11
\$\begingroup\$

I would like some help from anyone who could let me know any flaws or improvements in my design they can find and help me reconsider any last minute design fixes before I send my new project off to be mass produced. This is the second version of the design but the changes I made were extensive.

So basically I'm building an Arduino shield where the primary use case is to be a SWR meter with an LCD display (not depicted here as that is a seperate shield). But it provides all the functions of a Vector Network Analyzer (VNA) and as such goes well beyond your typical SWR meter.

Because the primary use case here is to act as an inline meter on a transmission line with an existing transmitter its not designed as a typical VNA would be using mixers. The only thing it doesn't have that a VNA would have is a function generator, as it relies on the transmitter to do that, and the directional coupler would be external. However I have designed it in a modular way so not only can the directional coupler be swapped out for one the user prefers or more suited for their setup, but it can also be configured to work more like a traditional VNA as well. With the additional of an additional shield with an in-built low power directional coupler and a sine wave generator it would be possible to also pop on this other shield and effectively have a handheld VNA instead. Being modular there are also several other possibilities for how this device can be configured including as a remote SWR meter so you can have a meter at both the transmitter and the antenna to properly calculate feedline loss, or to understand how the complex impedance of your antenna changes with frequency.

Because of its role as an in-line meter in an existing antenna system it also provides functions a traditional VNA would not, specifically the ability to analyze properties of the transmitter such as accurately determining the true RMS under modulation or precisely determining the frequency the transmitter is transmitting on. Perhaps in the future I may add other features as well either in software of hardware.

Some specifications:

  • Operates 1 MHz to 500 MHz

  • Can measure input signals from -52dBm to 0dBm (adjust external directional coupler and attenuator to handle any power transmitter).

  • Inputs are 50 ohm matched but if building yourself you can switch out different resistors to match different impedances

Note: This is an open-source project and free for anyone else to replicate my work. At some point I may sell kits and/or the printed PCB to people to make it cheaper than needing to pay to get your own PCB printed. So I'd like to make sure if I provide PCBs they have been scrutinized and tested first.

The link to the projects source can be found here for anyone who wants to access the actual files.

https://git.qoto.org/roes/roes-hardware/

Here is a picture of the schematic:

enter image description here

Here is an older picture of the UI in demo mode. The data on the screen is intentionally bogus, and the glitches in the rendering have since been fixed.

enter image description here

This is a four layer board so youll have to see the GIT repo if you want to pick apart my layout in detail. But for now I will share the front side and back side so you can at least get a sense of my placement of the chips and my use of shielding.

Front:

enter image description here

Back:

enter image description here

\$\endgroup\$
12
  • \$\begingroup\$ Minor note, you can right-click those global labels in KiCAD and choose between input, output, and bidirectional for added clarity. I'd suggest making 3pcs and selling to others at cost to gather some real-world usability data. \$\endgroup\$
    – rdtsc
    Sep 8, 2020 at 12:36
  • \$\begingroup\$ Yes I have 3 PCB prototypes coming in already and intend to do just that. I havent selected who will get the other two yet though. \$\endgroup\$ Sep 8, 2020 at 12:37
  • \$\begingroup\$ Have you built a prototype (before full run?) \$\endgroup\$
    – Andy aka
    Sep 8, 2020 at 12:41
  • 1
    \$\begingroup\$ @Andyaka Well yes and no, V1 I had printed off in small batches and tested, but it was significantly different. This version now I've soldering up prototypes of each section of the schematic independently using manhatten style construction to ensure each component works, but havent ran the whole thing together yet. I did order 3 PCB though that are on their way which I will run as prototypes first before I do a production run of course. \$\endgroup\$ Sep 8, 2020 at 12:43
  • 1
    \$\begingroup\$ @alex.forencich yes that is true, but the point is an SWR meter is still going to be effected my modulation in much the same way a complex-valued SWR meter would. So far all the early testing suggests its fairly accurate. \$\endgroup\$ Sep 13, 2021 at 23:17

1 Answer 1

1
\$\begingroup\$

A couple of thoughts. Bear in mind I am not an RF expert.

  • Your chips work at very high frequency which makes it important to decouple them. One question is about the stackup. I would advocate two GND planes on L2 and L3, very close to the surface layers L1/4 bkth of which contain signals and power routing. That way you prevent that any signal/power cross couples into any other signal/power.

  • All your chip supplies are directly connected to the power supply nodes, with some local parallel decoupling caps, but without series decoupling. Therefore, while HF currents will be drawn from the local caps, they will be also drawn from the power distribution network, including capacitors of other IC. This creates cross-coupling and can lead to parasitic C-L-C oscillations in your signal bandwidth (10s.. 100s of MHz). Both of this can be tackled by adding series decoupling to the IC supplies, i.e. a series L. This forbids high frequency currents from traveling across the power network. Only alllw sub-MHz currents there. A small chip L of about 1 uH should be fine. Better yet for your high frequencies would be a lossy chip bead.

  • The schematic would be nicer and easier to read if you made logical analog symbols for the amplifiers instead of boxes with labelled pins. Also consider a multisheet schematic. I find it rather congested.

\$\endgroup\$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.