You are here


19 posts / 0 new
Last post

We have had desense problems on 2 node installations now.  One from a 900 Mhz paging transmitter and one from a 1200 MHZ ATV transmitter.  We purchased some chinese filters that cover the entire 2.4 band and are good enough for channel -2.  It solved the problem for the paging xmtr and hopefully will work for the ATV.  However, it would be ideal if we could get or make a filter for channel -2 that was 10 Mhz wide.

Any suggestions?


k1ky's picture
10 Mhz would be pretty tight

At that frequency, that's would be a fairly tall order.  If that one filter is adjustable, you could possibly cascade several of them in series in order to tighten the bandwidth.  Can you provide a link to the manufacturer or source for what you are using?


here is the one we used after putting it in a box with n connectors. 

KE2N's picture

Telenochek makes this kind of filter at semi-reasonable prices
search their ebay store for "WiFi filter"

Due to "group delay" issues near the filter skirts you want the filter to be 2-3 times the width of the channel you are receiving - so 40-60 MHz wide for 20 MHz channel BW - centered on the desired channel.

Unfortunately, for MIMO you need TWO of them ...



Here is the one we have used that is packaged in a box.  Thanks to N8NQH.
Another that doesn't require any fab is Mini-Circuits VBF-2360+ for about $35 (although we haven't tried thi one yet)

Image Attachments: 
S/N Graph help

What does everyone make of this plot?
We have some theories, but would like some other opinions

Image Attachments: 
AE6XE's picture
N8JJ,  there must be some

N8JJ,  there must be some other signal interfering (noise).    Here's a probable explanation...   Put into context that the SNR measure is taken in the 802.11 radio in what's called the 'baseband' and occurs after conversion from Analog to Digital and after 3 AGC steps in this Atheros chip.    The goal of the design is to minimize the data error rates, not maximize the SNR at this stage in detecting and decoding the digital signal.   Here's what can happen:

1)  An interfering signal affects receiving and detection of this neighbor signal and the error rate shoots up.
2)  The radio responds by trying to increase the noise floor parameters to drown out the interfering signal in the receiver and changing AGC values.  This would be shown as a (can be large) drop in SNR.  Note this noise floor determination in the receiver affects receiving all the signals (it's not neighbor specific).
3)  A program I put into wakes up every 1 min to detect significant SNR changes.  (look at /tmp/rssi.log for history and to correlate with the SNR chart).  The receiver can become stuck with settings that make it deaf to one or more neighbor mesh nodes (a big problem).  It might wonderfully tune out the noise and better receive one of the neighbors (very desirable for a client locking onto an AP), but then also tune out other neighbor signals we really care about receiving in our mesh network :) .
4) When this condition is detected, this program tells the radio to do a normal operation -- to scan an adjacent channel -- which causes the radio to recalibrate the noise floor and other parameters.  Hence this can trigger another spike in the SNR chart.   Note AGC values and Noise Floor thresholds are all being changed to track and minimize data error rates, so the SNR spikes can show up/down--not necessarily intuitive.

If this noise or interfering signal continues, then the radio can go in cycles like this.  While the SNR chart looks unusual, it's only a real problem if the data throughput is degrading and affecting normal use or if some neighbors start to drop in/out.    To resolve, this troublesome signal needs to be mitigated in some way.  Add a filter, change channel, adjust the antenna to a directional if an omni, etc.



Some more info:
This node is an omni Bullet on channel -2 at 170 feet.
It's about 2 miles from Wright-Patterson AFB and the base is LOS from the node.  They have a squadron of C-17's there.
We have a filter on order, but don't want to climb the tower any more than necessary.
We have verified that the ATV repeater on the tower at 1255 MHZ causes the S/N to drop 20-30 DB but it's not as weird as the plot
My theory is the a/c transponder responses are causing the sharp spikes.  They are at 1030 MHz.

AE6XE's picture
The aircraft transponders

The aircraft transponders reply on 1090 and can have up to 500W of EIRP.   If 1030 is the magic frequency, airplanes do xmit on 1030 to query other airplanes directly.  This is 'TCAS'--I'm a pilot in one of my other vices.    The pilot then has a view of all the airplanes around them that have transponders with mode 'A' and 'C' capabilities that are turned on and have replied.   This comes in real handy in crowded LA airspace.



1090 is even worse since it's closer.  I always get 1030 and 1090 mixed up.  You are right the transponder can squitter without an interrogation. 
I haven't been able to catch a C-17 over the site but they frequently to touch-and-go training and do the go around in the direction of the node.
We have had a similar problem with a 900 Mhz pager transmitter.  I asked Ubiquity if they had any kind of front end filter on any of their models and they said no.
If they did, we would buy that one.

KE2N's picture
channel filters

Another source of filters is DCI in Canada.
They have band pass filters and "channel filters"

for example

They will tune filters to a custom frequency if you want.  I have one of their filters that I use on 2304 weak signal. Its one of the WiFi band pass filters re-tuned for the bottom of the band. 

There is a weather resistant version of the filter as well.

These are serious filters and priced accordingly. But they do the job.

N8NQH's picture
new sytle

Thanks to AREDN for adding that "charts" feature, we were able to document this interference we were experiencing (well... its effect on the Mesh nodes operation) easily.

those Chinese micro-strip filters sure helped us out, about 95% elimination of the negative effects caused by  very-near high-power transmitters; though the time and expense of placing then in a sealed enclosure was somewhat a hassle.  The filter is installed between the LCom antenna and the Bullet.

Later we were lucky enough to get ahold of a commercially sold filter for evaluation purposes:

it already has "N" M & F connectors, and housed in a weatherproof enclosure; truly plug & play.

This other filter is not a micro-strip type; it has a very steep bandpass plot, and the lower frequency end of the pass is at 2.380ghz; no problem for Mesh ch -2.

Bill WA8APB and N8JJ took the  commercial filter out to one of our installations; the location where a 900mhz paging transmitter was totally wiping us out (its gain antenna was around 15 feet away from our Bullet M2HP titanium and LCom vertical) and did extensive A-B comparison testing between the commercial and chinese filters.

This commercial filter ($71) had less insertion loss, and totally removed any issues from that paging transmitter.  The Chinese filters had done  a great job too, but there was still some minor "dent" in the Bullet M2HP performance when the paging equipment was active.  This commercial filter removed all the harmful components, no intrusion into the 2.4 Bullets performance at all  now (with less insertion loss to boot).

We just bought 4 of them, model FT-BPF2400MF

if anyone would like to try one out, I have a loaner filter I can send you.

email me:

Cavity Filters

I took a chance and bought a cavity filter on the hope that I could retune it to channel -2.
We were successful! (although you need an expensive network analyzer to do it)  See the attached spectrum plots.  This is a 25 MHz filter as opposed to the top trace which is a ceramic whole band filter.  I haven't tried to seal it and install outdoors yet.

Image Attachments: 
N4RT's picture
SCADA transmitter interference on Channel -2

Hello all,

We are experiencing what appears to be desense on an AREDN node installed on a 160 foot water tank in our network.  The LQ and NLQ on any node that it "sees" is constantly cycling up and down.  LQ or NLQ might be 100% one minute and 0% the next.  Even when we catch the LQ high (greater than 70%) we can't seem to connect to that node.  Something is definitely interfering.  We've replaced the entire node with new hardware thinking we had a radio problem (This is a Rocket M2 on an AMO-2G-13 omni antenna) but the new hardware works exactly the same.  The node that we removed from the tank is up and running at my QTH with no problems.  So this is definitely site related.

The only other transmitter on the water tank is a SCADA 900 MHz transmitter.  I'm wondering if anyone else has run into a problem trying to locate a 2.4 GHz node near a 900 MHz SCADA system?  If so, did you see similar problems and, if so, what did you do to mitigate them?

I read some of the info in this thread regarding filters and, perhaps, that's what we need to do with this node.  What filters work best?  Some of the links given in the previous posts are broken and don't work.

Any info on this problem would be appreciated.

Thanks & 73,

Ron N4RT
South West Alabama Mesh Alliance (SWAMA)


The cavity filter from Teletronics is the best. (See my post above)  It comes with N type connectors, so it depends what kind of node you are using.
I have tried it on one channel of a Rocket with adapters.  It's not rated for outdoors, so I sprayed it with a sealant after tuning it.
You also need a spectrum analyzer to tune it. (Although one vender offered to tune it for $50 bucks or so.)

N4RT's picture
Node RFI is not from SCADA

Further investigation using a spectrum analyzer (RF Explorer) at the node site I discussed above shows that we have a strong digital signal of some type on 2399 MHz.  We drove around trying to DF the signal and found that it was strongest at a cell tower site about a quarter mile away from the water tank that our node is installed on.

This appears to be an AT&T cellular site but this signal is very definitely in the amateur radio allocation.  Anybody else have problems with cell towers radiating signals in our amateur radio allocation between 2390 and 2450 MHz?

Wondering what to do next?


Ron N4RT

K5DLQ's picture
I'd start with calling the

I'd start with calling the interference guy at the ARRL.  He usually has good suggestions.
(Ed Hare,

EDIT: email address corrected

Cell Tower Interference to 2.4 GHz Channel -2

Ron, N4RT and Joe, AE6XE:  What N4RT is describing has the exact same symptoms as the interference we are experiencing here in Layton, UT.  In our case, the firmware detects the offending signal as noise (per AE6XE) and compensates by raising the noise floor by about 10 dBm (from the default -95dBm to, in our case, about -86dBm), causing even close-in nodes to not appear as Current Neighbors, rendering Channel -2 virtually unusable.  We, too, have a local cell tower which we suspect may be the source of the problem.  We've used the AirView spectrum analyzer (recommended by Joe) which is native to the AirOS and see a broad strong signal toward the upper end of Channel -2, as well as a weaker and narrower signal just below Channel -2.  We've taken each of the nodes on our Channel -2 out of service and the problem still exists.  Ron, we definitely feel your pain!  Now to track down the cell providers and see if we can have a dialog.  K5DLQ, your idea is a good one.  Ron, do you want to make the call, or should I.  73, bill n7ie .. .

k1ky's picture
2.4Ghz Filters - revisited

Are there any good 2.4Ghz filter options available these days to mitigate overload products at Channel -2?  We recently are experiencing some new nearby interference sources that are now rendering our tower mounted nodes (Nanostation and Rocket Sector) virtually unusable at channel -2. I believe it could possibly be a strong signal on channel 6 nearby.

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer