1. Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.
    Dismiss Notice

USB Oscilloscope

Discussion in 'Members Lounge' started by arhi, Apr 19, 2008.

  1. arhi

    arhi Member

    Joined:
    Apr 17, 2008
    Messages:
    887
    Likes:
    12
    Location:
    Belgrade, .rs
    Hi everyone,
    I searched for some "recommendation" on what is the best value for the money in usb oscilloscope world, and got nowhere, so this might be the best place to ask.

    to get the answer to the question, i know i need to put something about "why the hack i need it" so here it is:

    - i come from the software world, and playing with hardware is hobby
    - i mostly play out with PIC* (16F*, 18F*), usually 10-20MHz
    - i play with some analog devices (mostly for building sensory devices to connect with pic controlled brain)
    - i live in europe, but do not have problem getting device from usa or anywhere in europe
    - as this is only hobby - i do not want to spend too much money (few hundred eur max)

    So far, I found
    - hobbylab
    - picotech
    - parallax

    so before i order, I would like to hear some opinions / experiences and maybe some other recommendation

    thanks in advance
     
  2. dknguyen

    dknguyen Well-Known Member

    Joined:
    May 6, 2006
    Messages:
    6,763
    Likes:
    72
    ONLINE
    The only one you are going to be able to get for a few hundred euros is a PO scope or one from Parallax. YOu aren't going to be able to get a "real" scope unless you spend $800USD (how many euros is that)?

    The problem arises from the fact that if you want to observe a square wave of F MHz, you need an oscilloscope with at least F*10MHz bandwidth. And that gets expensive...fast. Remember, that you need a sampling rate that is MORE than your bandwidth so that your bandwidth is the limiting factor. If the bandwidth is 1Mhz and the sampling rate is 1Mhz, well you can only really observe a 100kHz signal (sinusoid, not a square wave) accurately because your sampling rate is the limiting factor.

    It's almost "all or nothing". You pay a lot of money to be able to observe those digital signals, even of low frequency since you need a bandwidth at least x10 the frequency of the signal (and a sampling rate that is 10x the bandwidth). So even for 115kbps RS232 or 100kHz I2C, you'd need a 1MHz bandwidth- which you may or may not be able to find since it's such a strange bandwidth number. As soon as you want to measure an oscillator on your PIC, you need 20MHz (since it's a 20MHz sine wave). Now if you wanted to measure the a 10MHz SPI signal then you suddenly need a 100MHz bandwidth, although you could probably see it somewhat properly with a 60MHz bandwidth scope.

    I think it has to have at least 20MHz bandwidth with a sampling rate 10x that to be useful. That would let you observe the 1-2MHz SPI signals, audio, ultrasonic , UART, and some motor PWM (but not the transients!), and an average oscillator on a PIC. But if you want to observe the fastest square waves a PIC can output, forget about it, you're gonna need to spend the money to get a 100MHz 1Gsps scope. The bad one is the motor transients, you are observing a square wave (so you need x10 the bandwidth just to be able to see the square wave) and then you want to see the transients which requires even more bandwidth because they are such fast events. But if you were going to get a PicoScope 2205 for $590, you might as well just spend $800 to get a Tektronix TDS1001B which has twice as much bandwidth and is a REAL oscilloscope with knobs. USB scopes only outprice real scopes on the low end (where there are no real scopes). But I find that a USB scope comparable with a real scope costs about the same.

    I suppose you might be able to get by if you have less than frequencyx10 for a digital signal if you're only interested in the digital part of it and not the actual analog waveform - but then you might as well get a logic analyzer for that.

    Here is another one. It's quite similar to the Hobbylab the one you've posted. http://www.hvwtech.com/products_view.asp?ProductID=650

    So the question is...do you really want to be able to observe accurate waveforms of digital signals? If not, and the fastest analog signals you are working with are 20Khz or less, just get one of those $200 USB scopes with 200kHz bandwidth and a built-in logic analyzer. If you were only interested in observing the analog waveforms of the slowest digital signals (like I2C or UART), you might go with the PicoScope 2203, but it costs 50% more than the $200 scopes and doesn't come with a logic analyzer (which would probably be more useful to you than observing the analog waveform of the digital signal anyways). It is the only scope that can display the analog waveforms of those I2C and UART signals correctly though, but is that worth the extra $100 when the logic analyzer can be got for $100 less?

    I think:
    -The Picotech ones are overpriced
    -THe Parallax one has no extras (especially no frequency analyzer or logic analyzer), but it does have a better sampling rate than the PO scope or HObbylab scope. Notice that it has a sampling rate 5x it's bandwidth.
    -The PO scope and hobbylab scope are full of features at a good price, BUT take note that they say the bandwidth is 200kHz and the sampling rate is 200kHz giving an effective bandwidth of 20kHz.

    So if the only analog you are working with is audio, then the PO scope or hobbylab might be good since you can use the logic analyzer for anything digital. The parallax one covers a lot, but no logic/spectrum analyzer so your life is a bit more difficult but you can get by doing things "Manually". The cheapest Picotech is okay for analog and would cover most things (like the non-RF analog and serial interfaces slower than 100kHz), but like the Parallax scope, no logic analyzer so your life is more difficult. If you're going for a Picotech 2205, spend another $150 and get a real Tektronix scope.
     
    Last edited: Apr 20, 2008
  3. arhi

    arhi Member

    Joined:
    Apr 17, 2008
    Messages:
    887
    Likes:
    12
    Location:
    Belgrade, .rs
    dknguyen, thanks a bunch for this assessment ... I knew I should post the question before I go and purchase something on my own :)

    After reading your post (two times actually) I decided not to go with usb osc at all. For ~300E I can get (locally) used oscilloscope with everything I need. (and as for the logic analyzer, I can always use the emulators - I mostly use ISIS).

    Thanks again.
     
  4. dave

    Dave New Member

    Joined:
    Jan 12, 1997
    Messages:
    -
    Likes:
    0


     
  5. simonbaker

    simonbaker New Member

    Joined:
    Aug 8, 2009
    Messages:
    3
    Likes:
    0

    I don't understand why you need a sampling rate 10*bandwidth?

    I guess it depends on meaning of bandwidth -- if very slow roll off, like 2nd order filter, OK. But if "wall" filter -- why such high sample rate?

    -SB
     
  6. dknguyen

    dknguyen Well-Known Member

    Joined:
    May 6, 2006
    Messages:
    6,763
    Likes:
    72
    ONLINE
    Sample rate and bandwidth are not the same thing. The bandwidth is how much of the analog input signal actually gets through the front-end amplifiers properly. The sample rate is the number of points actually recorded on the output end of the front-end amplifier. Suppose you had a bandwidth of 100MHz and a sample rate of 200MHz. You can sample a single point of a 100MHz sinusoid just fine since it gets through the front-end amplifier...but when recording the signal as a whole, you sampling slowly at only 200MHz. So only two points per period appear on the oscilloscope. It would look like a randomly jagged waveform and not the sinusoid input signal (the sampled waveform would still be "unique" to the input sinusoid because it satisfies Nyquist sample criterion , but it "uniqueness" is not the same as accurate). 8-10 samples per period is when it starts looking like an actual sinusoid you are sampling.

    It has nothing to do with the definition of bandwidth or the filter. The example holds true whether or not the frequency cutoff is ideal or very slow roll-off. It's the same as using a high bandwidth measuring device, but only looking at the readout and recording it every so often.
     
    Last edited: Aug 9, 2009
  7. tariq7868

    tariq7868 New Member

    Joined:
    Mar 19, 2009
    Messages:
    116
    Likes:
    3
    i also had a good mind to buy a usb oscilloscope... but many thanks to dknguyen and the details & fact... i know usb scope wont be much of use for me and my needs..
     
  8. Mikebits

    Mikebits Well-Known Member

    Joined:
    May 24, 2008
    Messages:
    6,217
    Likes:
    175
    Location:
    San Diego, Ca
    ONLINE
    A nice 200MHz analog scope can be had for cheap, and will serve most hobbiest needs.

    Most USB scopes are not worth the cable they are attached to.
     
  9. tariq7868

    tariq7868 New Member

    Joined:
    Mar 19, 2009
    Messages:
    116
    Likes:
    3
  10. simonbaker

    simonbaker New Member

    Joined:
    Aug 8, 2009
    Messages:
    3
    Likes:
    0
    But I think you just need to be smarter about reconstructing the original waveform from your sample. Your analog filter has removed any frequencies about 100 mHz, therefore, oversampling does not get you any more information. With the proper "inverse filter", you can reconstruct the exact waveform that made it through your 100 mHz filter with 2X samples (Ok, that assumes brick wall filter -- and I would hedge a few more samples, like 2.5, but the point is the same).

    For example, someone might say -- sample 10x so you can see the shape of the 100 mHz signal, in case it is a square wave, etc. No -- it cannot be a square wave, or anything but a sine wave -- because it is at the upper edge of your filter -- no higher frequencies to give it any shape except sine wave.

    Now I know in reality you are right, that we see gigahertz sampling for scopes, but I think maybe because scopes try to use very slow roll-off amplifiers to get as much signal beyond rated bandwidth as possible. Also probably special amps that try to maintain linear phase relationship so only make time delay instead of phase distortion. I'm not sure. But just want to be clear there is no theoretical need for more than 2x sampling if have brick-wall filter, but yes in real world it is done. I wonder if 10x really needed though.

    Regards,

    -SB
     
    Last edited: Aug 9, 2009
  11. tariq7868

    tariq7868 New Member

    Joined:
    Mar 19, 2009
    Messages:
    116
    Likes:
    3
    it is right tht oversampling doesnt give u any more information , but higher the sampling the higher will be the accuracy in wave formation ... sampling is like collecting information(or in other words the current value of the voltage at a certain time) ,.. so if the oscilloscope can collect as many infos or voltage values in a smallest time ,.so as the accuracy goes higher..

    so it depends on our whether if in a case i want to detect even a very small spike or anything like a very smallest change in the waveform .. i would go for a higher sampling rate of oscilloscope...
     
  12. simonbaker

    simonbaker New Member

    Joined:
    Aug 8, 2009
    Messages:
    3
    Likes:
    0
    Don't forget we're talking about oversampling 10x times the maximum bandwith of your input amplifier/filter. Let's say your input analog filter is 100 mHz, meaning that is the highest frequency that can pass through your oscilloscope probes and preamp. If any small spike can get through your input filter, then the highest frequency of your spike cannot be greater than 100 mHz (assuming perfect filter). You can reconstruct all the frequencies in a 100 mHz band using 2X sampling and the appropriate inverse filter. If you can reconstruct all the frequencies, then you will see your spike exactly as it came through the input filter. It's all in the math, but does require careful processing of the samples.

    I think there is good reason to oversample 10x though, and I think it is probably because the input filter of an oscilloscope is not sharp but rolls off slowly and really includes frequencies well above the "cutoff" frequency, but I'm not sure.

    Regards,

    -SB
     

Share This Page