VNA test results

[Joe]

 Since the latest version Hermes-Lite board supports duplex operation

I downloaded and tried both the Ham VNA and  Hermes VNA software

and just messing with the programs the following observations.

Both program will run the current board and had no problems

discovering it. I used a homemade directional coupler along with my Short-open-load for calabration.

both programs do a reasonable job displaying reflection and Smith charts as only

about 40db of range is required however when running THRU measurements I found the displayed

range was just slightly better tan 40db. I'm certain that the AD9866 can do better than this

and I don't know if it's a software limitation or cross talk due the the board layout.

Also keep in mind that the MAX frequency is 30MHZ.

The above is my finding others may have different results 

[John Laur]

This is mostly conjecture based on my understanding, but to get better dynamic range we would have to sweep each bin in multiple passes. The first pass would be with the preamp disengaged; then a second pass could be made with the preamp engaged for the frequency ranges where it can be safely engaged without causing ADC clipping. A calibration step to characterize the preamp itself would also be necessary. So I'm thinking it's a software problem, albeit making the algorithm much more complicated than we might initially suspect. Maybe more sophisticated software already can do this.

I agree with you though that we could probably see better than 40dB of dynamic range without having to resort to trickery with the preamp. I think steve was seeing about 52dB, so I do not know what is going on with your setup. Perhaps this is a limitation of the drive power being unable to fully saturate the ADC (due again to the existing software being written for the capabilities of the hermes/angelia/orion hardware). It may also have something to do with the bit errors that you and Steve identified in the first release of the full duplex code. I would have expected we could get closer to 70dB by default, and perhaps a lot more from process gain even without employing the preamp. Since we have an easily accessed onboard attenuator and preamp, I would think the Hermes-Lite may eventually be more capable than the Hermes as a VNA if the software is extended.

It looks like HamVNA runs either with the "attenuator" disabled or optionally enabled at the user's request. It seems to run with drive fixed at 0, and it does not adjust the step attenuator value that I can see. There is some kind of funny interaction with the attenuator/preamp/atten value that I do not understand. It detects the ADC overload condition reported by the RTL should it occur.

I am not sure whether these baselines used by HamVNA are appropriate for Hermes-Lite. It appears that in VNA mode the hermes-lite drive controls are remapped by the RTL to limit the maximum Tx power to about 15mW, but with drive=0 perhaps we are still under-driving.  It also appears that in VNA mode the RTL sets some limits set on the preamp; maybe to prevent possible damage?

Here is where HamVNA initializes the Hermes. It sets attenuator=off, preamp=on, vna=on, drive=0

https://github.com/VE3NEA/HamVNA/blob/aee5a1ee8adc5fd534b549da98e71cf3b00db293/VnaCli.pas#L185-L186

Here is where HamVNA enables the attenuator value when the checkbox is marked.

https://github.com/VE3NEA/HamVNA/blob/aee5a1ee8adc5fd534b549da98e71cf3b00db293/VnaCli.pas#L316-L317

I finally had time to build my 1.22 this weekend, so I am looking forward to experimenting with the VNA now as well. First step is probably to see if I can get the tools to compile HamVNA and then dust off the part of my brain that hopefully still holds some Pascal syntax. Is any of  this Delphi stuff freely available?

73, John K5IT

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

[Joe]

John,

I used a HP355D 10db switchable attenuator for the test. below -40db the VNA didn't make 10db steps

although it would to -50. I'm uncertain if the THRU CAL function can calibrate out any cross

talk that maybe occurring due to proximity  of the traces and transformers.

I have a DG8SAQ VNA which I use as my comparison. I will measure the TX signal

level and post.

Joe   wa9cgz

[John Laur]

Ah ok; well that kind of thing is a bit beyond my pay grade. I have seen reports when used to make reflection measurements with Hermes that calibrating at the end of a transmission line the antenna measurement fairly accurately remove the effects of the line.. So I would be fairly confident that the results will be good enough at removing the effects from the PCB. Obviously anything between the ADC/DAC and the DUT could compromise the instrument's dynamic range even if you can calibrate to delete its effect from the measurement.

I have not looked up the specs of the AD9866 receiver. Steve did not say whether he just ported the code and kept the sasme limits as used on Hermes or made specific adjustments yet for the Hermes Lite hardware. I assume maybe a little of both.

73, John K5IT

[Joe]

After a few more test  using Hermes VNA and checking the 20 db attenuator

box the average range now -60db and at lower frequencies somewhat better approaching 

-75db but getting worse as the frequency is increased. By removing the input and output

terminations the cross talk looks like almost -80 db but once the TX (output) is terminated

the cross talk is about -60db. I also notice a small glitch at about 2.7MHZ I'm not

sure of the cause maybe core type (I'm using 61. All the above is using the Hermes VNA program

which has thru transmission calibration as far as I can determine the HAM VNA doesn't 

have a this option for thru calibration just reflection.

I measured the TX output  power at about +2dbm.

Joe   wa9cgz 

[John Laur]

OK This looks closer to spec; the datasheet lists Tx/Rx isolation (with IOUTG pins) at 77dB with attenuator at 0db so its likely we cant do better without switching the vna mode entirely to the IOUTP. Still that is fair performance so far.

Does the absolute Tx power decrease as frequency goes up? Is the loss in dynamic range at the upper end the fault of the Tx or Rx side?

73, John K5IT

[Steve Haynal]

Hi Joe,

Thanks for all your testing and feedback. This is very useful and will help with v2.0. Did you recompile to use the 61MHz oscillator in full-duplex?

I have noticed a bit of noise floor increase on the higher frequencies too for some boards. Haven't tracked down exactly why yet.

I don't think the power output is at maximum now for the VNA mode but varies across the spectrum. Ideally, we should be putting out the the maximum acceptable by the RX at every frequency without clipping. Now, and like the Hermes, the power output is fixed across the sweep. A more sophisticated VNA would adjust the power output so this is true across the spectrum. And as discussed earlier, we want to do several sweeps with the RX LNA at different levels so that we increase the dynamic range. So much to do and so little time!

There are some VNA detail on the software wiki page.

73,

Steve

KF7O