SEDM a hal comp to emulate Sinker EDM

[tjtr33]

New Readme is a tutrorial.
Chheatsheets are used to plan work.
PDF Etables are pritable references.

[tjtr33]

At 3/27/26, 23:15, I got email saying build failed.
It eas titled "[LinuxCNC/linuxcnc] PR run failed: Build CI - sedm a comp that emualtes Sink EDM"
The page I am on now says "All checks have passed... No conflicts with base branch"
I thinl all is OIK, and the email was stale. Thank you for looking.

[tjtr33]

I asked Google the following
"My co-worker say mey documentation reads poorly. What can I do if no English readers or speakers are available to review?"
I will also follow those suggestions,. Beginning with "read aloud".

[BsAtHome]

Please stop closing your pull request and then opening a new pull request right after closing the old one. It does not improve your code and makes review harder.

You still have to answer the questions and comments from #3877.

Additionally, I'd like to know how you created the .ENC files (all the files in configs/sim/axis/sedm/Etabs/).

[tjtr33]

Hello B
This iis to answer your 2nd ( and most recent post)

<Please stop closing your pull request and then opening a new pull request right after closing the old one. It does not improve your code and makes review harder.

Yes, stopped. I was doing that because it was the only way I knew to force the tests.

<You still have to answer the questions and comments from #3877.
you should see those answered in prev msg here on this page.

<Additionally, I'd like to know how you created the .ENC files (all the files in configs/sim/axis/sedm/Etabs/).
...in summary,
a LLM was creaated ( this was 1980 or near)
The data was curve fit to a formula where OberFlache was proportioinate to joulses
Initial currents ere set accorfong to area.
Finsl settings were set accccorging to surface finish
intervening setting wer walked dow recuing joules till finla finh reached.
thn tesated ( size with Busch&Lombe Opticalc omparator, finisg with Mitutoyo reflectometer)

the looooong story if you like....

data from many manufacturers was examined. All these data use different terms to specify amps volt ontime off time. Instead they used 'channels,' Frequency Duty Cycle Tau and more). All the data werre tanslated into standard units and their overburns examined. As is ecpected, N joules profuve thge same overburn ( oberflacke) and surface finish ( Ra Tz Hmax VDI)
S I has a relation ship for a single powewr setting Vurrent * seconds * disjcharge voltage si proportinal to overburn and surface finish.
How the initial current was chosen. The ideas is called Current Density and is like wire gauge 'ampacity'.
The urve is not linear as mopst texts explain, but is asymptotic * lik the bottom right corner of old CRT TV.
So the cross sectional area specifies the initial amperes setting.
The initila ontime is spedific tp the electrode material.
For graphite 80 to 120 uS ( very fine grain to coarse grain materials)
For Copper 2000 to 50 uS ( huge castings to spinnerettes ).
The joules dor the roughing was now known ( any size any tool material)
The final setting could be found using the houle to finish equiations.
Now the final setting was known.
The entervening stages were a 'walking' algorithm... split thje amps, keep the ontime, then split the ontime, keep the amps until the desired surcae finish wass reached.
Extremely small cuts required the ED pocess to attain the surface finish, wjile larger work coul d be polished easier because larger and time was saved. (Mirror dinishes are acievable by EDM but they take a very lomg time and specific work material, in general not worth it)

so to produce the E tables:
Make an LLM of cutting data, curve fit that to derive formulae, build tables per size and matrial,
then test a LOT. These tables were used by customers around the world. The manufacture's code and tables were replaced,. The suite was and use from ~1990 to presen. The manufactureres are out of business.
This is all origibal work and I never copyrtighhted it or signed any NDAs/
It was afe bbecause it was not understood . How about you? :-}

Tomp

[tjtr33]

Wpw I just asked Google:
"In sink EDM what is the relationship of joules and 'overburn'.
Woe, I spent a lot of years discovering what it replied.
https://www.google.com/search?client=firefox-b-e&channel=entpr&q=In+sinke+EDM,+what+is+the+realtion+of+joules+to+%27overburn%27+%3F&sei=gMXIadO2Af6r0-kP0cLM0Ao
I havew more W's for Gemeni. ..

[tjtr33]

The last 3 commits explained: I could not push, Git said there were files on remote that were not on local. So I made local BU, then ''git fetch origin; git reset --hard; origin'; git clean -d --force', Then found the missing commits in my BU. Then i did 'git add .; git -m msg' each content and 'git push origin sinkEDM' each step individually. I got no errors. I will not add or delete any files direct;y on GitHub again."

[andypugh]

Sorry Tom, but we discussed this in the online meeting yesterday and we don't think that it belongs in the LinuxCNC main repo.

. There is nothing compiled in here, so anyone can download the files to a stock LinuxCNC config and it will just work. It doesn't need to be integrated with LinuxCNC to work.
. It's an awful lot of files.
. We are unclear on the copyright / IP of the tables.

As an alternative, can I suggest keeping it as a standalone GitHub repository for interested parties to download? If you are on the forum, then you could post an announcement there pointing to it.
You could possibly also mention it on the Wiki under "contributed components"

I am also not comfortable with doing motion control outside the realtime layer. A better architecture would be to have the Python code create a much smaller set of parameters to drive a C realtime component to perform the oscillation. We might accept the C component as part of LinuxCNC, and link to your GitHub repository in the component docs.