Notch or Band Pass

gatekeep · #1 (**permalink**) 01-02-2013, 10:48 AM

Ok, so I'm having a moment. And I hope someone could help me here. I have one of those cheap chinese duplexers and I'm nearly 95% sure its a notch-type duplexer.

The duplexer is very similar to this: 6 Cavity Duplexer for Radio Repeater UHF Free Tune | eBay

I'm attempting to retune mine because its wayyyyy out of tune, and at the moment I don't have a spectrum analyzer, so while I know it won't be tuned perfect it will be tuned better then it is (don't worry its being used for amateur and not commercial purposes).

If it *IS* a notch duplexer, just to be sure, the LOW side is tuned to notch out the transmit frequency and the HIGH side is tuned to notch out the receive frequency correct?

kayn1n32008 · #2 (**permalink**) 01-02-2013, 11:07 AM

Wirelessly posted (Mozilla/5.0 (BlackBerry; U; BlackBerry 9780; en-US) AppleWebKit/534.8+ (KHTML, like Gecko) Version/6.0.0.600 Mobile Safari/534.8+)

Yes it is a notch filter.

The high side is tuned to notch out the low frequency and the low side is tuned to notch out the high frequency.

If the repeater transmit frequency is higher than the receive frequency, the transmitter is connected to the high side and the reciever is connected to the low. If the repeater transmit frequency is lower than the receive frequency, it is connected to the low port and the reciever is connected to the high port.

If the reciever and transmitter are connected to the duplexer backwards you WILL burn out the finals on your transmitter and the reciever will be deaf.

gatekeep · #3 (**permalink**) 01-02-2013, 11:11 AM

Should have used an example for my brain, being sleep deprived due to New Years isn't helping....

(Yes I'm using GMRS freq's for the example)

High Freq (Rx): 467.725
Low Freq (Tx): 462.725

The high side of the notch would be tuned to notch out 462.725 and low to notch out 467.725?

kayn1n32008 · #4 (**permalink**) 01-02-2013, 11:18 AM

Wirelessly posted (Mozilla/5.0 (BlackBerry; U; BlackBerry 9780; en-US) AppleWebKit/534.8+ (KHTML, like Gecko) Version/6.0.0.600 Mobile Safari/534.8+)

Quote:

Originally Posted by gatekeep

Should have used an example for my brain, being sleep deprived due to New Years isn't helping....

(Yes I'm using GMRS freq's for the example)

High Freq (Rx): 467.725
Low Freq (Tx): 462.725

The high side of the notch would be tuned to notch out 462.725 and low to notch out 467.725?

Yes. The low side is tuned to notch out 467.7250 and the high side is tuned to notch out 462.725

gatekeep · #5 (**permalink**) 01-02-2013, 11:19 AM

Thanks, much appreciated. Now to take a nap and perhaps, attempt to fix this thing later.

gatekeep · #6 (**permalink**) 01-02-2013, 1:38 PM

Hmmm....so after playing with this thing (no I didn't touch the current tuning on it) for a bit and running some tests with a watt meter, it appears the low/high are inverted?

The 3 cavities on the low side are notching 462.725 and passing 467.725, and the 3 cavities on the high side are notching 467.725 and passing 462.725.

kayn1n32008 · #7 (**permalink**) 01-02-2013, 8:09 PM

Wirelessly posted (Mozilla/5.0 (iPad; CPU OS 6_0_1 like Mac OS X) AppleWebKit/536.26 (KHTML, like Gecko) Version/6.0 Mobile/10A523 Safari/8536.25)

On the low side are you getting high or low swr at 462.725?

gatekeep · #8 (**permalink**) 01-02-2013, 9:33 PM

High SWR, its not passing power on the low side for 462.725.

Using my available tools right now, I've managed to get this thing pretty much retuned roughly. The low side is now passing 462.725 with low SWR [the best I can measure right now, putting out nearly 4.5W at 5.6W in] and is rejecting 467.725 completely. The high side I've fixed as well it rejects 462.725 [barely, though but I'm too tired to continue screwing with this thing tonight] and passes 467.725. I suspect to get this perfect I'll need to put it on a spec. analyzer eventually.

kayn1n32008 · #9 (**permalink**) 01-02-2013, 11:03 PM

Wirelessly posted (Mozilla/5.0 (BlackBerry; U; BlackBerry 9780; en-US) AppleWebKit/534.8+ (KHTML, like Gecko) Version/6.0.0.600 Mobile Safari/534.8+)

Just remember, any adjustment you make one cavity will effect all the other cavities' resonance. Once you think you have it go back to the other side and double check it. Remember that being a notch filter it rejects ONLY the frequency you notch out +/- 50KHz and will pass all others. If you want to see what the insertion loss is, you subtract the power coming out from the power going in. If you putting 50w in and getting 43 or more watts out call it good. But do not be surprised to see as little as about 38w out. 38ish watts out for 50w in is just under 1.5db loss. 43w out for 50w in is around 0.75db loss and .75db loss through a duplexer is pretty decent. I have tuned Sinclair Res-Loc's to around .1db of loss but 0.5-1.0db is 'normal' and acceptable.

kayn1n32008 · #10 (**permalink**) 01-02-2013, 11:20 PM

Wirelessly posted (Mozilla/5.0 (BlackBerry; U; BlackBerry 9780; en-US) AppleWebKit/534.8+ (KHTML, like Gecko) Version/6.0.0.600 Mobile Safari/534.8+)

You will never get 0 loss. There will always be some power loss through a duplexer. If you can get the power loss to 7w or less for 50w in, congrats!!! You have properly tuned the duplexer and there is not real need to put it on a service monitor. Although with a tracking generator/service monitor you can 'see' the notches. When you get a chance post your power in/power out numbers.

VE6RHS · #11 (**permalink**) 01-03-2013, 3:57 PM

On the mobile "notch" duplexers, you want to tune for maximum rejection - not maximum power transfer!

Connect a signal source tuned to the high frequency to the low-pass port and adjust the tuning rods (on the low pass side) for maximum REJECTION of the high frequency (tuning for the deepest notch) while recieving on the high-pass port.

Then reverse the setup - connect the signal source tuned to the low frequency to the high-pass port and adjust.

This is easily done with a spectrum analyzer & tracking generator, but you can do it with a transmitter set on ultra-low power and a signal strength meter (S meter) in a receiver (remember that the transmitter will see a VERY high VSWR during the tuning, so keep the power very low). You may have to add 20 or 30 dB of attenuators to the receiver to knock the signal down initially - but once you get the notches tuned you can lose the attenuators.

Once both halves of the beast are tuned, double-check everything when installing the duplexer in regular service. I've been caught several times mixing up the high & low ports and damaging a transmitter...

kayn1n32008 · #12 (**permalink**) 01-03-2013, 4:37 PM

Wirelessly posted (Mozilla/5.0 (BlackBerry; U; BlackBerry 9780; en-US) AppleWebKit/534.8+ (KHTML, like Gecko) Version/6.0.0.600 Mobile Safari/534.8+)

Is that not what I told him Ray???

LoL (one of my mentors)

gatekeep · #13 (**permalink**) 01-04-2013, 8:25 AM

I didn't have time to play with it yesterday. I might try tonight, but I need to get larger attenuators.

So the question remains then, should I not care about the power output of the rest of the frequencies that are being passed? I should be only concerned with the amount of loss on the notched frequency?

(I'm well aware of the VSWR conditions when tuning a duplexer like this.)

kayn1n32008 · #14 (**permalink**) 01-04-2013, 8:57 AM

Wirelessly posted (Mozilla/5.0 (iPad; CPU OS 6_0_1 like Mac OS X) AppleWebKit/536.26 (KHTML, like Gecko) Version/6.0 Mobile/10A523 Safari/8536.25)

You can look at other frequencies, but it is rather pointless. You are only really concerned with the repeater transmit and repeater recieve frequencies, so really those two frequencies are the only ones to worry about. Also while tuning terminate the other port (when tuning the low side put a dummy load on the high side and when tuning the high side put the dummy load on the low side) with a dummy load.

gatekeep · #15 (**permalink**) 01-04-2013, 10:25 AM

Quote:

Originally Posted by kayn1n32008

Also while tuning terminate the other port (when tuning the low side put a dummy load on the high side and when tuning the high side put the dummy load on the low side) with a dummy load.

I am using a rather low input wattage 50ohm load. Just out of curiosity (assuming that transmitting in the air wasn't a problem) couldn't one also load the opposite port from the one your tuning with a real 50ohm antenna?

Quote:

Originally Posted by VE6RHS

On the mobile "notch" duplexers, you want to tune for maximum rejection - not maximum power transfer!

Connect a signal source tuned to the high frequency to the low-pass port and adjust the tuning rods (on the low pass side) for maximum REJECTION of the high frequency (tuning for the deepest notch) while recieving on the high-pass port.

Then reverse the setup - connect the signal source tuned to the low frequency to the high-pass port and adjust.

This is easily done with a spectrum analyzer & tracking generator, but you can do it with a transmitter set on ultra-low power and a signal strength meter (S meter) in a receiver (remember that the transmitter will see a VERY high VSWR during the tuning, so keep the power very low). You may have to add 20 or 30 dB of attenuators to the receiver to knock the signal down initially - but once you get the notches tuned you can lose the attenuators.

Once both halves of the beast are tuned, double-check everything when installing the duplexer in regular service. I've been caught several times mixing up the high & low ports and damaging a transmitter...

So, I'd have the S-meter or power meter or measurement tool on the OPPOSITE port from the one I'm tuning instead of common/antenna? In this case the common/antenna would be loaded with a 50ohm dummy? This is quite a different method from the one I've learned and the methods shared in articles on repeater-builder.

kayn1n32008 · #16 (**permalink**) 01-04-2013, 3:25 PM

Wirelessly posted (Mozilla/5.0 (iPad; CPU OS 6_0_1 like Mac OS X) AppleWebKit/536.26 (KHTML, like Gecko) Version/6.0 Mobile/10A523 Safari/8536.25)

In theory you could, but if it is mis-tuned it would not be a 50ohm load.

petnrdx · #17 (**permalink**) 01-04-2013, 3:47 PM

I am sorry, this is likely to start a storm, but...
You really won't get good results tuning a reject duplexer with a watt meter and transmitter.
I have worked on dozens of these "flat pack" duplexers and hundreds of duplexers, filters and combiners.Its kind of what I do for a living.
And I am sure other people on here have done more than me. I am a bit surprised others have not joined in here...
I am not convinced there is a "standard convention" for labeling. Sometimes I have found the "High" and "low" labeled ports to be the PASS, and other brands the REJECT.
Of course the same holds true for TX and RX.
So, I always check them for actual reject and determine which should be which.
MOST of the time you will find the interconnecting semi-rigid coax between the individual "cavities" (resonators if you like) will be LONGER on the lower frequency half. Shorter on the higher frequency half.
MANY of the duplexers brought to me were "tuned" upside down.
You will never get one to operate in specs when tuned upside down.
I use a vector network analyzer or a tracking generator to tune these things.
MUST you have a VNA or TG to tune one?
NO you can use a simple signal generator and tunable receiver.
The VNA or TG are just faster and maybe a little more accurate. Not enuf to matter in most cases.
So, if I were you, I would find someone that has an old Signal generator and use a programmable RX as the receiver.
Then tune each "half" for the greatest REJECT of the opposite frequency.
As another poster stated, terminate the unused port with a good dummy load. ( you could use your antenna, but the antenna is rarely a real 50 ohm load so it is not recommended)
Never tune for maximum transfer of power ( best pass ) on this type of duplexer.
When you try to use a transmitter, you WILL get stray coupling direct from the transmitter ( thru the case ) direct to the receiver ( again, thru the case ) and not be able to "see" the reject notch thru your device and cabling.
Bear in mind that you are looking for a reject measurement in the 70 dB and greater range.
If you had radios that had 80 or 90 dB "isolation" or "shielding" you might be able to get a duplexer of this type to partially work, but I am doubtful of the results.
You might get something to "work" but it won't be near as good as you would get with 30 minutes of tuning with a simple sig gen and a tunable ( programmable ) receiver.

gatekeep · #18 (**permalink**) 01-04-2013, 3:56 PM

Quote:

Originally Posted by petnrdx

As another poster stated, terminate the unused port with a good dummy load. ( you could use your antenna, but the antenna is rarely a real 50 ohm load so it is not recommended)

That was a theoretical question out of my curiosity because I've never heard of it being done that way, I am using a 50 ohm load for this.

petnrdx · #19 (**permalink**) 01-04-2013, 4:26 PM

It is a good theoretical question.
And I would think it would be better than using nothing.
I have a selection of quality loads with each of the most common connectors ( BNC, Type N, TNC, UHF, Mini, male and female) that I have collected over the years to be sure I have the one that I need.
I would think somebody near you has the stuff to help you tune that thing up.
I do them for people around here all the time...

gatekeep · #20 (**permalink**) 01-04-2013, 4:51 PM

Quote:

Originally Posted by petnrdx

I would think somebody near you has the stuff to help you tune that thing up.
I do them for people around here all the time...

Well "New York, USA" is misleading...I'm not in the city, I'm on the island out towards the east end, its like the sticks out here, there are a few shops, but I'm mainly tuning this for the educational experience.

On the topic of the second question, which is the proper procedure to ensure best rejection? Measurement via the common/antenna port and Tx/Dummy Load on the Lo/Hi ports, Measurement/Dummy Load via the Lo/Hi ports and Tx on common/antenna, or, the suggested method above of Tx on the Lo/Hi, Dummy on Antenna/Common and Measurement via opposite Lo/Hi port?

EDIT: Forgot a whole sentence...