Does the 1x oscilloscope probe setting slow down MHz digital signals?

Question

I was debugging what I thought was a weak LVDS driver fed with a 10MHz clock for hours when, out of other options, I've set my scope probes to x10 and although the signals were showing some ringing they looked fine otherwise.

Which should I trust, and if I should trust the x10 reading, why do the scope probes flatten the signal like that?

Here is what I am talking about:

With x1 setting:

With x10 setting:

Note1: The clock is active by groups of 8 cycles so the pause is normal.

Note2: Both probes are calibrated on the internal calibration signal of the scope.

Note3: As can be seen on the second snapshot, the coupling is DC and no bandwidth limiting is configured.

Note4: The signals are generated by the DS90LV027 and not connected to anything (not even a termination resistor). It's supposed to have less than 1ns rise time.

Note 5: Oscilloscope probes = PP510

Answer 1

Not only does the probe have lower bandwidth when in 1:1 mode but there is much greater capacitance load on the source when in 1:1 mode.

That extra capacitance will significantly lengthen the rise and fall times.

I couldn't find the probe bandwidth in the Siglent documentation but the input capacitance is about 100pF in 1:1 mode compared with about 20pF in 10:1 mode.

This is because the 10:1 attenuator is in the probe head itself so the input signal does not have to drive the cable capacitance with the full signal. In the case of the 1:1 probe the input signal has to drive the cable capacitance (20-30pF/foot) plus the input capacitance of the scope itself (usually 15-20pF).

P5510 Brochure

This reseller of the Siglent product gives the bandwidth of the 1:1 position as 6MHz in the 1:1 setting.

P5510 bandwidth in 10:1 and 1:1 setting.

I only use the 1:1 setting with small signals that need the most sensitive capability of the scope. Even there I often use a direct wired coax rather than the scope probe as it is can give better signal integrity and reduce noise coupling from other sources.

Answer 2

Kevin has told you what the issues are when you use a 1:1 probe, high capacitance. Even with a 10x probe setting, the probe capacitance is around 15 to 20 pF which is enough to slow the rise time (depends on the LVDS driver impedance or current drive). To get a better idea of what's happening, you would use an active differential FET probe, very pricey probes.

The observed rise time will be limited by the combined bandwidth of the oscilloscope and probe. You can estimate the rise time limitation of your measurement system if you assume a first order low-pass filter. $$ t_r = {0.35 \over Bandwidth} $$

The PP5510 probe is a 100 MHz rated probe, thus the probe will limit the observable rise time to 3.5 ns.
You don't mention the model your oscilloscope, however, but I'm guessing that the bandwidth is 100 MHz which gives a 3.5 ns rise time limit. If the of the scope and probes are both 100 MHz, you'll get a rise time slightly longer than 3.5 ns.

With 4-channel scopes, you'll get best performance if you use channels 1 & 3 (disable 2 & 4) or 2 & 4 (disable 1 & 3) since the ADCs are generally shared between channels 1 & 2 and 3 & 4.

How you ground your probes is something you need to be mindful of. Fast edges require making the ground lead length as small as possible to get good fidelity, something that your waveform don't have. You need to remove the clip-on probe tip and run a ground connection between the ground ring on the probe to circuit ground.

Answer 3

Yes, it does. If you have a probe with 10x/1x selection, there is usually no reason to use the 1x mode due to bandwidth being absurdly small, unless you know you must use it.