0
\$\begingroup\$

*I searched about the correct oscilloscope and probe to measure the a signal with a certain frequency. I read some documents and white papers about this topic. I have heard some "rule of thumbs" about the measurements : You should use scope and probe with a BandWidth of at least 5 times the Bandwidth of the signal to be measured. i.e : If I want to measure a signal of 200MHz differential DDR3 clock line, I need a Scope + Probe at least 1GHz BandWidth. OR, Rise Time / Fall-Time duration of 200MHz clock signal is much more important parameter than frequency of signal for correct Scope + probe selection?

*Another one is measurement options of Scope. If I try to measure above 200MHz, 50 ohm differential Clock lines from the input pin of DDR3; Which coupling should I select in the Oscilloscope? AC or DC? And how much termination should I select in the oscilloscope? 50 ohm or 1 Megaohm ? And why?

Thank you,

\$\endgroup\$
  • \$\begingroup\$ You need to measure with a higher frequency of the signal to avoid aliasing ie getting a harmonic of the signal not hge true signal. \$\endgroup\$ – Solar Mike Jan 24 '18 at 19:49
  • \$\begingroup\$ To measure 200MHz, you have to look at the sampling rate: 2Gsp/s is a minimum, in order to get 10 samples per signal. At 1 Gsp/s you will have only five. The "MHz" rating is the limit at which the quality of the signal is still valid for proper measurement. It should therefore be higher than that, say 300MHz, but it depends on your budget and how often, and how bad you need it. IMO buy the best you can reasonably afford. \$\endgroup\$ – Fredled Jan 24 '18 at 22:53
  • \$\begingroup\$ What exactly do you need to measure? That rule of thumb might simply be a result of the fact that a x MHz clock will have harmonics at every odd multiple of x MHz – and if you don't capture at least the first three harmonics, then what you see only remotely resemble your actual signal, and you won't be able to tell much other than "yep, this signal is periodic". \$\endgroup\$ – Marcus Müller Jan 24 '18 at 23:55
  • \$\begingroup\$ I dont need to measure a certain signal now. Just, I want to learn how should I approach to the measurement of a signal. Thank you. \$\endgroup\$ – doner_t Jan 25 '18 at 21:03
2
\$\begingroup\$

When choosing the right scope and probe for high-speed measurement, first consider: signal amplitude, source impedance, rise time, and bandwidth.

If rise time is an important measurement, then both low probe capacitance or low (near matched) impedance methods will work. Then for rise time,tr the DSO BW needs to be greater or faster the your desired measurement BW=0.35/tr . This is for 10% to 90% of the rise time not 64%.

When using RC with R being driver RdsOn in Ohms and probe capacitance, RC=T can be converted to 10~90% rise time using tr=2.2T then converted to bandwidth limitation using probe capacitance. It is useful to know high speed low voltage "74ALCxx" CMOS is ~ 25 Ohms nominal and 5V logic is 50 ohms nominal.

If accurate rise time measurement is critical, then active probes are the more important than the DSO specs. http://info.tek.com/www-abcs-of-probes-primer.html

But to measure all the parameters of DDR memory, I would give serious consideration to a low cost used 8GHz Logic Analyzer with < 1pF probes such as TLA5202 with 68 channels and 125 ps resolution with variable threshold. ( market value $1500 used)

enter image description here

\$\endgroup\$
  • \$\begingroup\$ Thank you. I have dowloaded the doc. It is good. But how would termination selections effect on the measurement? For which measurement should be selected 50 ohm termination, for which measurement should be selected 1Mohm termination on the oscilloscope? \$\endgroup\$ – doner_t Jan 26 '18 at 20:50
  • \$\begingroup\$ these probes are 10M < 1pF , use that. If all you had was 15pF 10:1 probe then I would use an R divider with < 1pF then 50 Ohm termination. Load capacitance is a significant delay factor in memory on rise time. But for a power supply, definitely AC couple and 50 Ohm termination, for 1ns arc rise time , I use only 50 ohm termination \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jan 26 '18 at 21:04
1
\$\begingroup\$

Rise time/fall time is really what matters, so if you want to do things properly that's the spec to look for. All of the "rules of thumb" and bandwidth recommendations are derived from the rise time.

We have a ton of probing tips and other oscillosocpe on our Keysight Labs Youtube channel, if you're looking to learn I highly recommend it. Here's a playlist to get started: https://www.youtube.com/playlist?list=PLzHyxysSubUk48NPCudDMNEzdHF3xReke

Your input impedance should match, but in general if you're measuring over ~750 MHz you should use the 50 ohm path. You'll get higher bandwidth and lower noise.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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