Table of contents

RoadTest: PICOSCOPE 5444D MSO -  USB Oscilloscope

Author: avnrdf

Creation date: 16 Nov 2018

Evaluation Type: Test Equipment

Did you receive all parts the manufacturer stated would be included in the package?: True

What other parts do you consider comparable to this product?: Link Instruments MSO-9412, Hantek 6004 Series, Hantek DSO3000(A) Series, TiePie Handyscope HS6 DIFF, CleverScope CS328A-XS, Keysight U2702A

What were the biggest problems encountered?:

Detailed Review:

Thanks to PicoTech and the element14 community for letting me test the PicoScope 5444D MSO!

I've used a couple of USB scopes: $50 ADALM1000, $99 OpenScope MZ & a $279 Analog Discovery 2. The PicoScope 5444D MSO is a big step above all of those: it has a 1G/s (max) sampling rate, 200Mhz of bandwidth, 8/16bit flexible resolution, a dedicated AWG and a lot of memory (upto 500MS) - the specifications & price put it in the territory of professional benchtop DSO/MSOs.

1. Unboxing

Element14 shipped the PicoScope box inside a larger box, with shipping paper for protection.

The PicoScope box is simple, but pretty stiff. The PicoScope unit is at the top, and the probes and other accessories are beneath it.

The PicoScope is well built, and the rubber casing at the front and back of the unit grips the surface well. The scope & MSO inputs are on the front, while the AWG output, USB 3.0 and power supply connectors are on the back.

Accessories: The probes, the digital MSO cable & test clips (for the logic analyzer), locating sleeves and power supply (the data sheet says that a 5V 3A PSU

is included on 4 channel models like this one).

While the power adapter, USB cable and MSO wire harness were of good quality, the test clips could use some improvement.

As I've highlighted in the gallery, the test clips that Pico has bundled seem to be the identical to the cheap ones that are available online and cost around $0.1 in small quantities (without the soldered wire).

The build quality/finish isn't good and they feel flimsy. They do not open and retract as well as the hooks that Saleae and Hantek include with their products. While the quality of the probes isn't a deal-breaker, PicoTech could include better ones considering the cost of the unit. Element14 sells a wide range of probes, and Sigrok.org has a nice comparison of test hooks/logic probes.

The oscilloscope probes are of much better quality, and come with a couple of accessories. The hook of the retractable hook tip is thicker than other probes that I've used, which makes it slightly more difficult to use, though on the other hand it might be more durable. The retractable hook tip doesn't 'click' in when attached to the probe like some other probes I've used, so it ends up being a little loose. It stays on just fine, but doesn't feel as sturdy as Tek probes.

The ground lead is a crocodile clip with some what seems to be heat-shrink tubing to insulate it. Many probes use this (expensive ones as well), but I personally prefer the molded rubber insulator that Tektronix ships with their probes (even the low end 50Mhz $27 ones) since the tubing slips off the clip very easily, exposing the crocodile clip.

2. Product Overview & Initial Impressions:

The marketing material highlights the following features:

• FlexRes: flexible 8 to 16-bit hardware resolution
• Up to 200 MHz analog bandwidth
• 1 GS/s sampling at 8-bit resolution
• 62.5 MS/s sampling at 16-bit resolution
• Up to 512 MS capture memory
• 16 digital channels on MSO models
• 130,000 waveforms per second
• Built-in arbitrary waveform generator
• 18 serial decoding protocols as standard
• Up to 200 MHz spectrum analyzer

• FlexRes allows the user to reconfigure the scope hardware to increase either the sampling rate, or resolution. If you're working on fast digital circuits (LVDS etc),you can select 8bit resolution at a high sampling rate, but if you're doing low frequency audio/analog work, you can bump up the resolution to 12, 14, 15 or 16 bits, but at a lower sampling rate.

• DeepCapture memory allows this model to store upto 512 million samples of data. PicoTech says that "it can capture waveforms over 500 ms long with 1 ns resolution. In contrast, the same 500 ms waveform captured by an oscilloscope with a 10 megasample memory would have just 50 ns resolution" - it needs to sample fast enough to measure the difference in nanoseconds, while still having enough of sample depth to sustain this rate over a period of around a second.

• MSO models come with 16 digital channels.
• Advanced digital triggering - trigger the analog channels based on pulse width etc., and digital channels for a certain input pattern.
• Arbitary Waveform Generator - use standard signals, or generate your own wave-forms with up-to 14-bit resolution at 200MS/s
• PicoSDK - Users can write their own drivers and interface to third party software like LabVIEW & MATLAB.

The product & software guides contain detailed descriptions of the features:

PicoScope® 5000 Series FlexRes® Oscilloscopes (Product Page)

PicoScope 5000D Series Data Sheet

PicoScope 5000D Series User’s Guide

PicoScope 6 User’s Guide

PicoScope 5000 Series (A API) Programmer’s Guide

Triggering a PicoScope signal generator using the PicoScope API functions

PicoScope USB Oscilloscope Quick Start Guide

PicoScope 6 Frequently Asked Questions

PicoScope 6 Oscilloscope Software Training Manual

Beginner’s Guide to PicoScope

Competing Products:

The $3000 price tag puts the PicoScope 5444D in the territory of DSO & MSO models from manufactures like R&S, Tektronix & Keysight.

I also found a couple of other manufacturers that sell similarly priced USB MSOs:

#Hantek also has the 6004 Series & DSO3000(A) Series.

Comparing the PicoScope 5444D MSO to other USB models, the specifications vary across the chart, but the PicoScope is probably the best all-round tool. While some cost less & offer isolated inputs, the PicoScope matches the sampling rate, resolution and offers more sample memory, plus it has an AWG & digital inputs for serial decoding.

Compared to the bench-top MSOs in that range, the PicoScope has a lower sampling rate (for similar bandwidth), but offers much more sample memory and comes bundled with serial decoding & an AWG, while the MSOs might require an upgrade to activate those features.

3. Setup:

I downloaded the PicoScope 6 installer from the Picotech website. The Windows installer was 178 MB, and installation was straightforward. A USB driver is also installed.

The PicoScope doesn't have a power button, so simply plugging it in turns it on.

It works even when the DC power jack is unplugged (the unit will draw power from the USB 3.0 port), however, channels C & D get disabled. I've also used it with the USB cable plugged into a standard USB 2.0 port (rated for 500mA - not the charging type) and it worked (with channels C & D disabled) fine.

Each probe comes with a compensation tool, and the PicoScope has a probe compensation test point on the front panel. The probes I used didn't require any manual compensation.

4. Teardown:

Opening up the PicoScope was very easy. I removed the flexible rubber and flipped the unit over, which revealed 4 screws. Removing the 4 screws allowed me to lift the bottom plastic cover, revealing the bottom side of the PCB.

To access the the top (interesting) side of PCB, remove the two screws that secure the top plastic cover to the PCB. That's it - no warranty seal, glue, security screws etc.

The main ICs are a Xilinx Spartan 6 LX25 FPGA, an Analog Devices HAD1520 ADC, an Analog Devices AD9733 DAC, Cypress Semiconductor FX3 SuperSpeed USB Controller & 256MB of Micron DDR3L-1866 RAM.

5. Using the PicoScope 5444D MSO (Software overview, features & hardware testing)

I tested out the PicoScope 5444D MSO by sampling different signals in order to test out specific capabilities of the device, as listed below:

Analog (Oscilloscope):

• High frequency measurements (>100Mhz) from FPGA I/O: LVCMOS, TMDS & LVDS.
• Audio signals from a 3.5mm jack to test the 16-bit mode.
• Bridged audio amplifier output.
• Power supply ripple measurements (measure noise + frequency spectrum/FFT).
• MOSFET switching waveforms
• GPS receiver 1PPS output to test sample depth.
• Serial decoding using the analog channels.
• Waveforms generated from the PicoScope 5444D MSO and Analog Discovery 2 AWG.

Digital:

• FPGA I/O
• Protocol Decoding of SPI, UART & USB 1.1

PicoScope 6 (software):

The UI of PicoScope is simple (like other similar software): most of the area is used for plotting waveforms, and the remaining is used for toolbars to control modes & acquisition and displaying measurements. The UI design feels a little dated compared to Saleae Logic, PulseView & Waveforms (Digilent), but it works well enough.

The topmost toolbar controls the mode (Oscilloscope, Persistence or FFT), time scale, sampling rate, resolution, zoom and buffer replay. Below that, you have controls for each individual channel: the vertical range, coupling of each channel & probe settings.

The display space starts with one 'view', and users are free to add more later - the choices for types of view are 'Scope', 'Spectrum' and 'XY'.

The 'views' are usually arranged automatically, but users have the flexibility of displaying everything in a grid (upto 4x4). Views can be dragged between the cells

of the grid, and adding more than one view to a cell automatically creates tabs within that cell.

Scrolling changes the timebase of the selected view, and users can define their own keyboard shortcuts for a variety of tasks. Unfortunately, there is no quick shortcut for adjusting the vertical zoom - you need to manually click the control, which takes a couple of clicks.

Something along the lines of 'mouse-wheel scroll' adjusts the horizontal timebase, and 'Ctrl+mouse-wheel scroll' adjusts the vertical zoom would make navigation a lot easier. Dragging/panning is also possible, but on the whole navigation is slower than a bench-top scope - though PicoScope could fix this with an update.

The toolbar at the top lets you select the acquisition period, number of samples to capture (which is set at 2 giga-samples, but the software will reduce this based on the no. of active channels etc.), resolution & current waveform buffer (an acquisition is split up into multiple 'buffers').The 'Properties' tab on the right displays the actual sampling rate & interval, no. of samples & resolution.

The trigger control is at the bottom.

Math Channels:

The Math Channel Wizard lets users perform mathematical operations on the sampled data. The resultant waveform is plotted and measurements can be performed on these waveforms as well.

These are the signals used to test the math channels:

• Slide 1 plots the output of a bridged audio amplifier. The positive (A) & negative (B) channels have been plotted, and the math channel was used to plot A-B. Since the PicoScope inputs are not isolated/floating, the ground lead cannot be connected directly to any of the outputs, which are at around 7.2V DC. To measure the resultant output (since the +ve and -ve outputs of a bridged amplifier swing 180 deg out of the phase), the ground leads of the channels need to be connected to the ground point, and probes to each of the channels. The Math function A-B is used to calculate the effective output.
• Slide 2 plots the same data as slide 1, except this time the audio signal was a triangular wave. I tried using the low pass filter math function - the plot colored black is the a low-pass filtered version of A-B (f<1kHz).
• Slide 3 plots the TMDS output of a FPGA which I'm trying to use to generate LVDS in order to drive a LCD panel. Channels A & B are the positive & negative differential pairs from the FPGA (terminated to 1.5V by a 50 ohm resistor). The third trace is A-B, which is the differential voltage swing, and the fourth version is the differential voltage shifted up & biased around the common mode DC voltage ((A-B) +  average(A,B)). The math is done automatically using the math functions & some of the inbuilt measurements were applied (max, min, average etc.)

Triggers

The PicoScope supports quite a few triggering methods. While they're not something that users might use everyday, they're quite powerful when it comes to debugging troublesome issues. All of the triggering modes have been explained with the help of examples in the PicoScope 6 training manual.

Slide 1 shows the trigger configuration. The mode, type, edge, voltage, pre-trigger and delay can be configured. If the 'rapid' trigger mode is activated, the last option selects the number of rapid acquisitions.

The main trigger modes are none (free sampling), auto (wait for trigger, else capture), repeat (continuously capture on trigger) and single.

PicoScope also supports ETS (Equivalent Time Sampling), which can be used to increase the effective sampling rate above the actual hardware rate when sampling repetitive signals.

When a small timebase is used, a 'rapid capture' mode is enabled, which PicoScope claims reduces the time between waveforms to as low as 1 microsecond due to the way acquisitions and transfers are handled.

The rest of the slides show the advanced trigger modes.

Custom Probes

PicoScope 6 lets users define their own probes, which is useful when working with transducers. Compensation can be performed using a linear equation or lookup table.

Once a probe is created, it can be activated for any channel. Some probes are built into the software - for using the 1x,10x etc. settings with the bundled probes, and other probes that PicoTech sells.

PicoScope SDK

WIP as publicly available SDK does not support the 5444D MSO. I emailed PicoTech support, and they've sent me a pre-release version, which I'm currently trying out.

Will update this section soon.

PicoScope Support Toolbox for Matlab Instrument Toolbox

WIP as publicly available toolbox does not support the 5444D MSO.

Oscilloscope:

FlexRes

FlexRes allows users to select high resolution, high sampling rates or a compromise between both. It supports a peak sampling rate of 1GS/s in the single channel 8-bit mode, and a peak resolution of 16-bits in the 1-channel 62.5MS/s mode.

The next couple of slides are a comparison of the different modes (high sampling rate @ a low resolution vs lower sampling rate @ a higher resolution).

• Slides 1 & 2 plot the voltage of the +5V output rail of a SMPS in 8-bit & 16-bit mode.
• Slides 3 & 4 plot the output of a 3.5mm audio jack in 16-bit & 8-bit mode.
• Slides 5 & 6 plot MOSFET switching waveforms (microcontroller output, gate drive, and drain to source voltage) in 8-bit & 16-bit mode.

Spectrum Analyzer

The spectrum analyzer offers many options: FFT spectrum bins, maximum frequency (bandwidth), scale type, windowing functions etc. The 'measurements' available in the spectrum analyzer mode also differ from the oscilloscope: THD, SNR etc.

Since resolution of each 'bin' of a FFT depends on the sampling rate and number of samples, the main parameters that need to be considered when using the spectrum analyzer would be the bandwidth & spectrum bins (in addition to the windowing function etc.).

One feature that I liked about PicoScope 6 was that it displays (in the Properties bar on the right) the no. of bins (which the user can select), bin width (resolution of each bin) and time gate (acquisition time) - which means that for a given acquisition setting, the user does not need to manually calculate the resolution and acquisition interval.

The test signals I used were square and sine waves from the PicoScope's AWG, and the AWG of the Analog Discovery 2.

Persistence mode

Persistence mode is useful for plotting the variance (jitter) & noise of signals.

PicoScope 6 comes with many options for using persistence mode (color types, plot type and so on). The persistence mode was easy to use and worked well, though I'm not sure why the the plot is blurred (as shown in the screenshots).

Signals I used to test out persistence mode were:

• 1PPS output of a GPS receiver.
• 1PPS outputs of two GPS receivers. Since the acquisition triggers on one output, the other one exhibits jitter.
• FPGA outputs: the data & clock lines. PicoScope 6 doesn't do clock recovery, so eye pattern generation will require a clock signal as a trigger.

{gallery} Persistence Mode
1 PPS outputs of 2 GPS Receivers overlaid

Sample Depth & Deep Measure

The large sample buffer allows the PicoScope 5444D MSO to sample data at a high rate over a long period of time. It can store 500 million samples, which are distributed among the active channels, which is a big plus over similarly priced scopes that can store samples in the lower million range.

Deep Measure is feature that calculates a bunch of parameters for every cycle in the acquisition period.

• Slide 1 shows a capture of the 1 PPS output of a GPS receiver. With a sample rate of 250MS/s (which works out to a sample every 4 nanoseconds), the 500 MS buffer allows the PicoScope to capture data for 2 seconds.
• Slide 2 shows the capture of the 1PPS outputs of 2 GPS receivers: there is a roughly a 60 nanosecond offset error (which is good considering that they're independent), and the PicoScope can measure this with 8 ns resolution over a period of 2 seconds - which is required when simultaneously measuring the jitter in the rising edge (time scale of a couple of nano seconds) and frequency (measured over a second), since the sampling rate needs to be high, but the scope needs to be able to store samples over the period of a 2 seconds.
• Slides 3 & 4 show DeepMeasure in action: PicoScope measures a bunch of parameters for every cycle in the memory, and lists out the calculated values of every cycle i.e. every square wave.
• Slide 5: Mask limit testing is similar to what's included with other scopes: the PicoScope creates the mask by capturing the signal first, and it displays the trace over multiple acquisitions, plotting any deviations.

Logic Analyzer & Protocol Decoder:

The PicoScope 5444D MSO has 16 digital input channels, and an inbuilt protocol decoder than can also decode data captured on the analog channels.

Protocol decoding can be done on any mixed combination of analog and/or digital channels as shown in slide 1.

Coming to the biggest problem with the PicoScope's MSO: The triggering options for the digital channels are limited since edge triggering is either rising or falling, and only one channel can be made edge sensitive. This was disappointing because much cheaper logic analyzers do not have this restriction.

I tested the protocol decoder with USB 1.1, SPI, UART & RS-232.

The PicoScope decodes packets of the selected protocol and lists them out, which makes it easy to find the packet you're looking for using the filtering function.

The large sample memory of the PicoScope is certainly a very big plus when it comes to debugging. The PicoScope software crashed a couple of times when the sample depth was set to high values (>100MS), and CPU usage increases for a couple of seconds as the protocol decoder decodes large streams of data.

As mentioned previously, protocol decoding can be performed on analog channels as well.

Waveform Generator:

The waveform generator comes with inbuilt waveforms and a arbitrary waveform generator.

The standard inbuilt waveforms have configurable frequency, amplitude (though it's limited to 2V peak), a sweep mode (increment/decrement frequency over time) and configurable triggers.

The AWG has a couple of options when it comes to creating waveforms:

• use a binary stream, draw a waveform, use an inbuilt wave and modify it, import a waveform or capture a waveform from an oscilloscope channel.

Other Features:

Features that I did not test out in detail are the Alarms, Macro Recorder & Reference Waveforms. I'm also working on Pico SDK & the Matlab Support Package.

I will update the review when I finish testing out the features.

Conclusion:

To conclude, the PicoScope 5444D MSO is a very capable piece of hardware. FlexRes and the deep sampling depth give it an edge not only over other USB scopes, but DSOs/MSOs in the price range as well. The argument that bench-top instruments are easier to use compared to their USB counter parts is valid when it comes to general navigation (since USB scopes win at data export & logging), but it's something that PicoTech could fix with a software update.

A couple of minor points that PicoTech could improve:

• General software usability: PicoScope 6 works well, but I think that PicoTech could improve it further.
  • improve the controls for navigation, zoom etc. and add more shortcuts.
  • Increase the number of rulers available to the user, and allow measurements to be taken between any two (or more) random points of the waveform.
  • Make the sampling rate, depth, acquisition period & mode (capturing to the hardware buffer or streaming in real-time) more transparent to the user, and allow the sampling rate to the configurable & lockable.

• Digital triggering for the MSO: Level (high/low) and edge (rising & falling) cannot be used simultaneously, and only one channel can be made edge sensitive. Significantly cheaper logic analyzers & USB MSOs implement digital channel triggers better than the 5444D MSO.
• Quality of the accessories: PicoTech should upgrade the logic probes (test hooks) that they bundle with the unit as they shouldn't be shipping with a $3000 product - especially since cheaper products ship with better ones. The oscilloscope probes could be improved as well (ground lead & a sturdier hook tip), but this is a comparatively minor issue.

What the PicoScope 5444D MSO gets right:

• High sampling rate, bandwidth & resolution
• All-in-one tool: The PicoScope includes oscilloscope, spectrum analyzer, logic analyzer and waveform generator functionality.
• FlexRes: The PicoScope 5444D gives users the flexibility of selecting high resolution & sampling rate depending on what is required.
• Advanced triggers: the analog triggers on the Picoscope offer many features.
• Deep Capture memory: the 500 million sample depth beats MSOs that cost a lot more than the PicoScope.
• Software features: math channels, custom probes etc.
• Logic Analyzer & Protocol decoder: these benefit from the large sample depth, allowing long streams of data to be captured, which is helpful when debugging. The protocol decoder also lists out the decoded packets, and can be used on the oscilloscope channels.
• Portability: The PicoScope packs bench-top level performance in a very portable package, in addition to being a tool with multiple features. 2 of the 4 channels work with a standard USB port, and PicoScope documentation suggests that all 4 channels will work with a USB port that can supply more than 1.2 A - which are found on most new laptops, making it a high performance portable lab instrument.
• Good documentation & support: Certain features were confusing at first, but the user manuals, support guides etc. and the documentation in general did a good job of explaining everything. PicoTech support responded in a couple of hours when I emailed them about the SDK, and the forums seem to be pretty active as well.