Building a 22 GHz network analyzer for under $1000

[arlo_g]

Looks like a fun project and you have some great parts to work with!

Your photo of the inside of an HP directional coupler is quite interesting. It looks like yours is similar to the 779D couplers. They claimed in the HP journal articles about those that the “arrowhead” metal triangles was important for directivity.     

I think that most high performance VNA’s  since the HP 8510 have used directional bridges for frequencies below mm wave.  Directional bridges can be built with coaxial baluns to deliver 1000:1 frequency ranges, while getting even a 50:1 frequency range with good directivity and uniform coupling out of a directional coupler is pretty heroic.   The Julius Botka “triaxial bridge” patents have a lot of good detail about the very high performance 8510 bridges.  Henrik Forsten seemed to get good performance to 6GHz from his directional bridge build, but I think that he hadn’t fully enclosed them, and higher frequencies are likely a challenge on several fronts.  Since you have nice directional couplers already, it makes perfect sense to use them.

If you wanted to bootstrap your way to a calibration kit, then a coax LRL calibration would be a good place to start.  A single “line” length (assuming you don’t count your 0 length through as a line) only gives a good calibration over an 8:1 frequency range though, so multiple lines, or hybrid schemes like LRM would be needed to cover your full range. 

[EggertEnjoyer123]

My plan in the future is to use two different paths, with one going to a high frequency directional coupler and one going into the classic return loss bridge with balun. An RF switch could be used to switch between the two depending on the frequency, which would let you cover both frequencies in the low megahertz and mmwave frequencies.

[EggertEnjoyer123]

I finally got the S11 measurement to mostly work. (For some reason my code doesn't work when I try sweeping from 5 - 22 GHz, but it does when I try sweeping from 18 - 22GHz. It also breaks if I try doing more than 201 points).

Calibration was done with the cheap calkit from the LiteVNA. I'm pretty sure it's not good at 22 GHz.

The setup is mostly similar to what I did before, except I'm using some random splitter in my collection to get the reference signal. I'm pretty sure the performance of the splitter isn't critical since any errors will just be calibrated out. The antistatic bag helps increase isolation between the two oscillator boards (I found in my previous tests that adding the bag increased isolation by at least 10dB).

I calibrated everything using the set, then I took off the calibration sets and remeasured them to determine repeatability. It looks like we're doing better than 0.5dB. I also measured a shorted transmission line, which seems correct (we should expect the line to go around the Smith Chart). The shorted transmission line is just two SMA barrels and the short from the cal kit.

I also measured a 10dB attenuator, with the other end open and terminated.

Finally, I verified that my setup was indeed sweeping from 18 - 22 GHz. The amplitude in the picture is a lot lower than expected since the signal had to go through a lot of coax to make it to the spectrum analyzer.

[wilhe_jo]

I can't really contribute too much but maybe this could get you a big step forward: https://github.com/scott-guthridge/libvna

13GHz would be a reasonable compromise for be.... That's right above where you can get troubles with 2.4G ISM radio.

But I tend to look into TDR for this because CML Logic seems fast enough for the "heavy lifting".

However, if you need some testing... I have a receiver that can do 26.5GHz and some anechoic room good for double digit GHz is coming soon.

Regards

[EggertEnjoyer123]

I fixed one of the software bugs and now I can sweep from 3GHz to 23GHz. Even though my coupler is rated for 12.4 - 18 GHz, there seems to be enough directivity left at lower and higher frequencies for my system to work correctly. The amount of signal coupled out definitely changes, but that doesn't matter and only the directivity matters.

Here are the new plots from 3 - 23 GHz. I also measured the return loss of a 6.2 GHz lowpass filter. I think I need better calibration standards since the LiteVNA ones suck at high frequency.

I also plotted the return loss of a 17.7 - 19.7 GHz isolator, and a WR18 waveguide adapter.

(In case anyone is wondering, I basically programmed my board to act as a LiteVNA, and I am using the NanoVNASaver application to display my result).

[EggertEnjoyer123]

I now have S21 working. An RF switch is used to switch between measuring S11 and S21.

The isolation sucks (maybe -30dB at some frequencies), but that is probably because I have no shielding. All my measurements have zero averaging so far. My C program takes 32 samples at 48 kHz (the IF is 12kHz) and just adds/subtracts every other sample to get the sine and cosine amplitude.

I'm also not sure why the S21 measurement isn't accurate below 6 GHz. I think it might be some weird programming error, because the points seem to be in the same place between sweeps if a 10dB attenuator is added, but without the 10dB attenuator the points jump around. It could also be that I'm saturating my ADC or mixer.

Here are my measurements for a 10dB attenuator, a 17.7 - 19.7 GHz isolator (both directions), a 6.2GHz lowpass filter, and a directional coupler which seems to be a 2-8 GHz coupler with -16dB coupling according to my LiteVNA (coupling and isolation were measured). The isolation is also shown. I wonder if it's possible to build a good RF shield using conductive filament.

[EggertEnjoyer123]

I tried wrapping everything in the tinfoil bags that you get when you buy SMD components. That didn't really help much.

The interesting thing is that the isolation goes up with averaging, so there is some random aspect. Averaging 10 times seems to improve the isolation by around 10dB (the two plots are from an uncalibrated setup, so you can't compare them with my previous results, and the actual isolation is around 5-10dB worse). You can see that the peak goes down, but at other frequencies the isolation is the same, which is what you would expect if the leakage path is deterministic. (If you have a 70dB attenuator connected between the ports, no matter how much averaging you do you will always get 70dB). I've ruled out the issue being the RF switch (which is only rated to 20 GHz) and the mixer leakage, because the problem remained even after disconnecting the reference signal and the reflected signal. It could be the power rails potentially, or maybe the control voltage for the amplifier. Touching the pins with my fingers had no effect though.

The isolation also "increases" after 22 GHz, but that's only because I run out of power to drive the mixers, so the conversion loss increases by 20dB.

[EggertEnjoyer123]

It seems like to get TDR to work, I need to have low frequency data, which I can't do right now without having to splice together S2P files from the LiteVNA.

In any case the software I'm using (NanoVNASaver) does not calculate the TDR correctly. I have submitted a pull request to fix the issue: https://github.com/NanoVNA-Saver/nanovna-saver/pull/715

The software takes the raw S11 data, windows it, and then does an inverse FFT to get the impulse respose from the S11 data. The resulting signal is complex, while the impulse response should be real, so the developer used the absolute value function to make it real. This is the wrong way to do it (for starters, a short and open would look exactly the same, since their S11 values are just negatives of each other, so if you take the absolute value you will get the same result). The correct way is to take the S11 data and append the conjugate of the measured data. Basically, for real signals, the FFT must be symmetric (in that the negative frequencies must be the complex conjugate of the positive frequencies). After adding terms to the end so that the negative frequencies and positive frequencies are conjugates of each other, if you take the inverse FFT, the signal is real, and you get the correct impulse response.

I probably need to design a better VNA which has a copy of the LiteVNA circuit, and some RF switches to switch between the high band (>2 GHz, which I have now) and low band (50kHz - 2 GHz). That way I can do TDR.

[EggertEnjoyer123]

I also have a major issue with thermals. At high frequencies I run out of power to drive the mixers, and as a result the conversion loss is highly dependent on the LO amplitude. As everything heats up it seems like the output amplitude goes down. The difference between hot and cold is quite a bit (like about 5dB), but it only seems to be an issue above 21 GHz.

I calibrated my setup, and then power cycled everything a few times. As you can see, for all frequencies below 21 GHz, nothing bad seems to happen. However, since after power cycling the board is cooler, the amount of LO power going to the mixer is higher, which means the conversion loss is lower. If the board is completely cold the peak reaches about 6dB, and drops below 0dB after the board is warmed up more. All frequencies below 21 GHz remain the same though since there is enough LO power to make the conversion loss roughly constant.

[joeqsmith]

...
The software takes the raw S11 data, windows it, and then does an inverse FFT to get the impulse respose from the S11 data.
...

From Brian and CMT, they use chirp-Z transform:
https://www.eevblog.com/forum/rf-microwave/vna-for-cable-characterization/msg5656049/#msg5656049

[EggertEnjoyer123]

The software I'm using doesn't seem to implement that. I read through the code for the TDR measurements and it just does the IFFT. Nothing fancy.

Anyways I swapped the two mixers and now I seem to have >50dB isolation across the entire band. I'm not exactly sure why this happened. The blue line is what I get when I connect a cable, and the brown line is with nothing connected. The difference should be the isolation. There is a peak in the blue and brown lines near 22 GHz but that is just because the mixers are running out of LO power, and one of them seems to be significantly better at lower amplitude than the other (probably because they're using different diodes). In my next design I'll probably pay $30 more and buy higher power amplifiers.

I'm not sure what's causing the remainder of the error. If I disconnect the ADC the noise floor is around -90 dB, so it's not the ADC. It's not the RF relay either.

Edit: my IF bandwidth is 1.3 kHz. Here is the frequency response of my "filter" (which is just correlating the sequences 1, 0, -1, 0, 1, 0, -1, 0, ... and 0, 1, 0, -1, 0, 1, 0, -1, ... with the received signal at 48 ksps). The IF frequency is 12 kHz.

[smaultre]

Why don't you try to buy some used HP? VNA 26-40GHz acquisition frontend?
And add some modern, generators, ADC, & PC-based software?

[EggertEnjoyer123]

Why don't you try to buy some used HP? VNA 26-40GHz acquisition frontend?
And add some modern, generators, ADC, & PC-based software?

Getting the S parameter test set for 22 GHz will probably set me back at least $500, and I'd still need to buy the PLL ICs which make up most of my cost right now. (Of course I get to save one IC because the HP test sets use sampling as opposed to mixing, but the cost difference is still quite a bit). The advantage of buying the HP test equipment is that I won't have to worry about shielding that much.

Anyways I seem to have found the main source of error, which is the splitter. It turns out that the performance of the splitter does matter and can't be "calibrated out." The splitter is way too lossy at 20 GHz, and unfortunately I don't have any that would work at that frequency. My signal processing also has issues, because if I measure the isolation with an oscilloscope I get about a 9dB improvement.

Here is the isolation with a lower IF bandwidth of 20 Hz. The filter frequency response for the IF is attached. The "peak" is at 20.93 GHz, where the isolation is only 50 dB.

I swept from 20 - 20.01 GHz and I got an isolation of 70dB. Using an oscilloscope, I got 81 dB (the 12kHz peak had a power of -56.4 dBm with port 1 and 2 disconnected and with 50dB gain, and the 12kHz peak had a power of -24.9 dBm when port 1 and 2 were conneected).

At 20.93 - 20.94 GHz, the isolation is under 50dB, but with the scope I got 58dB isolation. The power with port 1 and 2 disconnected was -46.4 dBm (with 50 dB gain), but the power with ports 1 and 2 connected dropped to -38.1 dBm. Because the power drops so significantly with both ports connected, and the isolation doesn't change that much (it was quite hard to measure because the amplitude of it is so small), I suspect the splitter to be the main culprit. Theoretically, the amount of leakage should be the same, as both the source and LO oscillator output the same amount of power at 20 and 20.94 GHz. However, the amount of received power with the two ports connected is significantly lower, so if we look at the difference, we will see a loss of isolation even though the leakage is the same.

I think the reason why there's a ~10dB difference between the measurements from the ADC and from the scope are because of low frequencies. My digital IF filter does not filter out low frequencies that well, and looking at the oscilloscope readings there is a huge amount of low frequencies compared to the amount of 12kHz signal. (This is because the DC offset of the IF changes every time the frequency is changed, causing huge amounts of ringing). With a better digital filter, I should be able to eliminate this issue. (The oscilloscope capture shows the IF output with nothing connected at 20.93 GHz).

[Gerhard_dk4xp]

Is there a reason why you ignore
<   https://www.analog.com/media/en/technical-documentation/data-sheets/adl5960.pdf   >
from post #3?

bidirectional couplers, 6 GHz synth input with /2 ... *4 frequency gearbox, mixers,
stepped if gain & if filters, interface to frequency offset osc, all controllable via SPI,
sync-able to others for more than 2 ports, all on one chip??

[EggertEnjoyer123]

Is there a reason why you ignore
<   https://www.analog.com/media/en/technical-documentation/data-sheets/adl5960.pdf   >
from post #3?

bidirectional couplers, 6 GHz synth input with /2 ... *4 frequency gearbox, mixers,
stepped if gain & if filters, interface to frequency offset osc, all controllable via SPI,
sync-able to others for more than 2 ports, all on one chip??

I have a few reasons for not using the chip.

1) The part costs $150. While it seems like I would be able to save money, this is not actually true. This is because the IC only does S11 measurements, and I need to do both S11 and S21 measurements. The IC does not have a source oscillator, so I still need to buy a LMX2820, and unless I buy another IC to do the S21 measurements (which brings my BOM cost to $300), I would have to buy another LMX2820 to generate an offset frequency to mix the received signal on the second port. Right now, my entire design cost me about 200 dollars in total (not counting extra parts I ordered). The two LMX2820s were $50 each, and I have 6 RF amplifiers at $5 each. Everything else on the board is <$5 and all the RF modules were bought for about $5 each from the ham radio swap meet.

2) The part doesn't allow for much experimentation. I'd like to eventually see if I can do harmonic mixing, which is used by the original NanoVNA and the newer LiteVNA to extend the frequency range at the cost of decreased performance. If I can get it to work on my design, then I might be able to increase the range to 40 GHz (that's a big if though). With my current design I can just swap out the directional bridge to achieve that, but I can't do that if I use the chip, since it runs out of directivity fast after 20 GHz.

3) The part does not support 12kHz IF frequency, so I'd probably have to choose a more expensive ADC instead of a cheap audio codec, and I'd need to do more data processing in the microcontroller, which means more programming (which I really don't want to do).

There are several advantages though.

1) With 2 ADL5960s, you can get all 4 S parameters without needing a transfer switch. A normal RF switch can be used to switch the source oscillator between the two ports. The downside is that you lose some dynamic range with S21 measurements. No matter if you use the ADL5960 or a mixer for getting the S21, you need to put an attenuator in front. Otherwise, the return loss of the mixer/chip would ruin your measurements because port 2 would not look like 50 ohms. For example, the return loss of an amplifier is highly dependent on the output impedance. The difference is that there is an additional loss due to the coupler inside. So if we look at the signal path before going to the mixer, in the current design there would be 10dB attenuation before the signal hits the mixer. In the ADL design there would be 10dB attenuation, but then the signal is not at the mixer yet. It has to go through the coupler, which adds another 10dB or so loss. Analog Devices does not share what the coupling loss is, but we can estimate because the insertion loss is -2dB. Assuming everything is perfect, the coupling has to be below 20 log (1 - 10^-0.2) which gets you -8.6dB coupling. However, this is assuming everything is ideal, and the actual number is probably 10.

2) You can get measurements faster with 2 ADL5960s. Right now I only have 2 mixers, but I have three signals: reference, reflected, and transmitted. Having 2 ADL5960s would let you downconvert all three signals at the same time.

3) You can do low frequency. My coupler runs out of coupling at megahertz frequencies, so I can't really get any data below 1-2 GHz.

Anyways, if someone hired me to design a mass producible 20 GHz VNA, I would probably use the chip. But I'm just doing this at home for fun.

[EggertEnjoyer123]

I did more research into the ADL5960, and I think my current design has superior performance when it comes to measuring S11 at least.

I read the entire datasheet and the reference design datasheet. You'd think that a multi-billion dollar company would be able to provide plots showing them measuring a decent 50 ohm calibration load with the chip, but that doesn't seem to be anywhere in the datasheet at all.

Anyways I'm basing my comparison on the following slideshow: https://www.analog.com/media/en/training-seminars/seminar-materials/01-20-ghz-small-form-factor-multiport-network-analyzer.pdf

On slide 26, we can see their measurements of (presumably) a 9.5 - 11.5 GHz filter. You can see that their S11 trace has a lot of noise at 20 GHz. Their graph is zoomed out a lot, but it appears to be around +/- 2dB. I'm getting pretty much zero noise to 20 GHz even with no averaging (see post 33). Of course, this might not be a fair comparison since I'm measuring a short that I calibrated the VNA with, while they're measuring a filter (which we don't know the actual S11 of). However, the fact that the points change so rapidly between close frequencies and the fact that it looks "random" as opposed to being smooth seems to point to this being a noise issue, and not an issue with the device they're measuring. If you look at post 30, where I measure a filter, the S11 curve seems smooth and while the S11 is obviously not correct (thanks to my bad calibration set), there seems to be no random noise.

For S21 I don't think a fair comparison can be made. A picture of their setup is on page 16, and you can see that they have no shielding. So their S21 is not the best that it could be with proper design.

I think my setup can also be improved more, but I can't find a good power divider at all in my collection. It would ideally be resistive and go up to a decent frequency (like 18 GHz)

[tszaboo]

Is there a reason why you ignore
<   https://www.analog.com/media/en/technical-documentation/data-sheets/adl5960.pdf   >
from post #3?

bidirectional couplers, 6 GHz synth input with /2 ... *4 frequency gearbox, mixers,
stepped if gain & if filters, interface to frequency offset osc, all controllable via SPI,
sync-able to others for more than 2 ports, all on one chip??

I have a few reasons for not using the chip.

1) The part costs $150. While it seems like I would be able to save money, this is not actually true. This is because the IC only does S11 measurements, and I need to do both S11 and S21 measurements. The IC does not have a source oscillator, so I still need to buy a LMX2820, and unless I buy another IC to do the S21 measurements (which brings my BOM cost to $300), I would have to buy another LMX2820 to generate an offset frequency to mix the received signal on the second port. Right now, my entire design cost me about 200 dollars in total (not counting extra parts I ordered). The two LMX2820s were $50 each, and I have 6 RF amplifiers at $5 each. Everything else on the board is <$5 and all the RF modules were bought for about $5 each from the ham radio swap meet.

2) The part doesn't allow for much experimentation. I'd like to eventually see if I can do harmonic mixing, which is used by the original NanoVNA and the newer LiteVNA to extend the frequency range at the cost of decreased performance. If I can get it to work on my design, then I might be able to increase the range to 40 GHz (that's a big if though). With my current design I can just swap out the directional bridge to achieve that, but I can't do that if I use the chip, since it runs out of directivity fast after 20 GHz.

3) The part does not support 12kHz IF frequency, so I'd probably have to choose a more expensive ADC instead of a cheap audio codec, and I'd need to do more data processing in the microcontroller, which means more programming (which I really don't want to do).

There are several advantages though.

1) With 2 ADL5960s, you can get all 4 S parameters without needing a transfer switch. A normal RF switch can be used to switch the source oscillator between the two ports. The downside is that you lose some dynamic range with S21 measurements. No matter if you use the ADL5960 or a mixer for getting the S21, you need to put an attenuator in front. Otherwise, the return loss of the mixer/chip would ruin your measurements because port 2 would not look like 50 ohms. For example, the return loss of an amplifier is highly dependent on the output impedance. The difference is that there is an additional loss due to the coupler inside. So if we look at the signal path before going to the mixer, in the current design there would be 10dB attenuation before the signal hits the mixer. In the ADL design there would be 10dB attenuation, but then the signal is not at the mixer yet. It has to go through the coupler, which adds another 10dB or so loss. Analog Devices does not share what the coupling loss is, but we can estimate because the insertion loss is -2dB. Assuming everything is perfect, the coupling has to be below 20 log (1 - 10^-0.2) which gets you -8.6dB coupling. However, this is assuming everything is ideal, and the actual number is probably 10.

2) You can get measurements faster with 2 ADL5960s. Right now I only have 2 mixers, but I have three signals: reference, reflected, and transmitted. Having 2 ADL5960s would let you downconvert all three signals at the same time.

3) You can do low frequency. My coupler runs out of coupling at megahertz frequencies, so I can't really get any data below 1-2 GHz.

Anyways, if someone hired me to design a mass producible 20 GHz VNA, I would probably use the chip. But I'm just doing this at home for fun.

I know that the nano and the liteVNA does this, but you really shouldn't use an audio CODEC for these sort of measurements. The DC/gain parameters of them is not characterized, or guaranteed. Same with INL or DNL. An industrial ADC will have all these parameters characterized and it will have a reference voltage, either built in or external. The reference voltage is something that has the same (meaning 1=1) importance in an ADC as the input. The audio codecs will only have this internal, with questionable accuracy. ADCs that will do 24 bit and 50-100KSPS are below 5 EUR in cost.

The ADL5960 is very interesting, but indeed you need one on every port of a VNA. I'm very tempted to attempt to build a VNA with it, but maybe after I'm done with the current projects that I'm doing.

[EggertEnjoyer123]

I bought some SMA calibration loads which are similar to the ones from the LiteVNA and sent them to my friend, who has a 26.5GHz VNA in his lab. Here is the link:
https://www.aliexpress.us/item/3256805982526182.html

It seems like the load is worthless (the seller claims 1.02 VSWR from 1M-6G which is laughable), and I should probably get my money back because the S11 is at -10dB at 6 GHz. The short and open seem to be good according to my friend's VNA, and there seems to be very little variation between two different opens and two different shorts, so you could probably use them for calibration up to 26.5 GHz.

The offset delay for the short is approximately 0.077 ps. The open seems to be nonlinear, but I'm sure someone here can help me get the C0, C1, C2, C3, and offset delay from the data.

[EggertEnjoyer123]

Here's my calculations for the coefficients for the open.

I got the following numbers:

C0 = 2.377218773138113e-15
C1 = -3.77944625e-46
C2 = 7.90828757e-36
C3 = -4.21470232e-48

Offset: 1.4244367406724633e-12 s

My Jupyter notebook is attached if anyone wants to check my work. I also attached plots of the regression vs actual data. My code sucks and all it does is try out all the offset values, and then it does a 3rd order regression, calculates MMSE, and chooses the offset which gets the lowest MMSE. I found out that low frequency data also has to be eliminated during the process.

Apparently the cal standards for the Anritsu 22N50 are these values:

Open: offset 17.83mm, C0=4e-15 F, C1=200e-27 F/Hz, C2=0 F/Hz^2, C3=1.1e-45 F/Hz^3
Short: offset 17.83mm, L0=0, L1=0, L2=0, L3=0

Looks like this agrees with what I found, namely that the short is perfect and the open has some fringing capacitance.

[EggertEnjoyer123]

It looks like everything works now and I get good results. I had to experimentally determine the capacitance of my 50 ohm load though, by hooking up a short transmission line and manually adjusting it until the ripples were small (and were equal on both low and high frequency). After that, I tested the filter and I was surprised to see that the S11 line was completely flat, which means that my calibration worked. I think the program has a bug where it would randomly write "50" to C0, but it doesn't actually put in 50 (otherwise my results would be completely different)

I used the same numbers from the previous post. Here are the short, open, and load after measuring with the calibration standards set. As you can see, they are all reasonable and match the results that my friend obtained before. The load is also somewhat reasonable (the VSWR is 1.32 at 18 GHz). This small difference is enough to throw the measurements off by 3dB though.

I guess good open and short standards exist and are fairly easy to find. Getting a good load might be extremely tricky though, and not everyone has a friend who works in a microwave lab.

[EggertEnjoyer123]

It looks like the software is buggy and randomly replaces my C0 value with 50 fF. Going to have to correct that one and then pull request it on Github...

You can see the difference between the actually measured open and my "calibrated" open. I bet the 55fF of capacitance goes away when I correct for this.

Edit: I changed it back to 2.377 fF and I deleted the 55 fF of capacitance from the load. As you can see everything works now, and I didn't even have to guess any values! The load I'm using is from SV Microwave, I found it in the dumpster at school and it has a blue dielectric instead of the standard white one. It seemed to have the lowest amount of reflections so I picked it to use for calibration.

It looks like this: https://octopart.com/sf8018-6005-amphenol+sv+microwave-30479673?r=sp but brass instead of silver. Anyways, for the price of $20 you can buy a <1.05 VSWR load to 18 GHz. Quite impressive...

Edit again: I missed the +0.01 * f part. Whoops. A 1.1 VSWR load is fairly inexpensive though: https://www.centricrf.com/terminations-loads/sma-terminations/sma-terminations-loads-6-27ghz-1-2-watts/c18s6-sma-miniature-male-termination-1watt-18ghz-vswr-1-10-s-steel/ and it's probably good enough for government work.

Also these are $15 each and judging by the typical performance data they should be much better than 1.1 VSWR: https://www.fairviewmicrowave.com/product/rf-terminations/coaxial-rf-terminations/sma-terminations/rf-load-1-watts-18-ghz-sma-male-st1827.html?srsltid=AfmBOooyBVU_AtmYB_hiXh7Weh9wFROYSK9GgAz4zkoQmdc0K1jBYl3j

[EggertEnjoyer123]

It looks like none of the Chinese loads I have in my possession can be used anywhere close to 18 GHz, when compared to my "good" load.

All of them have return loss under 10 dB at 18 GHz. The return loss of the load that came with my LiteVNA looks like the attached picture.

The only good way to get a decent calibration set to mmWave frequencies is probably by buying from a reputable company. I'll probably have to order a few of these to test soon: https://www.fairviewmicrowave.com/content/dam/infinite-electronics/product-assets/fairview-microwave/product-datasheets/ST1827.pdf

Also I've changed my mind on the ADL5960, and I think it's probably not as bad as the graphs in the presentation show. The design that Analog Devices chose to use in their slide deck (https://www.analog.com/media/en/training-seminars/seminar-materials/01-20-ghz-small-form-factor-multiport-network-analyzer.pdf) is laughably bad. The total cost of those parts is probably well into the thousands too when a decent VNA using their chips could be made for $500.  (I guess they didn't want to use TI PLL chips, but they have their own PLL chips and they could have just selected two of them instead of using the current overcomplicated setup). It looks like the LO and the source oscillator are at the same frequency, since the output of the upconverter they're using is differential so both the LO path and the source path must have the same frequency. This is stupid design because any leakage from the LO to the source will mess up your measurements. You can try to calibrate it out, but it's going to be impossible to get any good dynamic range out of it. I'm also not sure how they're getting frequencies below 6 GHz if their upconverter is only rated for 5.9 GHz. Anyways, some Keysight VNAs use integrated circuits to replace the coupler/mixer/etc, so I think the chip has potential. I'm honestly not sure why their application engineers never thought about making an actual well-designed VNA using their chip if they're going to spend millions developing it.

[Gerhard_dk4xp]

Your synthesizer price seems quite optimistic when compared to
an official distributor. When creating new things I can't prove
anything when using parts from unknown heritage.  1 hour loss
and the advantage is gone.

They only need a 6 GHz synthesizer: A 0.5 divider and a
times 2 and times 4 multiplier are included, and equal mixers,
and mixer LO drivers for the entire range.

IIRC someone in a neighbor tread declared an ADF**** as
king of the hill with regard to phase noise at 22 GHz, so they
probably don't need TI.


I must stop now. Ressurected from the Death after 3 days.
Probably Covid. Still don't dare to go downstairs.

Gerhard

[EggertEnjoyer123]

Your synthesizer price seems quite optimistic when compared to
an official distributor. When creating new things I can't prove
anything when using parts from unknown heritage.  1 hour loss
and the advantage is gone.

The LMX2820 is $54 on LCSC though, and I've already purchased three with no issues.

They only need a 6 GHz synthesizer: A 0.5 divider and a
times 2 and times 4 multiplier are included, and equal mixers,
and mixer LO drivers for the entire range.

The ADL5960 still requires an external source oscillator to generate the 0.01 - 20 GHz. If you look at the block diagram of the chip, there is nothing connected to RFIN. You have to set up your own oscillator and feed the output into the chip.

Also, an 8GHz LO frequency is still required, because the doubler works from 2-8 GHz and the quadrupler works from 4-8 GHz. In 4x mode the minimum frequency is 16 GHz, so if you want to set the LO to 14 GHz you must put in 7 GHz while having the LO circuit on x2. The LMX2594 is $21 on LCSC though and goes up to 15 GHz, so it's not really a problem.

IIRC someone in a neighbor tread declared an ADF**** as
king of the hill with regard to phase noise at 22 GHz, so they
probably don't need TI.

All of the chips have roughly the same jitter at 6 GHz (around 35 - 40ps). Not sure what the jitter is like at 22 GHz but given that all the chips use multipliers to get there I don't think there is much of a difference.

Anyways I'm now convinced that the ADL5960 isn't as bad as I first thought it was, and that the bad performance is probably due to their designers being bad. At least there isn't really anything fundamentally wrong with the chip based on the datasheet. I think I'll try using it in my next attempt since my current design would require the use of an RF switch to select between a low and high frequency path, and this chip doesn't. Without a low frequency path, it would be cumbersome to do TDR measurements.