|Product Performed to Expectations:||10|
|Specifications were sufficient to design with:||10|
|Demo Software was of good quality:||10|
|Product was easy to use:||10|
|Support materials were available:||10|
|The price to performance ratio was good:||10|
|TotalScore:||60 / 60|
PicoScope 2205A Review
by Gough Lui
I'd like to thank Pico Technology and element14 for choosing me to take on this particular RoadTest. I hope this review will be useful for the community, and any prospective buyers considering purchasing a USB Digital Storage Oscilloscope in the future. It's another mammoth production, so feel free to bookmark this and come back to it at a later time. If you like it - please leave me a like and rate the content! Also, feel free to visit my personal blog to find out what I'm up to.
An oscilloscope is one of those essential tools every hobbyist and engineer will need at some point in time. This might be because you are troubleshooting and servicing equipment using analog signals, such as audio, video, motor control and RF devices, or checking for the presence and amplitude digital signals on expansion boards, embedded devices, microcontrollers, etc. It's a piece of test equipment that is almost just as vital as a basic multimeter!
However, an oscilloscope was often a very expensive purchase that couldn't be justified for many hobbyists. In the days of cathode ray oscilloscopes (CRO), utilizing a cathode ray display, the units were expensive, heavy, hot and inflexible. There was no signal storage in most models, meaning only repetitive signals could be imaged, and triggering was accomplished with a basic comparator-style structure with level, edge and hold-off time adjustments. Making effective use of a CRO could be quite difficult for certain types of signals, and the cost would easily exceed AU$1000 for even basic units, out of reach for many hobbyists.
An increase in computing power, interface speed and the availability of high-rate analog-to-digital converters (ADCs) has led to the development of the digital storage oscilloscope (DSO). A DSO relies on sampling the signal with the ADC, storing it in memory, and processing it for display and storage. This allows for much greater flexibility in terms of one-shot signals (e.g. signal glitches, drop-outs) and for waveform recall, comparison, storage and export for logging purposes. It also allows the units to be small, lightweight, even portable. While initially more expensive, its advantages have led to the dominance of DSOs throughout most markets and applications.
Today, reputable standalone DSOs (replacement for CROs) and computer-hosted USB DSOs are available in price ranges from about AU$400 upward, putting them within the reach of many hobbyists. However, it is important to note exactly what you are getting for your hard earned dollars!
DSOs - Apples vs Oranges
Given that DSOs are becoming very popular, the market is literally bustling with different options at every price point. DSOs come in two major varieties - the standalone units which look like a CRO and operate independently of a computer (but may have abilities to be controlled by, or download data to a computer) and those which are reliant on the use of a computer as a display, to do the signal processing itself and store the data. This RoadTest will focus mostly on the latter.
Buying a DSO can be a difficult experience, as different manufacturers will specify only certain details about their products, or specify them under different conditions. It can be quite difficult to make a comparison.
Some of the more common specifications include:
- Number of Channels
- Voltage Ranges
- Number of bits
- Sample Rate
- Memory Depth
- Dedicated Digital Logic Capabilities
A closer look at these specifications will allow us to sort the real 'scopes from the toys! In fact, Pico Technology has an excellent write-up about this exact topic. I've taken a look at it, and I would definitely say that it's pretty unbiased as advice comes, but I'll try and explain some of it in my own words with my own experiences below.
The number of channels is important as it determines how many signals you can simultaneously measure. You might need to take into account how the grounding operates on the unit as well (often connected to USB ground, rarely isolated). For example, if you're measuring a signal (e.g. motor phase A) relative to another signal (e.g. motor common) of which cannot be grounded, you will use up two channels to measure both and perform a mathematical subtraction. Also, some synchronous serial protocols feature a clock and data line, both of which need a channel. This may be alleviated in some units if they feature an external trigger (so you can connect the clock line to the external trigger, freeing up a second channel for other uses).
The absolute voltage range will determine whether you can measure your intended signal. Many USB units only have 10-40V inputs, mainly for safety reasons as it isn't isolated from USB, which should never be exceeded. Standalone units can be found with higher CAT ratings and higher voltages, which might be important if you're working with high energy systems like mains voltage.
The number of voltage ranges is also important when you factor in the number of bits (resolution) as this determines the voltage granularity (e.g. how fine the steps in voltage are). For example, here's a table of the size of step versus number of bits and voltage range.
Depending on the level of voltage accuracy you need in your application, you may need more than the basic 8-bits of resolution in many products. This is where resolution enhancement can be used, as this technique oversamples the signal and averages the result to give you a better resolution. Unfortunately, this does not work at the higher sample rates and can end up distorting measured transitions in your signals, so "real" resolution is preferable.
Bandwidth and Sample Rate are two parameters that are often confused and sometimes appear to be used interchangeably but they aren't! Bandwidth normally refers to the frequency range of signals which can pass through the front end before a certain level of attenuation is reached and is one of the limiting factors to how you can resolve fast-changing signals. Sample Rate is the rate at which the analog to digital converters (ADCs) are converting the input signal. The Nyquist theorem states that you must digitize at least twice the rate of your highest frequency component to avoid aliasing. However, as the front end bandwidth is not exactly a brick wall filter, it's more common to sample at four to five times the rate of the highest frequency component.
For example, for a 100Mhz bandwidth, you should expect to sample at least 500Ms/s to be able to draw a good representation of the signal.
However, there are some units which claim 60Mhz and 60Ms/s. Unfortunately, this sample rate is way too low to resolve a 60Mhz signal, so what's happening? It turns out that there is a second metric to be concerned about which is equivalent time sampling (ETS). This works only with repetitive signals which are locked to a trigger, and works by "staggering" samples in order to build up the picture over several sampling passes. Thus, it may have 60Mhz bandwidth and 60Ms/s sampling rate, but it may also have a 1Gs/s ETS which would allow it to resolve signals up to 60Mhz but only if they are repetitive.
So keep this in mind - despite the word "equivalent" being thrown in there, real sample rate and equivalent time sampling rate is not equivalent. Only the real sample rate works for you when you're capturing one-off signals!
Memory depth is also another important parameter, as a USB DSO isn't usually constantly sending data from its ADC to the computer. The USB link itself isn't fast enough for that, and it would over-burden the CPU. Instead, the oscilloscope is normally "sitting" and waiting for a trigger, after which it captures a certain number of samples into its memory buffer which is then sent to the computer. This can happen periodically (when it's free running) or on every trigger (when trigger is configured). Think of it as taking snapshots of the signal periodically and sending it through. The one exception to this is for slower sampling rates where it is possible to stream the data through the USB link.
Memory depth is important as it is a reason why high rate ADCs may not be put to good use. For example, if you have a high rate ADC in your oscilloscope but only a tiny memory buffer, you can only have it sampling at a high rate only at the shortest timebases. At longer timebases, the ADC will be configured to a slower rate, so the buffer covers the whole screen, and as a result, when you zoom into the signal, you won't have the detail because it's sampling at a lower rate! Generally, more is better, as this cannot be upgraded in most DSOs.
Finally, you might be interested in any digital logic capabilities such as serial decoding, or dedicated digital inputs for mixed-signal DSOs. This may allow you to free analog channels for purely analog uses while keeping an eye on the state of serial transfers from microcontrollers to peripherals.
Unfortunately, the basic specifications for a DSO reveal only a fraction of what it would be like to actually use the DSO in practice. Computer-connected DSOs are ultimately dependent on the software to function, and the design of the software (and sometimes, the computer it's connected to) has a big impact on the ergonomics of using the DSO. Further to this, there are many un-written specifications which aren't publicised by many manufacturers which can be important to your measurement ability.
There really is no template to designing a DSO software package. Different manufacturers have different levels of functionality, different tool layouts and different limitations in inbuilt functions for recording, measurement and display. Some of these limitations can be really annoying under the right circumstances, especially if they impede your ability to effectively use the instrument.
DSOs - My Adventures to Date
I suppose the main way to solidify the fact that not all DSOs are equal, is to recount my experiences with DSOs to date.
My first unit was a PoLabs PoScope Basic 2 with a single probe and USB lead, which I purchased in 2008 at a very expensive AU$249 (it was really worth AU$70 or so). Not knowing any better, I trusted a local electronics shop's description that it offered a "real professional oscilloscope experience" on a computer. It was also quite small - about the size of a packet of cigarettes. It wasn't until I actually delved into the specifications and began using it that the flaws became apparent.
This unit featured two analog channels at a paltry 200kS/s. There was no claim of the actual bandwidth (in fact, it has nasty aliasing for signals above ~32kHz). It touts a spectrum analyzer ability as well. It claims to be 10-bits, although it wasn't mentioned that it only really has a single voltage range of +/-20V, meaning a voltage resolution of only about 39mV. It also had the capability to do 16-channel 8Mhz digital logic and serial decoding through a DB-25 connector, although I would have to build a cable for it. There wasn't much inside, except for a SiLabs C8051F32x, MOSFET relays and opamps. Its functionality was mostly software based.
Even then, it could be suitable for audio frequency signals and data logging were it not for the fact that the software was very unstable. Despite many revisions by the author, when Windows 7 was introduced, there were many issues getting it installed and working. Newer software revisions often had regressions leading to new bugs and strange behaviour (e.g. sampling stops on its own, crashes, signals turn to random noise, pieces of signals get chopped off and moved elsewhere). The data logging feature also managed to log to a custom format, for which a converter has to be used, but in later versions, the data logging feature never worked correctly at high sample rates. Long capture runs would also cause the software to crash.
Further to this, even basic elements such as triggering was very limited. It was easy for the trigger (implemented in software) to continually "retrigger" on a train of data, or miss triggering altogether for short pulses and with no waveform buffer storage it was impossible to go back and view a signal that had been "just missed".
It was pure and utter frustration, and contact with the author of the software didn't resolve many of these issues. Further to this, there was and is no support for Linux nor Windows 8.
However, the software did have some merits. Despite the "decorative" channel knobs and buttons, the main functionality was all within reach, as were the cursors. Rescaling the graph was easy, and the window could almost be of an arbitrary size. Very few necessary controls were buried in the menus.
I also did buy a SeeedStudio DSONanov2 for AU$99 which I loaded up with with BenF Firmware. This is an open-source, pocket sized unit, which has its own power source and records to microSD cards. It's specified as a single channel at 1MS/s at 12-bits with 200Khz bandwidth, which was a little better than the one above. It also features a relatively limited square wave signal generator that can do frequencies of 10Hz to 1Mhz in steps, with no control of output voltage (roughly 3v). It comes with "probes" which are merely 3.5mm mono plugs to clips/pins.
The addition of BenF firmware makes a big difference to its usability, and it now has many measurements similar to those available from benchtop DSOs.
I suppose, for the price, it's not bad if one can respect the limitations, but with the 200khz bandwidth, signals can and do alias which makes interpreting the result tricky at times.
I can safely say that both these units are merely toys when it comes to getting some serious measurements made. Low sample rates limit their practicality in some circumstances, and they can hardly be described as reliable or enjoyable to use. The problem of aliasing and limitations in voltage resolution can cause issues even for measuring something as simple as a raw 38khz IR signal.
Despite this, it's interesting to note just how many products are still on the market with such low specifications claiming to be a USB Oscilloscope. With the money I've spent above, I might have afforded something decent.
My next purchase was to be a little more informed. I chose a (decently pricey) Velleman PCSU1000, which was a two-channel, 60Mhz, 50Ms/s (1Gs/s ETS) USB oscilloscope with 8-bits resolution and 4k points buffer. This was a much bigger unit, about a dictionary in size. It featured a much more decent bandwidth and sample rate for what I needed, and the software was much more stable.
The software interface looks rather dated, but has features and buttons in all the right places. Most of the important adjustments are one-click away. Unfortunately, there wasn't much choice of window size (you can't resize it!), and many of the measurement functions were hidden behind menus. Also note the sample rate is less than the bandwidth! That's a Nyquist warning right there - you need the sample rate to be twice the bandwidth or more to avoid any possibility of aliasing. Instead, the high bandwidth rating is just stating the "quality" of the front end inputs, and instead, relying on a repetitive signal and the use of ETS to make use of that bandwidth. It's likely only good to about 10Mhz for one-shot use.
I also found the voltage calibration of the unit and noise to be quite poor. Eventually, due to some transient failure in the internal range-switching micro-relays, start-up calibration would fail outright. I contacted them and they supplied a few replacements, but not enough to fix the problem once and for all.
The software itself was also only capable of very basic edge trigger, and it was not until the later versions of the software that waveform storage was available. Data-logging was possible, but only at a rate of 100hz, which was very low.
I was happy to use it to measure waveforms - it's always been my go-to if my PoScope Basic2 can't do it. It also featured a persistence mode, which made some die-hard fans of CROs more "at home". Unfortunately, there was no Linux support for it either, and the lack of signed drivers means issues installing it under Windows 8 as well.
Looking back, I definitely fell for some of the traps when it came to buying a USB DSO, and paid too much for too little. Technology has also evolved somewhat, and much better units are available at much better prices. While a careful look at the specifications can often help avoid some of the traps, there are others related to software and functionality which are much harder to spot. How would you know just by looking at the specs that the software was unstable or the windows not-resizable?
I've been "stuck" using these units for my own work (mainly because it's expensive to get another one), and it's been a little frustration of mine. On the bright side, "at least I've got one". At the university, where I do my PhD research, we could afford to get a Rigol DS1102E, an 8-bit, two-channel, 100Mhz (1Gs/s) DSO which I have access to if I was desperate. Unfortunately, it's not on my desk at home!
About the Pico Technology PicoScope 2205A
Pico Technology is a UK based company that has been focusing on computer-based test-equipment and data acquisition since 1991. Their PicoScope 2205 was introduced in 2008, which the 2205A featuring two channels with 25Mhz bandwidth, 200MS/s sample rate, 8-bits resolution and a 2MS/s, 4k point, arbitrary waveform generator with amplitude of up to +/- 2v. It features digital triggering, integrated decoding of many serial protocols, as well as math and measurement functions too numerous to list.
In fact, it's one of the most fully specified USB DSOs I've seen to date. It is part of the PicoScope 2200 series Entry-Level USB Oscilloscopes and is listed at GBP199 on their website.
The unit connects to a host computer via USB 2.0 (backwards compatible with USB 1.1, and also usable at USB 2.0 rates on a USB 3.0 port) and is powered from the computer. It is small and light, measuring 142mm x 92mm x 19mm and under 200 grams - smaller than most paperback books. It is currently supported under Windows XP SP3, Windows Vista, Windows 7 and Windows 8, both 32 and 64-bit editions.
Unboxing the Hardware
The product comes in a sturdy outer cardboard carton.
Inside, neatly packed, is a Quick Start Guide and the Oscilloscope unit in a second cardboard surround.
Beneath the Quick Start Guide, a CD is provided with PicoScope 6.7.40, Datasheets, User Manuals and SDK 10.5.0.11. Two Pico Technology MI007 60Mhz probes are provided as well. Other probes are available at the Pico Technologies Accessories site.
There's a wall poster which illustrates the software features available, which makes a great reference if you're a new user to PicoScope.
There is also a reminder to provide feedback about your Pico product, as Pico Technology are donating money to charities on behalf of users who provide suggestions to improve the software.
Nice to see is a quality, fairly long (2-3m) USB cable is provided with the package.
The main unit is very light and compact, barely being the height of the BNC connectors! In-between the connectors for Channel A and B is the activity LED which lights up red on power, and blinks for data transfers, giving a good indication of whether the product is working. In fact it's so light, I might even have to put a book on it to stop it from moving around!
The rear of the unit has a single USB B connector which is used for data transfer and power.
The underside has a quick list of important specifications - 8 bits at 200Ms/s with 25Mhz bandwidth, 16k memory and arbitrary waveform generator.
So, I have this nice piece of new, shiny hardware ... what's the first thing I do? Take it apart of course! I don't advise anyone do this (unless they need to repair it), and if you do, take care and make sure whatever inputs are unplugged to avoid damage and potential danger to yourself.
To open the unit, you can undo the four screws under the feet and lift the top cover off.
Visible is a metallized plastic film shield which wraps around the body of the PCB.
The PCB itself has a nice matte green solder resist finish, and their logo as well. It seems to be a PS2205 Rev 01 according to the datamatrix label, but V2 (Copyright 2013) according to the text on the PCB. The PCB itself is a multi-layer type, and has a clean symmetric layout.
The side view shows the other types of relay in use.
Visible for each channel is the use of:
- Fujitsu Technology B3GA4.5Z low signal relay
- Comus 3570 ATE Grade Relays
- An adjustment trimpot
- AD8065 FastFET Amplifiers
- Two HC4052 Dual 4-channel Analog Multiplexer/Demultiplexer
- AD8132 Differential Amplifier
These feed into an AD9288 8-bit 100MSPS Dual A/D Converter (I suppose that's where the 200Ms/s comes from). Data from the ADC is fed into an Xilinx Spartan 3E XC3S250E (250k gate, speed grade 4) FPGA clocked at 100Mhz. This is passed to a Cypress EZ-USB FX2LP USB Peripheral Controller (CY7C68013A) for interfacing to the USB bus. Configuration data for the Cypress chip is held by an ST 24C02WP 2kbit serial EEPROM.
It seems that the arbitrary waveform generator uses the MAX889S Inverting Charge Pump, and may otherwise be software controlled by direct digital synthesis from the FPGA (as I can't seem to see any dedicated waveform generator ICs).
The underside doesn't really have anything of note.
While the fundamentals to building a good USB DSO seem simple at first - condition the signal, feed it into an ADC, generate clocks for the ADC, look for trigger, do some data processing (maybe even compression) and send it through the USB link - the reality is that there are subtle differences in the quality of the execution that set it apart. For example - a lack of shielding in many units, questionable front end design which lets noise in (and thus obviates the benefits of higher-resolution ADCs), or poor/limited triggering algorithms that may miss the intended signal.
Further to that, the quality of components used can also be very important - for example, the electromechanical relays in my Velleman unit have been a constant source of trouble. The ones used on the PicoScope seem to be of high quality and of reputable manufacture.
Pico Technology's saying goes "Most oscilloscopes are built down to a price. PicoScopes are built up to a specification." I think it really shows in the teardown.
Installation and Usage Possibilities
While the unit is provided with a CD, it's highly advisable that you head to Pico Technology's website to download the latest version of software.
The CD was shipped with R6.7.40, the latest version beginning this RoadTest was R6.8.8 and this was upgraded to R6.8.11 at the end. It's rather interesting to see that they have such frequent updates as a positive sign of active support. The latest version of PicoLog was R5.23.0 and of the SDK was R10.5.0.28 at the time of RoadTesting.
Installation of the software was performed with ease, simply by following the prompts.
Thereafter, connecting the unit resulted in the automatic detection and installation of drivers.
This was tested both under Windows 7 (64-bit) and Windows 8 (32-bit) with no problems as the drivers were properly signed. While it is likely to be used as a replacement for a benchtop instrument, used with a desktop computer, the bus powered nature lends itself particularly to use in field service with a laptop or even a touch-screen 8" Windows 8 Tablet.
I can definitely say it works, although start-up is a little slower, but it opens up convenient possibilities of testing with a total weight under 1kg! It can also "untie" you from the power "earth" ground if necessary by using a battery operated laptop or tablet, but be very careful about what you touch!
Picoscope - Oscilloscope
The main software package used with the unit is PicoScope. This offers features of Oscilloscope, Persistence Mode and Spectrum Analyzer mode. The main interface uses most of its area to display the traces, with toolbars across the top and the bottom.
The top toolbar houses many important functions including the view mode - oscilloscope, persistence or spectrum. There is also a configuration button which changes for each mode. The lightning bolt icon automatically determines the best parameters to capture the signal (but most advanced users would know to stay away from "automagic" buttons). The Home button restores the start-up settings.
Time divisions can be selected from the drop down with values of 50ns, 100ns, 200ns, 500ns, [1, 2, 5, 10, 20, 50, 100, 200, 500 | us,ms,s] then 1000s, 2000s, 5000s per division. Time divisions of 2ns, 5ns, 10ns and 20ns are available when ETS trigger is selected. Zoom level can be selected from x1 to 100M in steps twice as large as the previous. Samples Control can be set from 50s to 2Gs, although this may not be honoured depending on hardware limitations.
The section across from it allows you to toggle through previous captured waveforms in the buffer memory, the compass icon bringing up the waveform navigator for a small graphical preview of the waveform. There is the regular pointer, grabber, and various zoom tools available on the top bar.
The bar underneath has channel configuration options which allow you to choose the probe type, configure resolution enhancement, and control scaling and offset. Importantly, it allows you to calibrate the zero offset for that channel with the probe shorted (you should do this first before using it).
For each channel, the voltage range can be selected from the dropdown from Auto, or +/- 50mV, 100mV, 200mV, 500mV, 1V, 2V, 5V, 10V, 20V. Coupling mode can be chosen from AC or DC. Auto voltage range selection does work, although it can be a little slow to change ranges and miss signals. Very slow unstable signals may cause the auto range feature to "flip" back and forth between ranges, so manual range selection may be preferable.
The final button on the top toolbar allows you to access the Arbitrary Waveform Generator/Signal Generator options including the waveform (arbitrary, sine, square, triangle, ramp up, ramp down, sin(x)/x, Gaussian, half sine, DC voltage), frequency, amplitude (250mV, 300mV, 400mV, 500mV 600mV, 700mV, 800mV, 900mV, 1V or 2V), offset (+/- 100mV to 1V in 100mV steps, +/- 10mV to 100mV in 10mV steps, +/- 0mV to 10mV in 1mV steps) and sweep mode - type(up, down, up down, down up), stop frequency (100mHz to 1Hz in 100mHz steps, 1Hz to 10Hz in 1Hz steps, 10Hz to 100Hz in 10Hz steps, 100Hz to 1000Hz in 100Hz steps, 1khz to 10khz in 1khz steps, 10khz to 100khz in 10khz steps), frequency increment (100mHz to 1Hz in 100mHz steps, 1Hz to 10Hz in 1Hz steps, 10Hz to 100Hz in 10Hz steps, 100Hz to 999Hz in 100Hz steps) and increment time interval (1ms to 10ms in 1ms steps, 10ms to 100ms in 10ms steps, 100ms to 1000ms in 100ms steps, 1-10s in 1s steps). Arbitrary allows you to define your own signal up to 4000 points at 2Ms/s, with ability to import and export from CSV.
The bottom toolbar allows you to control whether the oscilloscope is running or stopped, the trigger type (None, Auto, Single, Repeat, ETS), advanced trigger type (simple edge, advanced edge, window, pulse-width, interval, window pulse-width, level dropout, window dropout), trigger source, trigger edge, trigger threshold, trigger positioning and trigger delay.
The Measurements button allows you to add one of a variety of measurement options including AC RMS, Cycle Time, DC Average, Duty Cycle, Falling Rate, Frequency, Fall Time, High Pulse Width, Low Pulse Width, Maximum, Minimum, Peak to Peak, Rise Time, Rising Rate and True RMS.
The rulers popout allows you to configure options for the phase ruler features, which can be used when dragging out the green handles in the bottom right hand corner of the trace area. Cursors can be used by dragging out the blue handle in the top left corner, and the white box handle in the bottom left corner.
Finally there is a button that allows you to access a notes section where you can record notes.
Useful options accessible from the menu bar include:
The file menu allows you to save the current waveform buffer data as .psdata, .pssettings, .csv, .txt, .bmp, .gif (both static and animated), .png, .mat, .jpg, .pdf. It also allows you to print your captures and configure the startup settings. The option of animated GIF is a novel and useful feature.
The edit menu allows you to Copy as Image, Copy as Text, Copy Entire Window and access the notes panel.
The views menu allows you to configure views (e.g. add more tabs),divide your area into viewports, configure the channels displayed, and axes. Some of those features will be shown in testing that follows.
The measurements menu provides the same access to the measurement dialog when accessed from the bottom toolbar, with an additional option to change the text size.
The tools menu provides access to custom probes, math channel functions, reference waveform features, serial decoding, alarms, masks, macro recorder and the software preferences. Of most interest are the Math Channel functions.
The normal functions are provided, although you can define your own similar to the probes configuration. Also, the serial decoding feature, which provides support for CAN, I2C, I2S, UART, SPI, LIN and FlexRay.
Provided with the software is a high quality help document.
It's vitally important in a good piece of test equipment that the software complements the hardware and is at least equally well made. In this case, the PicoScope package is complete and featureful and in use, proved to be stable and reliable. The examples below will show you just some of the tests I have put it through.
Testing: Arbitrary Waveform Generator and Probe Calibration
One of the most important things to do, aside from calibrating the zero, is to check the probe compensation is correctly set. Luckily, as this comes with an AWG, we can use that to generate a 1khz signal to be used in adjusting the probe's trimmer capacitor when used at the 10x position.
When properly compensated, it should not have any overshoot or undershoot on the edge.
Testing: Input Stage Frequency Response
One thing that got me curious was the fact that the sampling rate is specified at 200Ms/s when the front end bandwidth was specified at 25Mhz. I wanted to see just how well it could resolve high speed signals. It's important to have already performed probe calibration and use the 10x probe setting to minimise loading and cable effects from affecting your signal.
As I didn't have any signal generators that could go above 2Mhz, I decided to get a little creative and use my Digilent Nexys 2 (Xilinx Spartan 3E 1200k) FPGA demonstration board to generate the signals instead.
By creating a design revolving around DCMs and counters to massage the on-board 50Mhz clock, and loading it onto the board, I was able to generate a variety of high-speed "nearly square" clock waveforms which could be measured by the PicoScope. I tried to use the digital spread-spectrum options on the fourth DCM to try and add some jitter to the signal, but it seems like the Xilinx DCMs aren't really effective at spreading, or I've got something wrong with the configuration.
It can be seen that at frequencies below 25Mhz, the waveforms remain reasonably square which assures us that the PicoScope has the claimed amount of bandwidth. At 25Mhz, the waveform amplitude has slightly diminished and appears to be "sine-like", which is an indication that the higher harmonics are being attenuated somewhat, as would be expected when one reaches the bandwidth limit.
However, the sampling frequency can be determined to be as stated, as increasing the frequency to 33Mhz, 41.6Mhz, 50Mhz, 66Mhz, and 83.3Mhz all failed to show aliasing in non ETS mode, although the 83.3Mhz wave showed some unusual shapes which may suggest we are approaching the Nyquist frequency. Even the measure function was capable of determining the frequency correctly on these samples, although as expected, the amplitude is unreliable.
Testing: Hard Disk Spindle PWM Signals
Something that might exercise the two channels on the PicoScope is to measure the voltage waveform over the spindle motor of a hard disk. It seems that the particular drive I've chosen uses a three-phase AC motor, being driven by a motor controller using PWM techniques. There is three phases and a common terminal connection - measuring over Motor Phase A using Channel A and Common for Channel B allows us to derive the voltage over that particular winding using the Math Channel A-B function.
It presents an additional challenge, as the PWM waveforms are rich in transitions. Regular simple triggering would place the waveform alignment in a relatively random phase - instead, windowed triggering was use only to trigger once the signal had dwelled outside for a particular length of time, to allow it to synchronize to the rising part of the AC waveform. The triggering was also fairly easy to configure, as approximate values can be configured in the dialogue box and refined by dragging the actual trigger icon to alter the thresholds and positioning. It's also possible to turn the drive off and back on, and record the waveforms as the drive spins up, producing a nice animated GIF from the buffer recordings (see attachment due to file size limits).
It also illustrates that the PicoScope has a decent Zoom function which allows for easy navigation by dragging around the inset window to control zoom amount and location.
Testing: Serial Decoder
The serial decoder was tested with UART data from a USB to TTL serial converter at 115200bps. No problems were encountered with the data decoding, although your buffer configuration may have an impact on how useful it can be for long continuous strings of data.
PicoScope - Persistence Mode
Another main mode which is mentioned by Pico Technology is the Persistence Mode which can be activated by clicking the appropriate button on the toolbar. This offers a "time stacked" view of your signal, where subsequent acquisitions are overlaid and coloured in different ways representing the frequency of occurrence, or the cumulative signal over long periods.
It also offers adjustable decay options, background and trace colour options to meet different uses.
It is useful in judging jitter in digital signals, as a way to plot eye diagrams from signal transitions or to get an estimate of the frequency of which a transient event occurs. Unfortunately, it seems that the plot lines are a little thicker in persistence mode, due to the way the phosphor emulation works, so it can be less accurate in some cases.
Testing: Switching Power Supply Transients
DC to DC switching converters can often introduce strong transients at switching points from incorrect choice of filtering components.
By using the persistence feature, we can estimate how "probable" a particular amplitude of transient would be from the colour it is highlighted in.
Further investigation can be performed by choosing to use two viewports, allowing you to view the scope and the spectrum at the same time (although the spectrum options aren't configurable). You could alternately open both views in the same viewport, resulting in multiple tabs shown. In this example, the cursors and phase ruler are demonstrated.
Testing: On-Air FSK Radio Signals
The discriminator output from a narrowband FM receiver tuned to a noisy FSK signal was connected to the oscilloscope. The regular oscilloscope view provides a view of the noisy signal.
By using persistence mode, it is possible to image all the bit transitions and determine the baud rate of the channel. It's basically drawing an eye diagram for us.
Testing: Artificial Jitter from Arduino
Another way to illustrate the utility of it would be to generate a periodic signal with added jitter. To do this, I decided to create a sketch called superrandom which calls a function called delayMicroseconds to purposely add 3-5us of delay.
Unfortunately, in reality, the additional overhead of calling the function and random add extra variances to the signal.
If we want to know what the real jitter is, we have to measure it! Estimating it from just the plain waveform is difficult, but once we activate persistence mode, it makes it a lot easier.
As it turns out, it's really about 13.96uS.
PicoScope - Spectrum Mode
The third mode of operation is the Spectrum Mode which allows you to examine the FFT of the input signal, similar to a spectrum analyzer. This allows you to see the frequency components of your signal. There are options for number of spectrum bins (128 to 1048576 in powers of 2), window function (Blackman, Gaussian, Triangular, Hamming, Hann, Blackman-Harris, Flat-Top, Rectangular), display mode (magnitude, average, peak hold), scale (log, linear) and logarithmic unit (dBV, dBu, dBm, arbitary dB reference to a Voltage).
The spectrum display width can be selected from 25Mhz down to 190.7Hz in steps of half the previous (i.e. 25Mhz, 12.5Mhz, 6.25Mhz ... etc).
In spectrum mode, the measurements available are Frequency at Peak, Amplitude at peak, Average Amplitude at peak, Total Power, Total Harmonic Distortion (THD) %, Total Harmonic Distortion (THD) dB, Total Harmonic Distortion plus Noise (THD+N), Spurious Free Dynamic Range (SFDR), SINAD, Signal to Noise Ratio (SNR) and Intermodulation Distortion (IMD).
Testing: On-Air FSK Radio Signal
The discriminator output from a narrow band FM receiver tuned to an FSK signal was fed into the oscilloscope, and the two frequency components were identified and their shift was displayed using the cursors.
One of the carriers is at 1.188khz, the other at 2.213khz making for 1.025khz shift.
Testing: Measuring the Pass-band Response of a Sound Card
I was interested in just how flat the passband response of my Creative Soundblaster DigitalMusic LX was, so I connected the left channel to Channel A, and the right channel to Channel B. Running software to generate a slow 10 minute full frequency range sweep, while using the spectrum mode in peak hold allowed the frequency response to be determined.
Note that, at the far side of the graph, the limitation of the oscilloscope starts coming into play as it has only 8 bits resolution and will not be likely to resolve the noise floor of a sound card with 16 bits or even 24 bits. It's important to know the limitations in your equipment before jumping to conclusions, especially in regard to the SFDR and Dynamic Range where the sound card may actually be better than the oscilloscope.
Testing: Transfer Function of a 2nd Order Op-amp Filter
A classic lab exercise in analog electronics is to try and plot the transfer function of a filter. In this case, I decided to use a uA741 and a few scrap parts to build a second order filter based upon the calculations at Okawa Electric Design.
Using a similar technique to the above, I measured the response of this filter using the AWG as the signal source. The 3dB cutoff seems to be a little earlier than expected at 1923Hz, although this may be because of the tolerances in the components.
The result seems to show consistent results in the broader scale, of 40.86dB loss after cut-off at about 30khz - noting the linear frequency axis.
After that, the slope reverses and seems to wander, likely due to measurement limitations due to the 8-bits resolution of the oscilloscope, and possibly even opamp noise contribution.
Software - Settings
Now that we've seen what the software itself is capable of, I thought it would be a good idea to take a look at the software preferences, as there is a wide array of settings which can be changed to improve the performance. In some cases, the default preferences don't unlock the full potential of the device.
General Tab - Waveform Buffer Maximum Waveforms setting can be increased up to 10,000, although it works best around 2,000-4,000 in most cases. By default it's about 32, which limits your ability to go backward through multiple captures of your signal. The Measurement Statistics Statistics Captures setting controls how many samples the measurement statistics are calculated from. If you wish for a better "averaging", say over 100 samples, you'll need to increase this.
Power Management - This allows you to limit the capture rate to save on processor usage and battery usage. However, for desktop computers attached to the wall, increasing the capture rate from 30 times per second to Unlimited can improve the speed of display updating for tough signals.
Sampling - The defaults generally work well here, but it allows you to change when the display updates and interpolates.
Keyboard - This allows you to define shortcut keys for certain tasks which can make up for the fact that many of the options are hidden behind menus, although it is often more intuitive to use the mouse.
Regional and Language - This allows you to change the language, although mostly, the default options work well.
Printing - It's nice of Pico Technology to recognize that these may be used in companies where it's important to watermark and credit the company performing the work - changing the settings from the defaults will allow for your screenshots and printouts to have a different credit line.
Colours - This allows you to change the colours used and the thickness of the lines. The defaults are generally okay, although if you have a preference, it will accommodate it.
Updates - Allows you to configure the update checking behaviour and disable collection of usage statistics if you're privacy inclined.
Pico Technology also provides a software package called PicoLog Recorder which allows you to perform longer term data recording with this PicoScope unit. While I had trouble using some of the Streaming modes, as it would lock-up the software with high CPU usage after only a handful of samples, the Real-Time Continuous option appears to work just fine.
In order to test it, I chose to hook up a 2v 500mA hobby solar panel which I left on my window sill as a way of logging light intensity (arbitrarily) on a cloudy day. In order to log with PicoLog Recorder, new settings have to be first set. In this example, I've selected Real-Time Continuous and Stop after the run completes.
The next dialog allows you to configure the sampling interval (approximately - actual sampling interval will depend on the computer) and number of samples. You can select a very large number of samples and abort the run before the samples are completed.
Then you have to select the device to acquire from. As alluded to by an earlier dialog, it seems to be possible to use multiple units together to make a powerful multi-channel recorder.
The next step is to configure the measurements themselves - i.e. the channels, parameters and name.
Click Add to add a measurement.
Fill in the required details and then click OK. Once you have configured all necessary channels, you must then select File -> New Data to create a .plw file to log to. Then you can click the Record button and watch it work.
After the recording finishes or you click the stop button, you can export the data as a .csv file for plotting in an external application. Out of scale values are denoted by asterisks in the output file.
Supplied with the PicoScope is a software development kit allowing usage in an extensive set of environments including C, Delphi, Excel, LabVIEW, Matlab, Visual Basic, Agilent VEE and Linux Console. Unfortunately, it doesn't seem the PicoScope itself is USB TMC compliant, so it will not be a drop-in replacement in applications where a USB TMC compliant oscilloscope is used and the applications you build will not be portable to other USB TMC based devices.
Using the PicoScope under Linux is also a possibility although, the software is considered a beta quality software as it doesn't support all the functions in every product, but at least it is available. I think it's important that more test equipment vendors work towards targeting Linux, where it is commonly used by developers and scientific applications alike.
Unfortunately, as I had too many other commitments, it wasn't possible for me to test these features, however I do very much appreciate their existence as they are sure to come in handy in future projects.
I would say Pico Technology deserves much praise. The PicoScope 2205A has demonstrated itself to be a solid piece of hardware, which appears to be designed with care. It's also supplied with regularly updated software, which mostly performed flawlessly throughout. The thought put into the large number of measurements, trigger and configuration options available really make this a lot more versatile than the other packages that I have used in the past.
However, as I've used different software packages before, one of my main gripes would be the hiding of options behind buttons and menus, and smaller buttons for many tasks. I can understand their motivation behind this, as it allows for the main display of the signal to be larger and clearer, however as computer displays have become larger and higher in resolution over time, this seems less important. It doesn't prevent you from doing anything, but it can make it a few more clicks than necessary.
When you're really trying to grab a transient signal, you really do want to have the main options at easy one-click reach, as it minimises fumbling around. It's the same reason dedicated DSOs have control knobs for time-base and vertical-scale, and buttons for start/stop and trigger, rather than burying it behind menus. If they could offer something with looks a little more like the Velleman interface (i.e. buttons for time-bases, buttons for voltages, buttons for trigger), it would definitely pay dividends in the ease of use especially on touch screen interfaces, such as those on Windows 8 devices.
It seems that Pico Technology are quite receptive when it comes to potential improvements, so I remain very hopeful to see this implemented one day. I also think it's quite socially responsible of them to make donations to charity as part of a motivator for people to contribute their suggestions - too few companies care about things aside from profit!
As it stands, for the price, this unit will meet the majority of hobbyists needs at a price they can afford as it strikes a good compromise on bandwidth, sampling rate, channels, number of bits and memory depth. It definitely performs honestly, from what I've seen in my measurements, and can be further integrated into projects with the availability of the SDK. The lack of USB TMC compliance, however, may make this a poor choice for industrial applications where you may already have USB TMC based applications and wish for a simple drop-in replacement.
However, if you regularly deal with RF, the 25Mhz bandwidth of this unit might be a bit limiting, or you require making precision measurements on audio equipment where 8-bits resolution isn't enough to discern the true signal-to-noise ratio of the equipment under test. Likewise, if you are adamant on decoding serial protocols which use a clock and data line, or need an external trigger source plus two channels of signals, you won't have any other channels to use. Likewise, longer data-trains may benefit from a larger memory buffer.
It comes down to choosing the right tool for the job, and as a general purpose tool, I think this unit will make many hobbyists happy. It's not like many microcontrollers will be putting out signals that it can't handle. But if you do need more, the entry-level 2000 series PicoScope range is available with two-channels and bandwidths up to 200Mhz at GBP499, right through to the top end 6000 series with four-channels and bandwidth up to 500Mhz at GBP4495 - all with the similar software and support.
From my two-months of using the PicoScope, I've been quite happy with it as I've always gotten the signal I was looking for, with a minimum of fuss. At no point did I feel that I couldn't trust my equipment - and that's a great feeling to have.
I'd like to thank Pico Technology and element14 for supplying the unit for review and for supporting myself and the community. Please visit my personal blog where you can find out what I'm up to - electronics, computing and more.