Building a 22 GHz network analyzer for under $1000

[EggertEnjoyer123]

Good day!

Hopefully someone else can explain (I suspect that the ferrite is probably special).

Minicircuits recommends using a high permeability ferrite. I would guess that Marki used something similar to B62152A0015X030. Quote from one patent (dual output autotransformer from 30 MHz to 3000 MHz or greater):

has a binocular core, preferably made from a high permeability ferrite material. The preferred core dimensions are approximately 0.096″ high×0.080″ wide. The wire used for the windings is two strand twisted 34 gauge and three strand twisted 36 gauge, both preferably 7 twists per inch

The length of the LO/RF twisted pairs should be slightly shorter than a quarter wavelength of the mid-range, adjusted for propagation speed. The characteristic impedance of the LO/RF twisted pairs should be about 100 ohms, since they are connected in parallel on the port side.

two resistors (there is one hiding behind the rings

Can you show in more detail the location of these resistors, what points they are connected to and the values?

The design is attached to this post. The resistors measure 22 ohms on a multimeter.

[EggertEnjoyer123]

Also I recently noticed this: https://hforsten.com/designing-a-low-cost-high-performance-10-mhz-15-ghz-vector-network-analyzer.html

It looks like his bridge has low directivity though. If we look at the graph titled "Uncorrected S-parameters with short on port 1 and load on port 2 with EM gasket. 10 Hz IF bandwidth." we can see that the red and black lines cross, which implies negative directivity.

I am able to get better results using the same design, but with 01005 resistors (I hate soldering those and I wasted so many) and 1mm rigid coax. My unoptimized results are attached to the post. The first picture shows the reflected output for a short and a load (you can see there is about 6-7dB difference). The second shows the insertion loss. My directivity sucks (and my return loss) but it is still positive up to 22 GHz which is promising. The coupling falls off at low frequencies because I had a soldering mistake where two conductors were very close but weren't actually soldered. I forgot to take screenshots but even after fixing the soldering, nothing changed at high frequency. Only the low frequency directivity got better. My return loss is also horrible and this bridge would probably be unusable because of that. One of the connectors is also missing, and that's because I intentionally did not solder it (otherwise it would be impossible to resolder the resistors). During testing I pressed it on. I have another bridge with the connector soldered, and it made zero difference.

I think I can optimize the design to use fewer 01005 resistors, and also have a better return loss. I am currently trying to improve the design by simulating in OpenEMS.

[rloc]

The design is attached to this post. The resistors measure 22 ohms on a multimeter.

Thank you! This is amazing, I will definitely check it out.

A long time ago I received a ZNB20/40 for repair with a burnt bridge. The ferrite material is 10K. The embodied thin-film resistors on a substrate had to be replaced with 01005.
https://patents.google.com/patent/US9823284B2/en

[rloc]

The DBM mixer in the high-frequency branch looks like a Giga-tronics 30037 with klopfenstein taper on the back of the PCB.

[EggertEnjoyer123]

I wonder why they seem to have made no attempt to match the broadband capacitor (since the pad of the DC blocking capacitor is a lot larger than the trace width, which means some matching should be needed to keep the return loss low).

I'm talking about the capacitor right below to the uppermost conical inductor on the first image you sent. It wouldn't have taken that much effort in an EM simulator to match the large pad to the thin microstrip trace. (Unless I'm missing a cutout on the bottom ground plane).

Interestingly the Keysight VNAs seem to not care about matching too... I attached their specs below.

[rloc]

Unless I'm missing a cutout on the bottom ground plane

A ground plane continuity issue seems like the most likely scenario.