blob: 6915f82c0292e1f0f7961eae93db14d532f9f210 [file] [log] [blame]
== Decoders
Protocol decoders are one of the key elements of PulseView's functionality. They take
input data that you acquired and process it in a way that results in a (hopefully) much
easier to understand representation of that same data.
In its simplest form, a protocol decoder (PD) converts a group of 1-bit signals into a
stream of n-bit events. This is exactly what the parallel PD does: it takes for example
8 logic channels and treats them like an 8-bit parallel bus, emitting annotations that
show the current state of the bus at any point in time.
=== Basic Operation
Another one of the protocol decoders available to you is the I²C decoder. It takes the
two I²C signals SCL and SDA (serial clock / serial data) and shows you the details of
the I²C communication without the need to evaluate the signal bit by bit yourself.
As an example, let's have look at one of the sample .sr files we keep around for
validation of the PD code base:;a=blob_plain;f=i2c/rtc_dallas_ds1307/;hb=HEAD[].
It contains the capture of an I²C master interacting with a[Dallas DS1307 I²C Real-Time Clock]
where the master repeatedly sets and queries the time of day. After loading and using
"zoom-to-fit", it looks similar to this:
Adding the I²C decoder by clicking on and selecting I²C from the list adds
a new decode signal to the view. PulseView tries to match existing signals to the signals
that the newly added protocol decoder needs to function, which is what happened here -
SCL and SDA have been automatically assigned and the PD has decoded the communication with
the default parameters. If you need to change the signal assignment or change the decoding
parameters, you can click on to do so.
When you zoom in, you now see that PulseView has decoded the I²C messages and displays
these annotations as part of the decode signal (note that we have zoomed in so far that
PulseView shows you the individual samples):
This is already very useful, and a massive improvement over counting out pulses on an
oscilloscope screen. However, sigrok allows us to go one step further with the use of
so-called stacked decoders.
=== Decoder Stacking
To add a stacked decoder we open the settings of the decode signal, go to the _Stack
Decoder_ menu ➊, and select the DS1307 decoder:
With the stacked decoder added, we can now see that PulseView has decoded the meaning
of the I²C commands, so that we don't need to bother searching the reference manual.
In this view, we can see that the I²C packet was a command to read the date and time,
which was then reported to be "10.03.2013 23:35:30".
In this example, we added the I²C and DS1307 decoders separately. However, when opening
the decoder selector window, you can also double-click on the DS1307 decoder and PulseView
will try to auto-resolve the dependencies needed to use this decoder. In case there are
ambiguities (e.g. when several different protocol decoders offer 'uart' output), it will
ask you to choose which one to use.
For a list of available and planned protocol decoders, you can[check the wiki].
=== Using Decoders on Analog Signals
If you're capturing data using an oscilloscope or import analog signal data from a file,
you'll quickly notice that protocol decoders don't give you the option to select analog
channels as inputs. That is because as of now, decoders only work on logic signals. You
can however convert analog signals into logic signals by choosing a conversion setting
from the signal setting popup.
Here, A1 has been converted using a threshold (with default settings) and A2 has been
converted using a Schmitt-Trigger emulation (also with default settings). Additionally,
the conversion threshold display mode has been set to _Background_ in the _Views_ settings
dialog. This way, you can tell how PulseView decided to change the logic signal level
as you can now visually understand where the ranges for high and low are placed.
Aside from the default conversion threshold(s), you can choose from a few common presets
or enter custom values as well. They take the form "0.0V" and "0.0V/0.0V", respectively.
=== Per-row Settings and Actions
Sometimes, you don't want to see all protocol decoder rows or all of the annotation classes
available in a row. To do so, simply click on the arrow or label of the row you want to
From that menu, you can either show/hide the entire row or choose the annotation classes
you want to see. Everything is visible by default but if you want to focus on specific
protocol messages or status annotations like warnings or errors, this should help.
Also, if you are examining really long traces, disabling annotations for the most-often
occuring class (e.g. bit annotations for SPI) then drawing performance will increase, too.
=== Binary Decoder Output
While all protocol decoders create visible annotations, some of them also create binary
output data which isn't immediately visible at the moment. However, you can examine it
by opening the Binary Decoder Output View as shown below.
Once opened, you need to select a decoder with binary output for it to show anything -
among which are I2C, I2S, EEPROM24xx, SPI and UART. Having acquired some I2S data and
using the I2S protocol decoder lets you have the sound data as raw .wav file data, for
Using the save icon at the top then lets you save this data either as a binary file
(in this case creating a valid .wav file) or various types of hex dumps. If you want to
only save a certain part of the binary data, simply select that part before saving.
You may have noticed that the bytes are grouped by color somehow. The meaning behind
this is that every chunk of bytes emitted by the protocol decoder receives one color,
the next chunk another color and so on. As there are currently three colors, the cycle
repeats. This makes it easier to visually organize the data that you see - in the case
of the I2S decoder, the header has one color because it's sent out in one go and
following that, every sample for left/right consists of 4 bytes with the same color
since they're sent out one by one.
=== Troubleshooting
In case a protocol decoder doesn't provide the expected result, there are several things
you can check.
The first check you should perform is whether the time unit in the ruler
is given as "sa". This is short for "samples" and means that the device didn't provide
a sample rate and so PulseView has no way of showing a time scale in seconds or
fractions thereof. While some decoders can run without timing information, or only
optionally make use of it, others may not be able to interpret the input data since
timing information may be an essential part of that protocol.
Another issue to remain aware of is that decoders need enough samples per protocol step
to reliably interpret the information. In typical cases the minimum sample rate should
be 4-5 times the rate of the fastest activity in the protocol (e.g. its clock signal).
If a protocol decoder runs but shows you annotations that don't seem to make any sense,
it's worth double-checking the decoder settings. One common source of error is the
baud rate. For example, the CAN protocol decoder doesn't know what baud rate
is used on the bus that you captured, so it could be that a different baud rate is used
than the one you set. If this is still not the reason for the malfunction, it's worth
checking whether any of the signals have been captured inverted. Again using the CAN
bus as an example, the decoder will decode the signal just fine if it's inverted but
it'll show data even when the signal looks "idle".
When a protocol decoder stops execution because of an unmet constraint (required input
not connected, essential parameter not specified) or a bug in the decoder itself, you
will be presented a static red message in the protocol decoder's display area.
In that case, you can check the log output in the settings menu. There you'll find the
Python error description which you can use to either adjust the configuration,
debug the decoder (and let us know of the fix) or create a bug report so that we can
fix it.
Further helpful knowledge and explanations on logic analyzers can be found in our["Learn about logic analyzers" FAQ item].
=== Exporting Annotations
If you want to postprocess annotations that were generated by a protocol decoder, you
can do so by right-clicking into the area of the decode signal (not on the signal label
on the left). You are shown several export methods to choose from, with the last one
being only available if the cursor is enabled.
After you chose a method that suits your needs, you are prompted for a file to export
the annotations to. The contents of the file very much depend on the option you chose
but also on the annotation export format string that you can define in the _Decoders_
menu of the settings dialog. If the default output isn't useful to you, you can
customize it there.
For example, the string "%s %d: %1" will generate this type of output for the DS1307
RTC clock protocol decoder: "253-471 DS1307: Read date/time: Sunday, 10.03.2013 23:35:30"
=== Creating a Protocol Decoder
Protocol decoders are written in Python and can be created using nothing more than a
text editor. You, too, can write one!
To find out how to go about it, please see our[Protocol Decoder How-To]
and the[Protocol Decoder API Reference].
If you do write one, we'd appreciate if you'd contribute to our project so that everyone
can benefit from your work.