Connecting external antenna and RF module via PCB

[rajo]

Hello

Regulars here could have been watching my path in designing my very first RF link. It took quite a long time (a few months, but doing some other stuff meanwhile), and I guess this is the final doubt.

So, the RF module is on the PCB. The external antenna connector is mounted on the chassis, and I want to connect the connector and the RF via mother board solder points, through a 50 Ohms coax.
I found a very interesting appnote from Wood and Douglas Limited, but there are some doubts I have to check on with experts before I pursue further.

Namely, from what I understand the track connecting RF module out pin and the RF feed solder point on the PCB should run over a groundplane. My question is: can I presume that this is the same groundplane which is used for eliminating the digital noise? Is the same ground plane used for radio and digital? Ot there are two (connected) groundplanes, one used for digital circuitry, and another for the radio?

In the same appnote they created a small ground plane paralel to the ground plane on the opposite side of the PCB. This groundplane is situated under RF module and the antenna connector (they use a PCB mounted connector, whereas I use soldering points). I think this is weird, since I believe these two paralel planes are not usual.

Thank you in advance for any clarification

 

[Buriedcode]

Hi rajo,

Just so that you know, my experience with RF is very limited, I know some 'theory' and I've research on the net/books the subject of RF design (for my final year project, BSc) so don't take this too seriously.

Reading your post a few times, I assume you just want to connect a 50-Ohm (SMA/B) connector to your board for an external antenna. Interestingly, I have a few 2.4Ghz modules which do not have the antenna connectors on the module, instead they use 1.27mm pitch headers, so I guess we are in the same boat.

About 'ground planes' in mixed-signal design. A while ago I designed a board for an ADC/dsp. Although it wasn't RF, the mixed-signal design still applies (needed SNR of >130dB). After much research (4 months, litterally) it seems that there is more than one answer. Some people suggest using a split ground plane, where you have two that are joined by a fairly large PCB trace at the point where the digital circuit ends, and the analogue begins. Others however, say this can cause more problems, and claim that using a 'single' groundplane for the entire design is the only way, as long as you seperate analogue and digital. If you've searched google, you would have noticed that nearly ALL references to this subject are focused mainly on 'ADC' design, where stray noise/interference can be picked up by very sensitive ADC's, regardless of their sampling rate.

However, the fact that your system is digital/RF, it will probably be far less sensitive to 'crosstalk' since your RF is >100MHz and your digital is <40MHz (I assume). So whether its your Tx board or Rx board, it doesn't matter, there are RF filters in place in the modules. Of course you may still get 'some' noise from your digital circuits, but I think that can be rectified simply by placing the digital stuff on one side, and the RF stuff (a single trace to the antenna connector) on the other, with a single ground plane under (or on top of) the PCB.

Its all about 'ground returns'. Just think about where a signal goes, on the PCB, and find the returns shortest route to GND.

I think this is weird, since I believe these two paralel planes are not usual.

Do you mean, a double-sided PCB which a GP on both sides? Well, most of my designs have that, it makes ground connections easier If you can, look at a PC's motherboard. Covered in traces surrounded by ground-planes on both sides (and probably several layers inbetween).

Or do you mean, two ground planes, on one side of the board, separated by a few mil? This is a clear sign of mixed-signal design, one for digital, and one for analogue.
Whatever you mean, I would say go with your 'app-note'. You can always email the company and ask, then, if it all goes wrong, you can blame them. App-notes are there to 'show-off' a company's product, they want it to operate at its very best so the design is probably perfect for your module.

One more thing, I really don't think that 'ground-planes' are an issue, but the width of the trace (along with PCB material, dielectric, thickness etc..) greatly influences its impedence. Correct me if I'm wrong here guys. Wrong width and you'll get insertion losses, so it'll lower the power from antenna to module, or visa-versa. The length should be short, because you don't want the trace to radiate, only the antenna. There are various equations for it, but I use a little prog called 'TXline'.

Anyway, as I said, I'm no expert, I haven't even built mine yet and I don't have the equipment to measure its output power, or to check for poor PCB design.
Good luck, and let me know how you got on.

BuriedCode.

 

[rajo]

Thank you BuriedCode!

It is not that I did not believe the appnote, but needed a confirmation.

Abot the paralel GP's. I was thinking about the GP's on the opposite sides of a PCB. If you say you have that practice (I have not) it is OK with me

I'll let you know what happens in the end

 

[jallem]

I think trying to combine RF ground and digital ground is not
convenient. In the case you mention, the main idea is the
final stage to have a nice transmission line as possible.
I would try to avoid any deviation from that. No matter
what, it must be a clean transmission line between the RF
module and the antenna.

 

[Buriedcode]

rajo, you are welcome

Having GP's on both sides of a double-sided PCB is all good, since both are the same signal GND.

I forgot to ask you one question. What frequencies are you working with? RF and digital. Looking at 'woodanddouglas.co.uk' I see they only do OEM modules in 150,450,868Mhz (all ISM bands in the UK I believe). I've only really worked with 2.4Ghz and above, where things are a tad more critical, but any RF work requires care attention to PCB design. In the app-note ' RF Connections to W&D Modules (AN 7000 0020)' I can't see anything worng with it at all. Obviously you don't have to copy that exact design, the microstrip line doesn't have to be at 90 degrees, but looking at the pin placement on the module, it might be convenient.

They've covered everything that I'm aware of, microstrip width (according to PCB materials) removal of sharp corners on the trace, ground planes, and a a shed-load of vias ensuring good connections between both ground planes. I do know that you shouldn't have any ground plane surrounding the microstrip IF the microstrip is to be an antenna (PCB antenna) but if you're going for an external, which you are, then you need the ground plane to connect the GND part of your BNC/SMA connector, just keep the ground plane a bit further away from the microstrip.

jallem, thanks for your input.

No matter what, it must be a clean transmission line between the RF module and the antenna.

I agree. But, the 'RF' area surrounding the transmission line will have its own ground plane. I can't see any real problems if any digital stuff is located within another ground plane, across the other side of the PCB (same side, just look at the PCB has having 'left and right' sections). The ground return for the RF will have a ground plane all by itself, so no digital signals need to enter it. But, the RF return may indeed go into the digital part, but sufficient decoupling caps wil handle that.
Of course, I suppose if you are after complete perfection, then yes, the RF should have a PCB all by itself, but I really wouldn't worry about it.

BuriedCode.

 

[rajo]

Hey BuriedCode.

So, you sniffed the very document I referred to.
My RF is at 433MHz, whereas my digital is at 8MHz max.

I do know that you shouldn't have any ground plane surrounding the microstrip IF the microstrip is to be an antenna (PCB antenna) but if you're going for an external, which you are, then you need the ground plane to connect the GND part of your BNC/SMA connector, just keep the ground plane a bit further away from the microstrip.

What about this: if I connect the GND solder point (where I solder my coax cable) and ground pin of the RF module (it is not a Wood&Douglas module, by the way, it has a different, SIL pinout) through the ground plane, but on the component side of the PCB, should I still have that ground plane at the lower side of the PCB? I mean, do you think they use the ground plane at the lower side (surrounding microstrip) to secure GND connection at the connector only? I would rather avoid it if possible.

So long

Edit:
I have a kind of missed the closing sentences of your post. Sorry for that. So, the ground plane at the lower side of the PCB is that special RF GP, right? By the way, what is it needed for? The transmission line consists of microstrip and the GP on the other side, ie digital GP, I think. This raises te other question since it sounds insane to me, to use the digital GP as a partof transmission line

Back to the RF GP. Wouldn't the RF signal also enter digital GP (the top side of the PCB) since the pins at the RF module are through hole? My next concern: my Rx and Tx modules both have only two GND connections: one RF GND and another supply GND, no dedicated signal GND pin. Referring to the figure at the appnote, may I connect the power supply GND to the RF GP as the guys did there? Would not it transfer radio noise to the signal ground return path?
So many questions...

 

[Buriedcode]

Hi rajo,

I'm starting to get out of my depth here, I'm merely a 'tinkerer' of all things electronics

I assume you are using a double-sided PCB, but to be honest, I've used cheap 433MHz sil modules on solderless breadboards, strip board, single sided FR-4 and I've had no problems (at least, none I can measure, it may have added more noise to the RF/digital parts). It didn't severely hinder the link, although, it has to be said, at 433Mhz, digital stuff can interfere with the reciever, reducing range and introducing noise.

Ok, I've just deleted about 40 lines, because I think I'm just over complicating it, just sat here thinking about it for a few minutes.....

Your SIL modules have 2 GND connections. One if for RF, and one is for the rest of the modules circuit (probably analogue). These are almost certainly connected together internally, so bascially they are the same pin, GND. However, if you think 'why have they duplicated a connection on another pin? Why add another pin? And why is it on the other side of the module next to the RF???'. I'm sure you've asked yourself that, but it's taken me ages to realise

The answer is, no matter how complicated you make the PCB ground planes, ultimately, both pins need to be connected to ground, the same ground. Now, the fact that RF GND is next to the RF out, away from all the other connections means that you have some sort of 'isolation' between RF signals on the PCB and 'other' signals. Have a solid ground plane, have your RF connector as close to the module as possible (well, about <2cm?). Both the module GND pins will be connected to the ground plane, but the 'other' signals (non-RF) returns via the ground plane to you main GND connection (battery??). The RF returns the same path, but because its over the other side of the module, it will go through the digital 'part', but the digital part cannot go through the RF part. God I'm confusing myself, here's what 4 minutes with 'paint' can do:

Sorry for going on about, i over-complicated things, damn my obsessive problem-solving.

Good Luck, hope this helps,

BuriedCode.

 

[jallem]

My RF is at 433MHz, whereas my digital is at 8MHz max.

At 8MHz you dont need to even worry about impedance control or
impedance matching at the digital part. You need to worry about
impedance matching at the RF section.

 

[rajo]

Thanks you both guys for being so patient. I am really sorry to perhaps be seen as a 'tinkerer' .
This is a whole new field for me and I am simply afraid I'd make a stupid oversight, prolonging the development period. Timing for producing prototype PCB's in my country is quite long (let's say three weeks, I guess it is shorter some happier placies) - so I prefer to be as sure as possible before sending the files to the workshop.

I am OK with all that was written here, although I am still 'a little' wondered by that GP surrounding the microstrip - whether it is really necessary or not.

Nevertheless, I am attaching a snapshot of a part of my receiver PCB, for an inspection - in case somebody is eager to take a look. Please notice that there is no RF connector on it - two solder points instead. Also, please notice that the microstrip is not surrounded entirely by GP due to space limitation. Should I change the design so the microstrip is entirely surrounded by the GP?

 

[Buriedcode]

Hi, sory for the late reply

Your PCB looks great to me.

I am really sorry to perhaps be seen as a 'tinkerer'

Heh, I was saying that I'm a tinkerer SO many people in here know far more than me, and I haven't got my degree yet. I'm just a practical man, theory is all good, but only useful if you can build things.

As far as grounding around microstrips......it really is a tricky subject. As I said, even the top PCB designers have different views.

I don't think that PCB will give you any trouble, and I doubt very much that it simply 'won't work'.

Good luck,

BuriedCode.