1

I'm using a power amplifier that requires an external blocking capacitor for the input and output ports and I'm trying to decide the best value to use. My understanding of this is that I choose a capacitance that is sufficiently large such that the 3dB frequency is a good deal lower than my lowest frequency of operation (answered here). I also imagine that it's important to choose a capacitor with the smallest possible ESL. Are there any other considerations that are important here? What about RF currents heating such small caps?

So, I'm thinking of using a 0402 100pF cap which, given my lowest frequency of 5.3GHz and min Rload of 6ohm, should put the corner frequency around 250MHz, well below the 5.3GHz. Does that seem sensible?

Autistic
  • 14,235
  • 2
  • 27
  • 65
MattHusz
  • 1,033
  • 9
  • 15

1 Answers1

2

You want an NPO Cap with a SRF >> 8GHz and ultralow ESR and ESL at 6 GHz.

Not any 100 pF 402 cap will do.

Supply noise filters depend on impedance ratios of series LR lossy ferrite and low ESR high SRF load caps and PCB characteristic impedance for Pwr/Gnd layers. These all affect the s21 attenuation response.

Update

At 6 GHz, perhaps not any 402 cap will do above 10 pF for self-resonant-frequency, SRF and geometric parasitics can cause a resonance with other shunt parts if not careful.

There is a geometric limitation with an aspect ratio of L:W=2:1 e.g. 402 603

  • ESL is fixed for any size of L:W conductors whether it is in thou", mm or an entire ground plane with an aspect ratio of 2:1, thus you would be searching for edge type MLCC or 1:2 aspect ratio caps or smaller parts like 0201 for a shorter length in smaller values < 10 pF.

  • This also supports using the thinnest dielectric layer between Pwr& Gnd layers but then many microvias must be used to reduce ESL.

  • C decreases inversely with electrode gap which is hidden in vertical layers and also affects ESL and C thus you must refer to the OEM datasheets or their simulation tools and then verify with your PCB layout that it does not compromise the performance.

  • Higher voltage ratings also reduce SRF values slightly

  • how to measure SRF

  • e.g. 0402 10 pF KEMET cap SRF = 2 GHz has an effective series inductance ESL of \$L=\dfrac{1}{{(2\pi f)}^2C}\$

    • this formula intuitively for a fixed SRF, you can see that if you triple f to 6 GHz, C must reduce by 9 , so 100 pF is never going to work well.

enter image description here

Tony Stewart EE75
  • 1
  • 3
  • 54
  • 182
  • I'd like to know what is incorrect about this statement. Does he want me to do his searching? Although it appears this is his 1st rodeo in 6Ghz and microwave design. – Tony Stewart EE75 Jan 12 '19 at 04:46
  • 1
    It was useful to me and I definitely do not expect you to do the searching for me. However, if you have any general recommendations for determining ESR/ESL at specific frequencies that would be great. I'm having trouble finding those numbers in the datasheets. And yes, this is my first RF design.. – MattHusz Jan 12 '19 at 05:24
  • 1
    @MattHusz it might not be on the datasheet, sometimes you can get more detail by searching the part number on the manufacturer's website. For 8 GHz you might need to look to specialty vendors like ATC or Johansson. Or a broad line vendor like Murata might have a specialty part. – The Photon Jan 12 '19 at 15:30
  • Kemet KSIM simulator may be a very good helper when you need some informations on ESR/ESL, it even provide S21 which is precisely what you are asking for. – Wireless Learning May 16 '22 at 13:52
  • 1
    Yes Kemet's tool is great for their parts https://ksim3.kemet.com/capacitor-simulation and is a good reference for testing others. – Tony Stewart EE75 May 16 '22 at 14:43
  • 1
    Thanks to @WirelessLearning for inspiring me to update my answer in general terms rather than a specific solution , but I concur with Photons OEM recommendations – Tony Stewart EE75 May 16 '22 at 15:33