frappy/doc/source/tutorial/tutorial.rst
Markus Zolliker e411ded55b worked on tutorial
Change-Id: I630e7379e35cc0146d58f9abe10cb19283191ca8
2021-01-21 08:42:43 +01:00

257 lines
11 KiB
ReStructuredText

Frappy Programming Guide
========================
Introduction
------------
*Frappy* is a Python framework for creating Sample Environment Control Nodes (SEC Node) with
a SECoP interface. A *SEC Node* is a service, running usually a computer or microcomputer,
which accesses the hardware over the interfaces given by the manufacturer of the used
electronic devices. It provides access to the data in an abstracted form over the SECoP interface.
`SECoP <https://github.com/SampleEnvironment/SECoP/tree/master/protocol>`_ is a protocol for
communicating with Sample Environment and other mobile devices, specified by a committee of
the `ISSE <https://sampleenvironment.org>`_.
The Frappy framework deals with all the details of the SECoP protocol, so the programmer
can concentrate on the details of accessing the hardware with support for different types
of interfaces (TCP or Serial, ASCII or binary). However, the programmer should be aware of
the basic principle of the SECoP protocol: the hardware abstraction.
Hardware Abstraction
--------------------
The idea of hardware abstraction is to hide the details of hardware access from the SECoP interface.
A SECoP module is a logical component of an abstract view of the sample environment.
It is one independent value of measurement like a temperature or physical output like a current or voltage.
This corresponds roughly to an EPICS channel or a NICOS device. On the hardware side we may have devices
with several channels, like a typical temperature controller, which will be represented individual SECoP modules.
On the other hand a SECoP channel might be linked with several hardware devices, for example if you imagine
a superconducting magnet controller built of seperate electronic devices like a power supply, switch heater
and coil temperature monitor. The latter case does not mean that we have to hide complete the details in the
SECoP interface. For an expert it might be useful to give at least read access to hardware specific data
by providing them as seperate SECoP modules. But the magnet module should be usable without knowledge of
all the inner details.
A SECoP module has:
* **properties**: static information describing the module, for example a human readable *description* of
the module or information about the intended *visibiliy*.
* **parameters**: changing information about the state of a module (for example the *status* containing
information about the state of the module )or modifiable information influencing the measurement
(for example a "ramp" rate)
* **commands**: actions, for example *stop*
A SECoP module belongs to an interface class, mainly *Readable* or *Drivable*. A *Readable* has at least the
parameters *value* and *status*, a *Drivable* in addition *target*. *value* is the main value of the module
and is read only. *status* is a tuple (status code, status text), and *target* is the target value.
When the *target* parameter value of a *Drivable* changes, the status code changes normally to a busy code.
As soon as the target value is reached, the status code changes back to an idle code, if no error occurs.
**Programmers Hint:** before starting to code, choose carefully the main SECoP modules you have to provide
to the user.
Tutorial Example
----------------
For this tutorial we choose as an example a cryostat with a LakeShore 336 temperature controller, a level
meter and a motorized needle value. Let us start with the level meter, as this is the simplest module.
Coding the HeLevel Driver
-------------------------
As mentioned in the introduction, we have to code the access to the hardware (driver), and the Frappy
framework will deal with the SECoP interface. The code for the driver is located in a subdirectory
named after the facility or institute programming the driver in our case *secop_psi*.
We create a file named from the electronic device CCU4 we use here for the He level reading.
CCU4 luckily has a very simple and logical protocol:
* ``<name>=<value>\n`` sets the parameter named ``<name>`` to the value ``<value>``
* ``<name>\n`` reads the parameter named ``<name>``
* in both cases, the reply is ``<name>=<value>\n``
``secop_psi/ccu4.py``:
.. code:: python
# the most common classes can be imported from secop.core
from secop.core import Readable, Parameter, Override, FloatRange, BoolType, \
StringIO, HasIodev
# the class used for communication
class CCU4IO(StringIO):
# on connect, we send 'cid' and expect a reply starting with 'CCU4'
identification = [('cid', r'CCU4.*')]
end_of_line = '\n'
# inheriting the HasIodev mixin creates us the things needed for talking
# with a device by means of the sendRecv method
# Readable as a base class defines the value and status parameters
class HeLevel(HasIodev, Readable):
"""He Level channel of CCU4"""
# define or alter the parameters
parameters = {
# we are changing the 'unit' parameter property of the inherited 'value'
# parameter, therefore 'Override'
'value': Override(unit='%'),
}
# define the communication class to create the IO module
iodevClass = CCU4IO
def read_value(self):
# method for reading the main value
reply = self.sendRecv('h') # send 'h\n' and get the reply 'h=<value>\n'
name, txtvalue = reply.split('=')
assert name == 'h' # check that we got a reply to our command
return txtvalue # the framework will automatically convert the string to a float
This is already a very simple working He Level meter driver. For a next step, we want to improve it:
* We should inform the client about errors. That is what the *status* parameter is for.
* We want to be able to configure the He Level sensor.
* We want to be able to switch the Level Monitor to fast reading before we start to fill.
Let us start to code these additions. We do not need to declare the status parameter,
as it is inherited from *Readable*. But we declare the new parameters *empty*, *full* and *fast*,
and we have to code the communication and convert the status codes from the hardware to
the standard SECoP status codes.
.. code:: python
...
# define or alter the parameters
parameters = {
...
# the first two arguments to Parameter are 'description' and 'datatype'
# it is highly recommended to define always the physical unit
'empty': Parameter('warm length when empty', FloatRange(0, 2000),
readonly=False, unit='mm'),
'full': Parameter('warm length when full', FloatRange(0, 2000),
readonly=False, unit='mm'),
'fast': Parameter('fast reading', BoolType(),
readonly=False),
}
...
Status = Readable.Status
STATUS_MAP = {
0: (Status.IDLE, 'sensor ok'),
1: (Status.ERROR, 'sensor warm'),
2: (Status.ERROR, 'no sensor'),
3: (Status.ERROR, 'timeout'),
4: (Status.ERROR, 'not yet read'),
5: (Status.DISABLED, 'disabled'),
}
def read_status(self):
name, txtvalue = self.sendRecv('hsf').split('=')
assert name == 'hsf'
return self.STATUS_MAP(int(txtvalue))
def read_emtpy(self):
name, txtvalue = self.sendRecv('hem').split('=')
assert name == 'hem'
return txtvalue
def write_empty(self, value):
name, txtvalue = self.sendRecv('hem=%g' % value).split('=')
assert name == 'hem'
return txtvalue
...
Here we start to realize, that we will repeat similar code for other parameters, which means it might be
worth to create our own *_sendRecv* method, and then the *read_<param>* and *write_<param>* methods
will become shorter:
.. code:: python
...
def _sendRecv(self, cmd):
# method may be used for reading and writing parameters
name, txtvalue = self.sendRecv(cmd).split('=')
assert name == cmd.split('=')[0] # check that we got a reply to our command
return txtvalue # the framework will automatically convert the string to a float
def read_value(self):
return self._sendRecv('h')
...
def read_status(self):
return self.STATUS_MAP(int(self._sendRecv('hsf')))
def read_empty(self):
return self._sendRecv('hem')
def write_empty(self, value):
return self._sendRecv('hem=%g' % value)
def read_full(self):
return self._sendRecv('hfu')
def write_full(self, value):
return self._sendRecv('hfu=%g' % value)
def read_fast(self):
return self._sendRecv('hf')
def write_fast(self, value):
return self._sendRecv('hf=%s' % value)
Configuration
-------------
Before we continue coding, we may try out what we have coded and create a configuration file.
The directory tree of the Frappy framework contains the code for all drivers, but the configuration
file determines, which code will finally be loaded. We choose the name *example_cryo*
and create therefore a configuration file *example_cryo.cfg* in the *cfg* subdirectory:
``cfg/example_cryo.cfg``:
.. code:: ini
[NODE]
description = this is an example cryostat for the Frappy tutorial
id = example_cryo.sampleenvironment.org
[INTERFACE]
uri = tcp://5000
[helev]
description = He level of the cryostat He reservoir
class = secop_psi.ccu4.HeLevel
uri = linse-moxa-4.psi.ch:3001
empty = 380
full = 0
A configuration file contains several sections with a header encloded by rectangular brackets.
The *NODE* section describes the main properties of the SEC Node: a description of the node and
an id, which should be globally unique.
The *INTERFACE* section defines the address of the server, usually the only important value here
is the TCP port under which the server will be accessible. Currently only tcp is supported.
All the other sections define the SECoP modules to be used. A module section at least contains a
human readable *description*, and the Python *class* used. Other properties or parameter values may
follow, in this case the *uri* for the communication with the He level monitor and the values for
configuring the He Level sensor. We might also alter parameter properties, for example we may hide
the parameters *empty* and *full* from the client by defining:
.. code:: ini
empty.export = False
full.export = False
However, we do not do this here, as it is nice to try out chaning parameters for a test!