We’ve discovered that if you have a high-altitude node or a node that is near a communications tower, you might need a filter. Filters will cut down interference from nearby sources and lower the noise floor, which increases your receive sensitivity. It can make a remarkable difference in your radio’s ability to hear incoming signals. Infrastructure nodes should definitely consider adding a filter as a best practice, especially for MeshOregon. There is a side discussion about the radios being deafened by loud incoming signals from nearby sources. Adding a filter will help, but you may need to be aware that this is an internal radio issue (firmware) and ask for help separately.
I was given the opportunity to do a side-by-side comparison test for three Meshtastic filters using a NanoVNA. These filters are useful for high-altitiude sites where the radio may be exposed to a lot of background interference from other communication towers. They both increase the receive sensitivity of the radio (by lowering the noise floor) and they also ensure that our radios cannot bleed over into adjacent frequencies and possibly cause interference to other Meshtastic users on other modes.
Test #1: Lothar (Airframes) cavity filter
Note: This filter was originally factory tuned to LongFast on 906.875 MHz. We manually retuned it in the field to MeshOregon on 918.5 MHz.
- Dimensions: 4.25″ x 4″ x 1.25″
- Type of connectors: SMA female
- Cost $130.
- Website: https://shop.airframes.io/products/lora-915mhz-filter
- Results of test: Usable bandwidth: 4.8 MHz
- Design frequency: 918.5 MHz
- Lower cutoff (-3dB): 915.9 MHz
- Upper cutoff (-3dB): 923.9 MHz
- Insertion loss: -2.27dB
- Test notes: This is physically the largest filter we tested. This has the highest insertion loss of the 3 filters we tested, but that may be due to our field retuning not being as well done as factory tuning. (You won’t notice a tiny bit like 0.5dB.) This is the most expensive filter we tested. Often out of stock, or on pre-order status.
- Comments: This filter performs well. We have several of them in service, including one at SKYLINE02.
Test #2: Mini ceramic filter (wehooper4/zbm/others)
Acknowledgement: This filter was donated by @pdxlocs for testing purposes. Thank you!
- Dimensions: 5/8″ x 1″ x 1/8″
- Type of connectors: IPEX / U.FL
- Cost: $25
- Website: https://zbm2industries.com/products/mesh-ceramic-filter
- Results of test: Usable bandwidth: 42.0 MHz
- Design frequency: 918.5 MHz
- Lower cutoff (-3dB): 892.8 MHz
- Upper cutoff (-3dB): 934.0 MHz
- Insertion loss: -1.53dB
- Test notes: This is physically by far the smallest filter we’ve tested. This is the least expensive filter. This is an extremely wideband filter, with band pass for the entire 900 MHz ISM band. Middle of the pack for insertion loss. Extremely well-suited to very small nodes where space is at a premium. Note: This filter is very wideband. The band pass is the entire ISM band, so you won’t get the benefit of filtering out other ISM and Meshtastic users on other modes / frequencies. However, it will filter out cell phone towers, 5G, and other out-of-band users.
- Comments: We have several reports of good performance in the field, including @pdxlocs.
Test #3: Baymesh4 (nullrouten) cavity filter
Note: This filter was originally factory tuned to LongFast on 906.875 MHz. We manually retuned it in the field to MeshOregon on 918.5 MHz. This filter is available to purchase professionally pre-tuned from Nullrouten.
- Dimensions: 5″ x 2.5″ x 1.75″ (including N connectors)
- Type of connectors: N female (SMA female also available)
- Cost: $95
- Website: contact nullrouten on Discord. Ask for model # baymesh4-918 for MeshOregon.
- Results of test: Usable bandwidth: 6.4 MHz
- Design frequency: 918.5 MHz
- Lower cutoff (-3dB): 914.5 MHz
- Upper cutoff (-3dB): 922.3 MHz
- Insertion loss: -0.74dB
- Test notes: This filter originally shipped as a LongFast filter and was field retuned by ziffzuh. Can be factory tuned to MeshOregon by nullrouten on request. This filter had the lowest insertion loss. Comes with excellent printed documentation of test data.
- Comments: We have several of these in service, including the excellently performing SYLVAN02 node.
Conclusion
- Filters have reasonably similar results. All three passed spec with no issues, and all performed their tasks as designed. So I decided to arbitrarily assign the following awards:
- BEST IN SHOW Nullrouten (Baymesh) cavity filter This filter is the middle in cost, but it outpeforms the other two on insertion loss. Excellent documentation. In service, we have one of these filters installed at SYLVAN02 and it offers outstanding performance.
- BEST VALUE Wehooper4 ceramic filter This is by far the least expensive filter tested. It also has an outstanding quality all its own: the dimunitive size. If you have an enclosure that is too tight for a Baymesh cavity filter, or your budget doesn’t permit the more expensive filter, consider this one instead. @pdxlocs reports very good operation in his testing. You won’t notice the very small difference in insertion loss. (I haven’t had a chance to install or field test this one yet.)
- BEST MUTT – UNDERDOG AWARD Lothar (Airframes) filter This filter won this award (and came in Third Place) due to the fact that manually retuning this filter is tedious. This is a very intricate and delicate operation and is best performed when you drank decaf. The cost is the highest of the three filters, and it’s also physically the largest – by far. These filters are often sold out and backordered, so you might be waiting a while if you want to purchase one. In service, this filter is currently being used at SKYLINE02 and performs its tasks admirably well. I have no complaints.
- Final results: Under typical conditions, all three filters basically perform the same in the field for most applications. Choose the one that best serves your needs, budget, and available enclosure space. There is very little difference in insertion loss between the filters. In practice, you won’t notice half of a dB of difference. If you are having interference issues from nearby nodes that are operating on different modes (i.e. Longfast), or have interference in or adjacent to the ISM band you’re fighting, the wehooper4 ceramic filter is not the best one to use for this purpose – you’ll want to go for a cavity filter for the narrower bandwidth.