S11 values > 0 dB are forced to be 0? (And few other questions)

[Dimitri]

Hi, Melchor!

First of all, thank you very much for this amazing unit and incredible support. SARK-110 really adds significant capabilities to electronics lab, especially amateur.

Question #1
The question is rather strange, I know. The absolute of reflection can not be higher than 1, yet it possible that due to maybe not perfect match load calibration standard,
the measured reflection can actually be higher than 1 (noise and some small matching error). So as I can get this far - the SARK replaces such S11>0dB values with
exact 0 dB (well 10^-15 or something in PLOTS, for us - exact zero).

I noticed, that when I perform an OSL calibration, the Short Load measure bars appear in red and reflection coefficient shows something like |Rho| = 0.92  and Phase(Rho) ~ -175.5 (Should be 1 and -180).
OSL Open load calibration values seems to be correct: 1.0 and -0.1. Match load values are also looks fine: 0.05 < -56 (yes, ideal is 0 and undetermined, bet hey, these values looks OK for an experiment)

So, the question is this due to SARK measuring algorithm (force values >0 to be exact zero) or maybe it is due to improper detector calibration (signal goes over the range of ADC maybe)?
Detector calibration values are the following (the path is SARK MCX -> little SARK default cable -> BNC -> Adapter to SMA -> calibration standards):

Short Load:
Voltage ~61000
Current ~116650

Open Load:
Voltage ~97000
Current ~115

Match Load:
Voltage ~75170
Current ~73050

The results of measured standards after the OSL calibration lools like this:

Short load (in the range <70 MHz the values are zeroed, phase looks OK though):



Open load (looks reasonable):



Nothing connected (a.k.a. open load with different offset length, hence non zero phase) in the region >120 MHz again these zeroed values:



Match load (looks fine to me):




Question #2

The SARK OSL calibration is performed with 400 measurement point, but then in SARK plots it is possible to set arbitrary number of points, even significantly more than there was during calivration.
How does SARK determines the values of calibration coefficients for frequency points where there are no cal data (between calibrated frequency points), interpolation?

Question #3

During OSL calibration, does SARK assume that the calibration standards are ideal and with zero offset? In other words if I have a SHORT load with non zero length from the measurement plane
(where the phase is supposed to be -180 for all frequencies) to the actual short contact, that means that SARK doesn't take into account such offset?

Thank you!

Dimitri.

P.S. Can not wait to see two-ports SARK with extended frequency range and SMA connectors! :)

[Melchor Varela]

Hi Dimitri,

I will try to answer your questions:

#1

The absolute of the reflection coefficient can be higher than 1, e.g. for active circuits. Also, the analyzer can get a reading higher than 1 because the reasons you stated.

The analyzer firmware forces S11 values >0dB to be 0dB and also negative values of Rs to be zero.

If you see the measure red bars during OSL calibration, it is likely that you will need to apply detector calibration. Nevertheless, the result of the OSL calibration should be fine even in this case.

The Voltage & Current readings for the different loads are OK.

#2

Yes, the calibration coefficients between calibrated frequency points are estimated via interpolation.

#3

Yes, the sark110 assumes ideal loads, which is a fair approximation for the supported frequency range. Anyway, I have plans to provide the capability fr specifing non-ideal cal. standards.

Best regards,

Melchor EA4FRB

--
You received this message because you are subscribed to the Google Groups "sark110" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sark110+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Dimitri]

Thank you, Melchor!

I often use various connector types, cabling etc. and thus different calibration standards. Also often I use external calibration (so, measure something first and then later calculate the proper reflection coefficient).
The problem is that sometimes I deal with very small differences of reflection coefficients which are near 0dB. The problem is that now it is not possible to get any data, I mean forcing measured reflection amplitude > 0dB to be exact 0 dB
is essentially throwing out any measurement results. Yes I know, the phase measurements are still there, but it is not enough.

Is it possible somehow to get RAW uncalibrated data without any zeroing? As for now there is no any way to make measurements near 0dB. I am not sure how many people actually dig that deep into calibration
and high sensitivity measurement, but it would be nice to have an option like "get RAW data" without any mathematical processing. Actually I deal with VNAs occasionally which may have +5dB reflection coefficient or more when uncalibrated and
it is OK, because, well, you usually know what are you doing. And of course full OSL calibration will deal with that discrepancy.

Thank you!
Dimitri.

[Melchor Varela]

Hi Dimitri,

I am considering modifying the firmware for removing the zeroing of the measured reflection coefficient. This in combination with selecting an uncalibrated OSL profile, will let you get raw uncalibrated data. I am going to investigate and I will come back you shortly.

Best regards,

Melchor EA4FRB

--

[Melchor Varela]

Hi Dimitri,

Please find the updated firmware with the removing of the zeroing of the measured reflection coeffiecient if > 0dB and also negative resistance values.
http://www.sark110.com/files/firmware

Best regards,

Melchor EA4FRB

[Dimitri]

Thank you, Melchor!

It works! Man, you work fast.
Best customer support ever.
Now I am digging into it, will conduct few experiments with various standards. Although it seems to be working fine.

Thanks a lot!