RF PCB Material Selection

For most 2.5 GHz or below wireless RF designs we just use FR-4. If the board gets big and loss gets really important then you can get better specified FR4 like the stuff from Isola.

The trick is the layer stackup. For multilayer boards you need to specify the layer stackup carefully as the space between the layers and traces sets the impedance of the traces.

I regularly use simple low cost boards for RF to 2.5 GHz, especially on low cost wireless designs. (I'd probably also use FR-4 for a 5 GHz wireless LAN design too as the RF portion of these things is quite small).

It would help to know what you are trying to do / design so that a better recomendation can be made.

HTH - Steve H.

Hi Steve,

thank you for your helpful reply.

I would like to make a SWR meter working on 2.45 GHz with a 50 Ohm track designed on a PCB.

Do you think that the FR-4 PCB could fit this requirement ?

There is a "Epoxyd hard fibre FR4" for sale in FARNELL manufactured by "ROTH ELEKTRONIK" under "RE1001-LF" reference.

Do you know where I could find the exact drawing to make such 50 Ohm coupled tracks on the FR-4 PCB ?

I have  a 2.4 Ghz generator with a 2W output RF power and I would like to use this device to check a microwave cavity supposed to work at 2.4 Ghz. Hence the SWVR meter...

Hope you could help.

Best regards

Richard

Richard - If you just need a coupler - I would buy a Minicircuits pre-made coupler - it's only $120.00 USD and you really can't even do a quick protoboard for that. Much less make a coupler with PC Material and an xacto knife. Look at the Model,

ZABDC20-322H

It handles 50 watts, has excellent specifications and is connectorized.

If your need is to put this on a PCB then you can get the surface mount version from Minicircuits also.

Regards,

Steve H

Hello Richard,

FR-4 Material is OK at 2.45GHz only if your radio can handle significant drift over temperature.  FR-4 is typically used for consumer applications only and the board stack must be tightly controlled.   You need to have expertise when working with FR-4 at 2.4GHz if you hope to get repeatable results.   You need to be careful to make sure your design is sound before you go into production.  You could develop a radio using 5 to 10 prototype boards and everything looks fine.  However you most likely will see a difference in performance in production.   FR-4 is not intended for radio products.

Since you are developing an SWVR meter you will need to use a different material in order to get any kind of accuracy in your SWVR measurements.    I would suggest you look at a ceramic material before you start.  Rogers corporation has been the standard for industrial and specialty designs for a very long time.    I would suggest you look at the RO4000 series.  It is the most cost effective product they have available.  However it will cost you 7 to 10 times that of FR-4 .   It is well worth the cost in order to get repeatable reliable results.

http://www.rogerscorp.com/acm/products/16/RO4000-Series-High-Frequency-Circuit-Materials-Woven-glass-reinforced-ceramic-filled-thermoset.aspx

I have been developing radio technology for the last 15 years.  It is not an easy task you are undertaking.   The design my well seem rather straight forward but you will see performance issues if you do not get the  transmission lines, PCB layout and grounding issues nailed.   In all radio design the PCB is a major part of the circuit.

If you need some help. I am available as a consultant. I can review your design, transmission line calculations and prototype if you would like a second opinion.

Good luck,

Chris Pinter

Senior RF Engineer

Pinter Electronics Consultants

http://www.pinterec.ca

Good advice Chris - We don't really know if he is building one or a thousand or what his accuracy requirements are, but here is what I know.

When working for HP building their Network Analyzers we used FR4 type materials up to around 6 GHz. These materials were called GeTek (Now Isola FR408HR) and they had tighter specifications on Er  (Dielectric constant) and Dk (Loss tangent) than plain old FR-4 (which as you suggested can be all over the place).

High accuracy devices need to be calibrated so the calibration would take care of the part to part variations due to the PCB's core material being slightly different – batch to batch as long as Isola type materials are used. Plain old FR-4 would be iffy indeed in production.

We still use these Isola type materials up to about 7.5 GHz today and they can be used to build very good test instruments.

Where I use Rogers materials is where I need lower loss – that is when I am building a spectrum analyzer front end from 3 to 10 GHz or some sort of radio receiver where noise figure (hence loss) is very important – then these more expensive materials are real problem solvers. But they are rarely needed for my work below about 3 GHz (Isola is low loss enough at 3GHz compared to everything else in my circuits – for my work).

Last year I designed and built a Bluetooth Tester (2.5 GHz) that had very decent specifications (0.5 dB accuracies) of both forward and reverse power measurements over time and temperature and it was built quite economically on Isola 408HR.

All my test circuits are built on very economically on plain old FR-4 (the Bare Bones Type) – I don't want to do this for production – but for test circuits it is very economical and I can do a lot of testing for very little cost. (I have a design Idea that is to be published in EDN later this month that shows how to do this).

For high volume consumer type items – I have yet to take apart a Bluetooth device, Wireless LAN or Cell Phone that didn't just use something like Isola for their RF board. So these very high volume devices do just fine to the 2.5 and even 5.9 GHz arena where cost is more important than a dB or so.

This is all not to say that anyone is wrong or right – but we all approach these things from our own “Required Specifications” and “Experience” point and that's what this is – my “Experience” point of view for what sorts of things I have been involved in.

Cheers – Steve H.

Hi Steve, I thank you very  much for all the information you provided about RF PCB and the ZABDC20-322H model.

The same for Chris Pinter from Pinter Electronics Consultants for the link to ROGERS Corp.

I will follow your advice.

Cheers

Richard

Steve brings up some good points,   I agree that if you can work around the calibration issues and you plan to use the product in room temperature than FR-4 material should be OK.

We did a study 3 years ago comparing FR-4 to Rogers 4000 and we found the following benefits when using Rogers.

1. The performance over temperature was much more stable.   The power transfer was consistent as the return loss remained consistant.   Rogers 4000 has a very low insertion loss so it does not fluctuate much.

2. Rogers 4000 has a low insertion loss so it is great when you want to build a low noise figure receiver.   We found that the LNA noise figure was able to come very close to our theoretical calculations, even at 6 GHz.

Rogers 4000 is the best choice when performance is critical.   You have a much better chance of success.   However, most products have used Rogers 4000 are typically low volume and high value items such as test equipment.  

For high volume low performance products such as consumer WiFi hot spot locator you will want to invest your money in engineering to develop a solution using FR-4.

It looks like we have a lot of really good ideas on this topic.   Please use what works for you.

Chris Pinter

I have got a PCB sample from Rogers corporation in order to make my VSWR-meter. Nonetheless I have got some trouble to "translate" figures from maths to real lines on the PCB.....

I worked with this calculator:

http://www1.sphere.ne.jp/i-lab/ilab/tool/ms_line_e.htm

As an example I have got these figures for a RF4 PCB :

Operating frequency : 2450.000 MHz
Dielectric constant :5.4

Substrate thickness : 1.600 mm

Microstripline impedance : 50 ohms
Microstripline physical width : 2.53 mm
One wavelength (360°) : 61.6 mm

Quarter wavelength (90°) : 15.4 mm

Hence my questions below ==>

It leads to a lambda= 15.4 x 4= 61.6 mm inthe PCB instead of 300/24.5 (122.4 mm) in the air, right ?

Does it means that the velocity factor is 61.6/122.4 = 0.503 ?

Is the length of the 50 Ohm line on the PCB a mandatory request or its length is pointless ?

In other words may the line be of any length ?

About the forward and the backward lines:

Do they have a 5O Ohm impedance as well or they could be of any width ?

Do they have any length or they must be as long as the main line ?

What is the depth of  the gap  between the main line and the backward and forward ones ?

Could anyone tell me where I could find some theorical work about this concern ?

I would like to make my own VSWR-meter to "refresh" a bit my knowledge in the field of RF since I didn't make anything for a lot of years in this field. I feel like an idiot.....

Thanks a lot to you all and, above all,

HAPPY NEW YEAR

Richard