First RF design, anything wrong with it?

[Daveedius]

Hi everyone. I'm designing an outdoor enclosure for my Helium miner and have designed this board to replace a traditional coax lightning arrestor (in order to reduce insertion loss).
It will go inside the enclosure along with the rest of the circuitry. This is my first RF board, so I want to make sure I'm doing everything right. Here's what I've done:

- The RF trace, a coplanar waveguide, has a designed impedance of 50ohm per Saturn PCB's PCB toolkit V8.06
- I've included a matching network (determine values with a VNA, correct?)
- Ground connections do not have thermal relief
- 4 layer FR-4 board, with a continuous ground plane under the signal layer
 
A few questions:
- Any pointers, recommendations, or notice anything I'm missing?
- I've never seen a lightning arrestor in this exact configuration, any comments on that?

[Joel_Dunsmore]

It's a bit unusual to hang anything off the RF line in the opposite way of the RF signal path.  Your gas-discharge is like some capacitance at the end of a stub. So you would have to model that bit in parallel with your antenna as you apply a matching network design.  Did you simulate the gas-discharge parasitic effect when doing your matching network?

[T3sl4co1l]

Just where is this thing sited, that it's subject to direct lightning?  And why can't the radio be inside a solid metal box, keeping it inside from lightning EMP, common mode, etc.?

Tim

[Ruthenium]

Just where is this thing sited, that it's subject to direct lightning?  And why can't the radio be inside a solid metal box, keeping it inside from lightning EMP, common mode, etc.?

Tim

The antenna need to be outside the box.   Won't work very well inside a shielded solid metal box. 

Seriously,  I would just sweep the gas discharge device and see what it looks like at 915 Mhz.  No need for ultra expensive test gear,  just a NanoVNA will give you a pretty good idea.

Most commerical coaxial lightning arrestors use Gas Discharge Tubes,  so I see nothing wrong with your layout, just need to check what capacitance the GDT is adding.

[Bud]

What are you trying to solve by moving the gas discharge lamp from an inline surge arrestor to this stub configuration? It is the same gas lamp, what insertion loss improvement does your design provide? You better use a standard inline coax surge protection device as they are mechanically constructed to handle significant surge currents.  The  thin  PCB trace from the antenna connector to the lamp in your design in case of an unfortunate event will evaporate before the lamp will finish arcing and a portion of the surge will toast the matching circuit and possibly the front end of the receiver. My recommendation is to stay with proved conventional technology.

[Ruthenium]

Just a few comments on the design,  you probably don't need the matching network, the LoRa transceiver should already be 50 Ohms and likewise the antenna should be 50 Ohms.

I'm curious as to why you are separating the ground on the antenna coax from earth,  I would just connect direct to earth you don't need to isolate from the earth connection.  But maybe I'm missing some other aspect of your installation. 

I'd also shorten up the traces to the GDT,   you don't want to accidentally end up with a 1/4 wave line to an open circuit on your antenna input.  LOL.

Also, there are SMD GDT's that might be worth looking at.

[neilhao]

Maybe quarter wave transmission line could be used for lightning protection.
Just connect the 915Mhz 1/4 wave line between antenna and the  earth, all frequency except band 915Mhz will be discharged to the earth. This method almost has no cost but PCB.

I'd also shorten up the traces to the GDT,   you don't want to accidentally end up with a 1/4 wave line to an open circuit on your antenna input.  LOL.

[Daveedius]

Thank you every one so much for responding - I've learned a lot from your responses. Didn't even cross my mind that the GDT was creating a stub. I'll make some revisions before sending it off to the fab house

[fourfathom]

I also wonder about the shield / ground disconnect.  If you get a nearby lightning strike your coax shield is likely to see just as much of a spike as the center conductor.  I'm not really familiar with the use of GD tubes, but might you need a second-layer of protection?  Those GD tubes trigger at around 70V (or more, I think) so you might need an additional limiter.  Or is the remaining spike brief enough that after it gets through the front-end tuned circuits there isn't enough energy left to cause a problem?

[Daveedius]

Went ahead & redesigned the board. Here's the summary:
- Earth & GND have been combined into a single ground plane
- GDT has been switched to surface mount & moved in order to remove the stub
- Added a second U.FL connector & a normally unpopulated zero ohm jumper so I can bypass the matching network if necessary

I also wonder about the shield / ground disconnect.  If you get a nearby lightning strike your coax shield is likely to see just as much of a spike as the center conductor.  I'm not really familiar with the use of GD tubes, but might you need a second-layer of protection?  Those GD tubes trigger at around 70V (or more, I think) so you might need an additional limiter.  Or is the remaining spike brief enough that after it gets through the front-end tuned circuits there isn't enough energy left to cause a problem?

Bourns makes a TBU chip for secondary protection but I didn't use it in this design because I'm trying to minimize insertion loss as much as possible. I found this pdf to be really helpful:
https://www.bourns.com/docs/technical-documents/technical-library/gas-discharge-tubes/publications/bourns_gdt_short_form.pdf

[gristleking]

Rad little project, and I love the geek factor here, but...is there really a need to avoid the insertion loss?  Even the cheap ones don't add much loss (.5 dB), and that loss doesn't appear to affect any earning outcomes.  Is there a specific reason for this?

Again, I think this is a super cool project and appreciate the effort!

[Daveedius]

Rad little project, and I love the geek factor here, but...is there really a need to avoid the insertion loss?  Even the cheap ones don't add much loss (.5 dB), and that loss doesn't appear to affect any earning outcomes.  Is there a specific reason for this?

Again, I think this is a super cool project and appreciate the effort!

Thanks man, big fan of you and your site; have spent many hours on there going through your posts. There's a few reasons, some driven by the fact that I'm gonna try sell the enclosures soon:

1. Cheaper, simpler, and I just think it's a lot neater overall.
2. To be honest, I'm not entirely sure if there is a need to avoid the insertion loss but its more of a gut feeling that I haven't tested yet. But for example, on my miner (a bobcat) I've eliminated the lightning arrestor, its own RP-SMA jack, and 1-2 U.FL connectors.

For testing it, I was thinking of using a GLAMOS and a VNA. Any thoughts on this?

Anyways, I've also attached some photos of my first build. Its been completely revised since but shows some of the parts that have been eliminated.