Online Amateur Radio Community Wiki

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
====== Messing about with filters for 868MHz ======
Meshcore has been coming along leaps and bounds and, until recently, Southampton proper had a dearth of repeaters so we lashed up "Soton WEST HBI Omni" (674cd415480a0de0aec84145eeedcae44c746d3ffaf1e16123babff712b2ab52) on 869.618MHz at one of our radio sites. Great location, covers everywhere but my oh MY, those LoRA boards do not like strong adjacent signals!

With a dummy load attached the RAK4631 board we are using has a reported noise floor of -118dBm. With the 11.5dBi Paradar antenna attached the noise floor jumped to -96dBm and inbound signals that should be strong-ish were reported to be around -10dB SNR, so Something Must Be Done!

{{:lora:lora:hbi-nofilter.jpg?600|}}\\
And strong adjacent signals we have, plenty of them. Television, 4G, and unfortunately some very loud IR-2030 86xMHz traffic.

Of course, Aliexpress to the rescue - plenty of things to try for comparatively little money.

===== £7 868MHz SAW filter =====
First port of call was one of the dinky SAW filters which are widely available for a few quid. Apparently they can handle 20dBm, and given our antenna gain we were only at 16dBm to stay legal so let's have a bash with one of those first.

{{:lora:ali-saw.jpg?600|}}\\
Looks pretty good, 2.4dB insertion loss isn't too hot but we have plenty of antenna gain. Passband is 14MHz wide, ish, and -40dB rejection is 30MHz wide so that should take out all the 4G and telly stuff.

{{:lora:hbi-sawfilter.jpg?600|}}\\
After throwing in the SAW filter we see the expected out-of-band 40dB rejection. Nice! The reported noise floor, which was somewhere around -96dBm, dropped to somewhere around -105dBm and inbound signals' SNR improved by about 10dB. We're onto something! But what's that large spike right next to our QRG?

{{:lora:hbi-saw-tight.jpg?600|}}\\
That large spike is four, essentially continuous transmissions. A cacophony of "Non specific short range devices" and very loud! Fortunately they are a few MHz away from our QRG. Perhaps we need a tighter filter...

===== £62 868MHz helical cavity filter=====
[[https://www.aliexpress.com/item/1005003857991485.html|These helical efforts are available]], shipped, for £62 from Aliexpress. Gotta be worth a punt right...

{{:lora:acasom-cavity.jpg?600|}}\\
They seem nicely put together and the screws aren't painted down to the point of impossibility. We have purchased a few of these by this point and they have turned up in a variety of conditions - one had clearly been adjusted with a pair of pliers! Oh well. Let's see what they look like.

{{:lora:cavity-before.jpg?600|}}\\
Not streets ahead of the SAW filter! A better insertion loss (1.3dB vs 2.4dB) but a similar bandwidth. In their out-of-box configuration they're not a huge upgrade from the SAW filter so let's see how they respond to tuning.

{{:lora:cavity-after.jpg?600|}}\\
After a mammoth, complete re-tuning effort we now have something which is actually useful! I managed to squeeze the bandwidth down to a few MHz at the cost of some insertion loss, steepen up the low-frequency knee and move the center frequency. Tweaking around the pass, notch and coupling screws afforded a filter which placed the knee on top of the signals we wish to reject while giving reasonable S11 and S21 characteristics on our operating frequency.

{{:lora:cavity-s21.jpg?600|}}\\
After some very careful fine-tuning markers 1-4, the frequencies which we would like to reject, were down to -15dB to -35dB. Not too shabby! S11 is -33dB, S21 is -2.4dB so we have a filter which we can compare directly to the SAW filter currently on site.

{{:lora:finished_cavity.jpeg?600|}}\\
Who said you can't put lipstick on a pig...

To be continued!!

-----
de G5RKT, March 2026