NanoVNA V2 / Forum

Note: this page is a mirror of https://groups.io/g/NanoVNAV2.
Click here to join and see most recent posts.

Port 2 Return Loss mod


martin.thornber 2021/04/20 05:53

This may or may not be of interest (first post here too).

Intro - just picked up a NanoVNA V2 which looks like the discontinued "2.8 V2 with metal case" version going by the official website.

It came from Ebay so am not sure of whether it is a clone etc. but seems to function fine and I updated to the recommended firmware (20201013).

Whilst evaluating it the first few days, I calibrated port1 to the end of a supplied SMA cable using the supplied standards and had a look at the full range return loss of the second port - it was only about 10dB or even slightly worse at 3GHz, so not quite meeting spec for some reason.

As the unit has a wonderful low cost, I have no qualms about tinkering with it! Looking at the schematic I saw at least one thing to investigate - the input series capacitor C8. The value of this capacitor in the schematic at 10uF is heavily skewed to low frequency operation and at any RF frequency over several MHz will be well beyond its self-resonance frequency and will effectively be an inductor thereby potentially degrading return loss - decided to place a low value capacitor in parallel with a larger value. Also I decided to move the pi-attenuator (3 resistors) closer to the SMA connector pin as this then somewhat isolates the external port from the capacitor. So in the end I swapped the location of the pi-attenuator with the capacitance. The resistors were re-used but the capacitor changed to a 2.2uF 0402 in parallel with a 100pF 0402.

*So Summary of Changes:*
C8 changed to the series resistor 120ohms
75ohm 0402 Resistors moved from R10 & R11 to either side of the 120ohm resistor (location C8), soldered to one of the SMA connector ground pads.
R9 then became a 2.2uF 0402 capacitor with an 0402 100pF cap piggy backed on top.
As I'm not sure of whether or not the switch input required a DC path at the input I also put a 511ohm 0402 resistor to ground at location R11

*Result* - port 2 return loss now >14dB at <=3GHz (though when I get access to a big VNA at work I'll try to confirm rather than self test by a cal using the supplied reference standards)

Other comments - I measured the resistor values (correct) and also checked the switch components were the correct type - ok. I also put a 100pF 0402 cap piggy backed upon C1 and C306 while I was in there (these are also 10uF capacitors in the RF paths). Also note I didn't do an exhaustive search of parallel capacitor values to use (lack of time so far) - it could be that 22pF, for sake of example, in parallel with the uF cap may give better results still since this gives even lower impedance at GHz frequencies than 100pF..

Apologies if this has been covered elsewhere already!

OwO 2021/04/20 21:44

The clones got the PCB stackup and materials wrong, and have bad port 2
return loss: https://groups.io/g/nanovna-users/topic/76231435

Original V2 Plus4 devices have close to 20dB typical port 2 return loss
up to 3GHz (15dB minimum).

Changing resistors and capacitors like that can work, just beware that
there is now a DC path to ground and DC should be avoided on port 2
after the mod.

OwO 2021/04/20 21:55

Yes a special high frequency capacitor is used to get good response from
50kHz to >3GHz, but your idea of stacking a small capacitor on the big
capacitor is a good one and works too.

martin.thornber 2021/04/20 07:31

Thanks for the replies - the clone's PCB stack being incorrect sounds like a plausible explanation for some of my poor return loss - given the pi-attenuator with the resistors (14dB loss by design = 28dB return loss min in theory at least) the 10dB return loss measured was quite surprisingly bad! Given that my return loss is still only ~15dB@3GHz with the 3 resistors almost right at the connector also means that there's still a small degradation due to the SMA connector thru-pin or SMA ground - perhaps the ground on the top/inner layers of the clone PCB is too close to the SMA-pin for example.

If your design used a specific broadband 10uF capacitor choice, maybe the cloners decided to save some $$ and fitted a different 10uf?

The capacitor changes I made helped me this time, though sometimes capacitors in parallel like this can interact badly and cause unwanted resonances, but this didn't happen on this occasion.

As for DC on the port - I usually test passive items or devices with a DC block already, but I will have to remember what I've done :)

This is the slightly messy mod:

rfclown 2021/04/20 15:54

I have the same experience as you. I guess I also have a clone, S-A-A-V2. I thought I was buying an authorized model, but I guess not. It came with very old firmware, and the newest firmware that will run correctly is 20200926.

I measured S11 of port two with a E5071C. I had planned on doing a similar mod as you did, when I get around to it. I planned on putting the attenuator just like you have it. The difference is that I was going to put the DC block between the attenuator and the SMA pin. I would remove that narrow trace and solder the cap (or paralleled caps) right at the SMA center pin (and cut the pin back).

The reason I put the project aside is that I was very disappointed that the software does not allow you to read corrected S parameters, just raw data. The only way to read corrected data is to use a program like nano-vnasaver. And when I plug a USB to my unit, I get crap in my high frequency measurements around 3.88 GHz (see picture with and without USB connected).

With an attenuator on port 2, it seems like I can make decent measurements if I cal it correctly. For open I use no standard, for short and load the supplied pieces are ok, and I would never use a "through" piece, just make sure the two reference planes are mating (and have an attenuator at the port 2 side of the plane). For the price, this seems fantastic. Too bad I can't read the good data since connecting USB screws it up. So either I have to figure how to fix the hardware, or I have to change firmware to do what i want. I didn't buy this to be a project, since I already have VNAs, but to have a handy portable unit.

n2msqrp 2021/04/20 21:10

I want to try the attenuator addition on my SAA2N. Is a 6dB attenuator
adequate or do I need a larger value?



Mike N2MS

> On 04/20/2021 6:54 PM rfclown@gmail.com wrote:

>

>
>

>
>

> I have the same experience as you. I guess I also have a clone, S-A-A-V2. I
thought I was buying an authorized model, but I guess not. It came with very
old firmware, and the newest firmware that will run correctly is 20200926.
>
> I measured S11 of port two with a E5071C. I had planned on doing a similar
mod as you did, when I get around to it. I planned on putting the attenuator
just like you have it. The difference is that I was going to put the DC block
between the attenuator and the SMA pin. I would remove that narrow trace and
solder the cap (or paralleled caps) right at the SMA center pin (and cut the
pin back).
>
> The reason I put the project aside is that I was very disappointed that the
software does not allow you to read corrected S parameters, just raw data. The
only way to read corrected data is to use a program like nano-vnasaver. And
when I plug a USB to my unit, I get crap in my high frequency measurements
around 3.88 GHz (see picture with and without USB connected).
>
> With an attenuator on port 2, it seems like I can make decent measurements
if I cal it correctly. For open I use no standard, for short and load the
supplied pieces are ok, and I would never use a "through" piece, just make
sure the two reference planes are mating (and have an attenuator at the port 2
side of the plane). For the price, this seems fantastic. ,_._,_

_._,_._,_

* * *

rfclown 2021/04/20 18:32

6 dB is what I initially grabbed for my measurements. It's a trade off. What ever you choose directly subtracts (dB for dB) from the dynamic range of the S21 measurement. If you don't need much dynamic range on the S21 measurement, the bigger the better. It seems that the cal is assuming a perfect 50 ohms on port 2. The higher the attenuation value, the less the deviation from 50 ohms if port 2 isn't 50 ohms (unless it's a crappy attenuator that isn't 50 ohms).

rfclown 2021/04/20 18:39

Also, I'd do:

port 1 - DUT - attenuator - cable - port 2

where DUT is "device under test". that way if the cable isn't great, the attenuator helps with that mismatch also. An attenuator is usually better over a wide frequency range (>3 GHz) than a cheap cable. Putting in the attenuator subtracts from the S21 dynamic range independent of where it is inserted, so might as well put it where it hopefully will do the most benefit.

n2msqrp 2021/04/20 22:45

Good point. Thanks



Mike N2MS

> On 04/20/2021 9:39 PM rfclown@gmail.com wrote:

>

>
>

>
>

> Also, I'd do:
>
> port 1 - DUT - attenuator - cable - port 2
>
> where DUT is "device under test". that way if the cable isn't great, the
attenuator helps with that mismatch also. An attenuator is usually better over
a wide frequency range (>3 GHz) than a cheap cable. Putting in the attenuator
subtracts from the S21 dynamic range independent of where it is inserted, so
might as well put it where it hopefully will do the most benefit.

_._,_._,_

* * *

Stephen Laurence 2021/04/21 07:02

Ladies and Gentlemen,

My ancient brain is getting a bit confused. I can understand the Port2 14db attenuator helping get an impedance match. I can understand a degree of complex impedance upset and leakage of unwanted signals might be due to the poor board layout near the sma connector.

But why put the attenuator ahead of the signal leakage, which is giving such poor dynamic range above 3ghz? The complex impedance upset does not degrade the actual dynamic range, or does it? Further, the sweep to 4.4ghz is a tad out of range of the specifications.

I must admit, however, that my second ssa2N bought 2 months ago has a dynamic range of 50db at 4ghz, without averaging; noticeably better than my first N, and my “Black and Gold”.

Is there any improvement to be gained to put the attenuator right up to the sma connector in a satisfactory vna?

Steve L

martin.thornber 2021/04/21 12:10

"Is there any improvement to be gained to put the attenuator right up to the sma connector in a satisfactory vna?"

If it's already satisfactory then maybe no meaningful improvement for you!

I only moved the 14dB attenuator (which is 14dB by design, not by me) closer to the SMA connector to see if it made the port return loss better (it did) - this may be useful if testing say an amplifier that is spec'd to run into 50 ohms - 10dB return loss can be quite a long way from 50 ohms as i'm sure you know..

I didn't move it to improve performance >3GHz (I didn't even know my unit could go >3GHz until I tried it!).

Stephen Laurence 2021/04/21 14:16

Thankyou.

I actually have four v2 devices (original 2.8”, two V2N and a black and gold) and they are all ok. The most recent addition two months ago has remarkable dynamic range even up to 4.4ghz -it is not a plus4. In the early days, I did consider changing that 14db attenuator down to 4db (a potential 10db increase of dynamic range), but OwO messaged to say that it was to reduce signal overload, rather than improving impedance matching.

I would be interested in other ways to improve them however. I have draped the first three with ferrite plates (cheap on Ebay and lucky to find them) on both sides of the boards which improved things a bit. I now have a flexible ferrite emi sheet I am going to try. However my eyes and hands may not be up to sma soldering, but I might try it on the 2.8” original.

I have however started to add a few additional decoupling capacitors around the board as a starter for ten, if only to reduce the phase “noise” at low levels.

Steve L

Erich Kaltenbrunner 2021/04/22 02:31

Hi Steve, you seem to buy a lot of things and solder SMA, I do so as well.
One thing I bought some years ago was a stereo microscope with 4x magnification, this allows me to solder SMA down to 0402 quite easily and 0201 with some effort, I’m 72.
Erich

Stephen Laurence 2021/04/21 23:32

Dear Erich,

I wish I had a binocular one. I bought one of the Chinese “microscopes” with a screen you look at. I have used it for all sorts of things (especially serial numbers on the back of iphones etc), but not for soldering smd yet.

One other useful thing I got (from China, in the good old days when were were in the EU) is a soldering iron in the form of heated tweezers. Both tips are heated and you pick up, position the component and solder it with the chip between the tweezers. I have not used it yet but I lent it to a friend who used it in anger to good effect.

I also have a hot air gun, but using it on circuit boards........ It is used to open Ipads and to heat-shrink insulation sleeves.

However, the first smd component to install is a smd esd diode on every input (including the Tinysa).

Steve L

To reply to this topic, join https://groups.io/g/NanoVNAV2

View this thread on groups.io