Formallyzing a capacitive tap

[tkamiya]

Hi,

I just made a capacitive tap so that I can monitor my transmitter's output. 

Let's say there are 3 connectors:  connector 1 and 2 is a through line, and 3 is a tap.
1 and 2 are connected via small length of 50 ohm coax.
3 is internally terminated with 50 ohm register.
between center pin of 1 and 3 are connected via 10pf capacitor.

Now, I started thinking.  I did this as a matter of course but how do I formallyze this circuit?  50 ohm termination is easy.  It is represented to outside as a 50 ohm impedance connection.  But what about 10 pf capacitor?  It will "pick up" a small amount of RF from the through line but where is "the other" capacitor to form a capacitive tap?
The line is 50 ohm.  So how do I arrive at what voltage to expect on pickup terminal via 10pf cap?

[w2aew]

Since the "load" on the tap is 50 ohms, the voltage appearing across that 50 ohm termination is going to depend on the frequency - since that determines the reactance of the series 10pF capacitor. What you've made is a high-pass filter...

When I think of a capacitive tap, I think of the typical way that an old-school RF station monitor (ham radio) was made - by using a small capacitor from the transmission line directly to the vertical deflection plates of an analog oscilloscope.  Since the deflection plates "look like" a small capacitor, the result is a capacitive voltage divider, which gives a relatively constant divide ratio vs. frequency.

[tkamiya]

Maybe I'm not explaining it well enough.  THAT is what I'm trying to do, a station monitor like setup.

[tkamiya]

Yeah....  that's the word I was looking for....  a station monitor.  The only difference between that and my setup is, that input is going to a vertical amplifier rather than straight into deflection, and there is a piece of coax between the tap and the monitor.  Think that will make the difference?  I could also make this a current sensing setup, so it's not frequency dependent.  Maybe that'll be a better approach....

[joeqsmith]

https://www.eevblog.com/forum/projects/home-made-pa-and-load/msg635226/#msg635226

I made something similar but load was 50 ohms rather than an antenna and I wanted something flat from DC up to about 100M.  The C is a small section of magnet wire.  The wire was tuned to match a section of coax (and scope) which forms the second part of the C and R.

[w2aew]

Yeah....  that's the word I was looking for....  a station monitor.  The only difference between that and my setup is, that input is going to a vertical amplifier rather than straight into deflection, and there is a piece of coax between the tap and the monitor.  Think that will make the difference?  I could also make this a current sensing setup, so it's not frequency dependent.  Maybe that'll be a better approach....

Get rid of the 50ohm termination on the monitor side of the tap.  Then you'll have nearly a capacitive voltage divider involving your series capacitor, and the input capacitance of the scope (and capacitance of the connected coax).  It won't be perfectly flat vs. frequency due to the transmission line effects, but will work quite well.  Your tap capacitor will likely need to be quite small, probably 10pF.

[tkamiya]

Thank you.  Yes, it's already 10pf. 

Oddly enough though, 50 ohm termination is there so that length of coax does not become a factor.  I guess I'm thinking wrong.  While at this, I'll just measure both ways and see it will affect the flatness.  If this is not acceptable, I can always go the magnetic sensor route with a toroid.

[tkamiya]

I just tested, and this is not acceptable at all.

TG input into port that goes out to a terminator.  Tap connects via 10pf to cable back to SA.  1Mhz to 300Mhz, there are SHARP dips and peaks kind of close by.  They easily vary by 30db.  Total crap....!  The result was little better with termination for the tap output but not by much.

Plan B2....

[tkamiya]

To make the matter worse, BNC connectors are partly responsible for these peaks and dips.  I have tons of inexpensive type.  They fit rather lose.  So basically the whole thing is trash.  I'll have to start with quality parts.  I'll just turn this into a return loss bridge, so this ratio stuff doesn't become an issue again.

I'm sooooo disappointed.

[tkamiya]

OK, I got the desired effect.

Input to output is a piece of coax.  Shield connected only on one side.
I had an unknown toroid.  9 turns of wire.  Coax through it.

Response is 10 to 150Mhz basically flat. 

With capacitance involved, I could only get the expected frequency response based on reactance of C.  I would imagine number of turns, and more importantly, the core will need to be re-examined.  But the basics is done.

I can make stuff and measure....  what a concept.

I just recently started building up my own lab.  I still cannot get out of the old habit of not knowing the result and being "ok" with it.

[Wolfgang]

If you want even more accurate results, why not use a Bruene bridge ?
Then you could also measure forward and reflected signals.
Another way would be a resistive tap, good from DC to above 70cm band.

https://electronicprojectsforfun.wordpress.com/rf-module-gallery/swr-bridges/a-bruene-bridge-directional-coupler-with-direct-rf-output/

and

https://electronicprojectsforfun.wordpress.com/rf-module-gallery/the-rf-splitter-and-combiner-gallery/a-resistive-tap-attenuator-for-rf-measurements/