Voltera Community

Printing RF Filters

mharris 2016-09-21 21:49:09 UTC #1

After a fairly successful RF impedance test, onwards to printing filters. I'm going to start with a simple quad open loop filter targeted at 915MHz. I've done no simulation, I just want to see what it might do - so all the measurements are just off the top of my head - i'll be happy if i get a passband around 900-940mhz, and 6dBm of loss. Its not an optimal design

It's currently printing!
RF Filter Test 1_Copper_Signal_Top.gbr (1.1 KB)

mharris 2016-09-21 22:04:43 UTC #2

And we're off to baking.

The bottom right loop had a little bit of a blob next to the bottom left loop, so i've tried to clean that up as best as possible with a knife being used as a scraper before baking.

mharris 2016-09-21 23:12:56 UTC #3

Well, it printed, and it's not as bad as it could have been!
S21:

Came out tuned for 900mhz, 5.26dBm loss... so clearly not as good as I'd expect out of copper on good quality FR4 (I expect this board is giving me around 2db of extra loss for this circuit) but for trying out filter shapes i'm pretty happy with the result.

3dB bandwidth of about 100mhz, not a bad little filter for just making up numbers!

S11:

And as a crazy question, how would this filter work as an antenna?! (just feeding the left side)

VSWR (S11)

A surprisingly decent antenna for 915 and 858mhz... works in european and north american ISM bands?? I find this slightly hilarious.. Who knows what the radiation pattern is like, i'm not going to pay for chamber time to test it.

mharris 2016-09-22 17:42:04 UTC #4

So I decided to simulate the filter in Sonnet to see how close to a simulation the print was. Using the top metal layer as a resistor with the resistance that Jesus had mentioned elsewhere (0.0012ohms/sq) it looks pretty darn close. I guessed the 'FR4' or whatever the board was made of was giving me about 2db of loss and that seems to hold with the sim.

The sim comes up curve wise very very similar to the actual result. It would be interesting to get some good quality FR4 or even RF material made up to spec and re-print to see if there is a performance improvement, but that is not something I have money to burn on

mharris 2016-11-10 00:02:14 UTC #5

I just had the opportunity to test this filter on an EMSCAN ERX+, which shows the electromagnetic energy on the board. This really nicely shows where the energy is going. We're looking at buying one (they are about CA$100,000) so we were able to get them to come by for a demo, as they are also a Calgary based company.

With the tracking gen set to -30dBm output at 931mhz, at 1.88mm resolution, with the gerber overlay:

Without Gerber overlay

The feedport is to the right. I was measuring -6.1dBm of loss on the network analyser across the filter at this frequency. The brightest red is -21.2dBm with the blue background at -36.5dBm.

For fun, here's Sonnet's simulation of the current on the board at 931mhz.

Here's the resolution that the EMSCAN EHX+ (much cheaper) offers:

The ERX+ is quite a beast. It can now do down to 60uM (0.06mm) scan resolution, but this takes a pretty long time!