Cavity filter not working?

General Discussion
#1

Trying to set up a repeater and looks like ive got issues with Desense.

Basically I got a cavity filter like this one here:
(link removed- Please read the forum rules) 65W Duplexer UHF 6 CAVITY DUPLEXER for radio repeater N connector UHF Duplexer…
And its running a pair of baofeng UV5R radios, with its TX set going through a Btech U25 booster.

Its working fine with signals in close range, but signals further away, while still coming in strong, wont receive once the set starts transmitting.
I did a test, and had the other person talking and as soon as I keyed the PTT button, it went silent, I could not hear any noise from myself transmitting, but it was cutting out.

Any ideas? Only thing I havent tried testing is reducing the bandwidth to NFM instead of transmitting on FM band.
Channels are UHF band 5 Mhz apart. High band is RX, low is TX.

#2

Well firstly 5MHz is a bit of a struggle for these smaller cavities, but doable to a usable degree with care. My guess is simply they didn’t tune it properly, or accurately. Do you have any test gear like a tracking generator with an analyser? Other common mistakes are not connecting the dummy load to the unused output while tuning the filter. Then, when you connect it, the filter curves shift. Experience shows that even if you ask a supplier to tune them before dispatch, they’ll be out by the time they arrive. If you have a decent power meter, fire rf through it from below the target frequency to above in small steps and plot where the peak is on each port, these should peak at your selected frequencies. My guess is that they wont.

1 Like
#3

OK, i was a bit worried about this.
I trusted that the supplier who offered to tune it for me had done it correctly.
I dont have any of the test gear to do it myself.
Are they quite sensitive to shock during freight?
I have an RF signal generator, frequency counter and oscilloscope and thats about it.
If I had a spectrum analyser, could I get by with that?
Seems there are a few people playing around with these on youtube.
I see they have other models with a higher number of cavities, do these work better?
The idea about using a power meter to measure the peaks could be useful. I do have an SWR meter that also doubles as a power meter, I might be able to figure that out.

#4

I had one of those Retevis repeaters and the duplexer in that is very similar. With these kinds of filters unlike a notch style filter, the throughput stays high then takes a dive quite steeply, and adjusting them just revolves around keeping the output constant, but the filter slope steep. The analyser without a tracking generator doesn’t help hugely but will let you take accurate measurements. The Retevis repeater is multi-channel, and advertised as such. The filters are really one channel - that tight for good performance, two perhaps at a push. What you can do with the analyser is fire a signal on one frequency through the filter then read the output and then see the other frequency at the same time - so you can see how well one gets through and the other doesn’t. Probably the simplest way to proceed is to draw a graph, or use excel to produce the response curve with how the filter is set now. Go through with a known power, with one port fitted with a dummy load, and another or your proper antenna connected and plot the output level at the smallest step your gear can produce. You’ll get a good idea where the slope starts and ends. Do this with the other port and the point where the two cross should be evident as the centre between the two points of maximum throughput. My guess is it will be visibly wrong. Be aware that once you loosen the lock nuts it is small adjustments, so this will be a very long process tweaking like this. You won’t have to repeat the entire thing if you make small adjustments at one time, but with three cavities each side, what often happens is the interaction between them, that is easy to see on an analyser with tracking generator, gets hidden. Adjusting one screw puts a bump AND a dip in the response. The next one might leave the dip the same but move or deepen the bump, and every adjustment of one screw does unexpected things. If it helps, the one in the repeater was clearly tuned for maximum rejection of either side of the receive frequency, but was also not quite as loss-less on the transmit frequency - so Power output was a little lower than ideal. That’s not as bad as making it deaf on receive - so a compromise.

1 Like
#5

OK, i will look at that. I might have a friend who can help me with this.
Would separating the channels further apart help things?
Would 10mhz still be too close?

#6

10mhz would be much better! P

#7

Just hooked up an RF meter on the antenna and TX’d on each port, it was out on each end, I have been able to change the frequency on the set in small increments and was able to get a higher power output on the power meter while tweaking.
It looks like the whole thing is exactly tuned out by 8mhz!
So im assuming that if the meter peaks on each band, then thats pretty much the exact frequency its designed to tuned for?
Will be interesting to test the unit out on these bands to see if it works.
Either way, its going to be a bit of a pain to align this up, but my friend should be able to help hopefully.

#8

You’re doing the right thing. You are looking for maximum attenuation of the tx frequency. On the rx port but with as little attenuation of that frequency and vice versa. Each tiny adjustment impacts the other so you repeat and repeat and eventually get there. The reason tracking generators are so useful is that you see the end product so when you tweak you can see interactions and strange peaks and troughs. I can live with tx power suffering a bit if Desense is low. Lots of the adverts mention a 10MHz separation because that’s easier to achieve. 6 not too trick but once you get to 5 it’s harder with more interactions. Keep tweaking! P

1 Like
#9

While sweeping by the meter, its peak seems to have about .3mhz where you cant see much movement of the needle. I dont know if this should be lower on a better quality filter or not, but ive at least found where each band is tuned to on this for the time being.

#10

Not really a ‘better’ filter but they broadly fall into the small high or low pass designs with a number of cavities for each or a band pass or band notch design. Tuning them gives a gentler result where the cavities overlap or a sharper result when they overlay. Hence why seeing what you are doing is quicker and more effective.

1 Like
#11

OK that makes sense.
Whats the story with the larger ones? Ive seen several that have 8 or more cavities, i take it that these do a better job?
I dont think I will worry if I can get this to with with 10mhz separation.

#12

Think of filters like these as audio tone controls, but they only go plus or minus 3 instead of ten. So if yo7 want lots of bass, you need more of them. These devices have a parameter called Q, a frequency, an amount and Q, the, ‘narrowness’ or the slope of the filter’s ability. More filters gives generally a better performance when tuned properly.

1 Like
#13

OK, i understand that. I do a bit of work in audio circuits so know how tone controls can affect the signal.
I would imagine that they would take quite a bit more work to tweak.
What kind of losses are typical through these for interest? If im feeding say 40W into, this, how much can I expect to loose?

#14

If they are tuned properly then they can be nearly lossless. Random tuning without test gear means you are tuning for maximum depth of notches or slope steepness and it’s easy to get the two sides mismatched as everything you do impacts somewhere else. You could live with transmit loss. Even 4dB or even a bit more is lovable with if the receiver is descended free and has minimal loss. What is bad is receiver side attenuation. Checking with a transmitter and sensitive meter can prove this to you.

1 Like
#15

I just tested the unit today on the bands that I measured it working at, but still having issues.
Whats weird this time is that the radio will transmit, but no audio is coming through the radio, if I stop talking on my set and release the PTT button, the radio on the repeater is still transmitting, but its just static, as if someone was holding down the PTT button, but not talking.
Almost like a feedback loop or something.

#16

Ok. The first thing to remember is that these radios are really not clean enough to really be used as a repeater. first thing to do is test them to get a clean bill of health. Disconnect the filter totally. Run a long cable from the one you are using for transmit to an antenna, stick it on low power and test that it goes into transmit and is passing audio from the receiver. Or just attach a dummy load to the tx radio and use a normal antenna on the other. A bit of distance between them should enable desense to be minimal. The filter itself cannot be the cause of an audio/no audio problem unless RF is getting into whatever you are using for the switching. You’ve not described what logic you are using to control the switching. If it’s one of those audio sensing gizmos, then I suspect it’s RF getting into that. With a filter in circuit, it has to act as a no pass filter on one frequency but a pass on another, and the reverse on the other port, so the RF is still there, so it can leak from poor screening on the in and out cables, and of course your Baofengs aren’t exactly filtered, so there will almost certainly be sproggies on the output which will be radiated and could get into the switching. You have a filter that you have now tuned well away from how it was setup, even if badly, so have you tried something like a cheap sdr dongle to actually see what is coming out? It might show a huge spike somewhere you’ve missed and this is what is getting back in. There is an art to repeaters, and often spaced antennas can make filtering so much easier. Single antenna working is very hard for a filter to do. I sold a few portable repeaters, but they claim they are multi channel capable, when the reality is they are not more than12.5KHz away, maybe 25KHz with the wind behind you. More than this and you need to tune the filter, and you know now how hard this is without the right gear.

1 Like
#17

Out of curiosity, what band is this repeater being used for?

1 Like
#18

Sorry its been a while.
The radio appears to be working OK on low power and a long cable to the antenna.
The duplexer unit im using is this here: Surecom SR-628 Cross Band 2-in-1 Duplex Repeater Controller

Its operated by the mic switch on each radio essentially, so when a signal comes on it relays it to the TX set and puts it through the microphone input.
I dont know if RF would interfere with this. Should I shield the whole box?
I have tried tuning this unit to 10MHZ seperation using a power meter and measuring the levels until it hits zero while tweaking the tuning rods.
I was told this is a quick and dirty way to do it without a spectrum analyser.
I was unable to tune it for 5mhz seperation and it was quite out of tune when i checked it, I might need to seperate more than 10mhz with this I wonder.
Anyway, i have found some better quailty filters out of some radio gear which I believe will probably work better and one of them is even stamped for the very bands I want at 5mhz seperation!

Around the 460mhz band.

#19

Quick and dirty will never be enough once you reduce the separation. With 10MHz, the performance can often be nearly as good with no filter and separated antennas! All you need is to tune the filter properly, you will have exactly the same issue with another one if you don’t have the test gear. It’s an exercise in frustration. What you could try is to put the filter into the receiver only. stick a 50Ohm dummy load in the other port and tune for maximum reduction of the transmit frequency. This means the filter adjustments, which interact severely, won’t be so extreme. doing it by trial and error means that you get a little extra rejection, think you’ve done well, but have wrecked what you did on the other frequency, and so on. Notching out the TX frequency might be easier. You’ll still have to check receive hasn;t been severely compromised, but it might work better - or just give the ■■■■ filter to somebody to tune, with your radios and pay the bill!

#20

finding someone is the problem, but a local HAM has been giving me tips, he may have a chance to take a look at it eventually, im happy to pay!