开云体育

Well, read that as "easy to prototype" ;) Right now I can easily assemble 5 prototypes in an hour or so (after JLC has put on all the passives that is) and as a design that we also might use internally and make variants of, this is important because some variants might only have 10 units built in total. Prototype assembly is otherwise very expensive.

Hey Pablo and Hey Larry,

It’s really cool that you two have an eye on our conversation ?
And thanks for your opinions! I wish you both a Happy New Year~

Hey Gabriel,

Thanks for giving me your development data! It’s always exciting to see someone else’s design style. ?

As a common advice (to everyone ?): If you want to filter your power supply, I highly recommend to not use those standard 100nF Caps! Yes, it looks like everyone is using them (and unfortunately those still get recommended in the datasheets). But at a RF-view they far from optimum.
The purpose of those Caps is to filter noise thus acting like a pass through at low frequency (DC) and shorting higher frequencies to GND as much as possible. At most frequencies these Caps are above their self-resonance frequency (SRF). That’s why you should use a SMD ceramic capacitor with as many internal layers as possible at your desired size to lower parasitics like ESR and ESL, for example, a 4u7 at 0402 or 10u at 0603. In plain theory they are not recommended at high frequencies due to their SRF. But if you measure them with a precise VNA you’ll find out that due to the internal layer stacking the parasitic inductance (ESL) is very low as many layers connected in parallel. Therefore, the capacitors impedance (even as it acts as an inductor) stays quite low at higher frequencies (several GHz) and obtain a very neat noise filter characteristic!
Oh, and please put those bypass caps as near as possible to the desired IC supply pin. In fact, not all ICs need to have the bypass cap directly next to their power supply pin, but for some especially RF-ICs this is essential!

1) Absolutely! This SMA connector style you are using has a really bad performance at higher frequencies. It’s not just the stub giving you a parasitic inductivity also the discontinuity inside the bended connector affects the port match (at least if you don’t want to pay some extra to have a perfectly matched bending)… But unfortunately, it’s the only connector type you can use as you need to maintain a really (really!) good mechanical resilience and a long bulkhead. (But just between the two of us: most of the high end VNAs do not even reach a 20dB return loss at their ports either…)

Concerning the RF input cap: Your DGS seems to be fine (just by looking at it). Am I right assuming the Cap can endure 6.3V (max 10V?)

2) Yeah, I already thought this might be the reason why. As JLC is not able to put those IC on the PCB you need to do it on your own. But even so you should put critical passive components at the same side as the ICs. (e.g. some bypass caps or this 50Ohm E-Cal match!) ? It nearly adds no extra time to the population process

RF-switches: That’s why those Infineon switches produce a negative voltage level internally~ Please check them out as you can also use them easily up to several GHz, which might come handy in further versions/designs ;D

3) Don’t you think the cut out might be a bit too much? :’D
And I had a look again at JLC’s part library. The 68Ohm 0402 is available now:

So maybe you can change the size and get rid of the DGS.

As you want to use only ‘standard’ components those 50Ohms resistors have typically a bad impedance behaviour at higher frequencies due to their high parasitic inductance… At least for the E-cal and reference 50Ohm match you should think about replacing those single resistors with 2x100 Ohm resistors parallel +DGS. (At high end application they even try to force their 50Ohm reference to be a good match by adding attenuators in front of it, as those internal used SMD resistors are most of the time not as good as the external calibration match is ?)
And once more: I highly recommend to put at least the E-cal match resistor directly next to the switch pad, as this would improve your internal calibration and add a bit more stability to your system!
4) Oh yeah, the reference match…! I truly remember the same experience meanwhile I developed my directional module ?.
To obtain a good performance at (very) high frequency it is necessary to mirror the signal pathway. Not only due to the not ideal match but also due to the uncertainties within the trace (and if you want to refer the internal match to the external calibration load /DUT also due to the mismatch given by the DC-Block).
But I do admit that inside your frequency range the given mismatches/discontinuities shouldn’t be to high and may just appear at higher frequencies (>2 GHz), thus the gain and phase shift might be not affecting the system so much.

5) If you catch the wrong balun type, the yield can be awful! ?
That’s the reason why so many devices out there need to be tuned before they get delivered to the costumer… As you can not afford this in this application a stable balun with a good yield is the rise and fall of your whole application.
Give those TC1-1-13M+ a try. As far as I know from my and others designs they should obtain a fairly good performance!
I also did a simulation with your resistor values. But instead of a 50Ohm and 0.3pF I used 2 times 50Ohm to balance the baluns and adjusted the tuning R&C! Look how good it turned out ?
6) If you use ferrites for debugging please be careful at 1GHz and above, as ferrites permeability is decreased significantly, and they start to act dielectric.
7) Having U301 switched to port 2 there is also a signal at the trace to U2 reduced by U301s isolation value. The datasheet gives an isolation between 15 to 20dB at 3GHz (with perfect termination!). The signal then gets reflected at U2 and propagates trough U301 back to your mixer with a total attenuation of roughly 25 to 30dB, whereas signals which got coupled into the trace between U301 and U2 are only attenuated by 15dB!

If you want to, I would suggest to change U2 into an low noise amplifier. For example BGA2800, as this LNA just needs 3 caps to be up and running. It is also pretty cheap and concerning your application you can surely run this amplifier up to 4GHz! ?
The amplifier can be used exactly the same way as the switch by turning it on and off. The isolation between U301 and U1 is better than 20dB when turned off. In addition, it would amplify your couplers output signal, hence improving your SNR. This comes in handy as your LO signal from mixer is just 27dB damped (Isolation between RF-IN and LO Port) and interferes with your S11-signal. Furthermore, thanks to LNA’s output match with better 20dB, the mixers does not see any bad impedance match coming from the baluns!
8) To add some stability to your system it would be good, if you would add some DC-Block Caps between mixer and BB-Amplifier (like those 10u 0603) depending on your lowest IF frequency.

Hopefully this helps you to solve your problems soon~
I wish you all the best! ?

-Melanie

Wow, Fantastic discussion. I, and I'm sure may others, enjoy reading these "RF" design topics. A lot of first hand experience and know how from practical designs feed back.
Thanks a lot for sharing this.
Jean

Hi,
I agree with Jean and Larry - keep going with this discussion - as for me it already helped much in my pcb design ;)
Mel: THANK YOU for the tips and explanation, this is going to be very neat analyzer!

--
Slawek/SP9BSL

Hi Gabriel,

As an experiment, you could try to make a more symmetrical balun by placing the core vertically over the pcb and making sure the wires have equal length. Another request, could you publish the schematics in pdf? I'd like to study them and give some feedback. I don't use gEDA, I stick with KiCAD, Pulsonix, Ultiboard and Altium.
Happy 2020!, Reinier

Op 6-1-2020 om 08:40 schreef Gabriel Tenma White:

はぁ (ha), that did the trick. I did the surgery on only one leg so there is still some imbalance, but the directivity improved from 8dB to 16dB, on par with the performance with tuning (see attached). Don't think I'll ever convince TOKO or mini-circuits to build baluns vertically like this though, and it won't fit in the shield can anymore, so I think the tuning will stay for now.

Here's the pdf:

Good to hear that better symmetry is a solution. The problem with tuning as you do now is that it is frequency dependent. Maybe you can try to increase the pcb trace length to compensate for the length difference? You'd probably need to build a specific test pcb for that. I'll study the schematics and get back to you. I'm developing ideas to make a V3 like you (full 2-port, 6 GHz at least, filtered output signals, adjustable signal levels, >100 dB dynamic range, programmable DC bias on RF ports to investigate, 7 inch display, etc.) Lack of time is the main problem. My V3 would be more performance optimized than cost optimized.

Op 6-1-2020 om 11:21 schreef Gabriel Tenma White:

Hi Gabriel,
MiniCkts and Toko will build whatever you want - it all boils down to how much you're willing to pay...
...and yes, for a short manufacturing run like this, it's not worth it.

Unless...

If you can show them how the directivity is improved, maybe they will come out with a standardized part - who knows.

Just a thought.
Great work there !
Regards,
Larry

i found the jog wheel fine to use,just a badly made piece of crap tho!,would be fine if better made!

That is the screen i fitted to mine,works a treat,quite cheap as well.

Because of popular demand we replaced the jog switch with buttons (see attached).
Yes these are through-hole so should be able to take some force.
(Also note the SMA connector on the actual product will be longer than in the picture, in order to accommodate enclosures.)

Mel, thanks for the great help with this project (and to Gabriel-san, of course).
You should change your username on groups.io to Vignette Tsukinose April.
Sorry about the inside joke, at least you can google it :)

73, Mike AF7KR

Hey Gabriel and hey everyone,

First: Thanks to all of you and of course Gabriel in particular for this really nice and worthwhile discussion! I hope that we can gather some more experiences and ideas. ?

Concerning the caps: compared to 100n or 470nF caps those 4u7 have a lower resonance point and hence are able to filter lower frequencies too. So in total, they are really broadband!

Frankly I never had such radiation problems concerning the ‘normal’ SMA connector type. That’s why I’m a bit astonished that you had one at your design as 4cm distance between 2 SMA connectors (mid to mid) gave me one roughly 90dB isolation. Maybe your connectors are a bit closer…

Funny that we nearly had the same ideas! You surely did a good and detailed prototyping, wow! ?
But concerning the caps between mixer and BB amp: maybe a smaller cap may do the trick? A smaller cap is faster at its charging and discharging but still blocks DC.

I’m sorry that those TC-1-13Ms do not solve your problems. If your looking for a transformer having a compensation for its asymmetry within the component may you can have a look at those:
MABA 011082 or TCM1-63ax. Unfortunately, the S-Parameters are just starting from 5MHz and 1MHz. But within the balun simulation they still had a nice performance at those low frequencies. Of course, they might be a bit more expansive but as they are built up symmetrically you may not need to build in two but just one inside your directive coupler…
May I ask why the OIP3 is important to you? In my opinion to ensure a good performance the system needs to run at the linear region of the amplifier so values like P1dB might be more important than OIP3…?

And Wow to your plans building up a ‘proper’ VNA! That’s amazing! I wish you all the best for it ? But programming (modelling) a FPGA is totally different then a microcontroller. I thought the new SMT32 generation have some specials like higher clock and special filter setup internally etc so maybe those might be already fast enough to shift the IF frequency into a proper value.

As you’ve asked at which project I needed to work on an directional coupler: I did my master thesis at Rohde & Schwarz at their VNA department and developed a directional module from 9kHz to 20GHz (actually it had been more than 30GHz at the end) using PCB material. ?
At the moment I’m just a normal RF-Engineer with several different projects from schematic to layout design and from simulations to antenna setups. But I’m far from being able to lead a whole ne VNA development :’)
Although, maybe I might build up a small directive coupler if I have some spare time just to do some measurements on my one !

And ones more: Thank you very much for sharing your knowledge with us! ?

Hello,

I'm new here on the list, been a nanovna user for a couple of months now, otherwise a long time VNA user, from HP8410 times onwards.

Just to add my 1.9999 cents...

I find nanoVNA really great. It's main advantage for me is that it is a self contaiend unit. There have been previous amateur VNA constructions, but you always had to connect a computer, a power supply, a coupler, mess of cables, etc.

The small size is also great, it is in fact the first device that lets me measure the antenna match on a handheld device - no need to stick ferrites on a cable or other funny contraptions. In my younger years, I often had to lug something like an HP8753 up a telecoms tower, and then the mains cable drum... Even an FieldFox or SiteMaster is big and heavy when you are climbing a precarious ladder.

So regarding display size - maybe if I live to be 90, I will have a problem reading it. But now, at sixty, I can read it very well, without any problems. Also no problem using the touchscreen with a fingernail. So if you make a bigger screen version, keep making a small display (like current) version too.

SMA connectors - some people have expressed concern about their lifetime. However, that lifetime is specified for every time fastening with a torque wrench, for full performance at 18GHz. Tightening by hand (most of the time) and for 3GHz, the lifetime will be AT LEAST 10X more cycles. Also, you can put those small male/female extenders on, to square the lifetime. Negligible at 3GHz. BNC have no repeatablity and N are HUGE (in nanoVNA terms), so I vote for SMA to stay.

My suggestion would be: keep everything as-is, just extend the electrical parameters, like frequency range, dynamic range, directivity...
And please do not litter this little gem with bluetooth or similar crap. A RF measuring instrument works much better withouta built-in QRM source!

Marko Cebokli

A RF measuring instrument works much better withouta built-in QRM source!

The power inverter, microcontroller, USB and display are already RF sources.
BlueTooth should energize only between sweeps.

Tried a TCM1-63AX (had these in stock), it gave really strange S11 graphs and not good directivity. Might investigate it further.

Crosstalk between microstrips or grounded CPW is low in theory, but my experience is it turns to crap as soon as you have nearby metal objects to reflect the radiated signals, or put the board in a metal enclosure. The LCD is mounted right above the PCB so the whole thing forms a nice waveguide for leakage to travel. The remaining bits of leakage seen in the pictures is still due to radiation (maybe not the SMA connectors but shield can leakage) because I can affect it by putting my hand near the board. Past designs didn't achieve good system dynamic range even with shield cans because of the remaining leakage from the SMA connector center pin, so switching to this style of connector (and having the connector footprint in the shield can) was the only way to fix it.

The receiver linearity is important because nonlinearity causes errors that can't be removed by calibration. For example the IAM-81008 mixer has P1dB(in) of -15dBm and IP3(in) of -6dBm, but if you operate at -25dBm (which is 10dB below compression) your third order error power is -6 - (-6 - -25)*3 = -63dBm, which is 38dB below the signal. That's a EVM (error vector magnitude) of 1.25% which is just on the edge of being acceptable. Nonlinearity doesn't just generate harmonics, it also causes amplitude/phase error in the fundamental signal. The rule of thumb is at least 20dB below IP3, and also at least 10dB below P1dB. The way to check for linearity error is to measure a short length of low loss coax (after calibration) and check that it circles the smith chart as expected. I think there was a thread here that showed errors in the current Nano in this setup because of the low IP3 mixer (SA612) used. I'll try the BGA616 for the gain block which has good enough IP3 and P1dB.

I've done FPGA based VNAs before at a different company and I find it much easier to deal with than a microcontroller. All timings are deterministic, and you can coordinate things to happen at cycle accurate times with respect to the reference clock. Spartan 6 starts at $4, so as soon as the required MCU gets close to that price I'd just switch to the FPGA :) Optimal IF frequency is somewhere between 1 to 5 MHz (based on ADF435x noise skirt).

I'd like to see your coupler design; can you post the title of your paper?

On Thu, Jan 9, 2020 at 01:13 AM, Gabriel Tenma White wrote:

Crosstalk between microstrips or grounded CPW is low in theory, but my
experience is it turns to crap as soon as you have nearby metal objects to
reflect the radiated signals, or put the board in a metal enclosure. The LCD
is mounted right above the PCB so the whole thing forms a nice waveguide for
leakage to travel.

Hi Gabriel,
I found that simple ESD foam acts as quite good EMI absorber, give it a try and insert some of it between/above the cans.

I've done FPGA based VNAs before at a different company and I find it much
easier to deal with than a microcontroller. All timings are deterministic, and
you can coordinate things to happen at cycle accurate times with respect to
the reference clock. Spartan 6 starts at $4, so as soon as the required MCU
gets close to that price I'd just switch to the FPGA :) Optimal IF frequency
is somewhere between 1 to 5 MHz (based on ADF435x noise skirt).

what type of ADC do you plan to use?


--
Regards,

Slawek/SP9BSL

I can confirm that carbonized foam is quite effective at UHF and above...

73

Glen K4KV

The power inverter, microcontroller, USB and display are already RF sources.
BlueTooth should energize only between sweeps.

These are not intentional sources, just some weak unintended radiation.
Bluetooth has a *real transmitter* at 2.4 GHz, where the new nano is likely to be used.
Besides slowing down the measurement, the other side of the link does not know when "between sweeps" is.

Marko Cebokli