Hospital communications issue

[jonwienke]

As long as the transmitter sees a 50 ohm load, it'll be OK. But a simple T splitter with an antenna on each port of it will present a 25 ohm load to the transmitter.

No, because the coax between the transmitter and the T still has 50Ω impedance. The power is evenly split on the output side of the T, but the input side is still 50Ω.

 

[buddrousa]

If I remember electronics school it is 25 ohms

 

[prcguy]

If you use a simple coax T adapter with two 50 ohm things on two ports it will present a 25 ohm load to the third port of the T. That will be a mismatch to 50 ohm coax feeding the transmitter or duplexer of a repeater. Use a high power 50 ohm divider and all is good.
prcguy

No, because the coax between the transmitter and the T still has 50Ω impedance. The power is evenly split on the output side of the T, but the input side is still 50Ω.

 

[ramal121]

If someone uses a $10 tee to combine two antennas to a transmitter in lieu of a proper power divider deserves the results they get. At least in my book.

 

[TampaTyron]

I think you guys are mixing up the terms here, in the shops I have worked in...... power divider and splitter mean same in my world. Neither of those is a simple coax T. However, I have repaired several large systems where instead of 50/50 dividers, the installers user coax Ts. I was third vendor to look at it when it wouldn't pass commissioning. Telling the installer to replace 300 or so Ts with dividers (large fiber fed DAS) was sad and hilarious at the same time. Dont get me started on the mobile antennas with no ground plane pushed through ceiling tiles........TT

 

[jonwienke]

If you use a simple coax T adapter with two 50 ohm things on two ports it will present a 25 ohm load to the third port of the T. That will be a mismatch to 50 ohm coax feeding the transmitter or duplexer of a repeater. Use a high power 50 ohm divider and all is good.
prcguy

You are conflating DC circuit behavior with RF transmission line behavior. If you were right, co-phased CB antennas would always have terrible SWR because of the massive impedance mismatch.

The load on the transmitter is always 50Ω because the feedline between the transmitter and T is still 50Ω.

Power splitters are used instead of Ts to isolate the ends of the T from each other, not because they are required for impedance matching.

You can verify this with a simple experiment. Connect 2 dummy loads to a T, and measure SWR, then compare that to the SWR of the a dummy load connected directly. It won't change much.

 

[prcguy]

A CB co-phasing harness for two antennas is made of critical lengths of 75 ohm coax making up most of a Wilkinson power divider so the two 50 ohm antennas present a 50 ohm match to the radio.

A standard cal method for SWR meters is a T adapter with two 50 ohm loads making a 25 ohm load to cal the SWR meter for a 2:1 reading.

Bottom line is a "power divider" or "RF splitter" IS required as an impedance matching device between two 50 ohm things connected to a single 50 ohm transmission line. You should know this stuff.....
prcguy

You are conflating DC circuit behavior with RF transmission line behavior. If you were right, co-phased CB antennas would always have terrible SWR because of the massive impedance mismatch.

The load on the transmitter is always 50Ω because the feedline between the transmitter and T is still 50Ω.

Power splitters are used instead of Ts to isolate the ends of the T from each other, not because they are required for impedance matching.

You can verify this with a simple experiment. Connect 2 dummy loads to a T, and measure SWR, then compare that to the SWR of the a dummy load connected directly. It won't change much.

 

[speedway_navigator]

Those of you new to the business should check out the Telewave ANTPD24. If any other pros have other mfgs they prefer let me know.

 

[ElroyJetson]

No, because the coax between the transmitter and the T still has 50Ω impedance. The power is evenly split on the output side of the T, but the input side is still 50Ω.

I really think you need to spend a little time experimenting with the aid of a network analyzer before you say silly things like that.

If it WERE that simple, then every RF engineer and antenna specialist is wrong.

Maintaining constant impedance as seen from the transmitter is a bit more complex than I think you realize.

 

[Mike_G_D]

You are conflating DC circuit behavior with RF transmission line behavior. If you were right, co-phased CB antennas would always have terrible SWR because of the massive impedance mismatch.

The load on the transmitter is always 50Ω because the feedline between the transmitter and T is still 50Ω.

Power splitters are used instead of Ts to isolate the ends of the T from each other, not because they are required for impedance matching.

You can verify this with a simple experiment. Connect 2 dummy loads to a T, and measure SWR, then compare that to the SWR of the a dummy load connected directly. It won't change much.

Sorry, prcguy is quite right here. Whether it's DC or RF, the same basic equations still apply (they just get more involved with vectors in AC assuming you have reactive components).

Ideally, at resonance, you are supposed to see a purely resistive load so you can "assume" a simplified parallel load at first approximation. So, if you have two purely resistive loads of 50 Ohms each connected in parallel you will see a combined load of 25 Ohms (1/(1/R1+1/R2)). Obviously, if you get reactive components you need to deal with those but, for the sake of simplicity in this case we'll assume a pretty resonant load.

Dummy loads are, or should be, purely resistive, at least to a first approximation when used with their recommended bandwidth; there are no "intended" reactive components (just resistors, which, yes, when you go beyond a first approximation they will have reactive components but, within recommended frequency limitations, should present a nearly pure resistance with negligible reactive components).

In a well matched system consisting of source, transmission line, and load, if the characteristic impedance of the line is equivalent to the source and load impedances then the length of the line should not affect the impedance of the system (again, to a first approximation). Once your source and load becomes significantly mismatched then the length of the line and its characteristic impedance can become significant because the line becomes a matching element - a distributed component. As prcguy stated, those cophased antenna designs use this fact to make them function correctly. That is why they use 75 Ohm cables as the lines between the antennas and the T junction to "compensate", for want of a better term, for the two "ideally" 50 ohm loads combined in parallel.

Now, if you define a 2:1 mismatch as "not significant" then, yes, when you look at a simple T terminated with two 50 ohm loads with a VSWR meter you may judge that as "acceptable". Depends on your system design. But, generally, you try and keep things below that type of mismatch in most designs.

And, of course, receive only systems that have broadband antennas can present widely varying load impedances to the receiver. But that is less of a concern than a power transfer system as in a transmitter and power amplifier design.

-Mike

 

[speedway_navigator]

I think these Smith charts should help show dummy load being mostly resistive.
The 5 Watt load has a 40 dB return loss, the 20 Watt is 28 dB.
The last 2 charts are the 5 Watt load on 3' N to N jumpers.

 

[jonwienke]

Do you have a similar chart showing 2 50Ω dummy loads connected to a T?

 

[Mike_G_D]

Do you have a similar chart showing 2 50Ω dummy loads connected to a T?

I'm really curious, what exactly do YOU expect to see in that case?!

In fact, in rereading your earlier post, I'll propose the following:

1) Properly calibrated, maintained, and operated vector network analyser showing a Smith Chart display normalized to 50 Ohms with a coax T, basic T with no matching components just to be thorough, connected directly to the S11 port with the two other ports of the T terminated in 50 ohm dummy loads.

2) As above but with a length of good quality 50 ohm characteristic impedance coax cable between the T and the S11 port of the NA. Let's say 3 feet or so to start but I'll be generous here and let you choose whatever you want so long as the length doesn't add so much line loss at the test frequency that it attenuates your return loss readings and gives you funky results!

I know what I would expect but I'm really interested in what you would expect?!

-Mike

 

[jonwienke]

At this point, I don't have an opinion, I would just like to see the charts for both your proposals, 1 and 2, and if you're feeling generous, a variation of case 2 where the T is removed, and one of the dummy loads is connected directly to the short length of coax used in case 2. The exact coax length isn't critical as long as it is short and isn't an odd multiple of 1/4-wavelength electrically.

 

[speedway_navigator]

I have to order a T, mine has the center conductor cut back to make an Iso T.
I thought I had some BNC Ts from old time networking days but must have thrown them out.

 

[speedway_navigator]

Found a BNC T.

 

[Mike_G_D]

Yeah that looks about right although I'm a little surprised that the reactive component is a tad more than I expected; the real part makes sense at about 26 ohms and the reactive component is likely due to imperfections in the connectors and/or cabling or maybe some calibration issues (or maybe the load is less than mostly resistive at 150MHz).

-Mike

 

[prcguy]

Any length of cable that amounts to a measurable fraction of a wavelength between a vector analyzer and resistive load other than 50 ohms in a 50 ohm system can skew the vector analyzer off the purely resistive path.

I use two 50 ohm loads on a T adapter connected to SWR meters to calibrate them for 2:1 mismatch, because that is two 50 ohm resistors in parallel equaling a 25 ohm load when not used with a Wilkinson or other matching type power divider. If you connect the T adapter with two 50 ohm loads to a common Bird 43 wattmeter it should show 11% reflected power or 2:1 VSWR at all frequencies where the T adapter is not contributing to the VSWR.
prcguy

Yeah that looks about right although I'm a little surprised that the reactive component is a tad more than I expected; the real part makes sense at about 26 ohms and the reactive component is likely due to imperfections in the connectors and/or cabling or maybe some calibration issues (or maybe the load is less than mostly resistive at 150MHz).

-Mike

 

[Mike_G_D]

Yes, absolutely, but I was initially assuming that the system would have been calibrated WITH the length of cables. In other words, you null out the cables' influence first before attaching the load or DUT. But then, I guess we did also discuss what a length of cable would look like so (as part of the DUT) so in that sense, this paints the right picture.

I assumed the first test was JUST the T with the two 50 ohm loads so the test cables' effect would have been "zero'ed out" which is the way I always started things. Then, we can start adding cables of specific lengths, etc., which remain part of the system being tested (to be measured) rather than part of the test set itself (compensated for in the calibration).

I always prefaced my measurements with as exact a description of my test setup as possible including a list of all calibration steps, etc. If speedway_navigator provides a setup diagram or exact description, then we can better analyze the results.

For example:

1) Was the T with loads directly attached to the S11 port or was it connected with a length of test cable?

2) The analyzer was properly calibrated to the S11 port with lab class test loads correct? These would be at least an open, short, and 50 ohm load.

3) If cables were used, were they included in the calibration procedure, zero'd out, or not compensated for at all?

4) A purest might even calibrate for the T connector; don't think this test needs to get that exact but it should be mentioned.

5) What type of 50 ohm loads were used on the T connector? The same lab quality used to calibrate the analyzer, something else, and what were their rated frequency ranges?

Without knowing the above, we can only speculate but my guess is pretty much what you've said concerning the cables (if they were not compensated for before the test).

-Mike

 

[speedway_navigator]

I put the BNC T on the VNA and had N to BNC adapters on the T, no cables.