How to select the decouple cap value?

[John_li]

are decouple caps needed

Hi friends,
I am newbie for the RF design. I want to design: TX---900M and RX---5.8GHz,my LO:3.9GHz (generated from MMIC transceiver),and I want to know how to select the power line decouple capacitor values for MMIC,Mixer(IF:1.9GHz),5.8 GHz LNA .How can I measure and observe signals that I should reject from the power line?
Thanks

 

[biff44]

decoupling cap srf

You can use an ac coupled oscilloscope on the DC bias line to see what is there. YOu can also use an ac coupled probe and a low frequency spectrum analyzer, but you have to be really careful to not blow up the spectrum analyzer when doing it, so you one do it if you are really looking for low level ripples.

Often there are switching regulators used, and they have a big ripple component that needs to be filtered out. Sometimes you need CLC or CRC multisection filters in the bias line.

For an LNA, probably a 1000 pf in parallel with a 1 to 4.7 uF will be all you need. If it were a power amp, you would need lots more, especially if it was a difficult waveform (OFDM, QAM).

 

[kspalla]

decoupling value

If you have two separate regulators for LNA and mixer then forget about the isolation from the power supply.

If not, follow as per biff44 and take care in layout design such that the power routing will be at common point or plane but not a serial line.

 

[John_li]

spectrum analyzer ac-couple site:www.edaboard.com

Thank biff44 and kspalla for help. I have a reference schematic and it uses some small values (except large values for power ripple) Murata components. In 5.8GHz LNA, it uses 3p (0402) for RF short, but I checked the Murata component library, 3p is for 4164MHz(SRF);Mixer places 22p(1485MHz--SRF) and 33p(1256MHz—SRF) for power line, and MMIC(TXVCO:3.9GHz and RXVCO:3.6GHz,50KHz is for reference frequency for RXPLL and TXPLL) uses 4.7p(3.476GHz SRF).I do not know why not place the SRF value capacitor for RF short, what happens during debug and how can I observe so high frequency signal in power line when debuging?

 

[John_li]

decoupling signal lines

Anyone can help me?

 

[John_li]

how to select the value of a decoupling cap

how can i measure hign frequency in power line?

 

[kspalla]

decoupling cap mixer

What we recommended is the nominal limit, which means if we place this much cap will enable the signal to de-cap effectively.
If the design is using lesser then this value means either is is working as part of the matching network or de-coupling the same but at low rejection.

how can i measure hign frequency in power line?

To this question place the proble in power line and read with Spectraum analyser.

 

[biff44]

decouple cap

You should post the schematic and layout if you want detailed answers. You are not providing sufficient information in your questions.

 

[John_li]

mmic properly decoupling

Thank kspalla and biff44 for always kindness. And now attached part of schematic and layout, and
whether is it enough for analysis?

 

[kspalla]

how to select values for decoupling caps

This is purely layout issue as the power lines are crossing the signal lines.
I could not found single 3pF giving complete decoupling in the pictures from the links. they are associated with either L or another big Cap.
Why do't you try to avoid this power lines crossing signal lines by routing them in inner layer?

 

[John_li]

0402 cap self resonant frequency

Thank kspalla for reply.For cost down,i have to use two layers.your suggestion is right,i used Spectraum analyser to observe,nothing i can find.Maybe it is for reject LO signal that produce the power line crossing.
And i have another question,if i ignore the capacitor,what will happen in the output port?

 

[kspalla]

decouple cap library

I suppose you should observe the power depending up on the LO drive level.
Its tricky with the leakage measurements.
Generally it has 60dB isolation but to check for higher isolation requirement you should have the GND probe proper and make sure that you are in low noise floor may be at -110dBm with change in RBW, and LNA frent end etc.

if i ignore the capacitor,what will happen in the output port?

As mentioned by biff44, if you remove low cap then you may not filter the high frequencies.

 

[John_li]

how to select decoupling caps

Thanks,kspalla.Both of your suggestions are helpful to me.