Open Design Oscilloscope Could be (Almost) Free

If you could only own one piece of test equipment, it should probably be an oscilloscope. Then again, modern scopes often have multiple functions, so maybe that’s not a fair assertion. A case in point is the Scopefun open hardware project. The device is a capable 2-channel scope, a logic analyzer and also a waveform and pattern generator. The control GUI can work with Windows, Linux, or the Mac (see the video, below).

The hardware uses a Xilinx Spartan-6 FPGA. A GUI uses a Cypress’s EZ-USB FX2LP chip to send configuration data to the FPGA.  Both oscilloscope channels are protected for overvoltage up to +/- 50 V. The FPGA samples at 100 Mhz through a 10-bit dual analog-to-digital converter ( ADC ). The FPGA handles triggering and buffers the input before sending the data to the host computer via the USB chip. Each channel has a 10,000 sample buffer.

There are also two generator outputs with short circuit and overvoltage protection ( +/- 50 V ). Generator channels have 50 Ohm internal impedance and also operates via the GUI using the same USB chip. The FPGA generates signals at 50 Mhz using counters, algorithms, or simple waveform data and feeds a DAC.

A 16-bit digital interface can be set as inputs or outputs. The FPGA samples inputs at 100 MHz. The output voltage can be set, but inputs are 5 V tolerant.

According to the developer, you can build the scope from the information provided by using free sample chips from the various vendors, only paying for the small components and the cost of the PCB.

We’ve looked at several low-cost scope options before. Labtool even boasts some similar features.

52 thoughts on “Open Design Oscilloscope Could be (Almost) Free

    1. When you feed sine waves to x and y of a scope you’ll get a perfect circle if the waves are in phase and of same frequency, you’ll get a ellipsoid or even a diagonal line for getting out of phase and you’ll get a Lissajous curve for getting a difference in the frequency of the waves. Hence a circle is a very special Lissajous curve.

      1. This helps to explains why a scope, especially a dual channel one, should be considered an important purchase also by beginners not only for its value as instrument but as teaching tool as well. With a few parts it can become a curve tracer, an instrument of huge usefulness:

        1. Is anyone actually using X-Y mode for any practical purpose (outside of school and James Bond movies from 70s)? I sure looks nice but seems way too imprecise. For quick component check multimeter, voltage source and resistor are good enough, for components pairing oscilloscope would be way too imprecise.

      1. Where are links to hardware? It says it’s under CERN license. There are only code drops, but no repository? It would be nice to explore it more, but when you browse the site you find it lacks details. ( Forum is empty ).

  1. More abuse of sample programs, with no plan to create a salable product. Using prototyping samples for personal use is dodgy enough, but recommending the kit to the world as “near free coz samples” is just plain rude :/

    Sample programs kinda need to exist so engineers can prototype a project without having to go through a corporate purchasing process that could take weeks per part. Very few vendors will provide samples to someone just because they found an open source design online and want to build themselves a free device without learning or paying for anything.

    1. I agree, it is a pain for the companies and becomes a game of who is the best con. The basic design looks interesting and I wish someone that had some insider connections could put together a kit. Too bad we don’t know of any company like that.

    2. Some a-holes ruined sample programs for me and everyone else who lives in Poland 10-15 years ago. They abused sampling programs of many companies and sold their samples on local auction site. When companies learned about that, they restricted sampling programs for Poland. Few even refused to send samples here unless one asked them politely via email. And not any email – it must always be company or university email…

      1. Silicon is cheap these days, but shipping out samples costs labour.

        TI and SiLab is so paranoid about samples these days, they pissed me off enough to avoid using their products.
        Our location only puts out around 20k pcs a month, but we don’t like getting dicked around.

        The returns on a prototype design process can take years to mature. This is why digi-key can still charge 3 times the low volume cost of a sample chip.

        I like Microchip, and used 6 pieces of their silicon in the current product run.
        Because there product line is very stable, and rarely EOLs a chip without a pin compatible replacement.

        If a company gets annoyed over a tax deductible promotional program for an 80 cent die, than I just assume they won’t be around much longer. Primarily, I am not going to risk $800k of the development budget on some key piece of unobtainium, or take abuse from some imbecile in marketing (every company has them).
        Fact, a chip will not be popular if its too unreliable, hard to integrate, or simply too expensive relative to other manufacturers.

        Many students develop a taste for inexpensive samples…
        You know, like the AVR demo kits formally known as Arduino…. how many did the company move again?

        If people don’t or won’t use your chips, than you are out of business in a few years.

        1. Last time I ordered samples from TI they send it in THREE separate overnight airmail FEDEX boxes … to Poland of all places. Was something like 100Euro to ship, free of charge for me. This tells me they either dont care too much about shipping, or are dysfunctional.

          BTW The whole sample kit consisted of four chips. They send three minor ones plus a letter informing me the main controller is not available ….. so contents of all three boxes went straight to garbage bin :/

    3. Jeez. I typically don’t ask for samples even for real business projects. I would rather order them from DigiKey and not have to put up with a bunch of calls from the company asking me for technical details of the project and when I will be placing my first $100,000 order.

      One time I asked for samples for a just released chip. It showed up over a year later and the guy was pissed because I used a competitor’s part in the mean time!

      I know all too well about places with impossible purchasing procedures and unrealistic budgets, so can understand why going with samples is the path of least resistance for many.

  2. The DSO quad (from Seeedstudio) sounds very similar to this, except it’s specs are not quite up to scratch.. 10MHz dual analog input, with poor linearity. A great little (and I mean little) scope for carrying around in a toolbox. Sure, it had hardware problems (low bandwidth, poor linearity and *terrible* buttons), but for a cheap storage DSO, especially for fooling about with Arduinos and the like – it was great!

    With care (plastic case version only, with specially modified probes), I’ve even used it on the mains.

    The hardware was open (schematics available), and so was the code. A few people developed their own code, which was much better than the code that came with it (which was equally as poor as the buttons, and virtually undocumented – like most Chinese equipment).

    It had truly great potential, but lack of interest saw it die.

    What the unit shown here needs is a nice enclosure and proper useability. Hopefully then it will make a mark.

    1. The Sainsmart DDS-140 is another nice budget option, software support is.. barely adequate – but it’s seemed for a while like something the community could hack into awesomeness.

      And don’t forget the Red Pitaya, it’s on the pricey side, but was marketed as “open source platform” too – they just never actually released schematics and such…

      There’s the NI/Digilent Analog Discovery too, which is quite similar to the ScopeFun, but again quite pricey, although at least well supported.

      Lastly, for those on an even tighter budget, is a nice alternative to the usual fare of sound card / arduino scopes. It runs great on the $1.50 STM32F103C8T6 boards from china, but maxes out at 2x 461kSPS, which while sluggish, can work wonders for a hobbyist, and on the plus side means you can get away with little to no interface hardware.

      1. Tnx for tomeko. I was thinking along the lines of getting some DS0138 boards from Ali (Not from jyetech because it’s no open source) and string them together with the lcd connector (Like pc104) This would get you as many channels as you like for USD17 each. Frontend is pretty decent for this pricerange.
        STM32F103C8T6 1msps (single channel)
        AC/DC/GND switch & 10mv/Div to 5V/Div Y Axis.
        200kHz bandwith with some opamps (TL084)

      1. Exactly, the problem with all “budget DIY / USB” oscilloscopes is that either their specs are ludicrously bad or their price is high enough that you can nearly buy a legit full-blown scope with ten times their bandwidth out of it. The only such instrument I ever saw buck that trend was the OpenBench Logic Sniffer which was actually astonishingly cheap for some astonishingly usable sample rates; just make sure you don’t need trigger repeat counters – although that’s a software issue that could easily (but never will) be corrected.

      2. Dropped the LCD! Are you serious! That is what makes the Quad useful!

        The display makes it a “real” multi-channel tool (toy?) that is *fully* self contained and fits in your shirt pocket (excepting the leads of course – just like your multimeter). Apart from the DSO Nano it’s the only device I’ve ever seen that meets this need. In fact it is so small & light that moving the leads twists the dang thing so you usually can’t see it. If anything it needs to be bigger!

        I agree with the WiFi / Bluetooth comment – that should be incorporated… but while retaining the LCD!

        Rigols are certainly *much* better for the price, but would you carry one in your pocket like you would a multimeter?

        For a field tech, there is not much out there (scope wise) that you can cart around without requiring a full-on tool case.

  3. I very much agree with Viper-7. That said, if this is actually a functional design (hardware & software) it’s a fabulous product. If the original designer doesn’t want to endure the agony of turning it into a product, I sincerely hope someone will. For a lot of people, getting access to the test gear you need for electronics is financially very challenging.

    1. IMO until you’ve acquired the skills to build a scope from kit or plans, you don’t need one. So not really seeing production of this as other than a convenience. Also, quick eyeball of the BOM makes me think it would end up around $300 anyway.

      1. That’s one of its biggest issues… The BOM costs for a one-off are going to tear the notion of cheap into shreds. I’ve been down the “the board’s cheap, it’s just a few parts, let’s build a project” road a few times in my years, and it’s *never* ended up being cheap. Lessons learned, but at least I had fun!

      2. Not only that, you quite often need to beg or borrow the same equipment for calibrating the prototype. If you see something strange on your DIY scope while debugging, is it caused by your scope or your project or both?

    1. this a million times over, one shouldn’t be able to request them without at least being a company or student, some companies are already doing it, often not even answering any email that looks “personal”, if one sends from a proper domain, be it company or university, one usually has an answer within the week.

    2. I’d say the design originator was “within his rights” to do it, since he’s publicizing the design and getting them potential part sales. But to encourage other builders to do it, no.

    3. This! Many, many, many times this! Stop abusing the samples programs. It’s safe to say that I’m disappointed with HAD for publishing this article without editing (in some fashion) the encouragement to abuse the samples. This is not what I’d normally expect from an Al Williams article.

      1. This. Because of such behaviour more and more companies only send samples to students or other companies, which is sad because sometimes a sample is the only way to get a specific IC for a hobbyist.

        And btw, 100MSPS and 10.000 samples is more a toy than a scope, even with 10 bit resolution.

  4. Hmm… It looks interesting, for sure, but there are so many 2 channel USB scopes out there already. I’d love to see a 4 channel version, to be honest. Even if it has to chop and reduce bandwidth to switch into 4 channel mode, I’d be cool with just having the option!

    A dedicated 2 channel scope isn’t actually all that expensive…

    If I’m gonna go through the work of assembling my own PCB and doing something custom… Make it worth my while. Give me 4 channels, and maybe a bank of digital inputs as well, for MSO operation. Now THAT would be worth the effort of DIY, if it still can be made for less than the cost of a 2 channel scope!

      1. Openscope is not 100Mhz bandwidth, it sample at 100Mhz. It is not the same. According to Niquist theorem the bandwidth is limited to half sampling rate or 50Mhz in this case. But this is a scope so 100Mhz sampling rate is really low. Monitoring a sinoïd à 10Mhz it means only 10 samples/cycle. If one monitor a PWM signal with 10% duty cycle and 10Mhz repetition rate it will catch a single sample per pulse. Not very informative about the pulse shape.

        To put things in perspective the Rigol DS1052E 50Mhz scope sample rate is: 1 GSa/s(each channel),500 MSa/s(dual channels). 10 times Openscope.

        1. Nyquist is sampling > half, not = half (exactly half produces a constant). And it needs to be a lot greater than half unless you can see sin waves from 2.0001 points per cycle. Oversampling by 10 starts to look OK. Not needed for a lot of the math if you sample enough periods, but visually needs a lot. And analysis of a single cycle needs oversampling by as much as you can get.

          1. Jacques1956: Not really. You can claim mathematically that bandwidth is < half the sampling rate but at close to half, the sampling period must be many cycles in length and the analysis applied to the whole data set – and that is only for a sine wave at that frequency. Anything else will have higher frequency components. Usable bandwidth for capturing one shot events is more like highest frequency is 1/7 or 1/9 of Nyquist and if you want sharp corners on digital signals, more like 1/21. You need to sample at several times the frequency of the ringing for detail. 100MHz sampling scope is OK for 10MHz or less signals. If it has good band limiting filters, higher frequencies will get smoothed out instead of aliasing.

          2. Another way to look at this is to look at # of data points per period. Nyquist criteria says that you’ll need at least 2 data points to fit a waveform. That’s not much to work with and about the only thing you could do is fit a sine wave.

            Now if you want more detail for your real life waveform, you’ll need more data points.

          3. You can fit a straight line with 2 data points and around 3 to fit a curve.

            If you assume that waveform is repetitive/identical over a number of samples, you could in theory try to reconstruct a sine wave with slightly greater than 2 data points per period over a large enough sample.

  5. You can get almost same functions with Analog Discovery (AD), but better specs (i.e. 14-bit ADC). Also based on Spartan 6 FPGA. I have used AD with great success, the softare is pretty good and the hardware of very good quality (designed by Analog Devices and Digilent).

    Possible project: if you can run a soft processor on the FPGA you can make realtime DSP effects. Easier with Scopefun, but doable on Analog Discovery with some hacking. Think on a arduino-like software interface but running at 100Mhz analog+digital sampling.

  6. > FX2LP
    >samples at 100 Mhz
    >10,000 sample buffer

    so like 10 year old handek usb scope. no no noooo, USB 3.0 is what, 6 years old now? Even SDR guys figured out by now that the best way is to slurp every single sampled byte over the fastest bus they can get their hands on and compute everything on the CPU/GPU. There are SDRs today that can sample at >1GHz (over PCIe).

    1. Exactly. Pretty darn bad. Lots of n00bs saying “take my money”, but 100Msps is very bad! Even my older “backup” scope, a 60MHz Tektronix TDS-210 from the 90’s, has a 1Gsps sampling rate. This is 1/10th of that… 10k samples is very low too. I’m not asking for the popular Rigol’s 24M sample buffer but it needs to be *way* more thank 10k. FX2LP is also pretty darn old and outdated too… It was an old, early USB 1.1 chip by Anchor Chips that updated when USB 2 came out about 15 years ago… These days there’s countless other options for USB.

      So right. If anything, it’s on par with the usual cheap chinese USB scopes — just a bit above the sound card “oscilloscopes”.

  7. Might be a potential problem in this design unless you case it and run a heavy cable – like braided – from BNC grounds to Power ground at the power connector. Tek, in one of their first low cost ‘scopes used aPCB trace and they had a huge rate of blown or evaporated traces from users bumping some high current source, like mains or a battery, with the ground clip.

      1. In Australia the chassis of a PC is tied to earth (in turn tied to Neutral in the fuse box). The chassis is tied to the USB port shield, and the PSU ground (all the black wires) also tied to the chassis.
        Laptops are a different beast. Some are floating, some are tied. Again in Australia, the two pin mains connected power bricks are usually floating, whereas the three pin power bricks *usually* tie the ground of the output to the earth. Internal to the laptop, most PSUs will tie one of the external supply legs (usually negative, but not always) to the PSU ground.
        To prevent blown tracks, use a tablet/laptop running off batteries… as long as it doesn’t go flat just as that one in a billion runt pulse shows up…

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s