DIY DC Block

[amitchell]

I had a need for a DC Block and some time to make one. This is what I came up with based on some similar projects.

Parts are as follows:

SMA Housing - https://www.digikey.ca/products/en?keywords=SMA-KIT-1.5MF

1uF 50V X7R Axial - https://www.digikey.ca/product-detail/en/kemet/C420C105K5R5TA91707200/399-13735-1-ND/6235471

10nF 50V X7R Axial - https://www.digikey.ca/product-detail/en/vishay-bc-components/A103K15X7RF5TAA/1103PHCT-ND/145937

Add SMA to BNC adapters as required.

[awallin]

did you measure the amplitude(and phase?) response vs. frequency?

would a 'plastic' capacitor be better?

[amitchell]

I did not measure. I tried to find a film cap that would fit but I could not.

I thought that the commercial units are a tiny enough size that a ceramic capacitor is likely what's in them. But, that is just speculation.


Sent from my iPhone using Tapatalk

[CopperCone]

I have a 12ghz bias t/dc block and its just a 0603 cap and a wound torroid in a heavy metal box.


You can buy quality high frequency ceramic capacitors on digikey that are rated to many ghz.

[DaJMasta]

I did not measure. I tried to find a film cap that would fit but I could not.

I think running some tests on a VNA would be revealing, but if it suffices for your applications, it's just fine.

What you have is basically what you'd expect from a DC block capacitor, but the capacitance values are very large, so it will work down to a lower frequency than most, but it will also probably not perform well at higher frequencies (maybe a hundred MHz or more), whereas SMA connectors normally being rated for many GHz of operation speed, you'd find a commercial block capacitor in that form factor to be rated to maybe 3GHz or more.  But if you don't require that, then it's not so much of an issue.

As mentioned, a small NP0/C0G capacitor (better dielectric) would probably be a better choice, especially for high frequencies, but the cost of an expensive DC blocking capacitor is the verification that the dimensions and parts in question produce an even frequency response over a wide passband - so it's sort of all about the testing and characterization.


Short of actually testing it, if you find your signal is fast (or has very fast edges) and you're seeing some attenuation or distortion, try switching to a smaller value NP0/C0G cap and see if it helps with performance.  If you've got parts of the signal down into the 100s of Hz range, then your 1uF cap is probably helping out passing that through a lot.

[amitchell]

I did not measure. I tried to find a film cap that would fit but I could not.

I think running some tests on a VNA would be revealing, but if it suffices for your applications, it's just fine.

What you have is basically what you'd expect from a DC block capacitor, but the capacitance values are very large, so it will work down to a lower frequency than most, but it will also probably not perform well at higher frequencies (maybe a hundred MHz or more), whereas SMA connectors normally being rated for many GHz of operation speed, you'd find a commercial block capacitor in that form factor to be rated to maybe 3GHz or more.  But if you don't require that, then it's not so much of an issue.

As mentioned, a small NP0/C0G capacitor (better dielectric) would probably be a better choice, especially for high frequencies, but the cost of an expensive DC blocking capacitor is the verification that the dimensions and parts in question produce an even frequency response over a wide passband - so it's sort of all about the testing and characterization.


Short of actually testing it, if you find your signal is fast (or has very fast edges) and you're seeing some attenuation or distortion, try switching to a smaller value NP0/C0G cap and see if it helps with performance.  If you've got parts of the signal down into the 100s of Hz range, then your 1uF cap is probably helping out passing that through a lot.

Thank you for the feedback, I mainly picked that SMA connector for its cleanliness and convenience.

The goal is to these for SMPS LF ripple and HF Noise measurements along with 50R termination and direct coax connection to the filter caps. My scope is limited to 200Mhz, hopefully this will be good enough. I do have various COG smd caps at my disposal if need be.

[DaJMasta]

Everything's a tradeoff, but this tiny image does a good job illustrating both the difference in cap size and dielectrics:



You can see that the 1uF cap would have a fairly low point where its impedance is lowest, and that the NP0 dielectric just has a lower impedance than X7R altogether.  Just from eyeballing the graph, maybe like 0.68uF or 0.47uF would be most suitable for low impedance from near DC (10s of kHz) to about 200MHz, but it's not that you'd see significant attenuation at 200MHz with the configuration you have now - probably only a bit of a reduction in a similar place to the scope's bandwidth rolloff.  Sounds like it will be suitable for what you're doing.