‘Scope on a Chip? Why Not a DAS Onboard?

In one of my previous blogs Put an Oscilloscope on a Chip: Why Not?, I wrote about the idea that we ought to be able to squeeze almost all the circuitry for an oscilloscope onto an IC. Then Aubrey Kagan blogged about the advantages of being able to trace signals from behind closed doors. We've also discussed analog front ends (AFEs) more than once. Let's combine these ideas and see what we get, and let's see if the result is cost-effective while we're at it.

For the scope on a chip, I proposed integrating gain stages, filter networks, perhaps multiplexers, and multiple high-speed ADCs on a chip. You'd likely want the microcontroller unit (MCU) there, too. For the AFE, there are gain stages, filter networks, certainly one or more multiplexers, and one or more low- to medium-speed ADCs on a chip. The AFE likely has a voltage reference and some amount of power-supply circuitry.

You can see how these have overlapping functionality. While making a scope on a chip may still be a little way off, AFEs are here now. If you carefully choose the functional blocks that are integrated onto a chip, it should be possible and practical to make a general-purpose data acquisition system (DAS) on a chip right now.

There should be a large market for such a part. If such a part were added to the devices that we buy by the millions (TVs, smartphones, computers, etc.), it would simplify troubleshooting and repair. Two other markets, while not as large, are medical diagnostic equipment and telecommunications equipment. These are smaller markets (fewer pieces built per annum) but the equipment is critical. If it goes out of service, companies lose money.

That means that companies might be money ahead even if the ICs I'm describing are relatively expensive (though I doubt they would be). When you consider the cost to travel to the site of malfunctioning equipment, do the trouble-shooting, order replacement parts, and effect the repair, you can see how the DAS IC is pretty inexpensive in comparison.

I'd expect that a DAS IC that could monitor several power-supply voltages and currents, MCU clock frequency, temperature, and various digital data-streams would prove to be a valuable addition.

What equipment have you designed and done troubleshooting on for which a part like this would be useful? How much money would it save you in the long run?

— Brad Albing, Editor-in-Chief, Planet Analog and Integration Nation Circle me on Google+

Related posts:

14 comments on “‘Scope on a Chip? Why Not a DAS Onboard?

  1. D Feucht
    November 19, 2013

    Having been writing articles on instrument integration, I “took a shine” to your idea. Perhaps the way this could be approached is somewhat incrementally, where IC suppliers start with what can be done at reasonable technical risk now. The main barriers toward a full DSO on a chip include:

    1. Input voltage range

    2. Bandwidth or risetime (speed)

    H-P has led in the development of wideband attenuator circuits that can handle several hundred volts of input. The attenuator problem is possibly manageable now, though most IC companies do not have instrument designers who know how to do it and would need to collaborate or acquire them.

    As for speed, this is what separates what are referred to as DASs from DSOs. Barry Harvey is one of the Planet Analog contributors who knows something about speed, and there are multiple Tektronix and Agilent people who would too. Speed is where the silicon rubber meets the road, though what mitigates the integration difficulties somewhat are that precision can be reduced to 7 to 9 bits for a typical DSO and two channels are also sufficient, which simplifies multiplexing.  By giving up channels and precision, this buys some design maneuvering space for speed.

    Tektronix has already (for decades) integrated many of these subsystem DSO building-blocks, but they are proprietary ICs (or ICs that Maxtek probably cannot release to a general market). However, Intersil or LTC or TI or ADI especially (having a penchant for ICs that show movement in the instrument-on-a-chip direction) could conceivably do some of this high-speed subsystem integration, as Tek has in the past. With high-speed bipolar device capability (in both BJT and MOSFET) processes nowadays, it seems only like a matter of time before some of these subsystems show up for sale to the Chinese DSO makers seeking to integrate their DSOs more fully in future generations of them.

  2. RedDerek
    November 19, 2013

    …Chinese DSO makers seeking to integrate…

    Ritek is probably quite close to this process. I have not pulled one apart to check it out.

  3. eafpres
    November 19, 2013

    Hi Brad–interesting line of thought.  It occurred to me that automotive is a market that for a long, long time has had special purpose data interaces (the OBD-II etc.) which has allowed the makers to make it hard to translate the data coming out of the port without having the right equipment and/or a computer and plenty of time.  I wonder if you had a standard DAS chipset how many of the sensors in a car you could handle?  If such an approach became really standard the market just in automotive is a few tens of million units per year.

  4. samicksha
    November 20, 2013

    I knew about TDS210 and now DSO in chip sounds interesting, but DSO costs widely and if impant on chip it will real suphosticated which may increase $$ on chip.

  5. Brad_Albing
    November 24, 2013

    @D_Feucht – H-P has led in the development of wideband attenuator circuits that can handle several hundred volts of input – that puts use much closer then to fabricating my hypothetical DAS system-on-a-chip.

    Barry Harvey is one of the Planet Analog contributors who knows something about speed – I'll ask him if he has any further thoughts on this now. He has [shall we say] a bit of spare time available now, so perhaps he'll blog about this.

    Perhaps we'll see a Chinese DAS-on-a-chip soon that we can just drop onto exising boards in the manner that I was hoping for.

  6. Brad_Albing
    November 24, 2013

    @RedDerek – if you do, take copious notes and shoot pix – we can use that for a blog or 2.

  7. Brad_Albing
    November 24, 2013

    @eafpres – agreed that the automotive market is one to concentrate on simply because the units per annum that you can sell can be pretty big. So some IC manufacturer should certainly pursue this.

  8. RedDerek
    November 24, 2013

    Unfortunately I do not have a Ritek scope, nor the funds to disassemble. Here is a link to that Rigol tear-down.

    This was from Sep'2012. Comments on the ADC are interesting.

  9. Brad_Albing
    November 24, 2013

    @RedDerek — I'l have a look. Thanks.

  10. jkvasan
    November 25, 2013


    Last year we had a project that could have become easier and realized faster with a DAS chip. It typically had 6 analog inputs and two analog outputs. It runs turbines and measures process parameters. A DAS could have come in handy and implementing a SPI interface with opto-isolation would have been simpler.

  11. Brad_Albing
    November 30, 2013

    @JK – maybe (if we're lucky) one of the semi companies like Maxim will make an IC to do what you need.

  12. Victor Lorenzo
    December 2, 2013

    @Jayaraman. On one side, it seems that general purpose DAS-on-a-chip are still considered 'not-general-purpose' applications, but on the other side its becoming easier to find specific purpose DAS chips that could (with some effort and making use of our best hardware hacker skills) be 're-purposed'.

    For low speed signals we can find several AFEs like ADI's ADAS1000 (some others from another manufacturers also exist).

    And for high speed signals there are several video AFEs available like: ADIs AD9822.

    Sometimes its a matter of using only part of the AFE's features.

  13. jkvasan
    December 9, 2013


    Maxim and such other technology companies can definitely make this a reality and my gut feeling says they are already on the job. One day we could get a sweet surprise news.

  14. jkvasan
    December 9, 2013


    You are absolutely right.

    May be there are chips available and all we need to do is just exploit the portion which is needed for us.

    A typical application specific chip could accomplish a functionality which is entirely different from its overall intended function.

Leave a Reply

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