duplexer mystery

DAL-COM · Post by **DAL-COM** » Sat Nov 22, 2003 7:14 pm

I checked out a duplexer the other day, a DB4975 with two TX cans at 440.925 and one RX can at 445.925. I could not get better than 3dB loss on the TX side with half-wave jumpers for the tee. So I called DB (Andrew) and they gave me the correct lengths. Long story short, someone else made the jumpers and put it on the air saying it only had a 1dB loss. It worked fine for a week.
Today I checked it for the owner and it has 3dB loss. I only had a power meter and dummy load. Tried retuning the pass band but no joy. Could this be an intermittent internal problem? Has anyone else seen this problem?

Nand · Post by **Nand** » Sat Nov 22, 2003 7:56 pm

http://www.txrx.com/products/pdf/genta_5.pdf

The above link to TXRX systems will give you some information about why your measurements may not be accurate. They also show how to get around this.

Also if the duplexer is not exactly the correct impedance or the cable is not exactly the correct impedance, inserting items in the TX to duplexer feed line will change the impedance as well and result in different readings.

I am not familiar with the duplexer in question, but generally you should be able to remove one section of a duplexer at the antenna T and it should not affect the other leg. If it does affect the other leg, then the cables leading to the T are not the correct length. The reason for this is that the other leg should look like a high impedance at the frequency of the opposite leg, just as if it doesn’t exist. And the same the other way around.

Nand.

radiomidwest · Post by **radiomidwest** » Sun Nov 23, 2003 8:50 am

The type of cable makes a big difference on tuning as well.

DAL-COM · Post by **DAL-COM** » Sun Nov 23, 2003 8:32 pm

Interesting article by txrx systems, Nand. I can believe that the duplexer cavities do not have input or output impedances that are 50+\-j0 ohms, so clearly there will be standing waves which makes power measurements tricky. It makes jumper lengths important also.
They say you should use two wattmeters and a dummy load to check a cavity, one before and one after the cavity. Each coupled directly to the cavity, without jumpers. But it seems to me you could do it with one wattmeter, and a series of barrels/couplers to make a line the same length as a Bird. Then just swap the Bird and the line to get input and output power. Asssuming you use top quality silver-plated N barrels etc.

BTW, What did you conclude from your survey of TX-cavity#1 jumper lengths?

radiomidwest · Post by **radiomidwest** » Mon Nov 24, 2003 4:09 am

Whenever posible, use the actual repeater cables to tune the duplexer.
Back in the early years, I scratched my head a few times, banged it on the wall a couple too. I tuned up a duplexer at the shop, went to the site, had poor ops, high loss, ect. Not until I used the exact cables to tune the duplexer did everything work right.

http://batboard.batlabs.com/viewtopic.p ... 782#110782

Nand · Post by **Nand** » Mon Nov 24, 2003 4:30 am

DAL-COM wrote: But it seems to me you could do it with one wattmeter, and a series of barrels/couplers to make a line the same length as a Bird. Then just swap the Bird and the line to get input and output power.

You would think so, but what if the velocity factors of the inserted Wattmeter and the barrels are different. This would affect the electrical length.

As for the “poll” results in the other post, I believe as many do, that the best choice is multiples of electrical ½ wavelength cables are best between the TX and the cavity. This is because a cable that long does not transform the impedance that the transmitter sees when connected to the cavity and therefore doesn’t induce additional errors. If everything is perfect, any length should be OK and often good enough.
For transmitters that are not happy with a 50-Ohm termination for some reason, other length or a tuner for this purpose would be needed as XMO pointed out.

Nand.

Nand · Post by **Nand** » Tue Nov 25, 2003 4:17 am

Looking over your original post, I noticed that there are only two TX and one RX cavity. This typically doesn’t create a lot of isolation at low loss.

More cavities tuned for low loss are preferred over fewer cavities tuned for higher losses.

Nand.

DAL-COM · Post by **DAL-COM** » Tue Nov 25, 2003 10:53 pm

I see that I made a typo in my first post- it was a DB4075 or possibly a 4076 but I don't know the difference.
Nand: You are right, to replace the second wattmeter you would need either a section of airline same length as the Bird, or a shorter piece of cable or coupler of lower velocity factor.
These 3-can duplexers do not have obvious adjustments for loss, maybe you can change the coupling loops?
They have the usual threaded rod to set the passband and a trimmer cap under a threaded cover to set the notches. I say notches because they have two, one each side of the pass. This makes them very useful as single cavities for fixing all kinds of problems on site.

Nand · Post by **Nand** » Wed Nov 26, 2003 3:44 am

http://www.decibelproducts.com/ds_pdf/d ... er-catalog

http://www.repeater-builder.com/badconn/dbcans.html

http://www.decibelproducts.com/pdfs/095064-000.pdf

Here are some links to help you pull out your hair.

Nand.

DAL-COM · Post by **DAL-COM** » Thu Nov 27, 2003 7:31 pm

I don't have much hair left to pull. After reading the Repeater-Builder site, I was ready to give up on the jumper lengths. Probably the thing to do is go find an identical unit and measure them.

kcbooboo · Post by **kcbooboo** » Mon Dec 01, 2003 7:01 am

I have three Cellwave/Phelps-Dodge 526-4 (or equivalent) 6-cavity pass/notch UHF duplexers, all rated for 350 watts and over 100 dB isolation. They work great. The jumpers between sections are 11-1/2 inches of RG-214 with male N connectors on them, EXCEPT that the two cables that go to the antenna's Tee connector are 12 inches long. With the velocity factor of 0.66 for RG-214, I cannot resolve the electrical or mechanical length of these jumpers. These cables are supposed to be good from 440-470 MHz. Lower freqs get longer cables, by 1/4 to 1/2 inch. Higher freqs get shorter cables, also by 1/4 to 1/2 inch. All these dimensions are from an internal chart I got from RFSystems specifically for RG-214 and these 526-4 units.

A new set costs $250. I just made my own for about $9/cable by buying some new RG-214 ($1/ft), some crimp-on male N connectors ($3.75 ea), and the proper crimping die for my Ideal crimping tool ($30). Took me a few hours with a sharp knife and diagonal cutters to make the six cables, but I did save a lot of money.

Bob M.