Verify Rev 5.2 HF Frequency Response

[AleksaBjelogrlic]

On a calibrated 5.2 unit, measure HF response at all ranges, every termination mode (50Z, 1M with and without probe), both ac/dc coupling, and all PGA filter settings

[AleksaBjelogrlic]

800mV Range (Attenuator off)

BW settings used are: FULL, 650MHz, 350MHz, 200MHz, 100MHz, 20MHz unless otherwise specified

1M

DC:

AC:

50Z

DC:

Below are comparisons of FULL, 750MHz and 650MHz

BW:

Flatness:

FULL: -7.32 - -7.90 = 0.58dB
750MHz: -7.49 - -7.90 = 0.41dB
650MHz: -7.57 - -8.06 = 0.49dB

Back to the standard settings:

AC:

1M w/ P500-010 500MHz 10x probe

DC:

FULL BW only, for reference values for each peak and dip:

Below are comparisons of FULL, 750MHz and 650MHz:

BW:

Flatness:

Back to the standard settings:

AC:

1M w/ TT-HX 312 350MHz 10x probe

Only testing DC for these next tests, it is clear that the coupling doesn't affect the HF response

FULL BW only, for reference values for each peak and dip:

Back to the standard settings:

350 MHz Flatness (this BW setting is recommended for this probe):

1M w/ TT-MX 312 250MHz 10x probe

Comparison of 350MHz and 200MHz BW limit:

1M w/ TT-LX 312 150MHz 10x probe

Comparison of 200MHz and 100MHz BW limit:

Looks like this probe functions very well to 200MHz in combination with our front end!

[AleksaBjelogrlic]

1V Range (Attenuator on)

BW settings used are: FULL, 650MHz, 350MHz, 200MHz, 100MHz, 20MHz unless otherwise specified

1M

DC:

AC:

50Z

DC:

Below are comparisons of FULL, 750MHz and 650MHz

Back to the standard settings:

Looks like the bandwidth filters will be a bit off with the attenuator on. The response looks good otherwise although it is 0.5dB-1dB different from the 1x path. We can't really bring the 1x path down to match this without changing this response likewise (the only components that effect the 1x path are shared with the 50x path). The only way we could improve this is by changing the attenuator circuit - which I would rather not do given it is so heavily constrained in every direction. This is the drawback of having an attenuator shared between 1M and 50Z mode! If the response with a probe looks good, I am willing to call the 50Z response good enough.

HF LC Dip caused by attenuator (note change of scale from 0.5 to 4dB/div):

AC:

1M w/ P500-010 500MHz 10x probe

DC:

FULL BW only, for reference values for each peak and dip:

Looks very similar to 800mV with this probe, expect the last peak (point 7) is 0.35 dB higher relative to the preceding dip (point 6)

Below are comparisons of FULL, 750MHz and 650MHz:

Back to the standard settings:

AC:

1M w/ TT-HX 312 350MHz 10x probe

Only testing DC for these next tests, it is clear that the coupling doesn't affect the HF response

FULL BW only, for reference values for each peak and dip:

Back to the standard settings:

350 MHz Flatness (this BW setting is recommended for this probe):

1M w/ TT-MX 312 250MHz 10x probe

Comparison of 350MHz and 200MHz BW limit:

1M w/ TT-LX 312 150MHz 10x probe

Comparison of 200MHz and 100MHz BW limit:

[AleksaBjelogrlic]

Gain Sweep (Attenuator On)

Observe the difference between PGA HG (black, first trace) and LG modes (blue, second trace)

0.3dB from point 1 to point 2 on HG, 0.59dB from point 3 to point 5 on LG -> HG is flatter, LG is peakier

Last trace's -3dB also 538 MHz, only 9 markers on this software for some reason

[AleksaBjelogrlic]

Gain Sweep (Attenuator Off)

Observe the difference between PGA HG (purple, first trace) and LG modes (teal, second trace)

0.27dB from point 1 to point 2 on HG, 0.46dB from point 7 to point 9 on LG -> HG is flatter, LG is peakier. The diff with attenuator off is smaller, 0.19dB between LG & HG versus 0.29dB.

The damn thing disconnected on me again and I lost all my saved traces, the cal, and frequency range settings - This is what makes people hate usb instruments and the shitty software that comes with them!

Anyway, this does give us a chance to see the peak just off screen of the above measurements (this is uncal'd so absolute accuracy is off, but you get the idea)

This is pretty gnarly, but far enough outside of our frequency range that we don't need to care (BW filters will also help)

I think we're either slamming into a rail due to offset at high gains (this thing isn't completely cal'd) or we're hitting a limit of the SNA setup - easy way to find out is to attach a 20dB attenuator to the tracking gen output and take the above measurement again.

Much better, can use the teal trace as a reference to the previous measurement. Most important thing to note is all those spikes were artifacts of gain compression or whatnot. I'm also thinking the peaking difference seen between LG and HG may be partly due to this

Here are the two highest gains used in the system with the -20dB attenuated TG output:

Looks a bit different than any of the other measurements, interesting.

Anyway, these gain sweeps are more of a spot check to see if bandwidths or flatness changed to be way out of spec, and I'm now satisfied that they will not.

[AleksaBjelogrlic]

Conclusion

Flatness looks decent across the board, can easily spec +/- 0.5dB in 50Z mode out to 350MHz. I'd like to improve it some more, but I won't let the brainworms win this time - gotta ship something and this is already quite good.

Bandwidth exceeds the rated 500MHz in all cases, allowing filters to be used to improve flatness on some settings.

I also don't see the same odd behaviour past 5MHz that I saw in #401, so that means that test passes as well. I am closing this issue with no recommended changes.

[AleksaBjelogrlic]

Reopening issue to redo testing with R1037 = 50R due to findings in #399

The main effect of this change will be to slightly reduce bandwidth, so we only need to test 50Z mode and 1M w/ 500M probe to see if we still meet the 500MHz spec

[AleksaBjelogrlic]

800mV Range (Attenuator off)

BW settings used are: FULL, 750MHz, 650MHz, 350MHz, 200MHz, 100MHz. This is changed from above since the 20MHz filter removes pretty much any variation in response. We will also not be testing AC vs. DC coupling since we have determined that there is no diffrence between the two aside from normal measurement variability (<0.05dB).

Oh and because I learned my lesson about testing the super sensitive front end circuit in it's final config (with shielding can and enclosure), here's the test setup:

1M - 5MHz - 4.4GHz

Spot check for peaking issues that might get down-converted

1M

50Z - 5MHz - 4.4GHz

Spot check for peaking issues that might get down-converted

50Z

This won't do

1M w/ P500-010 500MHz 10x probe - 5MHz - 4.4GHz

Spot check for peaking issues that might get down-converted

1M w/ P500-010 500MHz 10x probe

Not going to bother with the filters for this one

[AleksaBjelogrlic]

Ok, let's speed through the 1V range tests since we know we won't be using this exact set of values:

1V Range (Attenuator on)

BW settings used are: FULL, 750MHz, 650MHz, 350MHz, 200MHz, 100MHz.

1M - 5MHz - 4.4GHz

Spot check for peaking issues that might get down-converted

1M

50Z - 5MHz - 4.4GHz

Spot check for peaking issues that might get down-converted

50Z

1M w/ P500-010 500MHz 10x probe - 5MHz - 4.4GHz

Spot check for peaking issues that might get down-converted

1M w/ P500-010 500MHz 10x probe