Author Topic: Array Solutions VNA-2180 not-quite-review + teardown  (Read 438 times)

0 Members and 1 Guest are viewing this topic.

Offline D Straney

  • Regular Contributor
  • *
  • Posts: 51
  • Country: us
Array Solutions VNA-2180 not-quite-review + teardown
« on: April 19, 2019, 10:40:41 am »
I'd thought the Chroma AC sources were the last bit of interesting equipment around the lab at work to take apart, but turns out there's a bit more!  So while I wait for my yellow curry to cook, hear this tale of disassembly...

This piece of gear comes not from a big name like Tek/Keysight/Keithley/etc., but from a comparatively small company called Array Solutions, who seem to be focused on ham radio.  It's a 2-port vector network analyzer which works from 5 kHz up to 180 Mhz, and has a ridiculous ~$1K price for what it does.  I haven't used it myself so can't go into too much detail, but the co-workers who do use it love it - we mostly don't use it as a 2-port VNA, but as an impedance analyzer for measuring the characteristics of inductors and capacitors across a wide frequency range.  It seems you can produce plots of L+R (or C+R, or mag. and phase), and it's great for characterizing the high-frequency behavior of custom-wound inductors or checking tolerances on capacitors.

It's got some N-type-to-BNC adapters stuck on it for our common fixtures and things:

...and here's the inside, once you remove 4 simple screws (no warranty or calibration sticker, even!):

Nothing's on the underside except a crystal and a piece of coax.

Let's get the control section out of the way first: there's an MSP430F149 running the show...

...and it communicates to the host PC through an FTDI USB-to-serial board, with a couple fast (10Mbps) 6N137 optocouplers:

The use of a pre-built daughterboard (which also simplifies the USB interfacing hardware a ton) emphasizes the "one (talented) guy in a garage" feel I get from this company.

Both ports are protected by gas discharge tubes - these definitely seem like the right choice for RF input over-voltage protection: TVS diodes have higher capacitance, as far as I know, and more importantly have very non-linear capacitances which vary with voltage, and so would create all kinds of signal integrity problems when trying to do precision measurements.  There's also what looks like a reed relay, which may disconnect the port when not in use:

Now can I actually figure out how this works?  This is the circuitry by port A:

There's 2 AD9859 DDS chips (10 bits, 400 Msps), which likely generate the measurement frequency.  These are probably fed from the 100 Mhz oscillator nearby.  My first thought was that they were doing quadrature outputs, but one of the outputs runs through the piece of coax soldered to the back side of the board over to the port B circuitry - if it was using quadrature LO signals (probably for direct conversion, mixing straight down to DC) then it would probably need the two signals at port B as well.  I think one is for the measurement frequency, and one is for the (measurement - IF) frequency...will explain a little more later.
* Chains of inductors and caps just to the left of each DDS chip are probably their reconstruction/harmonic-suppression filters.
* The PCB-trace-inductor near each one is connected to the DGND pin.  Maybe this is some kind of scheme to avoid putting Ghz-level digital noise into the analog gnd?
The OPA2677 op-amp, near the bottom, has both a high bandwidth (220 Mhz) and a high output current drive (up to 500mA according to the datasheet), so it's probably what drives the signal on port A.  It's probably followed by some kind of lumped-element transformer-less directional coupler circuit before going into the port A connector.  I'm not an RF expert, and haven't done the math out but I strongly suspect you could calculate any parameters you wanted to by putting the drive signal through a series resistor (like those two big 25.5-ohms connected in parallel) then the load, and measuring voltages on both sides of the series resistor as a kind of indirect current sense.

Either way, those two signals then probably go into the two SA612A mixers at the left.  These have oscillators built in, but I suspect the oscillators aren't used and the LO signals come from the DDS; they may be used only for the mixers as they're pretty common easy-to-find parts if I remember correctly.  Anyways, this is where I think the second DDS signal is used: the measurement signals are probably mixed with a second (measurement minus IF) frequency and then filtered to produce a down-converted IF signal.

The IF signals are handled by the analog switches and lower-speed op-amps (TLV2462 only has a 6.4 Mhz BW) above; these probably do scaling and ranging, and then hand off the IF to the MSP430.  The ADC itself must be on the MSP430, as there's no external chip and the datasheet shows it has a 12-bit one.  I would've expected basically a direct-conversion receiver, or if directly digitizing the IF to extract gain and phase, at least a faster processor (or FPGA), but I guess if the IF is low enough in frequency that sampling rate and data processing aren't a problem then it can work well.

The port B circuitry is like a receive-only version of port A.  There's another fast op-amp (the OPA2822) without the high-current output, which probably receives/buffers the port B input signal.  Then there's another SA612A mixer, with the coax bringing the LO signal from one DDS, and then a similar group of low-speed op-amps and analog switches.

I don't know how accurate my guesses are, but overall I'm kind of impressed with the design: it seems like one of those "push the hardware to its limits"/"do a lot with a little" sort of things, and I never would've expected what's effectively a low-frequency SDR back-end with an MSP430.  Seems like mostly pretty standard, cheap hardware used in creative ways...or maybe I just haven't dived into the VNA design scene enough.
The following users thanked this post: Melt-O-Tronic, capt bullshot

Offline capt bullshot

  • Super Contributor
  • ***
  • Posts: 1446
  • Country: de
    • Mostly useless stuff, but nice to have:
Re: Array Solutions VNA-2180 not-quite-review + teardown
« Reply #1 on: April 24, 2019, 04:30:30 pm »
I don't remember the name of it, but I've seen some open source hardware / HAM enthusiast built VNA / antenna analyzer that uses even less HW than this one.
The trick is basically open/short/load calibration, one can use that to mathematically eliminate all kinds of HW insufficiencies.
Safety devices hinder evolution

Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo