diff --git a/docs/device_object.rst b/docs/device_object.rst
new file mode 100644
index 00000000..65fda83b
--- /dev/null
+++ b/docs/device_object.rst
@@ -0,0 +1,173 @@
+Device Object
+===============
+
+The device object is a representation of the device under test and allows fast access to parameters and measurements.
+Conceptually it is very similar to the parameters dictionary but offers the advantages of a class.
+In analogy to the paramters dictionary, each device has terminals, which again have terminal parameters. Again, terminals represent
+functional parts of the real devices such as gates or ohmic contacts and parameters the physical quantities that can be changed or
+measured for each of those terminals, such as voltages, currents or temperatures.
+
+
+Creating a Device
+-----------------
+
+######################
+Creating manually
+######################
+
+TODO
+
+
+
+###################################
+Creating from parameters dictionary
+###################################
+
+The recommended and probably fastest way to create an instance of the QumadaDevice object class is to generate it automatically from
+an existing parameters dictionary. All dictionaries (independently if created manually or loaded from a json or yaml file) that
+are compatible with any measurement scripts are valid.
+
+.. code-block:: python
+
+	from qumada.measurement.device_object import QumadaDevice
+	parameters = {
+		"ohmic": {
+			"current": {
+				"type": "gettable",
+				"break_conditions": ["val > 100e-12"]
+			}
+		}
+		"gate 1": {
+			"voltage": {
+				"type": "dynamic",
+				"setpoints": np.linspace(0, 1, 50),
+			}
+		}
+	}
+
+
+	device = QumadaDevice.create_from_dict(parameters, station=station, make_terminals_global = True, namespace = globals())
+
+
+This code will create a device with two terminals, "ohmic" and "gate". The "ohmic" has one parameter "current" and
+the "gate" one parameter "voltage". As for the parameters defining measurements, the parameters have to match the Qumada conventions
+and be in the whitelist for valid parameter names. Additional attributes such as "type", "setpoints" or "value" have no direct influence
+when creating the object but are stored and can be accessed later e.g. for running measurements. In particular "values" are NOT directly
+set for parameters for safety reason. Creating a device has no direct influence on any instrument settings or applied values.
+
+The argument "make_terminals_global" adds all created terminals to the namespace provided in the "namespace" argument for easier access.
+In this example, all terminals are added to the global namespace, such that it is possible to for example directly access the ohmic's current
+by calling "ohmic.current()" instead of "device.ohmic.current()". Be aware, that automatically adding variables to the global namespace is not
+always a safe practice, make sure you are familiar with possible issues arising from this before you use the feature. To minimize the risks,
+QuMada will not override existing variables and will return an exception if terminals cannot be added to a namespace.
+
+.. note::
+
+	Blanks in terminal names are replaced by underscores, the variable refering to the gate is consequently "gate_1".
+	However, the name in the list of terminals (device.terminals) remains "gate 1". As for measurement scripts, the labels
+	of mapped QCoDeS parameters are relabled to the terminal name, to give e.g. more meaningful names in plots. An underscores
+	would be annoying in this case.
+
+An existing QCoDeS station (cf. QCoDeS documentation and QuMada tutorial on measurements) can be added as optional argument during the creation process
+(or later via "device.station = station") to make its components available for the mapping process.
+
+###########################
+Updating an existing device
+###########################
+
+It is easily possible to add or remove terminals and parameters to or from the device.
+Terminals can be added by calling the built-in method "add_terminal(terminal_name)" and be removed with "remove_terminal(terminal_name)".
+Parameters can be added by calling "terminal_name.add_terminal_parameter(param_name)" and be removed with
+"terminal_name.remove_terminal_parameter(param_name)".
+An oftenmore convenient method is to simply load an updated parameters dictionary.
+If you call "device.load_from_dict(param_dict)" all parameters and terminals from the dictionary that were not already in the device are
+added automatically. Note, that it is usually necessary to repeat the mapping process for the new parameters.
+
+
+
+
+Mapping
+-------------
+
+
+As for measurement scripts it is required to map the device object to the available measurement instruments. The QuMada mapping method is
+compatible with the device object.
+
+.. code-block:: python
+	map_terminals_gui(station.components, device.instrument_parameters)
+
+Opens the mapping GUI. The mapping is stored in device.instrument_parameters and it is possible to pass an existing mapping to
+map_terminals_gui to reuse an existing mapping. As for measurement scripts mappings can be saved to files and be loaded again.
+It is recommended to pass the QCoDeS station to the device (cf. "Creating from parameters dictionary"). In this case the mapping
+can simply be done via "device.mapping()", which opens up the mapping GUI without asking for additional arguments.
+
+
+Using the device
+----------------------
+
+###################################
+Parameters and simple measurements
+####################################
+
+With the mapping done it is now possible to use the device, its terminals and parameters.
+"device.terminal_name.parameter_name()" calls the get command of the mapped instrument parameter, "device.terminal_name.parameter_name(value)" sets
+it to the value. If the terminals were added to global namespace, they can be called without the device. As QuMada is tailored for experiments with
+gated quantum dots where the most accessed parameters are gate voltages, the voltage parameter can be directly accessed by just calling its terminal,
+e.g. "gate_1()" will return "gate_1.voltage()" and "gate_1(1)" will set the voltage of gate 1 to 1 V. For all other parameters (even if there is only
+one parameter for a certain terminal) it is required to access the parameters explicitely. Also, this works only for calling the terminal.
+If you try to access other attributes or methods of the voltage parameter you still have to call it explicitely. E.g. "gate_1.setpoints" will not return
+return the setpoints of the voltage!
+
+It is possible to print all voltages of the device by calling "device.voltages" for a quick overview.
+
+"gate_1.voltage.ramp(target, ramp_speed)" can be used to ramp to a certain value, "gate_1.voltage.measured_ramp(target)" will automatically
+start a new measurement (in the currently active QCoDeS database and for the currently active experiment container) ramping from the current value
+to the target. This offers a very quick and intuitive way to record measurements based on the current device working point.
+Note that there are a couple of optional arguments for the measured_ramp method to specify the starting point, the number of points, if the measurement
+should be buffered and its name. For details look into TBD.
+
+To quickly benchmark a devices stability it is possible to record a timetrace with device.timetrace(duration), 2D scans centered at the current working point
+can be recorded with device.sweep_2D(slow_param, fast_param, slow_param_range, fast_param_range). Again, both feature multiple additional arguments and can
+be buffered.
+
+In all cases mentioned so far the working point of the device is defined by getting the current values of all mapped parameters.
+Values and setpoints defined in the parameter dictionary are not used for measurements started with built-in methods to avoid confusion. However, only parameters of
+type "gettable" are recorded in those measurements. All other parameters are temporarily set to "static" except for the parameters that are to be ramped
+in the 1D or 2D sweeps, those are temporarily set to dynamic. To record a value that was not specified to be "gettable" when the device was created can simply
+be set to "gettable" by changing its type:
+"device.terminal.parameter.type = 'gettable'". If you do not want to record a parameter set it to "".
+Values from the parameters dictionary are stored in device.terminal.parameter._stored_value to distinguish them from device.terminal.parameter.value which is
+the current value of the parameter. However, in case you want to use the values and setpoints from the parameter dictionary instead of the one specified
+in the function call of measurement scripts, you can set the argument priorize_stored_values to True.
+
+Another important feature is the possibility to save and load device working points. To store a certain configuration as your default working point,
+use device.save_defaults. This stores all parameter values (of parameters that can be set). With device.set_defaults() you can reset it to the stored
+configuration. Alternatively you can use "device.save_state(name)" and "device.set_state(name)" to store and set multiple working points with
+custom names. They can also be accessed via "device.states" in case you forgot the name. Be aware that the set commands currently set the
+parameters instead of ramping to them, which can endanger your device if it is sensitive to voltage jumps.
+
+
+###############
+Safety features
+###############
+
+Maximum parameter ranges can be defined via
+
+.. code:: python
+
+	device.terminal.parameter.limits = [min_val, max_val]
+
+Those limits are then added to the validators of the underlying QCoDeS parameters.
+
+.. note::
+	Those limits are not checked in buffered measurements! Use them only as additional safety feature and do not rely on them!
+
+It is also possible to directly add a "limits" keyword to the parameter dictionary, limits are automatically applied if the dictionary
+is used to create a device object.
+With
+
+.. code-block:: python
+
+	device.terminal.parameter.locked = True
+
+you can look parameters. They cannot be changed (on the device object level) until unlocked again.
diff --git a/docs/index.rst b/docs/index.rst
index 2ef6b6e6..e168c422 100644
--- a/docs/index.rst
+++ b/docs/index.rst
@@ -9,6 +9,7 @@ Welcome to QuMADA's documentation!
    installation
    tutorials
    examples
+   device_object
    mapping_gui
    metadata
    qumada
diff --git a/src/qumada/measurement/device_object.py b/src/qumada/measurement/device_object.py
index 71c9f9e8..8c747b5c 100644
--- a/src/qumada/measurement/device_object.py
+++ b/src/qumada/measurement/device_object.py
@@ -94,8 +94,7 @@ def save_state(self, name: str):
         """
         Saves current state (inclung types, limits etc) as entry in the tuning dict with name as key.
         """
-        for terminal in self.terminals.values():
-            self.states[name] = self.save_to_dict(priorize_stored_value=False)
+        self.states[name] = self.save_to_dict(priorize_stored_value=False)
 
     def set_state(self, name: str):
         self.load_from_dict(self.states[name])
@@ -272,7 +271,8 @@ def timetrace(
         )
         mapping = self.instrument_parameters
         map_terminals_gui(station.components, script.gate_parameters, mapping)
-        map_triggers(station.components, script.properties, script.gate_parameters)
+        if buffered is True:
+            map_triggers(station.components, script.properties, script.gate_parameters)
         data = script.run()
         return data
 
@@ -290,6 +290,7 @@ def sweep_2D(
         buffered=False,
         buffer_settings: dict = {},
         priorize_stored_value=False,
+        restore_state=True,
     ):
         """ """
         if station is None:
@@ -304,13 +305,13 @@ def sweep_2D(
                         parameter.type = "static"
             slow_param.type = "dynamic"
             slow_param.setpoints = np.linspace(
-                slow_param.value - slow_param_range, slow_param.value + slow_param_range, slow_num_points
+                slow_param.value - slow_param_range / 2.0, slow_param.value + slow_param_range / 2.0, slow_num_points
             )
             slow_param.group = 1
             fast_param.type = "dynamic"
             fast_param.group = 2
             fast_param.setpoints = np.linspace(
-                fast_param.value - fast_param_range, fast_param.value + fast_param_range, fast_num_points
+                fast_param.value - fast_param_range / 2.0, fast_param.value + fast_param_range / 2.0, fast_num_points
             )
             temp_buffer_settings = deepcopy(buffer_settings)
             if buffered is True:
@@ -338,7 +339,8 @@ def sweep_2D(
             )
             mapping = self.instrument_parameters
             map_terminals_gui(station.components, script.gate_parameters, mapping)
-            map_triggers(station.components, script.properties, script.gate_parameters)
+            if buffered is True:
+                map_triggers(station.components, script.properties, script.gate_parameters)
             data = script.run()
         except Exception as e:
             print(self.states["_temp_2D"])
@@ -607,9 +609,11 @@ def measured_ramp(
         self,
         value,
         num_points=100,
+        start=None,
         station=None,
         name=None,
         metadata=None,
+        backsweep=False,
         buffered=False,
         buffer_settings={},
         priorize_stored_value=False,
@@ -626,7 +630,16 @@ def measured_ramp(
                 if param.type == "dynamic":
                     param.type = "static"
         self.type = "dynamic"
-        self.setpoints = np.linspace(self(), value, num_points)
+        if start is None:
+            start = self()
+        if backsweep is True:
+            if buffered is False:
+                self.setpoints = [*np.linspace(start, value, num_points), *np.linspace(value, start, num_points)]
+            else:
+                logger.warning("Cannot do backsweep for buffered measurements")
+                self.setpoints = np.linspace(start, value, num_points)
+        else:
+            self.setpoints = np.linspace(start, value, num_points)
         temp_buffer_settings = deepcopy(buffer_settings)
         if buffered:
             if "num_points" in temp_buffer_settings.keys():
@@ -651,7 +664,8 @@ def measured_ramp(
         )
         mapping = self._parent_device.instrument_parameters
         map_terminals_gui(station.components, script.gate_parameters, mapping)
-        map_triggers(station.components, script.properties, script.gate_parameters)
+        if buffered is True:
+            map_triggers(station.components, script.properties, script.gate_parameters)
         data = script.run()
         return data