No Script, No Fear, All Opinion
RSS icon Home icon
  • EEVblog #306 – Jim Williams Pulse Generator

    Posted on July 6th, 2012 EEVblog 17 comments


    Testing of the Jim Williams designed pulse generator from Linear App Note 47:
    http://www.linear.com/docs/4138
    On a 500MHz Agilent 3000X series oscilloscope.
    The build is by Munaut.S
    http://people.osmocom.org/tnt/hw/pulse_gen/

    Be Sociable, Share!
     

    17 responses to “EEVblog #306 – Jim Williams Pulse Generator” RSS icon

    • Does anyone know where the nice PCB mounted BNC connector can be obtained from / what it’s part number is?

      Thanks all,
      Richard

    • Dave, could you measure also the Rigol with it? Especially after the 100MHz mod, it seems that the bw of the scope is a lot better than 100MHz – it would be nice to know for sure.

      Regards,

      Jan

    • Did you use the 50 ohm mode of the scope ?
      if not, this will give you a better waveform (the small pulse seen when zooming could be the reflexion? or it’s a generator artifact)

      If you use the “high impedance” input (which is not really at those frequences), using a meter cable keeps the reflections some 20 ns off your pulse. Could be that with a direct connection you have a less nice signal ;)

    • Dumb question : is using a more modern transistor with a higher Ft and a low inductance SMD package going to improve that pulse ?

    • another dumb question: If the pulse is so short that it gets falling almost as soon as rising finishes, how can you make sure that you are seeing its full amplitude, vs an amplitude being attenuated by limited speed of input circuitry, frequency limits, etc?…

      Since the breakdown voltage varies from thansistor to transistor, and depending on the temperature, one way I see this can be done is by measuring the breakdown voltage of the transistor at DC first, so the maximum output swing will be known. Such as temporarily disconnect the capacitor.

      In general, I’d rather want to see some plateau at the top. Is there a way to modify this particular circuit this way?

      • I agree with tchicago. I am not convinced either.

        I believe that the formula applies to a step response with a sharp corner, not a narrow pulse with a rounded edge. The high frequency respons of the scope is seen in the sharp corner.

        My opinion is that the test shown overestimates the scope’s bandwidth. This has been my experience (albeit limited) with my scopes.

        I would like to see the bandwidth confirmed using a good signal generator and measuring the peak to peak of a sine wave, or using something closer to a step waveform.

      • Yes, there is away to modify it in this way, and I will be trying this soon.

    • your download links are broken on chrome on OS X

    • I think these measurements may require a step response, not an impulse.

      Also, Be VERY CAREFUL with this type of avanlanche pulse generator. There are HIGH VOLTAGES possible on the instrument under test input, especially in case of transistor failure.

      I think Jim Williams was targeting (perhaps) more “relative” measurement of scope probe bandwiddth where the probe may be in 10X mode and capacitively coupled. I would be very leary about putting this pulse generator directly on the input of an instrument without being very careful.

      As for getting a step, use a divider. Micrel (good luck getting parts from them) has some pretty fast little logic gates.

      Regards, David in Jakarta

    • Regarding the problem with the touchiness of the transistor: I think whats happening ist that you connect the collector (which is connected to the transistor case) via your body to ground. Since the voltage source has an impedance of 1MegOhm you’re basically shorting the whole thing. – Or to be more precise: Your building a voltage divider with your body and the 1Meg so that the maximum voltage drops below the breakdown voltage.

      Proposal for solution: Don’t touch it :-D

      • Different layout. Everyone copies the Jim Williams schematic. Only so many variations on the PCB layout before they all look quite similar.

    • Indeed, one should be very careful with this type of avanlanche pulse generator. There are HIGH VOLTAGES possible on the instrument under test input…

    • Sorry, but this is not a valid way to determine the scope bandwidth. You just measure the rise/fall time of the pulse generator, not the one of the scope. Once the generator rise time gets lower than the scope’s rise time, you’ll see an attenuated pulse (but with a rise time shorter than what the formula would tell you).

      The formula you give applies when you have a STEP signal input, i.e. the signal must remain high for long enough that the scope input settled out.

      So as others on this comment list have suspected, the measurement completely overestimates the scope bandwidth.

    Leave a reply