Capturing glitches

Discussion in 'General Electronics Chat' started by theamber, Jul 4, 2008.

  1. theamber

    Thread Starter Active Member

    Jun 13, 2008
    318
    0
    Hi, I have read that some Digital Storage Oscilloscopes DSOs decrease their sample rate depending on the settings of the time/Div.or the signal bandwidth. How can you tell? Also when calculating for example the chances of getting a glitch of 2ns of width random and intermittent, you have to take into account the 10 divisions across the screen plus the sampling rate of the DSO?. Also in this case is it good to use the persistance mode?. I know that Some DSOs have some special fuctions to capture glitches like to wait until the glitch happens, but how can this be if they are limited by their sample rate does this has to do something with the memory I am confused about this. Thanks.
     
    Last edited: Jul 5, 2008
  2. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    That and aliasing is one of the problems with digital Oscopes, which is what I think you might be refering to (acronyms are always a bad idea here, without defining it once at least). The real problem is you can't tell as far as I know, which is why there are still analog oscopes out there (including capture types) for big bucks. If people didn't need them they wouldn't make them.
     
  3. theamber

    Thread Starter Active Member

    Jun 13, 2008
    318
    0
    I have seen that the limitations on their sampling speed are determined by the speed of the ADC analog-to-digital converter, and the length of time over which an event may be sampled is limited by the size of the acquisition memory that receives the output from the converter.
    I think DSOs have some limitations over analog storage scopes in both adquisition and display modes and analogs when viewing very fast signal components, it is difficult to view spikes or glitches due to the dim scope illumination. If you really want to see the whole picture is hard you may need both. I think just to know when the glitch happens or if it happens should be the most important things anyways. I think is like trying to tight a particular screw you need both types phillips and flat head to do be more versatile.
    The thing is how you calculate the chances in percentage of getting a glitch of a particular width by having the info the DSO manufacturer tells you.
    By the way this scope is really nice.
    http://www.iti.iwatsu.co.jp/e/products/ss/ts_top_e.html
     
  4. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    If you are looking for glitches in logic, then a logic analyzer is probably superior. The frequency of the glitch has a lot to do with this, too.
     
  5. The Electrician

    AAC Fanatic!

    Oct 9, 2007
    2,281
    326
    Actually, digital storage scopes have bright displays for even the shortest captured glitch.

    The IWATSU scope is very impressive, but have a look at the pdf Data Sheet at this URL (http://www.home.agilent.com/agilent/product.jspx?nid=-536902766.536905504.00&cc=US&lc=eng) which you can find in the Key Specifications box, about half way down the page.

    The Agilent DSO6104 can capture 250 ps glitches and has a 1 GHz bandwidth. It seems comparable to the IWATSU in performance.
     
  6. Wendy

    Moderator

    Mar 24, 2008
    20,766
    2,536
    It all depends on the sampling rate for digital scopes. Betcha the Agilent is a high dollar model, you almost always get what you pay for with test equipment.

    The real problem is if a glitch falls between samples it is missed. I have to agree with the logic probe being a good idea as an alternative.
     
  7. theamber

    Thread Starter Active Member

    Jun 13, 2008
    318
    0
    You are correct this difficulty has been overcome recently by digital phosphor oscilloscopes that have a better variable intensity or intensity grading than regular CRT type.
    I have found a web page that helped me with some of my doubts about DSOs.
    http://www.rocketroberts.com/techart/sigproc.htm
     
  8. The Electrician

    AAC Fanatic!

    Oct 9, 2007
    2,281
    326
    The Agilent DSO6104 is about $12000. We don't know how much the IWATSU costs, but it's a 1 GHz analog storage scope. I'll bet it's more than the Agilent.

    The Agilent can capture a 250 picosecond glitch. A glitch smaller than this will indeed be missed by the Agilent, but I think the IWATSU will miss it too, and in that case it's not an issue of falling between samples. A glitch that would fall between samples on the Agilent 1 GHz scope would exceed the bandwidth capability of the scope, so or course it would be missed.

    But, you can't have a glitch smaller than 250 picoseconds in a system with limited bandwidth, systems made of opamps and typical CMOS or TTL logic such as are typically discussed on this forum. Therefore, the Agilent DSO6104 would catch any glitch that circuits discussed on this forum would generate. I think the 100 MHz version of that scope can catch 1 ns glitches; even that would suffice for most forum circuits.
     
  9. The Electrician

    AAC Fanatic!

    Oct 9, 2007
    2,281
    326
    This web page has outdated information. The Agilent DSO6104 has 8 megabytes of trace memory and captures 100,000 waveforms per second.

    It samples at 4 Gigasamples per second. It would not give a display such as figure 2 on that web page; it would look more like figure 1.

    The lower cost 100 MHz version samples at 2 Gigasamples per second and only costs about $5500.

    The author of the web page cited wants to be able to export the data to a program such as MATLAB. The Agilent scope has a USB socket on the front panel and can easily export the trace data.
     
  10. theamber

    Thread Starter Active Member

    Jun 13, 2008
    318
    0
    He said that: "This example illustrates one pitfall that can occur to the "careless" operator of a digitizing oscilloscope!" when he was refering to figure 2.
    You are right about the page being a bit outdated but it serves it purpose of explaining the workings behind that of digital scopes.
    About the USB connectivity; what the author means is that the lack of memory in DSOs is limited because of their inability to display all the points on their screens and that he guesses that was the reason manufacturers do not supply more memory. He then points out that more memory will be usefull for other uses like printing or exporting waveforms to other mediums where raw data can be displayed better like excel or matlab in order to have a more complete picture of the waveform. I guess that is one of the reasons that now days the scopes screen are getting larger.
     
  11. The Electrician

    AAC Fanatic!

    Oct 9, 2007
    2,281
    326
    That's not all he said. Right in front of Figure 2, he said "Using a digitizing oscilloscope on the same waveform as was measured with the analog oscilloscope would likely generate in image on the display similar to the one shown in Figure 2 below:"

    That may have been true of the scopes at the time he first wrote his comments, but it's not true of current DSO's, such as the Agilent. I don't think the best current DSO's are "likely" to generate such an image. The newest "digital phosphor" scopes are more "likely" to generate an image like Figure 1, even if the operator is careless; they default to that kind of display.

    To correctly put Joe's phrase in context today, one should change to the past tense and say "...the lack of memory in DSOs WAS limited because of their inability to display all the points on their screens and that he guesses that was the reason manufacturers DID not supply more memory." It's no longer true that they do not supply more memory; today they DO supply more memory (at least the newest ones do; that's what the customers demand today).

    I understand that Joe believes that the reason the DSO's of the time had limited trace memory was because he believed that a DSO can't display more than about 1000 points, but I don't think that was the reason. I asked manufacturer's reps at the time why they didn't supply large amounts of trace memory in the older DSO's to which Joe is referring. They didn't say that it was because they couldn't display all the points on the screen. It was because large amounts of extremely fast memory were quite expensive at the time.

    Every power supply engineer wanted a large trace memory, because if you had it, you could trigger on a catastrophic event and then look back in time to see what went wrong. I certainly wanted it, but it just wasn't feasible in the past, even though everyone could see the advantage of it.

    The new "digital phosphor" scopes CAN display all the points on the screen, in the same sense that an analog scope could be said to display "a million points" on the screen. They can't be seen separately as individual points, but their effect is to create a "filled in" envelope, in the words of Joe Roberts. The "digital phosphor" technique could have been used in the past if there had been enough trace memory to make worth doing, so I don't think I agree with Joe's assessment that inability to display all the points made a large trace memory useless. As I mentioned in the previous paragraph, the ability to look back in time is one of the very most useful features of the modern DSO; I wouldn't want to be without it. And even if it weren't possible to see all the points on the screen, the ability to zoom in on a portion of the entire trace memory at a time preceding the trigger point would justify a large trace memory. I think Joe is mistaken about the reasons why the DSO's available in the past, when he wrote his comments, didn't have large trace memory.

    The information on his web page is outdated. Most of the shortcomings of DSO's of the time that he describes have been overcome. However, they are things every user of a DSO should be aware of. A person may have to use one of the older DSO's; not everybody has access to the very latest scope. But I think Joe should update the page and indicate that major improvements have been made.


    As far as USB connectivity goes, even though the author believed the reason for the small amount of trace memory was that the screen couldn't display it, he still saw a use for a large amount of memory; he said:

    "Sure, the display cannot show 10 million data points, but the ability to export that amount of data on some type if magnetic media would be an extremely handy feature! The beauty of having access to the "raw" data is that the user can then process and display the data in any number of ways (using MATLAB, Excel, etc.)."

    I simply pointed that the Agilent can, in fact, export 8 million data points.
     
  12. theamber

    Thread Starter Active Member

    Jun 13, 2008
    318
    0
    I have to agree with you somehow but the reason manufacturers did not add memory it looks logical that it was because of memory speed limitations of that time. I could guess about 4-5 years ago the time Robert is taking about, Digitizing scopes where in the ten thousand dollars range and available memories where not that expensive.
    I agreed with you that modern DSOs are getting much better.
    I have found a video that I had on an Agilent CD but it can be seen here:
    www.agilent.com/find/mso600. Specially shows Agilent's 6000 Megazoom III as a giant improvement and also the 100k waveforms/s capture the largest in the industry. This technology allows the Agilent DSO to display a Z axis intensity gradation that would of display the AM modulated signal on Robert's page with no problems. Also the and 4th. dimension the waveform update rate would enable infrequent anomalies like glitches to be captured by this unit faster that others which have less waveforms captures per sec. like for example the Tektronixs MSO (Mixed signal scope) line only captures 35000 waveforms per second. Still sampling rate is limited by higher settings of the time div and limiting memory, a reason of why the increment of waveforms per sec capture. Agilent also claims their intensity gradation actually exceeds that of analog scopes :eek: (Agilent is 256 intensity levels). But as you said earlier 250ps is a quick glitch even for todays standars.
    By the way the post started, if any one could tell me how to calculate the probability of capturing a glitch because I do not know:
    • Product data sheet: 35,000 waveforms per second.
    • Update rate = 18 waveforms per second with
    10 Mpts and digital channels turned on.
    Probability of capturing the infrequent glitch = 0.09%
    after running for 10 seconds.
    • Average time to capture just one glitch = 128 minutes.
    or this:
    • Product data sheet: 100,000 waveforms per second.
    • Update rate = 95,000 waveforms per second with auto
    memory and digital channels turned on.
    • Probability of capturing the infrequent glitch = 99% after
    running for 10 seconds.

    • Average time to capture just one glitch = 1.5 seconds.
    Thanks.
     
    Last edited: Jul 7, 2008
Loading...