linuxcnc-ethercat

Driver for Beckhoff EL3xxx Analog Input Devices

The lcec_el3xxx driver supports a wide range of analog input devices and replaces a number of older single-purpose drivers.

It currently supports 4 basic families of devices:

• EL30xx 12-bit analog input modules
• EL31xx 16-bit analog input modules
• EL32xx 16-bit temperature sensor modules
• EM37xx 12-bit pressure modules

See the source code or device documentation for precise details on which hardware is currently supported. In general, we aim to support EL3xxx and the equivalent EJ, EP, and EPP modules, but this is not yet complete.

How to file a bug

Please file an issue if you have any problems with this driver, including:

• hardware that should work but doesn’t.
• similar hardware that should probably should be supported.
• unexpected limitations.
• features that the hardware supports but the driver does not.

EL30xx modules and issues

The EL30xx devices are fairly basic. To configure them, add a line like this to your ethercat.xml file:

    <slave idx="10" type="EL3068" name="D10"/>

This should create a bunch of lcec.<MASTERID>.<SLAVENAME>.ain-<CHANNEL>-<TYPE> pins in LinuxCNC. The pin types include:

• bias – settable to adjust the value returned in val.
• error – true if the hardware has logged an error.
• overrange – true if the input is over range.
• raw – the raw input measurement from the device.
• scale – settable to adjust the value returned in val.
• underrange – true if the input is under range.
• val – the current measurement on the input, with scale and bias applied.

Examples:

     8  float I/O             0  lcec.0.D10.ain-0-bias
     8  bit   OUT         FALSE  lcec.0.D10.ain-0-error
     8  bit   OUT         FALSE  lcec.0.D10.ain-0-overrange
     8  s32   OUT             0  lcec.0.D10.ain-0-raw
     8  float I/O             1  lcec.0.D10.ain-0-scale
     8  bit   OUT         FALSE  lcec.0.D10.ain-0-underrange
     8  float OUT             0  lcec.0.D10.ain-0-val
     8  float I/O             0  lcec.0.D10.ain-1-bias
     8  bit   OUT         FALSE  lcec.0.D10.ain-1-error
     8  bit   OUT         FALSE  lcec.0.D10.ain-1-overrange
     8  s32   OUT             0  lcec.0.D10.ain-1-raw
     8  float I/O             1  lcec.0.D10.ain-1-scale
     8  bit   OUT         FALSE  lcec.0.D10.ain-1-underrange
     8  float OUT             0  lcec.0.D10.ain-1-val

Even though this is a “12-bit device”, Beckhoff returns results in the range of -0x7fff - 0x7fff, which is 16 bits of precision. For devices where negative answers don’t make sense, this range is actually 0-0x7fff (or 32,767).

The value of val is raw / 32767 * scale + bias. Note that this means that val ranges between 0.0 and 1.0 unless scale and bias are changed.

EL31xx modules and issues

Note: It is possible that some older EL31xx devices (for example, EL3104 revision r16 or r17) may not work with this driver. If you have hardware that fails to load with a PDO error, please file a bug (see above for directions).

EL31xx devices should match the description of EL30xx devices above, with one minor difference: there is also a sync-err pin that will report on distributed clock sync failures.

EM37xx pressure modules and issues

The Beckhoff EM3701, EM3702, and EM3712 should all be supported. They are very similar to EL30xx devices, except the pins are named starting with press* instead of ain*.

EL32xx temperature modules and issues

The driver now supports Beckhoff EL32xx temperature modules. In general, we should be able to support all of the EL32xx temperature modules, and probably some of the EL33xx modules, if anyone has hardware to test them with. File a bug (see above) if you find unsupported hardware.

The temperature modules have a few optional settings in ethercat.xml to control temperature-specific featues:

    <slave idx="16" type="EL3204" name="D16">
      <modParam name="ch0Sensor" value="Pt100"/>
      <modParam name="ch0Resolution" value="High"/>
      <modParam name="ch0Wires" value="2"/>
      <modParam name="ch1Sensor" value="Ohm/16"/>
	  ...
	</slave>

There are 3 <modParam> settings for each channel:

• chXSensor: controls the temperature sensor type attached to the channel. Various sensors have different resistance curves. Supported types are:
  • Pt100: Pt100-type sensor, -200 to +850C
  • Pt200: Pt200-type sensor, -200 to +850C
  • Pt500: Pt500-type sensor, -200 to +850C
  • PT1000: Pt1000-type sensor, -200 to +850C
  • Ni100: Ni100-type sensor, -60 to +250C
  • Ni120 Ni120-type sensor, -60 to +250C
  • Ni1000 Ni1000-type sensor, -60 to +250C
  • Ni1000-TK5000 Ni1000-type sensor that reads 1500 Ohm at 100C, -30 to +160C
  • Ohm/16 No sensor, read resistance directly. 0-4095 Ohms.
  • Ohm/64 No sensor, read resistence directly. 0-1023 Ohms.
• chXResolution: controls the resolution of returned results.
  • Signed: returns results with 0.1C per bit. Default on most EL32xxx hardware.
  • High: returns results with 0.01C per bit. Default on some “high resolution” hardware. Limits range, only recommended for Pt100 sensors. Automatically changes the default value of scale.
• chXWires: Number of sensor wires. Please read Beckhoff’s documentation for your device, as no device supports all of these without additional hardware. In general, additional sensor wires reduce errors with long cable runs.
  • 2: used for 2-wire sensors.
  • 3: used for 3-wire sensors.
  • 4: used for 4-wire sensors.

All of these settings are case insensitive.

At the moment, the driver does not attempt to read current values from the hardware. This may mean that results are incorrect on some high-precision (-0010, -0020, and -0030 device variants, mostly) devices. If you have such hardware and see this problem, either file a bug (above) or explicitly set chXResolution to High and verify that results are accurate.

There are minor differences between pin names for temperature devices vs generic analog inputs. Instead of ain, the pin names are prefixed with temp, and the temperature value is in temperature instead of val. Example:

     8  bit   OUT         FALSE  lcec.0.D16.temp-0-error
     8  bit   OUT         FALSE  lcec.0.D16.temp-0-overrange
     8  s32   OUT          2260  lcec.0.D16.temp-0-raw
     8  float I/O          0.01  lcec.0.D16.temp-0-scale
     8  float OUT          22.6  lcec.0.D16.temp-0-temperature
     8  bit   OUT         FALSE  lcec.0.D16.temp-0-underrange
     8  bit   OUT         FALSE  lcec.0.D16.temp-1-error

Selecting sensor type Ohm/16 or Ohm/64, or High resolution will result in the scale pin’s default changing so that the reported results are correct for the sensor type chosen.

The author has a EL3204 in his test lab with 4 devices connected:

• Channel 0: Pt100, set to high resolution.
• Channel 1: Ohm/16, with a 1k Ohm resistor.
• Channel 2: Pt1000, set to normal resolution.
• Channel 4: Ohm/64, with a 1k Ohm resistor.

Sample values:

     8  float OUT         20.84  lcec.0.D16.temp-0-temperature
     8  float OUT       987.125  lcec.0.D16.temp-1-temperature
     8  float OUT          19.7  lcec.0.D16.temp-2-temperature
     8  float OUT      995.0469  lcec.0.D16.temp-3-temperature

Notice that ranges are automatically adjusted for High resolution (channel 0) or Ohm/ sensor types (channels 1 and 3) so that the results are correct and do not need additional scaling.

Also, notice that Ohm sensor types do not change the pin name. It’s still -temperature, even though the value is now in Ohms.

This site is open source. Improve this page.