larpix-control 2.2.1rc2

Control the LArPix chip

larpix-control

Control the LArPix chip

Setup and installation

This code is intended to work on both Python 2.7+ and Python 3.6+.

Install larpix-control from pip with

pip install larpix-control

To return your namespace to the pre-larpix state, just run pip uninstall larpix-control. If you'd prefer to download the code yourself, you can. Just run pip install . from the root directory of the repository.

Tests

You can run tests to convince yourself that the software works as expected. After pip installing this package, you can run the tests from the repository root directory with the simple command pytest.

You can read the tests to see examples of how to call all of the common functions.

Minimal working example

So you're not a tutorials kind of person. Here's a minimal working example for you to play around with:

>>> import larpix.larpix as larpix
>>> from larpix.io.fakeio import FakeIO
>>> from larpix.logger.stdout_logger import StdoutLogger
>>> controller = larpix.Controller()
>>> controller.io = FakeIO()
>>> controller.logger = StdoutLogger(buffer_length=0)
>>> controller.logger.open()
>>> chip1 = controller.add_chip('0-1', 1)  # (access key, chipID)
>>> chip1.config.global_threshold = 25
>>> controller.write_configuration('0-1', 25) # chip key 1, register 25
[ Config write | Chip key: '0-1' | Chip: 1 | Register: 25 | Value:  16 | Parity: 1 (valid: True) ]
>>> packet = larpix.Packet(b'\x04\x14\x80\xc4\x03\xf2 ')
>>> packet_bytes = packet.bytes()
>>> pretend_input = ([packet], packet_bytes)
>>> controller.io.queue.append(pretend_input)
>>> controller.run(0.05, 'test run')
>>> print(controller.reads[0])
[ Data | Chip key: None | Chip: 1 | Channel: 5 | Timestamp: 123456 | ADC data: 120 | FIFO Half: False | FIFO Full: False | Parity: 1 (valid: True) ]

Tutorial

This tutorial runs through how to use all of the main functionality of larpix-control.

To access the package contents, use one of the two following import statements:

import larpix.larpix as larpix  # use the larpix namespace
# or ...
from larpix.larpix import *  # import all core larpix classes into the current namespace

Create a LArPix Controller

The LArPix Controller translates high-level ideas like "read configuration register 10" into communications to and from LArPix ASICs, and interprets the received data into a usable format.

Controller objects communicate with LArPix ASICs via an IO interface. Currently available IO interfaces are SerialPort, ZMQ_IO and FakeIO. We'll work with FakeIO in this tutorial, but all the code will still work with properly initialized versions of the other IO interfaces.

Set things up with

controller = larpix.Controller()
controller.io = larpix.io.fakeio.FakeIO()
controller.logger = larpix.logger.stdout_logger.StdoutLogger(buffer_length=0)
controller.logger.open()

The FakeIO object imitates a real IO interface for testing purposes. It directs its output to stdout (i.e. it prints the output), and it takes its input from a manually-updated queue. At the end of each relevant section of the tutorial will be code for adding the expected output to the queue. You'll have to refill the queue each time you run the code.

Similarly, the StdoutLogger mimics the real logger interface for testing. It prints nicely formatted records of read / write commands to stdout every buffer_length packets. The logger interface requires opening or enabling the logger before messages will be stored. Before ending the python session, every logger should be closed to flush any remaining packets stored in the buffer.

Set up LArPix Chips

Chip objects represent actual LArPix ASICs. For each ASIC you want to communicate with, create a LArPix Chip object and add it to the Controller.

chipid = 5
chip_key = '0-5'
chip5 = controller.add_chip(chip_key, chipid)
chip5 = controller.get_chip(chip_key)

The chip_key field specifies the necessary information for the controller.io object to route packets to/from the chip. The specifications for this field are implemented separately in each larpix.io class.

Adjust the configuration of the LArPix Chips

Each Chip object manages its own configuration in software. Configurations can be adjusted by name using attributes of the Chip's configuration:

chip5.config.global_threshold = 35  # entire register = 1 number
chip5.config.periodic_reset = 1  # one bit as part of a register
chip5.config.channel_mask[20] = 1  # one bit per channel

Values are validated, and invalid values will raise exceptions.

Note: Changing the configuration of a Chip object does not change the configuration on the ASIC.

Once the configuration is set, the new values must be sent to the LArPix ASICs. There is an appropriate Controller method for that:

controller.write_configuration(chip_key)  # send all registers
controller.write_configuration(chip_key, 32)  # send only register 32
controller.write_configuration(chip_key, [32, 50])  # send registers 32 and 50

Register addresses can be looked up using the configuration object:

global_threshold_reg = chip5.config.global_threshold_address

For configurations which extend over multiple registers, the relevant attribute will end in _addresses. Certain configurations share a single register, whose attribute has all of the names in it. View the documentation or source code to find the name to look up. (Or look at the LArPix data sheet.)

Reading the configuration from LArPix ASICs

The current configuration state of the LArPix ASICs can be requested by sending out "configuration read" requests using the Controller:

controller.read_configuration(chip_key)

The same variations to read only certain registers are implemented for reading as for writing.

The responses from the LArPix ASICs are stored for inspection. See the section on "Inspecting received data" for more.

FakeIO queue code:

packets = chip5.get_configuration_packets(larpix.Packet.CONFIG_READ_PACKET)
bytestream = b'bytes for the config read packets'
controller.io.queue.append((packets, bytestream))

Receiving data from LArPix ASICs

When it is first initialized, the LArPix Controller ignores and discards all data that it receives from LArPix. The Controller must be activated by calling start_listening(). All received data will then be accumulated in an implementation-dependent queue or buffer, depending on the IO interface used. To read the data from the buffer, call the controller's read() method, which returns both the raw bytestream received as well as a list of LArPix Packet objects which have been extracted from the bytestream. To stop listening for new data, call stop_listening(). Finally, to store the data in the controller object, call the store_packets method. All together:

controller.start_listening()
# Data arrives...
packets, bytestream = controller.read()
# More data arrives...
packets2, bytestream2 = controller.read()
controller.stop_listening()
message = 'First data arrived!'
message2 = 'More data arrived!'
controller.store_packets(packets, bytestream, message)
controller.store_packets(packets, bytestream2, message2)

There is a common pattern for reading data, namely to start listening, then check in periodically for new data, and then after a certain amount of time has passed, stop listening and store all the data as one collection. The method run(timelimit, message) accomplishes just this.

duration = 10  # seconds
message = '10-second data run'
controller.run(duration, message)

FakeIO queue code for the first code block:

packets = [Packet()] * 40
bytestream = b'bytes from the first set of packets'
controller.io.queue.append((packets, bytestream))
packets2 = [Packet()] * 30
bytestream2 = b'bytes from the second set of packets'
controller.io.queue.append((packets2, bytestream2))

fakeIO queue code for the second code block:

packets = [Packet()] * 5
bytestream = b'[bytes from read #%d] '
for i in range(100):
    controller.io.queue.append((packets, bytestream%i))

Inspecting received data

Once data is stored in the controller, it is available in the reads attribute as a list of all data runs. Each element of the list is a PacketCollection object, which functions like a list of Packet objects each representing one LArPix packet.

PacketCollection objects can be indexed like a list:

run1 = controller.runs[0]
first_packet = run1[0]  # Packet object
first_ten_packets = run1[0:10]  # smaller PacketCollection object

first_packet_bits = run1[0, 'bits']  # string representation of bits in packet
first_ten_packet_bits = run1[0:10, 'bits']  # list of strings

PacketCollections can be printed to display the contents of the Packets they contain. To prevent endless scrolling, only the first ten and last ten packets are displayed, and the number of omitted packets is noted. To view the omitted packets, use a slice around the area of interest.

print(run1)  # prints the contents of the packets
print(run1[10:30])  # prints 20 packets from the middle of the run

In interactive Python, returned objects are not printed, but rather their "representation" is printed (cf. the __repr__ method). The representation of PacketCollections is a listing of the number of packets, the "read id" (a.k.a. the run number), and the message associated with the PacketCollection when it was created.

Individual LArPix Packets

LArPix Packet objects represent individual LArPix UART packets. They have attributes which can be used to inspect or modify the contents of the packet.

packet = run1[0]
# all packets
packet.packet_type  # unique in that it gives the bits representation
packet.chipid  # all other properties return Python numbers
packet.chip_key # key for association to a unique chip
packet.parity_bit_value
# data packets
packet.channel_id
packet.dataword
packet.timestamp
packet.fifo_half_flag  # 1 or 0
packet.fifo_full_flag  # 1 or 0
# config packets
packet.register_address
packet.register_data
# test packets
packet.test_counter

Internally, packets are represented as an array of bits, and the different attributes use Python "properties" to seamlessly convert between the bits representation and a more intuitive integer representation. The bits representation can be inspected with the bits attribute.

Packet objects do not restrict you from adjusting an attribute for an inappropriate packet type. For example, you can create a data packet and then set packet.register_address = 5. This will adjust the packet bits corresponding to a configuration packet's "register_address" region, which is probably not what you want for your data packet.

Packets have a parity bit which enforces odd parity, i.e. the sum of all the individual bits in a packet must be an odd number. The parity bit can be accessed as above using the parity_bit_value attribute. The correct parity bit can be computed using compute_parity(), and the validity of a packet's parity can be checked using has_valid_parity(). When constructing a new packet, the correct parity bit can be assigned using assign_parity().

Individual packets can be printed to show a human-readable interpretation of the packet contents. The printed version adjusts its output based on the packet type, so a data packet will show the data word, timestamp, etc., while a configuration packet will show the register address and register data.

Like with PacketCollections, Packets also have a "representation" view based on the bytes that make up the packet. This can be useful for creating new packets since a Packet's representation is also a vaild call to the Packet constructor. So the output from an interactive session can be copied as input or into a script to create the same packet.

Miscellaneous implementation details

Endian-ness

We use the convention that the LSB is sent out first and read in first. The location of the LSB in arrays and lists changes from object to object based on the conventions of the other packages we interact with.

In particular, pyserial sends out index 0 first, so for bytes objects, index 0 will generally have the LSB. On the other hand, bitstrings treats the last index as the LSB, which is also how numbers are usually displayed on screen, e.g. 0100 in binary means 4 not 2. So for BitArray and Bits objects, the LSB will generally be last.

Note that this combination leads to the slightly awkward convention that the least significant bit of a bytestring is the last bit of the first byte. For example, if bits[15:0] of a packet are 0000 0010 0000 0001 ( = 0x0201 = 513), then the bytes will be sent out as b'\x01\x02'.

The Configuration object

The Configuration object represents all of the options in the LArPix configuration register. Each row in the configuration table in the LArPix datasheet has a corresponding attribute in the Configuration object. Per-channel attributes are stored in a list, and all other attributes are stored as a simple integer. (This includes everything from single bits to values such as "reset cycles," which spans 3 bytes.) Warning: there is currently no type checking or range checking on these values. Using values outside the expected range will lead to undefined behavior, including the possibility that Python will crash or that LArPix will be sent bad commands.

Configuration objects also have some helper methods for enabling and disabling per-channel settings (such as csa_testpulse_enable or channel_mask). The relevant methods are listed here and should be prefixed with either enable_ or disable_:

• channels enables/disables the channel_mask register
• external_trigger enables/disables the external_trigger_mask register
• testpulse enables/disables the csa_testpulse_enable register
• analog_monitor enables/disables the csa_monitor_select register

Most of these methods accept an optional list of channels to enable or disable (and with no list specified acts on all channels). The exception is enable_analog_monitor (and its disable counterpart): the enable method requires a particular channel to be specified, and the disable method does not require any argument at all. This is because at most one channel is allowed to have the analog monitor enabled.

The machinery of the Configuration object ensures that each value is converted to the appropriate set of bits when it comes time to send actual commands to the physical chip. Although this is not transparent to you as a user of this library, you might want to know that two sets of configuration options are always sent together in the same configuration packet:

• csa_gain, csa_bypass, and internal_bypass are combined into a single byte, so even though they have their own attributes, they must be written to the physical chip together

• test_mode, cross_trigger_mode, periodic_reset, and fifo_diagnostic work the same way

Similarly, all of the per-channel options (except for the pixel trim thresholds) are sent in 4 groups of 8 channels.

Configurations can be loaded by importing larpix.configs and running the load function. This function searches for a configuration with the given filename relative to the current directory before searching the "system" location (secretly it's in the larpix/configs/ folder). This is similar to #include "header.h" behavior in C.

Configurations can be saved by calling chip.config.write with the desired filename.

Once the Chip object has been configured, the configuration must be sent to the physical chip.