[SOLVED] 30-520MHz PA output terminated into filter causing power drop

[ahsoopk]

Hi everyone,

I am trying to sort out a design that has following line

Pin -> VGA -> gain block -> driver amp -> PA -> Filter -> directional coupler (coupled signal fed back to VGA control) -> diode PIN switch -> Relay -> Pout

Now problem is that PA can deliver and does deliver 48dBm when filter is not in loop but as soon as the filter gets in the loop power drops back to 43dBm at some frequency points. One more important thing, Filter consists of 5 different filters that are switched in by relays controlled by micro-controller. PA's last stage has a BALUN at it's output to provide the match.

My understanding and conclusion is that because BALUN works on wavelength theory and it's impedance transformation heavily depends on the impedance seen by it on both sides. Now in this case filter is providing a dynamic impedance that is changing with frequency rapidly and so is the match seen by the PA hence it doesn't deliver the power on those particular frequencies. But if this is the case than how can i solve this and if this isn't the case than please guys help me to solve it,, all ideas are welcomed and deeply appreciated.

Kind regards and many thanks

 

[BigBoss]

If you use the filters directly at the output of the PA, it's normal because every filter bank may have different input impedance and it shall disturb the optimum load condition of the PA.
Because you're working at pretty high powers and optimum load matching circuits are generally very sensitive and a single Balun circuit can not provide a proper matching during whole band.
But I have a question... How do you expect a constant power during this large band with a single Balun circuit ?? Because an optimum load impedance may vary very radically from point to point and working in this large band 30-520MHz would be very tough from point of view of optimum loading.Balun circuit may give the optimum load condition either at a few frequency points or utmost in relatively a narrowband..
Can you share your thoughts with us ??

 

[vfone]

Because this is pretty wide band design I think the PA use a wideband toroid transformer BALUN (and not a microstrip or TL one), and this BALUN doesn't work on wavelength theory.
In this case the output impedance of the PA should be pretty constant vs frequency. If is not constant and if it varies with frequency, the only way to solve the problem is to tune/match the input impedance of each filter to the output impedance of the PA.

 

[chuckey]

I think that you have to check the directional coupler and the filter as individual items to see which is causing the problem. Your balun will be a problem , but only at high frequencies or very low frequencies. You do not say at which frequencies the dips in the amplitude response occurs, as the directional couplers are roughly tuned to their frequency band.
Frank

 

[ahsoopk]

Dear all,

Last two days were very interesting and learnt something that should have been quite obvious to me but I think sometimes we all dig ourselves too deep that forget the basic concepts.

Here is the thing, PA output is a BALUN based matching circuit that changes its characteristics with frequency. Filter is tuned using VNA which is a 50Ohm system and filter ports are tuned based on that impedance.

Now we put two (PA+Filter) together and it works on some frequencies and doesn't on others, so solution is "tune the filter while its connected to the PA i.e. you are tuning the filter under large signal impedance condition of the BALUN that is connected at the input of the filter rather than a perfect 50Ohm system under small signal conditions that was VNA. Doing so, I have achieved the desired results or should i say close enough that I can live with it.

Now last problem that I am having is that VGA+Gain block should deliver around +13dBm at the input of the PA driver to get 47dBm out of the PA this happens perfectly fine when unit is open but as soon as the unit is lidded power drops from 47dBm to 40dBm at some frequency points and mostly above 250MHz. One thing is important to mention that VGA is not far away from the output connector/relay and its one of those VGAs that has BALUN at its input and output. I thought of following solution and would like your second opinion to have a piece of mind

***Move the VGA+Gain block away from the output connector/relay more towards the front end of the PA and enclose it in its own solder down cavity.

Please your suggestions are very valuable and have help a lot to solve the first issue.

Many thanks gents.

 

[vfone]

I think when the lid is in place, the circuit that is affected is not the VGA+Gain_block, but the filter itself is de-tuned.
When you tune the high frequency filter you have to take also in consideration the "lid in place situation".