tripler & PA will affect frequency deviation?

[cybq]

frequency multipliers in fm radio - tripler

I made a tripler and PA to transfer 800MHz FSK signal to 2.4GHz band. But the output frequency deviation of some chnnels changes! i.e. not 3 time as value of 800MHz. Why? and the output deviation to be affectd by
shield cover location. This will affect the sensitivity.

 

[flatulent]

how to measure fm deviation using the bessel null

Something is wrong. You cannot fool mother nature. Theory and practice says that the FM deviation, phase modulation angles, and phase noise will scale as the multiplication ratio.

I suspect that there is some measurement problem in your data taking. Are you putting just one carrier in and changing its frequency for each measruement?

Are you filtering the output so that there is only the 3x component? If you have several harmonics at once, your measuring instrument may not work well.

 

[nandopg]

frequency deveation

Dear friend,
The frequency multiplier effect over an angle modulated signal is to increase the deviation exactly by the multiplier ratio.
What , for sure, is happening in your application is the VCO sensitivity (MHz/Volt) before the multiplier is changing with the frequency.
You can check that applying a test tone to the modulator with amplitude and frequency for to the first Bessel,s null. The modulator should be tuned at the lowest frequency.
Then you tune the modulator for the mid band frequency. The null observed should be there (the test tone should be the same as before). If not you change the frequency of the test tone to get the null back. Write down the new test tone frequency for the null.
Now repeat the measurement for the highest channel frequency.
The difference between the frequencies for the first null in each case should be appearing x3 at the tripler output, this is, the sensitivity of the VCO is changing over the tuning range and this is why you are getting a change in the deviation at the tripler output.

NandoPG

 

[webdog]

the output deviation to be affectd by
shield cover location. This will affect the sensitivity.

This sounds like selfoscillation! It can oscillate near your operating frequecny and show up as sidebands. I have seen things like this in class C amplifiers. Try to tune the input or output and notice if it affects the sidebands. If so, you have an oscillator

Trying to resolve these things without a spectrum analyzer can be very time consuming and really frustrating....

Good Luck
/WD

 

[cybq]

tripler & PA affect frequency deviation?

This problem disturb me for so long during my developing a RF module for 2.4GHz wireless project. I don't think I fool mother nature. What I discripted is true. I only want discuss about this strange phenomenon which I met and share my experience because I have now resolve it.

1. If the 800Mhz is input without modulation(CW), the output frequency is no problem, i.e. 3 times the input.
2. If not connected to the match net to tripler, the deviation of every channels(low, middle, high band) is correct. This means that the VCO sensitivity over all frequecy is no problem. And I surely affirm that the tripler and PA both have some degree affection to the deviation, as well as the shield cover location.
3. There is no oscillation because of the sheild cover location.
4. I think there are no noise coupled to PLL's offchip loop filter for the enough distance from tripler and GND protect.

At last, I can only conculude that the only propable reason is the reflect signal from the tripler. I resolved the problem through the following efforts.
1. Redesign the tripler with higher convertion gain to replace the previous tripler with the aid of ADS.
2. Use ceramic filer to replace the previous microstrip inter finger filter, which with less insert loss and more accurate 50ohm in/out impedance for convenience of following match design.
3. Shut a capacitor at the receiver input match part after the T/R switch.

Now, everything is OK. Hope the above discussion will helpful for all friends. Lucky!!

 

[nandopg]

Dear CYBQ,
What you described is a problem of VCO frequency pulling, what means VCO sensitivity changing over frequency.
If you measured the VCO deviation before the tripler with a 50Ohms load connected to the VCO you can have a reasonable flat sensitivity.
Now if you measure the sensitivity of the VCO over the channel frequency range but with the VCO loaded by the tripler input, for sure you will have a bad sensitivity caused by a poor input return loss of the tripler.
Anyway what you did is right: to improve the input return loss of the tripler.

NandoPG

 

[yellowtooth]

Mr XIE,

Have U solve the problem now ?

Can U explain it ?

 

[flatulent]

deviation measurement

How did you measure the deviation?

 

[cybq]

deviation measurement

My frequency deviation ways as following.
First, make it to work in CW mode to display it CW output on spectrum analyser.
Second, with "1" (high level) or "0"(low level) to modulation the signal, display the new frequency. so the diffrence is what the frequecy deviation.

 

[flatulent]

tricky

This method can be tricky. You have to temporarily make the bit rate very slow compared to the deviation so that you get two clumps of spectral lines around the deviation.

If you can sine wave modulate you change the modulating frequency to get nulls of the carrier when J sub 0 (zero order bessel function) goes to zero.