Analog Discovery 2 channel scope, waveform generator, logic analyzer

Posted on Wednesday, April 25th, 2012 in utilities by DP

Digilent’s Analog design kit is a 100 MSPS 2 channel scope, a 2 channel 100MSPS waveform generator, and a 16 channel logic analyzer built into one. The pricing ranges from $99 for US-students required to buy one for a class, to $249 for non-academics.

Updated specs, thanks Torwag!

Two 50 MSPS, 5 MHz differential oscilloscopes
Two 50 MSPS, 5 MHz waveform generators
Two fixed power supplies
16-channel logic analyzer
Trigger in and trigger out
USB-powered (cable included)

Thanks Dsm! Via the contact form.

This entry was posted on Wednesday, April 25th, 2012 at 1:00 pm and is filed under utilities. You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.

58 Responses to “Analog Discovery 2 channel scope, waveform generator, logic analyzer”

  1. torwag says:

    Small correction.
    Two 50 MSPS, 5 MHz differential oscilloscopes
    Two 50 MSPS, 5 MHz waveform generators
    Two fixed power supplies
    16-channel logic analyzer
    Trigger in and trigger out
    USB-powered (cable included)

    • MLarsen says:

      First of all, I have bought a system and tested it with version 2.0.19 of the software:

      The scope samples up to 100 Mhz, not 50 MHz. I think that the 5 Mhz bandwidth limitation in the specs is related to the analog circuitry roll-off, but you can make fft spectra up to 50 MHz.
      In the network analyzer they limit the bandwidth to 10 MHz, where the roll-off is ~1 dB.

      The spurious free dynamic range is close to 90 dB (with carefull selection of channel range, sampling rate and averaging), the noise floor with averaging is more than 100 dB below full range.

      In my opinion, this system is a perfect tool for designing analog amplifiers (open loop gain and phase included), audio codecs etc.

      • Robin Getz says:

        >The scope samples up to 100 Mhz, not 50 MHz.

        Yeah, the original 50MHz specs, were based on the Rev A, where we had some power issues, (which were solved in Rev B) – the final production is Rev C, where the analog noise floor issues were tweaked.

        >I think that the 5 Mhz bandwidth limitation in the specs is related to the analog circuitry roll-off

        Yes – the scope inputs are +/- .1dB out to 5MHz on the units I have looked at. I’m a little more of a purist when it comes to measurement tools; if you put in a one Vp-p, I want to measure 1Vp-p – not 0.707 (which is the -3dB point). The typical -3dB point will be higher — but for most first touch experimenters (the target for the design – engineering students taking circuits 1/2, and electronics 1/2) 5 MHz is plenty.

        >In the network analyzer they limit the bandwidth to 10 MHz, where the roll-off is ~1 dB

        Which is a function of the function generator – to produce a 10MHz sine wave which is has low distortion products in it (as close to spectrally pure as reasonable), you need a 100MSPS DAC, …

        The key difference between this and many other similar tools, is the differential inputs on the scope channels. Measure current through a high side resistors is just a matter of putting the two probes across the resistor, setting up a math channel to divide the voltage by a constant number, and then set the units to Amps.

        >In my opinion, this system is a perfect tool for designing analog amplifiers
        >(open loop gain and phase included), audio codecs etc.

        Thanks – that was the target.

        I would be interested in what you think can be done with the Logic analyzer piece.
        There is an R-R DAC experiment done with the digital outputs:
        Lab #5:

  2. dsm says:

    In their initial advertisements, Digilent states that the oscilloscope channels and waveform channels specifications are up to 100 MSPS, but as torwag points out, the Digilent website states 50 MSPS each for two oscilloscope channels and two waveform channels. 10x oversampling (Nyquist-Shannon) drops this to 5 MHz. So clearly there’s some “specsmanship” going on.

    But still, as a learning tool for students, I suspect the cost and capabilities are just about right.

    The host facing architecture is probably similar to the OpenBench Logic Sniffer (albeit with a more capable processor) with an Analog Devices part for the analog circuitry and a Xilinx FPGA to tie everything together. The device doesn’t seem to have BNC connectors for the analog functions, so perhaps it uses the same connector and probe technology as the digital logic analyzer.

    I would love to borrow one of these devices in order to write a detailed review…

    Thanks for your time.


  3. dsm says:

    The oscilloscope channels appear to use an Analog Devices AD9648 dual-channel, 14-bit, 105 MSPS analog-to-digital converter.

    The waveform channels appear to use an Analog Devices AD9717 dual-channel, 14-bit, 125 MSPS digital-to-analog converter.

    Linux® and OS X® versions “soon”.

    Thanks for your time.


  4. Brian says:

    This makes me wish I had pushed teho AWG faster, but it will still be a much more feature rich AWG (150 MSPS vs 50, BNC outputs, more power, higher voltages, better freq calibration) than this solution, and will be affordable to people beyond just academics.

    • arhi says:

      @Brian, I hope your AWG will have 10MHz reference frequency input like most counters and generators do, otherwise it won’t be that useful

      As for the lack of BNC, with 5MHz bandwidth I don’t see a need for BNC, a simple twisted wire works great at those speeds…

  5. hayahaya says:

    Hmm, I thought digilent was agilent ! Almost got fooled.

  6. MLarsen says:

    I have played around for a couple of hours with the Pattern Generator and Logic Analyzer. I am not a trained user of those tools, so I don’t know what is normally offered.
    But I can see that it is possible to use the pattern generator for generating stuff like I2S signals and use the logic analyzer together with the analog cannels to check the output from an audio sigma-delta, and that the analyzer captures the clock fine (as you would expect), without loosing packages (as you would expect).
    Generating the control signals in the pattern generator seems to be the most time consuming part, but i think that the import/export facility makes it possible to edit them in other tools as well.
    All in all, I think that this is a fantastic ‘Lab-in-a-box’ solution, both hardware and software is thought through. In my opinion, this is a great tool for electronics students and hobbyists at an extremely low cost. Final remark: if you want BNCs or other connectors, the wires are connected with a standard header, making it easy to build your own front panel or box.

  7. Very impressive unit! I wish this had been available 4 years ago while I was Engr/Phys lab instructing circuits. I am more interested in the analog side. I like that the scope inputs are fully differential and (presumably) can be used in a similar way to a floating multimeter. I assume all 4 Gnd wires are identical. Another great feature is the low ranges available on the scope inputs .. very handy. Lack of BNC connectors might be an issue or the unshielded scope leads with 1 Mohm input impedance. I’ll report more on this on my web site.
    Regarding USB, testing on some systems caused problems with marginal USB voltage (the trigger level for a warning seems to be < 4.5V) .. particularly with USB extender cables … so don't use extenders.
    Future support question: is there potential to use this with say an Android (or a new WinOS MS) tablet? instead of a laptop?

    • Robin Getz says:

      >I like that the scope inputs are fully differential and (presumably) can be used in a similar way to
      > a floating multimeter.

      Kind of, sort of — if you know what you are doing.

      A floating multimeter, well, floats. It’s powered by batteries, and the power supply floats to the negative rail (normally).

      On the Analog Discovery – Things are a bit different. The inputs are +/- 20V with respect to GND pins on the Analog Discovery. Since it’s USB powered, that’s USB GND from on the PC. Now, a little background in motherboard design — USB ground can be just about anything :( Both Desktop silver boxes (Power supply), and laptop’s grounds can be much different than earth ground.

      If you need isolation – try something like the ADuM3160 or ADuM4160 eval boards. (Down side is they are Full/Low Speed – so you drop from 480Mbps to 12Mbps…. :(

      > I assume all 4 Gnd wires are identical.

      Yes – we saw some channel to channel cross talk on early versions without these, so they are there to provide some isolation on the actual instrument, as well as decrease the impedance to ground of the DUT.

      >Lack of BNC connectors might be an issue or the unshielded scope leads with 1 Mohm
      >input impedance.

      There is a BNC adapter designed – I have one on my desk, I’ll check to see where it is in the production schedule…

      >Regarding USB, testing on some systems caused problems with marginal USB voltage
      >the trigger level for a warning seems to be is there potential to use this with say an Android (or a new WinOS MS) tablet?

      Today – WinOS tablets should be doable. Anything that runs standard windows drivers should be OK.

      Android is on the list, but I wouldn’t expect anything this calendar year – maybe next summer…


      • Robin, Thanks for the insightful answers above. Perhaps a note on the product web site discussing issues with noise and ground loop considerations, particularly when using the scope to measure external gnded and other products would be useful.

        Considering the main target audience (EE/Phys students) and considering how students can do unprecictable things in building circuits (my experiences), how robust are the scope inputs, PS outputs and WG outputs against abuse? How did the student focus groups fare in your abuse tests ?

      • Robin Getz says:

        >how robust are the scope inputs, PS outputs and WG outputs against abuse?

        Scope inputs will measure +/-20V, and will be damaged by anything over +/- 40V. If you plug it into the wall – it’s likey it will fail in a spectacular manner, including, but not limited to spontaneous rapid self disassembly, so don’t do that… :)

        PS outputs are both current limited – so I have not see any issues that hitting the software reset buttons don’t fix. Sometimes this causes a issues when the capactive load on the supply is too high – but it has to be really high…

        WG are also pretty robust – I haven’t see any issues when students have “learning moments” – by connecting the function generator outputs to the power supply outputs, or tieing them together, etc.

        >How did the student focus groups fare in your abuse tests ?

        The early Rev A’s showed a little issues – but they were resolved by Rev C, I have not heard any negative issues about the production version.


    • Hector Rios says:

      Can you just make isolation with a floating power supply in the laptop or see other problem?

  8. regarding analog BW of the scope inputs, I did a few spot measurements using a previously calibrated 10 MHz fn generator and for my Discovery unit, relative to 1 MHz:

    5 MHz: -0.05 dB
    8 MHz: -0.4 dB
    10 MHz: -0.8 dB

    • (for a sine level of ~ 1Vpeak)

      • Robin Getz says:

        >regarding analog BW of the scope inputs,

        Like I said above: “if you put in a one Vp-p, I want to measure 1Vp-p – not 0.707”

        We are in progess of doing a tear down, which describes the analog performance of the tool, and circuits (schematics and layout) used to get that performance – we just haven’t finished it yet…

        When it’s done – I will post it.


  9. Here is a simple optical impulse measurement of the Discovery:
    Optical Pulse Response.

  10. Here is a measurement of speed of light, which may be of interest to schools or colleges, with the Discovery scope :
    Impulse Response of Digilent Analog Discovery Scope

  11. Stephen Casey says:

    I have been trying out the Analog Discovery for a week or so now. It is a very impressive little box of tricks indeed. It’s really nice to be able to throw it into a laptop bag and use on the move. The maths functions and X/Y stuff, along with the Bode plotter, are really, really handy.

    The ability to drive digital lines to control a system, as well as the analogue outputs and inputs, makes debugging of mixed mode hardware a pleasure.

    Documentation on the API would be nice…

    So, how much are NI paying you to *not* release LabVIEW drivers?! ;o)

    Very well done for creating one of the best uses for my money I’ve found in a long time!

    Thanks a lot.


  12. Robin Getz says:


    Thanks – Digilent is working on making the API public.

    MathWorks have released a Matlab interface.

    So I know the API exists, and people are using it – it’s just that the documentation is a little weak now, but improving…


  13. Stephen Casey says:


    Thank you for an incredibly fast, and encouraging, reply. Do you know where the documentation may be found? Even weak would be interesting to see.

    Also, do you know if there is an automation interface to Waveforms? That would be very interesting, too.

    Thanks again for the reply, and the Analog Discovery – I really wish I’d had one of them when I was a kid!


  14. Robin Getz says:


    You would need to ask Digilent – it’s their product/API, I’m just a user (who helped out with design/beta the unit) too.

    I don’t think the API has been published publically yet, but other than me bugging them, I don’t think they have had many requests.


  15. Stephen Casey says:

    Thanks Robin,

    I shall bug them too, then!



  16. Any idea why I can’t see the “System frequency” drop down list in Discovery Device Manager?
    I asked here but no response.

  17. Robin Getz says:


    I answered back on the virtual classroom.


  18. Some of my results of Bode plotting with high-gain audio gear and also Bode to 10 MHz:
    Bode Plots with Discovery

  19. Smokey says:

    Any update on this stuff?
    I’m especially interested in the BNC board and the tear down/performance report,

  20. Stephen Casey says:


    Jack Ganssle has published a ‘mini-review’ in his Embedded Muse newsletter issue 233:

    I’m hoping that Dave Jones (the EEVBlog guy) might have one in his sights for a teardown and review in the future, too…

    I remain very happy with the Analog Discovery, and just keep buying them for people!



  21. Bruno says:

    @Robin and Stephen – any update on the API interface. Also, any insight as to what tools were used to create ‘WaveForms’.

  22. Robin Getz says:

    Waveforms is a .net application.

    No word on the public documentation — the API is all done. I will ask some folks again.


  23. since waveforms is in .net, it should be useable with Win8 pro tablets (but not the Win8 Metro tablets) ?

  24. ravi butani says:

    good news for who wish labview support for analog discovery … NI acquire digilent inc. on 27 Jan 2013… so i hope shortly we hv labview support for analog discovery….:)

  25. Smokey says:

    Bad news. Here is the response I got after asking about the API and schematics….. I’ll post this to the eevblog thread as well. sucks.

    “Unfortunately, there is no API available I believe we had planned on it at one time, but last I heard the idea has been tabled for now, with no timetable for picking it back up. I am not sure why. I do not believe we are planning on making the schematics available any time soon.
    Best Regards,
    Jordan Retz
    Digilent Inc.”

  26. I guess that means we aren’t going to get a teardown. Too bad. I was also looking forward to using the .NET api.
    I measured the output resistance of the W1 at 100kHz as ~ 0.2ohm (probably the direct output of an output opamp driver?). Output drive capability is ~ 46mA up to 1MHz but deteriorates to ~ 23 mA at 5MHz if you want a clean sine output.
    Here are a bunch of Bode plots for many popular op-amps in a standard 40dB NI circuit and the setup using the Discovery BNC adapter: Bode Plots

  27. Bruno says:

    Shame about the API, but there are several other similar devices out there from the likes of Data Translation and others that do ship with the API.

    And Ravi be careful what you wish for; NI purchased CEC who made my prefered software for Windows test and measurment applications and put it on a shelf.

  28. a basic characterization measurements of the Discovery: characteristics

  29. Smokey says:

    What devices specifically would you consider in the same category (price, features, performance) as the analog discovery?

  30. Syscomp’s Circuit Gear unit is similar but not quite as high a BW. It is aimed at the same educational/hobbyist market as Discovery. I have referred to their 10MHz WGM-201 function generator unit in some articles. There are a lot of support articles on their web site and the api is open.

  31. Some results on using Discovery to measure LED frequency response.

  32. Alexey says:

    So bad( I`ve realy been waiting for API… It was critical for me, and support were telling that it`ll be soon.

    By the way, Analog Discowery has very weak micro-usb connector. It holds only on soldered pads. And it broke down after a 2 month of carefull usage. Had to fuse together with FTDI pins.
    So be carefull with it.

  33. Discovery is the perfect compact USB scope for analog or digital audio design/testing. For example it isn’t difficult to design a free-space optical S/PDIF
    digital-audio link at audiophile-quality 96 kHz sampling rate for transmission distances of over 100′ using inexpensive red LEDs, lenses and simple electronics.

  34. Robi says:

    If I connect two analog discovery devices, can the software see and use both of them?

    • Attila says:

      Yes, you can use more devices on one PC.
      You have to connect to each device with different instances of WaveForms application.
      The trigger IO pins can be used for synchronization.

  35. Smokey says:

    Not that I should really be rewarding them with free advertizing for not releasing an API, but Newark is selling the units to everyone in the USA for $99 right now.

    • cephy says:

      This thing is selling like a hot cake, 26 in stock down to 1 in less than 24 hours. Somebody told me that the SDK is in beta testing, so I guess they would eventually release it. I can use this device with the latest version of matlab + data acquisition toolbox.

  36. Pierre says:

    I have to drive 3 analog discovery devices for my experiment, and I can trigger the 6 analog signals, but they are not completely synchronised, and because there are 3 instances of waveforms opened, each frequency slighlty shift (0.1Hz from one instance to another), so my trigger means nothing, because 10s after, the phases are no more respected.
    Is there a way to trigger during all the time?


    • attila says:

      The problem you are facing is due to the accumulation of the few ppm oscillator frequency errors.
      In order to synchronize the devices, you can generate digital pulse with pattern generator. Connect this digital IO pin to Trigger 1 pin of each device. Configure each channel of AWG: to Repeat Trigger: Yes (in drop-down next to the Run button); Trigger: External 1, Wait: none, Run: a bit less than the synchronization period; Repeat: infinite
      For instance: generate 100kHz digital clock as synchronization signal, 100Khz analog waveform, run time 9.99us. This will synchronize the devices after each period.

  37. Pierre says:

    One last question, using matlab, the only available method is the startForeground method, which generates an input array. But I quickly reach my memory limit considering the fact that I want an output during a long time. Is there a way to specify a known waveform (sine for exemple), and to give it only frequency, amplitude, and then it runs?
    Moreover, using this method, my rate is limited to 1e6 Hz, thus my waveforms are terrible at high frequencies, whereas using Waveforms, they are clean up to 5MHz.


    • attila says:

      The current MatLab support “streams” samples to/from device and this data transfer limits the maximum frequency. Repetitive, device buffered generation of a waveform and function generator feature is not (yet) supported.

      • toph says:

        Have you been able to find or come up with any way around this limitation? It would be a big help. Thanks!

  38. Pierre says:

    no, I gave up the use of Matlab and Analog Discovery because of the rate limitation.


    But I use them with Waveforms and it works really well.

  39. Robin Getz says:

    For anyone still wondering – it took longer than expected – but it’s now out:

    This document is a reference for the Analog Discovery’s electrical functions and operations. This reference also provides a description of the hardware’s features and limitations.

  40. Andi Hofman says:

    Hi !
    The headline of the article says Version 2, but described above is version 1 !

    So you do “analog discovery 2” no good ;)

    Specs of V2 here
    (note: the picture is wrong. Still got the old black case in. Should be green. But the text matches to Version 2 all right…) :

    For the stylish green case of Version 2 see digilents web shop ;) :

    Version 2 has 100MS/s and 30MHz BW (with BNC adaptor card) and other nice features. Labview etc..

    Only Box I know having 14Bit AD (!) and DA plus 16 digital ports (in: LA, or out: AWG) plus stackability via T1 T2 Triggering ports, several devices that get displayed in waveforms software, to one instance of the software, see end of pecs pdf.

    (If you compare to the lates picoscope 3000D MSO line, they claim 1GS/s, but share it between channels, so 2 analog and 2 digital ports in use leaves you 1GS/s / 4 = 250MS/s, not so far away considering you pay 5 to 10 times the price. Has anyone compared these against each other?)

    All I’m missing is USB isolation, but USB2 480MBit/s USB isolation is not viable for cheap for such a long time. Precisely, we’d have to wait for some device that incorporates the new chips from Silanna:

    (The chip alone will not transfer 5V / 500 or 900mA power. And for my USB port I want at least all the power USB3 warrants, 900mA. To not let it be eaten by the DC/DC module one has to strive for the highest efficiency possible right now. Seen circuitry >90% efficiency.

    My notebook is capable of charging (i-) devices with 2.1A, so not problem there.
    All that is needed is some chip demanding full USB3 power and an isolated DC/DC 5V / 5V Module with around 1A. Preferably I would like to see this in one affordable device.)

    Have fun with this poor man’s precision oscilloscope ;)


  41. Andi Hofman says:

    Hi !
    The links in the first post went wrong, forgot one quote, sorry.

    Specs of “Digilent Analog Discovery 2” here:

    shop with product description “Digilent Analog Discovery 2” :


Leave a Reply

Notify me of followup comments via e-mail. You can also subscribe without commenting.

Recent Comments

  • Joe Desbonnet: Ya, I can recommend the low melting point solder. I used brand 'ChipQuik' and it's amazingly easy to use.
  • Jerome: I need a new BusPirate for the Fablab ;) Many thanks!
  • Max: Seems like an unexpectedly violent way to remove the chip indeed. A hot air station should of course do the job just fine, but in...
  • jose: Part removal described here is pure butchery, the cheapest hot air station will do a fast and clean job removing the QFP, heat air to...
  • Cody: Yes please