6
\$\begingroup\$

I connected a BK Precision 4017B waveform generator to a Rigol DS1054 Oscilloscope.
BNC ⇽ Co-Axial-Cable ⇾ BNC. 8 MHz square wave generated.

Square wave looks rather distorted to me. See image below.

Also tried a 10 MHz oscillator 4-pin can and also get a distorted square-wave waveform.

What am I doing wrong? I kinda feel the waveform should be more “clean/square”.

8 MHz Square Wave - Rigol Oscilloscope

\$\endgroup\$
5
  • 4
    \$\begingroup\$ That doesn't look too bad for an 8MHz square wave on a 50MHz scope. That's only good up to the 5th or 7th harmonic. \$\endgroup\$
    – hobbs
    Commented Jul 10 at 19:24
  • 1
    \$\begingroup\$ Set your generator to 100 kHz and see if it looks better. \$\endgroup\$
    – gbarry
    Commented Jul 10 at 19:58
  • \$\begingroup\$ From photo, rise time is roughly 14ns. That's within the function generator spec of 30ns rise/fall time. And as @justme has pointed out, scope's bandwidth limit might possibly be engaged, making displayed risetime even slower. Your function generator's TTL output has a bit faster rise/fall time of 20ns, but you should use a 10X attenuator probe to monitor it. Seems you need a faster pulse generator to verify your 'scope rise/fall time of 7 ns. \$\endgroup\$
    – glen_geek
    Commented Jul 10 at 20:15
  • \$\begingroup\$ This is very helpful. I tried the 100KHz waveform experiment and I got a nice looking square wave. I will post the oscilloscope screenshot to this thread. \$\endgroup\$
    – Shep
    Commented Jul 11 at 0:37
  • \$\begingroup\$ If your scope does not have a switchable 50-ohm termination built in, you should use an external BNC in-line terminator at the input jack when using coax. \$\endgroup\$
    – Dave Tweed
    Commented Jul 11 at 1:42

2 Answers 2

12
\$\begingroup\$

That looks perfectly normal for your scenario.

You have a high speed digital signal and the coaxial cable has a lot of capacitance and the scope end is left unterminated.

The B symbol on channel 1 means also that something funny is going on, like you may have bandwidth limiting on that channel on, so it also looks weird because of it. Turn it off.

Also, the 4-pin crystal oscillator module with digital square wave output is not intended to drive 50 ohm BNC cables due to their impedance (50 ohms, and the capacitance). They generally have CMOS outputs that are meant to drive a short piece of PCB track going into some CMOS chip input. As the oscillator setup is unknown, the waveform can be bad if you simply wired some supply to it with short wired and put no bypass cap on the oscillator, and the signal and ground paths to the BNC cable are also unknown, any inductance from even short wiring will also distort the signal.

As it turns out if the scope is only 50 MHz then you can't expect to see a clean square wave. A clean square wave is a composite signal of sine waves with odd harmonics of the fundamental, in theory up to infinity. Your scope can approximately show only 5 first harmonics of the square wave signal.

Also a word about the BNC cable. It likely has standard 50 ohm characteristic impedance. Imagine time is slowed down and a signal like edge of an ideal square wave entering into one end of the cable. The signal starts travelling through the cable at about 0.66 times the speed of light and to the signal source with some voltage the current that enters the cable equals the characteristic impedance, i.e. a 50 ohm load. When the signal has traveled to the scope end of the cable, an unterminated scope looks like 1 Mohm impedance, almost like an open end, so there is an impedance mismatch or discontinuity, and basically at the signal voltage, current that matched the 50 ohms can't flow out of the cable, so the voltage at the scope end shoots up to double, and now the signal edge reflects back towards the source end of the cable and eventually hits the signal source. If the signal source is e.g. the CMOS output chip, it will mix in with the transmitted signal again and you see all kinds of weird waveforms that are just the steady state of square wave signals ping-ponging back and forth in the cable. In the case of the signal generator, it has a source impedance of 50 ohms, so it will dampen the reflections hitting back towards it.

\$\endgroup\$
3
  • 6
    \$\begingroup\$ In case it helps the OP visualise, this is a "square wave" band-limited to just the first and third harmonics. Looks almost identical to their waveform (bar some even harmonics in theirs). (image is own creation and free for use). \$\endgroup\$ Commented Jul 10 at 23:17
  • 1
    \$\begingroup\$ @TomCarpenter I was just about to post a comment on exactly the same point you made. In fact, I'd just caused LTspice to generate it for me. Then I saw what you wrote when scanning around to see if someone already caught it.... You did! I thought Justme should have put this up -- I still do think so. \$\endgroup\$ Commented Jul 10 at 23:24
  • \$\begingroup\$ Very helpful. Thanks! \$\endgroup\$
    – Shep
    Commented Jul 11 at 0:38
2
\$\begingroup\$

I did the experiment suggested by @gbarry and switched to a 100kHz square-wave and got something that looks like a clean square-wave. See oscilloscope screenshot.

All the analysis and advice that has been added to this thread makes a lot of sense. Very helpful.

enter image description here

\$\endgroup\$
7
  • 4
    \$\begingroup\$ If @Justme's answer helped you, you should mark it as "accepted" (a little checkbox next to his answer). \$\endgroup\$
    – Helpful
    Commented Jul 11 at 5:43
  • 5
    \$\begingroup\$ Your yellow channel still has the B symbol which means you have intentional bandwidth limiting on, which might have prevented getting a better waveform for the faster signals too. Please turn it off. \$\endgroup\$
    – Justme
    Commented Jul 11 at 5:54
  • 2
    \$\begingroup\$ Is this an answer, or an update to the question... or a sort of postscript? \$\endgroup\$ Commented Jul 11 at 14:20
  • \$\begingroup\$ Why is this not considered NAA (flag disputed)? Half of it is a "yeah that works" and the rest should be an addition to the question.... \$\endgroup\$
    – Dan Mašek
    Commented Jul 11 at 22:57
  • \$\begingroup\$ In answer to @Greenonline's comment. Yes, sort of a postscript because a member suggested an experiment and so I did it and posted the result. But I was only able to post a new image as an "answer" because I could not see how to attach an image to a "comment". My apologies if I have not properly followed the rules. \$\endgroup\$
    – Shep
    Commented Jul 12 at 21:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.