Low cost coupling for RF power detector

That should be OK. The 1K resistor may have a little shunt capacitance, which might peak the detector response a little. You can calibrate around that, or add a stuffing-option cap to ground at the detector input.

I'd put the DC block cap after the 1K pickoff. An ordinary 0603 cap should be good enough.

On a sunny day (Mon, 22 Feb 2016 09:35:03 -0800 (PST)) it happened Darol Klawetter wrote in :

If you know you are working into 50 Ohm, and SWR = 1, than just measure voltage with a simple diode detector. Calibrate your meter for non-linearity due to diode forward curve, either draw a scale (analog meter), or use a lookup table in a micro with ADC and some digital display / output.

To compensate for the capacitance of the 1k resistor, just narrow the main rf track a bit for a few mm on either side of the resistor. This will work very nicely to 3GHz. Also, as already suggested, connect the resistor to the rf track and then the capacitor afterwards as this will minimise the stray capacitance seen by the main rf path.

Choose a resistor package size such that the footprint of one end fits entirely onto the main rf track with little or no overlap.

John

Skyworks and others have surface mount schottkies that have about 0.2 pF capacitance and a few volts reverse rating. One of those might work.

The reason a directional coupler is used is to measure only the actual FORW ARD power, ignoring the reverse power. Sometimes both are measure and re ported.

If you use a simp[le voltage divider as you suggest, it will work fine for a matched load becasue the reverse power is small. But, if your load is m ismatched, the voltage at the point you measure will be a function of the phase angle of the reflection, so no longer an accurate indication of power .

If you don't care about any of that, your plan will work fine and is in fac t often used.

Mark

hm =

de=

ly =

nal=

er =

r t=

eri=

rov=

e m=

ara=

ll =

hig=

be =

sto=

eli=

's =

tec=

s?

ut.

Rather than use a simple diode detector, my plan is to use an RF power dete ctor IC that provides enough amplification of the input power to allow a lo w cost 8-bit ADC to be used for measurement. Unless the diode were connecte d directly to the RF main path rather than through the resistor, I doubt th e ADC will be sensitive enough to read the diode directly because my ADC's LSB requires 12.9mV to flip and the input power measured through the resist or will often be less than that.

Darol Klawetter

hm input of an RF switch on my PCB. To do this, I'm going to add a RF power detector to this input. Typically power detectors are not connected direct ly to the RF circuit that they measure but are connected through a directio nal coupler to avoid load mismatches on the measured path.

er (such as the Minicircuits SYD-20-33+) to sense the RF input level. Rathe r than using a coupler, I'm considering using a resistor and capacitor in s eries between the RF main path and the power detector input. The resistor p rovides a broadband coupling factor and would be around 1k ohms and would b e mounted within a few millimeters of the input pin to the RF switch. The p arallel combination of the resistor and 50 ohm input of the switch will sti ll appear as roughly a 50 ohm load to the microstrip transmission line (my highest frequency will not exceed 3 GHz). The capacitor blocks DC and will be large enough so that it's impedance is always small compared to the resi stor.

elivered to the detector (as do many directional couplers) but the detector 's sensitivity is sufficient to still measure the power levels I need to de tect.

s?

Thanks for help.

So is the that narrowing the track adds a bit of inductance?

On a sunny day (Tue, 23 Feb 2016 07:36:09 -0800 (PST)) it happened Darol Klawetter wrote in :

Very long time ago I made a couple of very linear detectors with something like this: + + | | e PNP | b -- [ ] 100 c Q1 | | R1 |-----| | [ ] [ ] 10k | | 10k | | | /// e

--||-------------- b PNP 1n c Q2 |---------- out | | [ ] 1k === | R2 | 100n /// /// The thing has a gain of 10, and is reasonably linear. Q1 biases Q2. Uout is close to zero and temp compensated at 0 Vpp RF in. At .1 Vpp RF the output is .5 V DC or something. The gain is R2 / R1.

I guess these days the chip solution is simpler. I used it, without the 100 nF, as 3 tracking detectors for R,G,B video in a color system I demonstrated,

It mostly reduces the trace capacitance. You can neck down the main trace at the resistor pickoff junction.

I do that with an x-acto knife, using TDR to see how I'm doing.