Definition of isolation and pulse response time for a 3-way power splitter

[Jiho]

I will use 3 way power splitter and power detector.

1. This is power splitter data sheet. In this data sheet, there are different isolation values. What does it mean?? Also in that point, I wonder definition of isolation at power splitter.

2. This is power detector data sheet. In this data sheet, Each pulse response rise/fall time are 400/10 or 800/400. What is difference between 400/10 and 800/400? What does it mean?

 

[trurle]

Regarding isolation, isolation is defined between each pair of output ports. It may matter if you have significant reflection at outputs - in this case imbalance of isolation is converted into imbalance of transmission.
Regarding pulse response, listed time is the 10%-90% transition time at output if rising or falling step input is applied. Fast response on falling edge likely indicate circuit difference (no pmos or zero-recovery diodes) - may be at cost of higher power dissipation.

 

[Jiho]

Thank you for your answer. But I have three questions.

1.What does isolation 1-2 in the data sheet? I can't catch meaning of number(1-2).
2.What happen if isolation becomes higher or lower??
3.Does fast response on falling edge mean small number of fall time? If I want to get low power dissipation, then 800/400 is better than 400/10??

 

[trurle]

Thank you for your answer. But I have three questions.

1.What does isolation 1-2 in the data sheet? I can't catch meaning of number(1-2).

1-2 mean isolation between output ports 1 and 2

2.What happen if isolation becomes higher or lower??

higher isolation mean better power balance. 16dB isolatiom realistically mean about 0.05dB power imbalance off the specs at worst -10dB output reflection

3.Does fast response on falling edge mean small number of fall time? If I want to get low power dissipation, then 800/400 is better than 400/10??

yes, fall time at output is faster in 400/10 device
yes, 800/400 likely mean device with lower power dissipation

 

[Jiho]

Thank you. I absolutely understand about pulse response time. But I still have two question.

1.What does isolation between output ports 1 and 2 mean? Is ratio of isolation between port 1 and port 2? Would you kindly explain me isolation 1-2?? I know isolation mean -10log(Pout/Pin). But I confuse what 'isolation 1-2' means. What are output power and input power at 'isolation 1-2'??

2.How can I calculate power imbalance off 0.05dB from 16dB of isolation. Does table exist??

 

[trurle]

Thank you. I absolutely understand about pulse response time. But I still have two question.

1.What does isolation between output ports 1 and 2 mean? Is ratio of isolation between port 1 and port 2? Would you kindly explain me isolation 1-2?? I know isolation mean -10log(Pout/Pin). But I confuse what 'isolation 1-2' means. What are output power and input power at 'isolation 1-2'??

2.How can I calculate power imbalance off 0.05dB from 16dB of isolation. Does table exist??

Isolation 1-2 is the power of input 2 needed to produce 0 dBm power at output 1
Regarding reflection to imbalance calculation, it is not on standard tables. No short explanation here, you need to understand the conversion of logarithmic (dB) scale.
In linear scale
imbalance=(delta(loss)+delta(isolation)*delta(reflection))/loss
where delta is worst-case discrepancy between ports
for example, delta(loss) for 750 MHz at table 1 is
10^-5.26/10-10^-5.32/10=0.00409=-23.9dB

 

[sophiecentaur]

What does isolation between output ports 1 and 2 mean?

Here is an alternative / additional view:
I am assuming that you can handle dBs ok.
If the ports are not totally isolated from each other, any power entering one port will result in some power from the other port. If one port is mis terminated then the reflected power will end up mixing with the wanted signal from the other. The only way to quantify this is to show the isolation. That, of course, doesn't tell the whole story because the relative phases could be anything - the vector sum or difference being the maximum possible error.