Advertisement

Blog

Put an Oscilloscope on a Chip: Why Not?

After seeing Dennis Feucht's excellent series on the design of a Z meter on a chip (see a list of his blogs here), I decided another piece of test equipment we need is an oscilloscope on a chip. I won't go into an analysis as detailed as the one that Feucht produced. Instead, allow me to provide an overview in order to evaluate the practicality of such test equipment on an IC by looking at the subsystems required.

(Source: Wikipedia)

(Source: Wikipedia)

To build a 'scope, you will need one or multiple high-speed ADCs, one per vertical input channel — so probably two or four. In front of the ADC, you need one or more gain stages and switchable filter networks to set the channel bandwidth. So far, this can be done at the chip level. You will also need a switchable attenuator, but at least part of that will likely be impractical to integrate because it will be subjected to high voltages. I don't like to mix low-voltage, low leakage, high speed device processes with high-voltage processes. But at some point, this may become practical.

You will need an accurate time base to mimic the function of the horizontal sweep — except that there really isn't a horizontal sweep like in a CRT (cathode ray tube) based 'scope. All you need to have is large amounts of digital memory and processing to take the values from the ADC and light up a pixel at the right place on the LCD or flat panel display. So with enough square millimeters of silicon, this is practical.

(Source: Wikipedia)

(Source: Wikipedia)

On the plus side, since the display is not a CRT, you don't need amplifier circuitry operating at a few hundred volts for the vertical and horizontal deflection plates, or an anode supply operating at a few thousand volts.

The trigger function should be straightforward. Look for a certain voltage (trigger-point). Then, perhaps, look for a certain rate of change of that voltage. Easy.

You will need the capability to accept input from the user, which means a bunch of I/O for the usual soft keys. So that means an MCU, an FPGA, and lots of port bits.

All of the above represents the important stuff, but it all seems practical. You could add additional functionality on the chip as desired. This could include power supply supervisory circuitry and touch screen interface.

Have you worked on 'scope designs before? What have I overlooked in this review?

Related posts:

61 comments on “Put an Oscilloscope on a Chip: Why Not?

  1. jkvasan
    October 15, 2013

    Brad,

    I would like the chip to have a composite video output. It could just drive a normal composite video monitor or a tv. This means the large amounts of memory to remember the pixels may not be necessary. Also, pin-count could be less as LCDs require a number of pins to drive them directly.

  2. Brad_Albing
    October 15, 2013

    @JK – OK, good point on the LCD. Might be best to use a completely separate IC to drive the LCD. Since driving a seperate monitor with composite video is an option that not everyone needs, it might be best to put that on a separate IC also.

  3. Scott Elder
    October 15, 2013

    Brad, While most EEs use an oscilloscope, in the context of your blog, it seems that “oscilloscope” is becoming an old word.  Perhaps all that is required is a simple digitizer IC that wirelessly streams the data to a generic standalone computing device with a display–here I'm thinking smartphone or tablet.  Then the world of app writers could be turned loose to build all types of GUI interfaces.

    In that situation, the chip is reduced to essentially an amplifier interfaced to an ADC interfaced to a wifi TX/RX.  That's sub $10 per channel.  Still would need the HV dividers, though.

    It seems that most of what is in an oscilloscope today is redundant.  Does one really need a separate display, computer, and memory for each instrument on a bench?

    Go here:  http://www.kickstarter.com  and search “oscilloscope”.  Four miniature 'scopes.  No single chip, but getting close.

  4. eafpres
    October 15, 2013

    Over on a sister site (All LED Lighting) I reviewed a spectrophtometer that interfaced to an iPhone (or iPod).  The module had the sensor and some supporting electronics, and all the processing, storage, display etc. was done on the phone.  You can see that here.

    Seems to me that starting with Brad's chip and putting enough in to get it into the phone and you could have a decent scope.  I saw a program on TV where they built an EKG module that snapped onto an iPhone and displayed the electrical signal of the heartbeat.  A scope is a little more complex, but if they can do that for the small signal of your body's electrical signals, then with some signal processing you could sample most things of interest.

  5. antedeluvian
    October 15, 2013

    Brad

    The PSoC comes close, although the frequency response is way down. It can handle a sample rate of 1MHz. I built my design around a PSoC5 and aside from the input conditioning everything was realized on the micro. I actually had a blog on MCC about it, but all that is left is a video on You-tube. I also submitted it to the Create the Future 2013 contest, where it was awarded an honourable mention in the electronics category and included in the top 100 designs.

    The PSoC allowed me to uise two ADCs simulateoulsy along with DMA, adjustable sample rate, built-in 11 channel multiplexer routable to either ADC all uder software control, gain control and with the logic create the trigger internally. Best part was when I found a problem with my cicuitry I simply reconfigured it intenally. Also allowed me a 6 bit logic analyser, thermocouple input, digital outputs, a 0-10V DAC and a PWM output, plus a few other things.

    Since the product is intended for a sealed cabinet all of these features are great, especially when they are all included on the micro.

  6. Scott Elder
    October 15, 2013

    Blaine – An extension of your idea would be for Apple/Samsung/others to start building differentiated smartphones/tablets.  One version could be for EE types with electronic I/O jacks like voltmeters, scopes, sig gens, etc.  Another for medical types with blood/temperature/ekg etc.

  7. eafpres
    October 15, 2013

    @Scott–that sounds good. The calculator industry did that well until computers obviated the need for most calculators.

  8. edartuz
    October 15, 2013

    STM32F303 contains 4x5msps 12-bit ADC, which can work interleaved for 20 msps or more (with decreased accuracy 8..10bits). Plus integrated opamps with PGA and comparator.

    Using integrated DAC and opamps also possible to create Z-meter on a chip.

    STM32F4xx contains DCMI camera module, which can continuously sample external ADC at 60..100msps.

  9. Davidled
    October 15, 2013

    There is no more tool box for technician who get app of smartphone with small extension cord. I saw one technician using app tool to capture frequency spectrum. For EE, Smartphone H/W configuration would be changed to meet a variety tool. App for stethoscope would be in the future.

  10. D Feucht
    October 15, 2013

    If a low-cost DSO is opened, such as a Tek TDS220, there are a few large ICs in it plus little else. The attenuators are still discrete but H-P (Agilent) has integrated some of that functional block. I didn't choose 'scopes because they have been far along the integration path for some time. I don't expect a scope on a single IC anytime soon because of the mix of high-speed analog and digital. Mixing wideband and sensitive analog circuitry with high-speed digital poses a crosstalk conundrum.

    In contrast, the Z meter ought to be done. I remember back at Tek when some older engineer in the '70s projected that microprocessors would be $2 each. This was when 6800s were $250 each. He predicted the low price based on the number of transistors on the chip. it was no more than high-volume TI calculator chips which were selling for a few $.

    Then Chuck Peddle from MOS Technology came around, offering the 6502 for $20 each. I jumped on it, as did Steve Wozniak. Apple succeeded; I didn't.

    DMMs sell for less than $10 US. In them is about the same amount of circuitry that a cheap IC-based Z meter would have. I can imagine a Z meter for $30. At that price, people will start using them to interface to sensors and use them for other novel applications beyond the traditional uses that marketing people base their projections on.

  11. D Feucht
    October 16, 2013

    Brad, I re-read your article on DSO integration and what got my attention was this:

    “The trigger function should be straightforward. Look for a certain voltage (trigger-point). Then, perhaps, look for a certain rate of change of that voltage. Easy.”

    I got a chuckle out of that, having co-designed the Tek M202 trigger generator IC. The problem is analog, all right, but along the time axis. After the sweep on an analog scope, or memory fill of a trace on a DSO, the next trace is acquired, but that occurs only when holdoff of the next sweep (analog) or ready for next acquisition (DSO) occurs. If the trigger event (from a comparator; that part is what is easy) coincides with the release of holdoff, setup times can be violated and the sweep started a little late caused by transition delay in the sweep gate. (DSOs have corresponding digital problems.) The result is trigger jitter and a fuzzy trace from successive traces that don't quite align in time.

    There are solutions to the triggering problem but they are too much for a description here. They can be found in my book (D. Feucht, http://www.scitechpub.com), Designing Waveform-Processing Circuits.

  12. Pho99
    October 16, 2013

    Have you seen the Red Pitaya board on Kickstarter?  It features two 14bit ADC's and a 125MSPS sample rate on both channels simultaniously — It runs Linux and Interfaces to iPads and Androids

  13. Pho99
    October 16, 2013

    Have you seen the Red Pitaya board on Kickstarter?  It features two 14bit ADC's and a 125MSPS sample rate on both channels simultaniously — It runs Linux and Interfaces to iPads and Androids

  14. RedDerek
    October 16, 2013

    Brad, using a CRT is outdated. Use an LCD on chip method. Then insert projection method to enlarge the final display. Otherwise, integrate onto glasses, like GoogleGoggles and have it minuturized. Granted a true scope-on-chip would have limitations.

  15. RedDerek
    October 16, 2013

    Brad, using a CRT is outdated. Use an LCD on chip method. Then insert projection method to enlarge the final display. Otherwise, integrate onto glasses, like GoogleGoggles and have it minuturized. Granted a true scope-on-chip would have limitations.

  16. Brad_Albing
    October 16, 2013

    @Scott – you're right – my focus here is pretty narrow. This could be done with a small box with ADCs and a USB port or a WiFi connection. Run the device on batteries and communicate via WiFi and you've got your galvanic isolation for doing high-volatge measurements.

  17. Davidled
    October 16, 2013

    I wonder how LCD interfaces on the chip. We might overview D Feucht book (Designing Waveform-Prcessing circuits) to design the circuit that all data would be transferred to LCD. The book describes the all signal conditional circuits between analog domain and digital domain.

    The website is below:

    http://www.scitechpub.com/catalog/product_info.php?products_id=350

  18. Brad_Albing
    October 16, 2013

    @Scott – I'd buy one of those.

  19. Brad_Albing
    October 16, 2013

    @D Feucht – I don't expect a scope on a single IC anytime soon because of the mix of high-speed analog and digital. Mixing wideband and sensitive analog circuitry with high-speed digital poses a crosstalk conundrum . I actually knew that, but I also know it's a good way to open up the topic and discuss if it's possible/practical or ever will be to integrate everything.

  20. Brad_Albing
    October 16, 2013

    @antedeluvian – like the iPad [type] app that I liked (another comment, above), this sounds pretty cool. Will it ever be a commercially available product?

  21. Brad_Albing
    October 16, 2013

    @edartuz – with the ADCs in those two versions (the 'F303 & 'F4xx) – 4 X 5MSPS or 60-100MSPS – that's pretty powerful. A lot of very nice data acquisition could be done.

  22. Brad_Albing
    October 16, 2013

    @Pho99 – have not seen it, but that sounds like it's worth looking into and maybe providing assistance to.

  23. Brad_Albing
    October 16, 2013

    @RedDerek – I know – not really going to use a CRT – that was just for historical perspective.

  24. Brad_Albing
    October 16, 2013

    @DaeJ – Yep, Mr. Feucht knows what he's talking about – he and his published works are excellent resources.

  25. D Feucht
    October 16, 2013

    @Brad – You have a point. Eventually (maybe) everything except the DSO display and power supply will be integrated. There is no basic principle I know that would rule it out. And putting the digital on the same chip would cut down on interconnect radiation and crosstalk, so ultimately it is not a bad idea.

    With the specialization of IC processes we have today, it would be (as you pointed out before) highly suboptimal. However a few-IC solution, reminiscent of the 1950s “All American 5” electron tube solution to “integrated” (consolidated) AM radio design, is nearly here now and is not bad – not at $400 US for a 100 MHz scope.

  26. D Feucht
    October 16, 2013

    @Pho99 – Yes! The designers are from the particle detector world. What really got my attention is that the product is OPEN SOURCE. Bravo! I look forward to the return of the days of disclosed technology, when buying equipment means that you have full ownership of it because you know what you bought. I hope those guys succeed.

  27. Scott Elder
    October 16, 2013

    Dennis,Brad — I'm not convinced that the digital and analog mixed create a problem.  Scopes aren't 16-bit vertical accuracy instruments.  8-9 bits is good enough–at least for my eyes (512 vertical points).  I think at that level of resolution, the digital noise can be managed.

  28. D Feucht
    October 17, 2013

    @ DaeJ – You write:

    “… to design the circuit that all data would be transferred to LCD.”

    You've hit on one of the key driving factors in DSO advancement. Scopes are capable of showing only segments of the giant bitstream of continuous data on the screen, and most often repetitive segments of waveforms. The trigger system has the sole purpose of deciding how and when to grab the segments.

    In analog scopes, the fraction of the total stream that you see on the screen is small, especially at high sweep rates, and this has led to Tek “phosphor” technology, of trying to use high-speed DSP to grab a larger fraction of all that goes by. All of it is only captured when you have infinite memory and time to look through an eternity of waveform data. Nobody even wants to do that, so triggering continues to be important, even in DSOs with high memory capacity.

  29. Netcrawl
    October 17, 2013

    @Daej great link thanks for the share, accurate and fast circuit methods, very useful in doing circuit analysis and design. Perefct for problem solving and begineers.

  30. samicksha
    October 17, 2013

    @pho99: Thanx for sharing info, although i havent seen it yet, but sounds great… read about same seems worth…Clearly this is some hacker-grade stuff and if you don't know what you'd do with a Red Pitaya you're probably not the right market.

    Source: Techcrunch

  31. antedeluvian
    October 17, 2013

    Brad

    Will it ever be a commercially available product?

    We are hoping to have it out early next year.

  32. fasmicro
    October 17, 2013

    >>  Perhaps all that is required is a simple digitizer IC that wirelessly streams the data to a generic standalone computing device with a display–here I'm thinking smartphone or tablet.

    I think that is technically a wireless scope. The key problem with that is the degradation of data when you use the wireless approach to stream the data.

  33. fasmicro
    October 17, 2013

    >>  Does one really need a separate display, computer, and memory for each instrument on a bench?

    No – scopes can be integrated into our PCs and laptops. But the main problem remains how to stream the data points to the processing element. Wireless has not been very optimal and that is what we still build these redundant units of dislays. Alternatively, we need to revaluate the probes in scopes and see how they can be made to become plug and plays like in USBs to work with PCs so that specified apps in tablets can be used to acquire and process data.

  34. fasmicro
    October 17, 2013

    @Scott, the driver will be the market size. You take a look at how many people will need scopes and then conclude if that will move market for Apple. They like to sell in millions and when that is not possible, they do not show interest. Everyone has been looking for the day we can have high quality scope experience via mobile devices. The problem has been that the people that can do that are not interested, at least, so far.

  35. Scott Elder
    October 17, 2013

    @fasmicro:  What do you mean by “degradation of data” when sending data wirelessly?  I can understand a bottleneck in the pipe, but its not clear why the data would be degraded.

  36. Davidled
    October 17, 2013

    Only thing that I can image is the air noise or data missing in the wirelessly depending on signal strength and cellular coverage. Also, in the most case, audio stream is highly distorted in the Bluetooth connection.

  37. Brad_Albing
    October 18, 2013

    @antedeluvian – Keep me in the loop on that.

  38. antedeluvian
    October 18, 2013

    Brad

    I am not sure if you would want to, but you could republish my blog from MCC (as you did with my Excel blogs) originally published 6/12/2013 “Trace Signals From Behind Closed Doors”

  39. Brad_Albing
    October 18, 2013

    @fasmicro – They like to sell in millions and when that is not possible, they do not show interest. Yep – so Apple won't do this – we'd need to find a smaller company that could use some more generic ASICs to build the product. Then, assuming enough units are sold, a custom device could be manufactured. Realistically, that's probably how this would happen.

  40. Brad_Albing
    October 18, 2013

    @antedeluvian – I'll go read it and move it over to Planet Analog.

  41. Brad_Albing
    October 18, 2013

    @D Feucht – Hmm… so you need a DSO that is “smart” enough to know what anomaly you're looking for so that it will trigger and store that anomaly (and of course whatever came right before). That would seem to be the only way to avoid storing infinite amounts of data and then scrolling thru it all to find the trouble.

  42. fasmicro
    October 18, 2013

    >> @Scott–that sounds good. The calculator industry did that well until computers obviated the need for most calculators

    And now typing “=” in google search box, you can do any type of mathematics you can imagine. That is a calculator that will dwarf anything one can have on his or her desk. With google, using the calculator in my PC seems remote.

    Yet, the scope and PC industry have two parallels. For the fact you are measuring something with one having to be really good, it may be challenging to get tablets to do the jobs of scopes without making scopes as tablets.

  43. fasmicro
    October 18, 2013

    >> I can understand a bottleneck in the pipe, but its not clear why the data would be degraded.

    Interferrence and emission. I think you can expose signals to those issues unless you will have a Faraday cage.

  44. fasmicro
    October 18, 2013

    >> we'd need to find a smaller company that could use some more generic ASICs to build the product. 

    @Brad, the problem with scope is that this is not something that needs to be fancy. The best electronics business is the one where people need to constantly upgrade. That is the economics. Another issue is the market size. Is scope business really that big? You see these big scopes from Tek going for $2,000 when Apple sells iPhone that has nothing inside $700. We hope one startup can find an opening and redesign the industry.

  45. eafpres
    October 18, 2013

    @fasmicro–it is of note that in the PC industry, some companies have made the effort to introduce specialized models.  A good example is Panasonic with their Toughbook series.

    Perhaps these would be a starting point for a tablet-based scope add-on.

  46. goafrit2
    October 18, 2013

    >> Perhaps these would be a starting point for a tablet-based scope add-on.

    The precision industry business is different. Instead of tablets having scopes as add-ons, I see scopes adding tablets as add-ons. It is very complicated to make precision systems and mixing it with consumer electronics will take a lot of efforts.

  47. goafrit2
    October 18, 2013

    There is the problem of collecting and constinuosly measuring things as some scopes do these days. Wireless solution must not only ensure there is no interferrence emanating from the envrironment, it must provide at least the same precision quality the corded one offers today. But look at it, while there is a huge advantage to move around lab to test things without being tethered to cables, we need to understand that the main cost factor in scope is the probe.

  48. goafrit2
    October 18, 2013

    @Brad: The main cost element in scope is the probe. The cost of the probe contributes up to 20% of the cost.  Having a tablet will not eliminate the need of scope and it will still be expensive. The electronics in scope is not the big problem. Take away the probe and the monitor, you can get it for few dollars. And when you cannot miniaturize the probe, there is no incentive to move it to the mobile tablet space. ECGs, EEGs etc are all being hooked to tablets. But scope is a different beast as it is a precision system that must be better that anything it has to measure. That creates problems on how you make it.

  49. eafpres
    October 18, 2013

    @goafrit2–I don't really agree.  I have been in a lot of industries in my career, and whenever somebody says “but our business is different ” you can expect they are a candidate for disruption.

    I agree with your other comment you need probes, but as you say the rest is inexpensive.  And you leave out that a high value-adder is the software–the software in the scope itself, as well as externally.  If that can be ported to a tablet and the hardware is a small module to interface probes, then it is likely to happen.

  50. goafrit2
    October 18, 2013

    >>  If that can be ported to a tablet and the hardware is a small module to interface probes, then it is likely to happen.

    Sure, it will eventually but I think the technology that will help in that porting is not ready. There are many things that can happen if we continue to advance in technology. Yet, the business of tablets and scopes are different. One is consumer with its focus on mass market while the other is not. But we will see as technology advances.

  51. eafpres
    October 18, 2013

    Hi goafrit2–“Yet, the business of tablets and scopes are different. One is consumer with its focus on mass market while the other is not. But we will see as technology advances.”

    I would have said the same thing, but iPhone and Apps changed everything.  As you said, doctors have EKG on iPhone, I have seen RF spectrum analyzer on iPhone, I reviewed myself a spectrophotometer for iPhone.  All of these would have said “the business of xxx is instruments while the iPhone is consumer”.

  52. goafrit2
    October 18, 2013

    >> As you said, doctors have EKG on iPhone, I have seen RF spectrum analyzer on iPhone, I reviewed myself a spectrophotometer for iPhone.  All of these would have said “the business of xxx is instruments while the iPhone is consumer”

    Absolutely. When you need electrodes to connect to measure ECG, EEG etc, that is a simple task. Until we can figure out a better way of making probes, it will take a while to get scopes in iPhone. Understand that people have tried to see how they can make ultrasound equipment interface with tablets. Unfortunately, no one has an alternative to the special probes used in ultrasound equipment. Technically, they are replacing ultrasound display and not the probe which is the main cost factor.

    Sure we will get there.

     

  53. D Feucht
    October 19, 2013

    Triggering is mainly for the purpose of capturing repetitive (periodic) waveforms at the same phase in their cycles so that when they are successively presented on the display, they are stable. This is accomplished with a comparator and an XOR gate following it to select the edge polarity corresponding to the slope of the waveform.

    Additional selective functions can be added, such as simple filtering (high-frequency and low-frequency reject, ac, dc) or video signal feature selection such as the vertical interval field. When i was in Tek Labs in the late 1970s, we were working on more advanced triggering capabilities involving counters, etc. Nowadays, some of these additional capabilities exist, such as selecting digital pulses that do not have valid logic levels (“runt” pulses – we used to call them “glitches”). Specialized instruments for acquiring data from various serial communication formats allow selection of features within data frames, etc. All of this is “triggering”. Yet for any of it, capturing successive traces for display requires some kind of synchronization for repititive waveforms.

  54. etnapowers
    October 21, 2013

    Many times , during my business trips at subcontractor facilities , I brought with me a portable oscilloscope. It worked very well but it was not integrated in the test board. the idea of an integration of an oscilloscope on a chip is very interesting , for sure it will have a great impact onto the testing yield enhancement at mass production stage.

  55. etnapowers
    October 21, 2013

    The board that contains the integrated oscilloscope circuitry, could be easily utilized as a general purpose board to be integrated to the circuitry useful for the testing of the specific product. This would create a reference circuit useful for many applications.

  56. Brad_Albing
    October 22, 2013

    @D Feucht – that's an excellent summary of triggering — types and requirements pertaining thereto. Thanks.

  57. Brad_Albing
    October 22, 2013

    @etnapowers – As you noted – and as @antedeluvian noted earlier – the scope is not integrated into the equipment that you want to monitor. Those small portable scopes are nice, but if the scope – or more generally a data acquisition system (DAS) – were integrated into the equipment, troubleshooting would get easier. Which means it would get less expensive to do. Sometimes, you could even troubleshoot from far away. So the additional cost of putting a DAS in every system you build would quite possibly be offset by lower maintenance/field service costs.

    Looks like there's another blog in there for me to write.

  58. Brad_Albing
    October 22, 2013

    @etnapowers – Hmm… yes, that could work.

  59. yalanand
    October 27, 2013

    The sampling head develops dual track and clasp circuits applied in a 65nm SOI procedure. The models are clocked asynchronously at lower frequency, and the time base is reassembled to create sub-picosecond timing purpose. 

  60. yalanand
    November 30, 2013

    If we integrate oscilloscope on to chip, exactly how much speed and  how many effective number of bits are  required  to convert the input signal  from analog to digital. If we can not achieve  required conversion speed it would be difficult.

  61. Brad_Albing
    December 2, 2013

    @yalanand – Probably 8 to 10 bits would be sufficient. For the conversion rate needed, it would simply depend on what you needed — just like any analog scope that you are familiar with. So perhaps 100MSPS (megasamples per second) or 1000MSPS — whatever you need and whatever you can afford.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.