length using the adjustable stop? Perhaps such a tweak is all that is needed to ensure the target edge angle is always maintained.

Or may be I've missed something and it is not as straightforward as I think  :-\

Andrew

I wish it were that simple.
What we've been doing for the last 2 weeks - grouped knives by blade thickness behind the edge: under 1mm, near 2mm and near 3mm, and sharpened each group from 10 to 20 degrees at 1-2 degree step, on two T8 and two T7 machines, in the standard position and edge-trailing using our Frontal Vertical Base.
The laser protractor numbers tell us that the ground angle we get depends on 4 factors: the angle, the thickness behind the edge, whether the blade is double-bevel or single-bevel, and whether the grinding is done edge-leading or edge-trailing.

Our formulas now factor in the difference between the double-bevel and single-bevel blade, and the blade thickness behind the edge. It is not the same as the thickness at the spine, so has nothing to do with modifying the knife jig to center the blades better. You can have a blade 5 mm thick at the spine but thin behind the edge. But of course, you still have to center the blade in the jig as practical.

The new computer algorithm in our software not only has formulas that factor in the blade thickness, but also follows the experimental numbers we've obtained for each edge angle from 10 to 20, ground edge-leading vs edge-trailing, to grind the correct target angle no matter what.
We've made a big change from a pure theoretical algorithm to the one fine-tuned by 100 of experimental sharpenings. This makes the match to the honing angle perfect. At the end of the day, all the man-hours we've put into this improvement are to better the deburring and sharpness.
Congratulations on your work and self-denial.
I think maybe some photos would make everything more understandable
with regard
Kindly yours

wootz

#18
Quote from: Dutchman on October 22, 2019, 09:55:33 AM
Thank you for mentioning the source :)

Ton, I believe in credit where it is due.
Your name is in the Acknowledgements of my book


wootz

Quote from: Jan on October 22, 2019, 01:07:19 PM
I have shaped several blanks into blades without observing differences between the desired and sharpened bevel angle provided that the projection length has not been shortened during blank sharpening.

I use my Excel script to calculate the distance between the USB and the wheel surface. My script works without approximations.

Jan

I did trial sharpening by Jan's formula / JVH _TormekCalc.xlsx with a knife 2.5 mm thick at the spine - the results:



If you have a laser protractor, you can see to yourself that neither Jan's formula nor _TormekCalc.xlsx by JVH grind to the exact angle, and the thicker the blade, the greater is deviation from the target.

While our updated Grinding Angle Setter and the FVB applet grind exact angle, as I've shown in the testing results.

Jan

#20
Wootz, exact estimation of bevel angle is not an easy task especially for hollow grinded blade. The angle at the tip and at the heel are different. We are interested in bevel angle at the tip.

Catra offers two categories of laser goniometers: hobby and commercial.
1)   Hobby laser goniometer "HOBBIGIONI" measures sharpened angle
to an accuracy of +/- 2 degrees
2)   Portable laser goniometer measures sharpened angle to an accuracy of +/- 1 degrees
3)   Laboratory laser goniometer measures sharpened angle to an accuracy of +/- 0.5 degrees

The commercial goniometers have focusable lasers, because without the knowledge of the laser beam diameter we can non make serious angle measurements, because we simply do not know how large segment of the edge we are measuring.

The statistical significance of your results is very low because the sharpened angle was estimated with too large error. 

Even the accuracy of lab laser goniometer ad3), which costs some 3000 Euro, is hardly sufficient to demonstrate the dependence of sharpened angle on the blade thickness behind the edge.

In 2016 I have built my homemade laser goniometer and I know in detail all the pitfalls that this method brings. The accuracy of my goniometer is between 0.5 and 1 degrees. The problem is not the accuracy of reading but the interpretation of the reflections from the edge. https://forum.tormek.com/index.php?topic=3104.0

Jan



wootz

#21
Jan, I knew you would be faulting the method because you are in denial of the fact that the blade thickness behind the edge must be factored into the calculation.

We've done a 100 of test sharpenings for last weeks to make our Grinding Angle Setter and the Frontal Vertical Base applets accurate. We do it for our customers, not for fun. Our updated software release is coming in 10-14 days. You then can get our updated applet, and see to yourself how well it works, not by your abstract estimation, but in actual sharpening.

Jan

#22
Vadim, I truly deeply admire your commitment to do these laborious experiments.

You are correct, I still have not digest the fact the sharpened angle depends on the blade thickness behind the edge. For me it is a new, almost revolutionary, hypothesis and that is why I am reluctant to accept it without detailed clarification. I do not see this thickness as a variable in the new Ton formula F9.

Sorry for my pedantic approach. Some time ago I worked for UN on screening and verification procedures and since that time I know how carefully each new hypothesis has to be analysed.

Because I have all necessary hw and sw tools I am conservatively reviewing and validating new procedures and formulas posted on this forum. You know that some two years ago I analysed the original Ton's formula for bevel angle and found that it is only approximate and can give slightly biased results. Last year Ton revised his approach, derived a new formula which is exact. I have validated also his new tables and found they are OK now.  I am happy about it, I see my mission in it.

Vadim, I wish you all the best with your sw upgrade. I am looking forward to see additional data supporting your new hypothesis.  :)

Jan

TorbenDenmark

Hi Jan,

Some time ago I made my own excel sheet to calculate these things. My calculations are not based on trigonometry but on vectors. I believe that my method does not introduce any errors.

You mention a new formula that you believe is correct Jan. I would very much like to compare my numbers with that. I have been searching this forum for 'ton', 'booklet' and more but nothing has showed up. Is this formula available for me? If so, can you help me find it?

BR
Torben

Jan

#24
Torben, you are welcome. The Ton Nillesen formula is of course available for you also. The original but updated Dutchman thread "Simple adjustment of the grinding angle" is here: https://forum.tormek.com/index.php?topic=1849

The correct formula F9 you will find in a booklet More math for the Tormek grinder A5 serial.pdf stored at dropbox.
https://www.dropbox.com/sh/zfzeuyyhors8oid/Yyq5S71LxR

Please let me know if you have some additional questions.

Jan

Dutchman

Quote from: TorbenDenmark on November 01, 2019, 02:45:43 PM
...
I have been searching this forum for 'ton', 'booklet' and more but nothing has showed up. Is this formula available for me? If so, can you help me find it?
...
See links in my 'Signature'  ;)

jvh

Quote from: wootz on October 26, 2019, 12:51:13 PM

I did trial sharpening by Jan's formula / JVH _TormekCalc.xlsx with a knife 2.5 mm thick at the spine - the results:



If you have a laser protractor, you can see to yourself that neither Jan's formula nor _TormekCalc.xlsx by JVH grind to the exact angle, and the thicker the blade, the greater is deviation from the target.

While our updated Grinding Angle Setter and the FVB applet grind exact angle, as I've shown in the testing results.

Hello,

With all respect to you work I have doubt about test procedure and result interpretation.

I have never noticed the dependency of the grinded angle and the thickness of the blade behind the edge. Math used in TormekCalc is precise and verified firstly in CAD simulations and then in real life. I have to say that I get the expected results every time I grind. It doesn't mean that small corrections aren't needed, but they are always caused by dimension changes or measurement errors.

The most common is the jig projection length shortening if you grind too much (under the axis). Wheel diameter change should be considered too in longer grinding.
Measurement errors also affect grinded angle (e.g. ±0,16 mm of the jig projection length changes angle by ±0,1%; ±0,14 mm of the vertical USB heigth changes angle by ±0,1%).

Next thing is laser goniometer accuracy (Catra Hobbigoni has accuracy ±2 °) and its scale division is very tight (low resolution). Interpretation of laser reflecting is another source of errors.

Last at not least your results in this trial sharpening (12° gave + error , 20 ° gave - error) don't make a sense. Result seems to me inconsistent and I cannot simulate/verify it.

Please do not misunderstand this post, it is not meant to be offensive. Just I didn't notice such problem and there aren't detailed information about testing procedure (used equipments, measuring accuracy for all parts etc.). Therefore I have a doubts and I see possible systematic errors. Of course I can be completely wrong, but I haven't found any convincing evidence yet.

jvh

wootz

#27
Ton Nillesen from Netherlands gave us mathematical description of knife sharpening on Tormek using an abstract model of a thin single-bevel blade.
Jan did a little tweak to Ton's formula, but this does not change the fact that it still describes the abstract thin single-bevel blade.
The _TormekCalc.xlsx by Jan JVH on this forum uses Jan's tweak as well, and still calculates grinding angle as if the blade is single-bevel and thin.

How on earth can this be accurate for a double-bevel blade?! For a thick blade? And especially for a thick double-bevel blade?

Formulas are giving a systematic error that Ton Nillesen explains in his 2018 booklet on page 6; and as Ton has shown, the thicker the blade and the lower the grinding angle, the greater is the error.
Why the 2 Jans persist in denial of what the author of this mathematical model himself says?

I've revamped my computer algorithm to take into account the blade particulars, and now our applets grind correct angle on double-bevel, single-bevel, thick and thin knives.

Grinding Angle Setter for Tormek
Released for Windows, MacBook, iPhones and iPads, and Android phones and tablets.
Details on our website.

Instructional video Software for Tormek >>

Frontal Vertical Base applet
Currently under review by Google Play and Apple Store. Will be released next week.

As soon as the FVB applet is released, I will email users of our software instructions on how to update both the Grinding Angle Setter and the FVB software free of charge.
... just in a few days...

jvh

#28
O.K., this is what's going on when grinding on any wheel - top of the edge has defined/counted angle (e.g. 20 °). Different angle is at the edge heel and it vary in dependency on thickness of the blade. This error is given by "roundness" of the wheel and can be eliminated by flat grinding on diamond wheels side only.

If you change grinding angle / heigth of the USB / jig protrusion length you will get better result at the edge heel but this will change the angle at the top of the edge accordingly.

All these errors exist but they are countable with pure math without any approximation and can be taken into account while grinding.

I still have doubt about your test procedure and result interpretation for some reasons. The main one is accuracy of used equipment. How you can see on Catra goniometer difference between angle 12 and 12,2°? The next big problem is height of the grinded edge. If you grind sharp angle or thick material you get very high grinded edge which makes wide reflection on laser goniometer. How do you evaluate such large reflection pattern with high accuracy? What is the accuracy of measurement of the jig protrusion length? What is the accuracy of measurement of the USB height? All these values (and many others) affect results and repeatability.

TormekCalc.xlsx
doesn't use Jan's tweaks at all. All calculations are made in the axis of the knife, results are valid for all double bevel knifes (symmetrical) at the top of the edge. For single bevel knives is possible to make a correction of dimension JC by changing diameter of USB.

Jan

#29
JVH, thank you very much for your drawing! :)  It nicely shows how the angle increases along the hollow. This angle increase is visible in measurement with laser goniometer as ellipsoidal widening of the dot reflected from the very edge.

Your drawing makes it clear that there is no assumption concerning edge thickness behind the edge in your approach. The same is true for my approach and also for the Ton formula F9.

Hopefully our reservations will help Vadim to reconsider his approach and mainly to explain the changes embedded in the sw upgrade more clearly. It is important for the sw acceptance by the grinder's community. It would not be wise to keep the sw upgrade as black box with some ad hoc correction formulas of unknown provenience and accuracy. The reference to the page 6 of Ton booklet 2018 is fully insufficient.

Jan