Measuring Splitter/combiners

[MavamQ]

Leave the unused port unterminated. It will still show about 3dB loss but with some amplitude ripple across the frequency range. At some points it will be a little more than 3dB and other points a little less than 3dB but an average of 3dB loss unterminated.

I don't have the equipment to measure this, I am working on it. I don't know how to interpret a an open port.
What is your comment about 1/2 the power going into the termination and that is why you measure -3db.

 

[prcguy]

If you want I will run the tests here and take pictures. Will an unterminated two way divider showing 3dB loss satisfy you? You could also take an engineering RF coarse and do the measurements in the lab and the instructor will tell you there is a 6dB increase combining in phase but you loose 3dB in hardware.

 

[prcguy]

I fired up some equipment on my garage floor to test the loss through a pair of 2-way splitters in both splitter and combiner mode.

The first pic is the 50MHz reference signal at 0dBm or 1 milliwatt I will be using from an HP8920 service monitor.



The second pic is feeding a 2-way splitter/divider and measuring its loss as a splitter with the unused port terminated which comes out to 3.26dB loss in this case.



The next pic is measuring the splitter/combiner as a combiner with only one port fed and the unused port terminated. The through loss in this case is 3.33dB.



The next pic is the same setup as a combiner but with the termination taken off the unused port. In this case the loss went up to 4.37dB and that will vary across the frequency range of the splitter/combiner. At some other frequency it could be less than 3dB and its due to reflections within the splitter that combine in or out of phase causing something different than the expected 3dB loss.



The next pic is with the setup changed where the source is now feeding splitter to create two in phase signals to send to the next splitter/divider and the power was recalibrated for 0dBm out of the splitter and cable which takes about 3dB more drive.



The next pic is feeding a splitter/combiner as a combiner with 0dBm at its input and only one port which has about 3dB loss. Remember this one.



Here is the same setup as last but the termination removed from the unused port to show its not necessarily the load that is absorbing half the power and the combiner still has about 3dB loss or 3.30dB in this case.



Now the final pic with the combiner receiving 0dBm into each port and each path has about 3dB loss shown in previous pics but the resulting combined power is 2.82dBm or close to 3dB gain with some other small losses.

Here is the point I am trying to help you understand. If each path through the combiner has 3dB loss and you feed it two signals at 0dBm, that gets knocked down to -3dBm of level from each input port which should give a combined output level close to 0dBm if combining in phase results in a 3dB gain. But we are getting close to 3dBm output level meaning the process is producing a 6dB gain for combining two identical signals in phase then we loose 3dB in the hardware doing it. Does this make sense?

 

[MavamQ]

Let me see If I understand the setup . Correct each sentence if it is wrong.
Your generator is outputting a 3db signal.
The splitter splits the signal and each output is at 0db.
Now you feed the outputs of the splitter into the combiner.
The output of the combiner is +2.82.
Isn't that the plus 3db I have said all along?
0db in on two inputs and 3db out that is +3db.
Mikek

 

[prcguy]

Let me see If I understand the setup . Correct each sentence if it is wrong.
Your generator is outputting a 3db signal.
The splitter splits the signal and each output is at 0db.
Now you feed the outputs of the splitter into the combiner.
The output of the combiner is +2.82.
Isn't that the plus 3db I have said all along?
0db in on two inputs and 3db out that is +3db.
Mikek

Yes but the first part using a splitter with 3dBm in to create two identical 0dBm signals doesn't matter, I could have used two separate signal generators providing 0dBm as long as they were phase locked. Its the end result of that being two identical signals at 0dBm each giving 3dBm out of the combiner that is in question. The combiner has 3dB insertion loss in each leg, internally knocking the two 0dBm input signals down to -3dBm each. Getting 3dBm out of the combiner now needs an explanation due to the fact the combiner has 3dB of loss making the combining process look like 6dB gain. That explanation is phase combining produces 6dB of gain and you can only realized 3dB of that due to hardware losses.

 

[MavamQ]

Yes but the first part using a splitter with 3dBm in to create two identical 0dBm signals doesn't matter, I could have used two separate signal generators providing 0dBm as long as they were phase locked. Its the end result of that being two identical signals at 0dBm each giving 3dBm out of the combiner that is in question. The combiner has 3dB insertion loss in each leg, internally knocking the two 0dBm input signals down to -3dBm each. Getting 3dBm out of the combiner now needs an explanation due to the fact the combiner has 3dB of loss making the combining process look like 6dB gain. That explanation is phase combining produces 6dB of gain and you can only realized 3dB of that due to hardware losses.

What? You already know the splitter has 3db loss. 3db into the input and 0db out of each output. So now, you go into the two combiner inputs with 0db each and they combined to +2.82db.

 

[prcguy]

What? You already know the splitter has 3db loss. 3db into the input and 0db out of each output. So now, you go into the two combiner inputs with 0db each and they combined to +2.82db.

I'm putting 0dBm into both ports of the combiner which has 3dB loss and getting 3dBm out combined. Yes the combiner has 3dB loss. There is 3dB gain missing. Where is it?

 

[MavamQ]

I'm have trouble with your power meter, First, db is a ratio, why not measure in power, aka dbm?
I would like to see is a signal measured in dbm. Those are easy to convert to voltage and then I can show the power in and the power out.
Can you do that with your equipment? Can you use set your sig gen to 0dbm, feed your splitter, verify -3dbm with one output terminated, test both if you like. Then connect the splitter to the combiner, I expect a reading out of he combiner of about -0.2dbm.
I do appreciate your time on this, one of us will learn something, I already have.
Thanks, Mikek

 

[prcguy]

I'm have trouble with your power meter, First, db is a ratio, why not measure in power, aka dbm?
I would like to see is a signal measured in dbm. Those are easy to convert to voltage and then I can show the power in and the power out.
Can you do that with your equipment? Can you use set your sig gen to 0dbm, feed your splitter, verify -3dbm with one output terminated, test both if you like. Then connect the splitter to the combiner, I expect a reading out of he combiner of about -0.2dbm.
I do appreciate your time on this, one of us will learn something, I already have.
Thanks, Mikek

I don't have a problem with it because its measuring and displaying in dBm. Every single picture shows the power level in dBm to an accuracy of .1dB. I happened to trim the reference input power in every case to within a tenth of a dB of 0dBm then used the "reference" feature to zero out the meter to 0.00dBm to get easy to read gain and loss, otherwise you would have to add or subtrace the fraction of a dB that I was off from 0dBm. Each picture besides the 0.00dBm reference shows actual insertion loss in -dBm and gain from combining at the 2.82dBm level. Since I used 0dBm as the reference you don't need any math to see the actual gain or loss. I could have used say 4dBm for my input but then you would need a calculator to see the actual gain or loss. For most of my career I lived in the dBm/dBw world and am very familiar with using the equipment.

In every picture where I'm feeding power into a splitter/combiner its 0dBm within .1dB.

 

[MavamQ]

OK then, I'll walk through it again. Let me know if any line is wrong.
In the description with the last picture, you have your sig gen set at +3dbm.
You feed it to the splitter and you get 0dbm out of each output.
you feed 0dbm into each input of the combiner.
out of the combiner you get +3db (+2.82)
+3dbm is equal to 2mW. That is the power into your splitter
0dbm is equal to 1mW. That is the power out of each out of the splitter.
You drove two 1mW signals into the combiner you got out 2mW. Double the power is +3db.
Two 1mW signals in and one 2mW out.
Nothing else, nothing is missing.

 

[MaineGuy]

prcguy is either a saint - or a fool at this point.

I'll go with the first option. A better man than I.

 

[MDScanFan]

Let’s take ZFSC-2-1+ as an example. Its insertion loss (mismatch + dissipative + isolation), per its data sheet, is around -0.3 dB. As a splitter, the output per port is around -3.3 dBm assuming a 0 dBm input. Total loss (input power - total output power) of the device is around -0.3 dB.

A passive splitter is a reciprocal device. Operating it as a combiner, apply in phase signals of -3 dBm per input. The combined output is around 0 dBm minus a fraction of a dB dissipative loss. For ZFSC-2-1+., the total loss (total input power - output power) is around -0.3 dB. This is also per conservation of energy.

What is disputed with this?

I believe an earlier comment may have implied there was an extra 3 dB combiner loss somewhere in there, when used as a combiner. If so, then what is the loss mechanism (dissipation, mismatch, etc?). And what happens to that 3dB loss when used as a splitter? Based on reciprocity, wouldn’t we then have -6dBm outputs when driven with a 0 dBm input?

 

[MavamQ]

Let’s take ZFSC-2-1+ as an example. Its insertion loss (mismatch + dissipative + isolation), per its data sheet, is around -0.3 dB. As a splitter, the output per port is around -3.3 dBm assuming a 0 dBm input. Total loss (input power - total output power) of the device is around -0.3 dB.

A passive splitter is a reciprocal device. Operating it as a combiner, apply in phase signals of -3 dBm per input. The combined output is around 0 dBm minus a fraction of a dB dissipative loss. For ZFSC-2-1+., the total loss (total input power - output power) is around -0.3 dB. This is also per conservation of energy.

What is disputed with this?

I believe an earlier comment may have implied there was an extra 3 dB combiner loss somewhere in there, when used as a combiner. If so, then what is the loss mechanism (dissipation, mismatch, etc?). And what happens to that 3dB loss when used as a splitter? Based on reciprocity, wouldn’t we then have -6dBm outputs when driven with a 0 dBm input?

Thank you, I think I have one person that agrees with what I have said.
The question has come down to does a splitter combiner, when used as a combiner lose 3db of signal.
I say no, prcman is saying a combiner has 3db loss. We have both tried to prove our point and we are not succeeding.
I hope when he tries looking at the power, ie 0dbm =1mW
I thought he proved it to himself in the last test of post 63 and then in #65 he is looking for the 3db again.
I hoping he ran the experiment I suggested measured in dbm so we can discuss mW in and mW out, things will become more clear.
If you feed 1mW into each input, you get 2mW out. For a gain of +3db. without any 3db of loss, but I repeat myself.

 

[MavamQ]

prcguy is either a saint - or a fool at this point.

I'll go with the first option. A better man than I.

Question MaineGuy, do you think a combiner has 3db of loss?

 

[prcguy]

Let’s take ZFSC-2-1+ as an example. Its insertion loss (mismatch + dissipative + isolation), per its data sheet, is around -0.3 dB. As a splitter, the output per port is around -3.3 dBm assuming a 0 dBm input. Total loss (input power - total output power) of the device is around -0.3 dB.

A passive splitter is a reciprocal device. Operating it as a combiner, apply in phase signals of -3 dBm per input. The combined output is around 0 dBm minus a fraction of a dB dissipative loss. For ZFSC-2-1+., the total loss (total input power - output power) is around -0.3 dB. This is also per conservation of energy.

What is disputed with this?

I believe an earlier comment may have implied there was an extra 3 dB combiner loss somewhere in there, when used as a combiner. If so, then what is the loss mechanism (dissipation, mismatch, etc?). And what happens to that 3dB loss when used as a splitter? Based on reciprocity, wouldn’t we then have -6dBm outputs when driven with a 0 dBm input?

The point I'm trying to make to MavamQ is the combining process actually includes a 6dB gain when combining two equal signals in phase and the internal loss of the splitter used as a combiner eats up 3dB of that combined gain leaving you with only 3dB useable gain from combining. You have correctly pointed out a splitter has 3dB loss when being used as a combiner. I put in two equal signals in phase at a level of 0dBm and got 2.82dBm out but we could call it 3dBm. If the combining process only had 3dB gain then I would get 0dBm out of a combiner that I fed 0dBm in both ports due to its internal loss of 3dB per port that would eat up my 3dB gain from combining. The 6dB gain theory for combining two equal signals in phase answerers for that 3dB lost in the hardware combiner but apparently its not discussed much, otherwise we wouldn't be here now.

 

[prcguy]

prcguy is either a saint - or a fool at this point.

I'll go with the first option. A better man than I.

I'm no saint, I just spent my entire 40yr career in RF related things and have taken engineering classes specific to what we are discussing.

 

[MDScanFan]

...The point I'm trying to make to MavamQ is the combining process actually includes a 6dB gain when combining two equal signals in phase and the internal loss of the splitter used as a combiner eats up 3dB of that combined gain leaving you with only 3dB useable gain from combining. You have correctly pointed out a splitter has 3dB loss when being used as a combiner...

That is not a correct interpretation of my post. I pointed out that as both a combiner and as a splitter the device has a ~0.3 dB total loss (total input - total output).

...I put in two equal signals in phase at a level of 0dBm and got 2.82dBm out but we could call it 3dBm...

This is consistent with my above statement. An ideal combiner would perfectly combine the two 0 dBm inputs and the output is therefore 3 dBm. Given the device is not ideal, there is a small amount of insertion loss. In your example, and my example above, it is ~0.3 dB insertion loss.

......I put in two equal signals in phase at a level of 0dBm and got 2.82dBm out but we could call it 3dBm. If the combining process only had 3dB gain then I would get 0dBm out of a combiner that I fed 0dBm in both ports due to its internal loss of 3dB per port that would eat up my 3dB gain from combining. The 6dB gain theory for combining two equal signals in phase answerers for that 3dB lost in the hardware combiner but apparently its not discussed much, otherwise we wouldn't be here now.

Doesn't the second sentence contradict the first? The first describes a 3dB gain vs that of a single input. Put in two 0 dBm inputs, which is equal to a total power of 3 dBm, and you get out around 3dBm. That is 3dB gain vs a single input, and it is consistent with conservation of energy.

If there are two inputs of 0 dBm (for a total input of 3 dBm) and if the device has an internal loss of -3dB, then, based on conservation of energy, the output would need to be around 0 dBm. You actually measure an output around 3 dBm, which implies the internal loss of -3dB cannot be real.

 

[prcguy]

That is not a correct interpretation of my post. I pointed out that as both a combiner and as a splitter the device has a ~0.3 dB total loss (total input - total output).



This is consistent with my above statement. An ideal combiner would perfectly combine the two 0 dBm inputs and the output is therefore 3 dBm. Given the device is not ideal, there is a small amount of insertion loss. In your example, and my example above, it is ~0.3 dB insertion loss.



Doesn't the second sentence contradict the first? The first describes a 3dB gain vs that of a single input. Put in two 0 dBm inputs, which is equal to a total power of 3 dBm, and you get out around 3dBm. That is 3dB gain vs a single input, and it is consistent with conservation of energy.

If there are two inputs of 0 dBm (for a total input of 3 dBm) and if the device has an internal loss of -3dB, then, based on conservation of energy, the output would need to be around 0 dBm. You actually measure an output around 3 dBm, which implies the internal loss of -3dB cannot be real.

No, a two way splitter or divider has 3dB loss plus a little more and typically about 3.1 to 3.5dB per port. A four way is 6dB plus a little more and so on. Any stated loss in a splitter spec that is fractions of a dB is the loss on top of theoretical which will always be there. That 3dB inherent loss in a two way splitter is what creates the mystery to some when using it as a combiner. The 3dB is still there per port as a combiner so whatever you put into the two way combiner immediately looses 3dB but the resulting combined power is actually 3dB gain. That seems to be the focus of this argument, how is it possible?

 

[MavamQ]

Hi prcguy, how is the fact that you drive the combiner with two 1mW signals and get out 2mW consistent with your answer that we lose 3db in the combiner?

I'm sure at some time there is some concept where you logically came to the conclusion that combiners have 3db loss. It maybe that when you add two voltages, you get 4 times the power or +6db. so the splitter must have 3db loss to end up with +3db. But a combiner does not add voltages in series, it puts them in parallel after an impedance conversion.

 

[MDScanFan]

At this point I believe we are in the same page with our understanding of power in and power out of splitters and combiners, which is good.

The contention stems from references to combiner internal loss, such as ‘If the combining process only had 3dB gain then I would get 0dBm out of a combiner that I fed 0dBm in both ports due to its internal loss of 3dB per port that would eat up my 3dB gain from combining’. Is internal loss referencing power loss?

If so, then simply tracking power out vs power in of a typical combiner shows there cannot be a ~3dB internal power loss term.

If ‘internal loss’ is not referencing power, then what is it referencing?