PDO variable 64 bit

One questions about variable can use with PDO.
On drive with 26 bit encoder , the PDO coomand position and encoder feedback position , if use float , is saturated with screew 10mm in 328 cm about.
The solution are 2 :
1- change gear numerator and denominator in drive
2-can declare in xml file a variable 64 bit

The 64 bit not is supported ?

Thanks

Ethercat 64 bit float is supported.
float-double-ieee
However, the ethercat drive delivers the encoder data in a certain datatype, and you can not change that.
For floating point there is the Ethercat REAL (32-bit) and LREAL(64-bit) data-types.
You convert them into a hal float (which is actually a 64-bit float, or double, since some versions back)
using the keyword "float-ieee" for REAL and "float-double-ieee" for LREAL.
On the other hand, for integers it is likely that linuxcnc is restricting by its 32-bit integers.
There are no 64-bit integers in hal (as far a I know is best to day).
 

For example , i now use :
<pdoEntry idx="607A" subIdx="00" bitLen="32" halPin="poscommand" halType="float" scale="838860.7"/>

for 64 bit float use :

<pdoEntry idx="607A" subIdx="00" bitLen="32" halPin="poscommand" halType="float-double-ieee" scale="838860.7"/>

is correct ?

No, unfortunately there isn't anything you can do about this. The first line will give you max number of digits.
The limiting factor is Ethercat's use of 32-bit float. Hal already uses 64-bit float.

If you in some way could increase number of digits from the encoder over Ethercat. That is, to make Ethercat use 64-bit floats for the encoder position, then the second line is the one to use, that will give 64-bit float in to hal.
But it usually isn't possible to increase to 64-bit output, only if you have exceptional control over the device like it is something you make yourself.

Ethercat master supports floats and double values. 

codeberg.org/skynet/realtime_app/src/bra.../DataTypeHandler.cpp

Yes it does. It works for 64-bit floats. Also lcec does support it. The whole chain works.
Actually, Scott added support for Ethercat 64-bit floats in lcec on my request.
I have tested with my own ethercat device with 64-bit floats. Works fine.
So it is "just" to make the device use another datatype - 64-bit floats. Unfortunately that "just" is impossible I guess.

Then the question is when do you actually need that much precision for the encoder?
This has been a limitation since day one, and things still work.

 

I use Mitsubishi MRj5 and it have encoder absolute value multiturn , without battery memory wirh resolution 26 bit.

With 26 bit you have much count for revolution and 32 bit pos-command hal pin, is full in 326 cm travel axis , with screew 10 mm pitch.

In drive parameter gear normally 1/1 , if you make a long machine with travel 10 meter , not can use 1/1 gear in drive , because after 326 cm moving , the pos-command return on -326 cm and linuxcnc return following error. For resolve now have increase gear in drive at 1/100 and can travel now 32600 cm. But normally is good have gear at 1/1 and use a 64 bit variable for command position and feedbak position.

i use debian 10 , with ethercat.
How can search path for this datatypehandler.ccp , for ceck if have support for 64 bit ?

If not have support , i can patch manually and recompile.

The sintax for use is :
<pdoEntry idx="607A" subIdx="00" bitLen="32" halPin="poscommand" halType="float-double-ieee" scale="838860.7"/>

or

<pdoEntry idx="607A" subIdx="00" bitLen="32" halPin="poscommand" halType="double" scale="838860.7"/>

You look at the drive's capabilities 

i use debian 10 with linuxcnc 2.9.pre0 and ethercat 1.5.

for use Pdo at 64 bit write in xml :

<pdoEntry idx="607A" subIdx="00" bitLen="32" halPin="poscommand" halType="double" scale="838860.7"/>

?