Keysight MSOX3000A series, shifted trigger point

[manupthehills]

Never noticed this before:
While verifying the user calibration results I noticed that the trigger point is shifted about 150ps ahead of where it should be (measured @500MHz).
Anyone else noticing something similar on their scope?
The two channels are perfectly aligned, and showing the same behavior

[bdunham7]

Can you verify that the trigger level is correct? 

[manupthehills]

Yes, it is. Measuring a 100kHz signal with the same amplitude the waveform is perfectly crossing the 0,0 point.
The shift can be seen only when close to the scope bandwidth limit

*BTW, the scope is perfectly in specs, all characteristics have been verified*

[2N3055]

On my 3104T there is measurement T@Edge that will exactly measure position of edge in respect to trigger point.
I don't know if 3000A has it.
Anyhow, here are numbers from my scope:



Looking at that, your doesn't seem too far off from mine.
I guess a self cal might be tried to see if it aligns it better.

EDIT: source has < 45ps risetime 10/90

[MarkL]

My MSO-X3104A has a trigger offset of about 240ps when displaying a 500MHz square or sine wave, but much less of an offset when displaying a 100Mhz square wave.

Interesting.

EDIT: I should add that the equipment generating the square wave is the same in both traces (ESG-D E4433 signal gen --> ADCMP820 demo board --> scope).  The only difference is a change in frequency on the signal generator.  The ADCMP820 has a rise/fall time (80/20) of 37ps (typ.) so should not be a factor between the two traces.

[manupthehills]

Yep, that's interesting.
I could verify the same effect with a 30MHz square wave with 1ns rise time; the shift goes down from 150 to 40ps

[Bud]

The trigger channel is just another analog channel through which the signal goes to the null detector comparator before it is passed to the processor. Because the signal channels and trigger channel are not ideally matched time delay wise and because the comparator needs some headroom to switch above noise level for a stable trigger and because of the frequency passband roll-off in the trigger channel that there is a delay which manifestates as a shift on the X axis on the display. The delay can be seen if to open the scope and probe the trigger comparator output against the signal channel at ADC input.
Compensation of the delay may be difficult because it is frequency dependent in general case, as the previous poster have observed.

[TheSteve]

Looks about the same as my DIY 1 GHz model.

[Someone]

Also watch closely when changing the input attenuation! there can be step changes in trigger position.

[Bud]

Looks about the same as my DIY 1 GHz model.

It also exists on the EDUX and 1000X scopes.

[MarkL]

On an old Tek TDS3054 scope there's a more consistent offset.

[2N3055]

The trigger channel is just another analog channel through which the signal goes to the null detector comparator before it is passed to the processor. Because the signal channels and trigger channel are not ideally matched time delay wise and because the comparator needs some headroom to switch above noise level for a stable trigger and because of the frequency passband roll-off in the trigger channel that there is a delay which manifestates as a shift on the X axis on the display. The delay can be seen if to open the scope and probe the trigger comparator output against the signal channel at ADC input.
Compensation of the delay may be difficult because it is frequency dependent in general case, as the previous poster have observed.

I would say that is good assessment.

I tried on other scopes that have digital triggering.. They are order of magnitude better..

[manupthehills]

All this is interesting and shows different approaches across the industry:

- KS 3000 series: they compensate the delay with the help of some "magic" without post-processing the waveform in software which would have slowed down the update rate of the instrument (clearly one of their primary goal, and one of the main reasons to own one of these scopes). In this case is slightly overcompensated.

- Tek TDS3054: likely no compensation

- Entry level scope (DS1202Z): looks like it's software compensated since there's no visible delay (see below). The screenshot also shows why (off topic), despite the excellent price/performances ratio, these devices have to be used with the "gloves" (something shown might not be there, something there might not be shown)