DATA acquisition workflow - Asyncroscopy Documentation

See more at https://github.com/bluesky/tiled.

AutoScriptMicroscope saves real AutoScript acquisitions on the microscope side and returns the registered Tiled key through Tango. asyncroscopy/data/data.py is the Tango data device for registering those files with the Tiled HTTP server.

The default format is one HDF5 file per acquisition event: each correlated output is a dataset, with parsed AutoScript XML leaf metadata as HDF5 attributes. Image acquisitions can instead write .tiff (Velox-compatible, one file per detector) via scan.output_format = ".tiff"; spectra and STEM data are always HDF5.

Acquisition commands that feed this pipeline include acquire_scanned_image, acquire_spectrum, acquire_camera_image, acquire_flucam_image, and acquire_scanned_data_advanced.

What a command returns — and how you read it back — depends on the format:

.h5 (default): returns one Tiled key, read nested. client[key]["image"]["HAADF"] (one sub-dataset per detector), ["spectrum"], or ["stem_data"]; single camera / flucam frames use image.
.tiff: returns a shared stem. Rebuild one key per detector as client[f"{stem}_{DET}.tiff"]; .read() returns the array directly (no nesting).

Notebook setup¶

Connect to the DATA Tango device once at the beginning of a workflow. The save_path directory should be visible to the Tiled HTTP server, and the microscope device should have data_device_address set to this Tango device.

import json
import tango

data = tango.DeviceProxy("asyncroscopy/data/default")
data.set_timeout_millis(120_000)
data.host = "10.46.217.241"
data.port = 9091
data.save_path = "/path/served/by/tiled"

Changing data.save_path creates the directory and restarts a DATA-managed Tiled HTTP server. Each acquisition is registered explicitly after it is written; DATA does not run a filesystem watcher. startup_scripts/run_servers.py sets the extended Tango timeout automatically.

Acquire as usual. With the default .h5 the return value is the Tiled key; with .tiff it is the shared stem (see the format contract above):

key = mic.acquire_scanned_image(["HAADF", "BF-S"])   # .h5  → client[key]["image"]["HAADF"]
# scan.output_format = ".tiff"                        # .tiff → client[f"{key}_HAADF.tiff"].read()

Server Roles¶

There are two data-related servers:

asyncroscopy/data/default is the DATA Tango device server. It belongs to asyncroscopy and bridges notebooks or microscope devices to Tiled.
http://10.46.217.241:9091 is the Tiled HTTP data server. It indexes and serves files.

startup_scripts/run_servers.py starts the DATA device and its managed Tiled HTTP server together. It also shuts down the managed Tiled server with the rest of the server stack. To inspect the active directory, use:

import json

json.loads(data.get_config())["tiled_server_serving"]

If the DATA device connects to an already-running external Tiled HTTP server, it manually registers acquisitions but does not terminate that external server during shutdown.

Direct Tiled access¶

We currently access the server in the notebook like this:

from tiled.client import from_uri

client = from_uri(“http://10.46.217.241:9091”)

list(client) # should print out some folders and files