Grinding software upgrade is coming

[wootz]

Jans, simply start your Tormek, and set the grinding angle using your formula and _TormekCalc.xlsx.
Grind a single-bevel blade and a double-bevel and observe difference between the resulting edge angle and the calculated.
Then grind 2 double-bevel blades of differing thickness behind the edge and observe difference between the resulting edge angle and the calculated.
Then take the thicker blade and grind it at 20 degrees, and at 12 degrees, and observe how the deviation from the calculated angle increases.

Then you probably will reread Ton's explanation with more understanding.

What Ton does not mention, but I've found out, is that grinding with the wheel rotation goes differently to grinding into the wheel and requires additional mathematics. Therefore, you will see that the angle calculated for the Frontal Vertical Base by _TormekCalc.xlsx deviates from real edge angle even more.

I understand that accepting that may be frustrating to you both, but we have to adjust maths to the real world, the vice versa doesn't work.

[Jan]

Vadim, in real life I am sharpening double-bevel, medium quality kitchen knives, which usually have the thickness behind the edge 0.4 to 0.5 mm. For those knifes I have never observed deviations of the actually grind angle and the desired angle. The accuracy of my angle measurements is cca 0.5 degrees. 

When the blade thickness behind the edge is larger than some 0.75 mm then the knife usually does not perform well, despite small BESS score, and is an adept for blade thinning.

For routine USB set up (over 80% of my cases) I use my double ended kenjig for projection length of 139 mm and bevel angle 15 degrees. New diamond wheels with fixed diameter made this even simpler.

Jan

[Dutchman]

... grinding with the wheel rotation goes differently to grinding into the wheel and requires additional mathematics.
...

Can you explain that?

[jvh]

Grind a single-bevel blade and a double-bevel and observe difference between the resulting edge angle and the calculated.
Then grind 2 double-bevel blades of differing thickness behind the edge and observe difference between the resulting edge angle and the calculated.
Then take the thicker blade and grind it at 20 degrees, and at 12 degrees, and observe how the deviation from the calculated angle increases.

TormekCalc calculations are made in the axis of the jig and the knife, so results are valid for all symmetrical double bevel knifes with defined angle at the top of the edge. Error given by material thickness is suppressed there completely.

For single bevel knives is a correction necessary to get "right" axis of the jig. For this you can change diameter of USB or jig diameter value and "virtually" move the jig axis to the right place. Then you get exact results for single bevel knifes with defined angle at the top of the edge. Error given by material thickness is suppressed by the jig axis movement.

What is still present is error given by "roundness" of the wheel – see picture Edges.jpg – which caused that angle increases along the hollow. As you can see, the error is bigger with thicker material.

The essential thing is that you cannot eliminate* this error when grinding on rounded wheel. The only thing you can do is shift desired grinding angle closer to the edge heel. Then you get the desired angle at the edge heel but angle at the top of the edge will be smaller accordingly. If you shift the desired angle to the middle of grinded edge then you get smaller angle at the top of the edge and higher at the edge heel.

In view of the above, it is necessary to ask: Which point on the edge is the right one for the defined angle? Top? Middle? Heel? Something between?
It's no problem to do it in TormekCalc – just change the jig projection length and you will get desired angle at the point where you want. Small corrections are sometimes needed while grinding because of change of the jig projection length, or measurement errors. It's really not dark magic or rocket science.

What Ton does not mention, but I've found out, is that grinding with the wheel rotation goes differently to grinding into the wheel and requires additional mathematics. Therefore, you will see that the angle calculated for the Frontal Vertical Base by _TormekCalc.xlsx deviates from real edge angle even more.

Sorry, without detailed explanation this doesn't make a sense.

I don't have any problem with grinded angles while using my FVB (regardless of direction of rotation). It should be noted here that all constants in TormekCalc have been set for my T-8 and FVB and may differ in other devices.

I understand that accepting that may be frustrating to you both, but we have to adjust maths to the real world, the vice versa doesn't work.

It's not about frustration, it's about precise inputs and corresponding outputs.
I made a robust testing on TormekCalc and I didn't find any problem with calculations. Here I write about some limitations and explain why it is so. TormekCalc is free and everyone can test it and check its outputs. I have no evidence that there is something wrong except your claim, which is not specifically substantiated, so I try to make it right.

I have indicated that I have doubts about the accuracy of your measurements, which affect the grinded angle and the subsequent interpretation of the results.
There is why:
1. Jig projection length (video): Minimal supposed measuring/reading error ±0,25 mm
2. USB height (video): Minimal supposed measuring/reading error ±0,1 mm
3. Wheel diameter (video): Minimal supposed measuring/reading error ±0,1 mm

All these minimal errors together (in the worst scenario) will change the desired angle by ca ±0,25° at the exact jig projection length. Errors caused by FVB constants and by hand grinding aren't included.

4. CATRA Hobbigoni protractor (video)
Scale division is poor, supposed reading error can be ±0,5°, big reflection pattern can be other source of errors. Declared accuracy by Catra "measures the sharpened angles to an accuracy of ±2°".

Therefore I am skeptical of the reported results which seems to me inconsistent and I cannot simulate or verify them.


*) Actually it's possible with flat grinding on diamond wheels or much more difficult way by changes of jig projection length while grinding.

[Jan]

JVH, I again agree with your arguments.

In view of the above, it is necessary to ask: Which point on the edge is the right one for the defined angle? Top? Middle? Heel? Something between?

The bevel angle is traditionally defined at the tip of the edge.


Jan

[wootz]

At my workshop we sharpened 3 types of blades, under 1mm behind the edge, 2mm and 3mm thick behind the edge,
from 10 to 20 degrees per side, at 1 degree step, registering the actual edge angle vs calculated.

We repeated this grinding into the edge, and grinding with the edge using our Frontal Vertical Base.

This way we collected a set of experimental data.
Looking at these data I saw regularities that I expressed mathematically and included into our computer algorithm.

We then made another round of the above 100 sharpening sessions to confirm the new algorithm on 2 Tormek-8 and 2 Tormek-7 machines, as detailed in the Software Testing on our website http://knifegrinders.com.au/software_testing.htm

Our computer algorithm is the most accurate and the best tested.
This way we've improved matching of the ground angle to the honing angle, and knives sharpened using our updated software and Frontal Vertical Base are sharper than ever. Everyone can now get hair-splitting edge on his Tormek machine.

Jans, you better put aside your pencil, switch off your computer and switch on your Tormek.

Another good news: last week Tormek agreed to supplying us directly with the XB-100 part for our Frontal Vertical Base. The first boxes arrive to Australia in 2 weeks, and we won't have interruptions in production as we used to when we sourcing this part through retailers.
We are also the very first in Australia to get Tormek Extended Universal Support US-430, to supplement our Frontal Vertical Base but not sell individually.

[jvh]

We then made another round of the above 100 sharpening sessions to confirm the new algorithm on 2 Tormek-8 and 2 Tormek-7 machines, as detailed in the Software Testing on our website http://knifegrinders.com.au/software_testing.htm

Our computer algorithm is the most accurate and the best tested.

Sorry, with all respect to you I have to say that it's just your statement without any evidence. Your test results looks like from ideal world, at least you should quantify the overall error (systematic + random) that occurs during your measuring and grinding. Otherwise your results looks suspicious.

I cannot verify your testing procedure but I saw in your videos (in production run!) some problematic parts with length and angle measurement which I pointed in my previous post.

In my opinion, these errors and used equipment do not allow you to work with the accuracy you declare.

That's all, have a nice day.

jvh

[Dutchman]

...
This way we collected a set of experimental data.
Looking at these data I saw regularities that I expressed mathematically and included into our computer algorithm.
...

So you adjusted your formulas to fit the experimental data, but without knowledge of the cause?
I have no problem with that, except that it is unsatisfactory.

[Jan]

I am very sorry to say, that the discussion reveals that the very laborious experiment was not prepared correctly. Design of experiments is extremely important because only correctly designed experiments can advance knowledge in natural sciences and engineering.

Offensive communication cannot establish validity and reliability of results or reduce the measurement errors.

Jan

[Jan]

Jans, you better put aside your pencil, switch off your computer and switch on your Tormek.

Vadim, my recommendation is you switch off all your Tormek machines, take your pencil, switch on your computer and rethink the design of your experiments. Otherwise it can easily happen that all your toil be in vain. That would be a great pity.

The nice CAD drawing prepared by JVH is an excellent starting point for a constructive discussion. I was dreaming about such a drawing for a long time, but my light CAD sw was not able to depict the strongly zoomed geometry of the edge and wheel accurately. Thanks JVH my dream has come true. 

Jan

[Ken S]

As forum moderator, I am getting concerned with this topic. You are all valued members of this forum. You have contributed much and I would not want to lose any of you. Please back away from this topic and let things cool down.

Ken

[Dutchman]

Wootz
Regarding the following statement, I can point out two reasons why there is a difference.
Quote from: wootz on November 04, 2019, 03:56:26 am
=> What Ton does not mention, but I've found out, is that grinding with the wheel rotation goes differently to grinding into the wheel

• The mounting of the USB is different in both cases. Either in the FVB or in the original position on the other side.
• You adjust the position of the USB relative to the bottom not to the center of the stone.
With this setting, the distance from the USB to the center of the stone should be the same in both cases to obtain the same grinding angle.
Your experiments however show that this is not correct and you have adjusted the "math" accordingly.
I recommend that you take a critical look at the position of the FVB, as the deviation must come from there.

However, a good method to set the USB gives the "KenJig".
There is no better way. 

[jvh]

However, a good method to set the USB gives the "KenJig".
There is no better way. 

Excellent advice for verification! 

I would like to add that this "KenJig" dimension ("T" in my drawing) stays the same for all types of universal supports (on the same wheel for the same grinding angle and jig projection length).

jvh

[Jan]

JVH, thanks for perfectly documented calculation of the kenjig length T. The simple concept works for all Tormek machines! You do not need to know all the constants of different Tormek models. I have checked your length T = 83.19 mm and can confirm that it is correct and identical with my calculations for the same input parameters.

The same distance you will get from the new Dutchman tables also. (77.19 + 6.00 = 83.19 mm)

I agree with Dutchman. Kenjig is for me king of jigs. It is probably the fastest USB setting. 

Jan

[Ken S]

The kenjig is a very simple device, easily made of cardboard or plywood in only a few minutes in a home shop. Its magic is derived from very solid math. I no not understand the math beyond a basic level. I am grateful we have forum members like you guys with a solid mastery of our math.

I have seen amazing innovation from the forum. I am convinced we still have much more to develop. I look forward to innovative cooperative thinking from all of you.

Ken