38 - 100W Power Sampler

 A previous article discussed attenuators, the three resistors within them and the need for some of them to be higher power than you'd typically find in your junkbox. Looking at signals up to a watt or two was ok with the attenuators made of smaller low power resistors and I use these to protect my NanoVNA and spectrum analyser. Once the power goes up to 10W you need more expensive resistors, which is why I bought a 10W 40dB attenuator. This allowed me to test my 10W linear Power Amplifier. But it was expensive - I paid nearly £30 for mine. 

What to do about testing amplifiers at higher powers? Luckily Wes Hayward W7ZOI designed a "Power Sampler" as part of his 6 page article in ARRL June 2001 entitled "Simple RF-Power Measurement" (Pages 38-43). In it he uses a simple resistive tap and creates a small box with three connectors, Power in, Power Out and Sample out. The sample is 40db below the input. The key to understanding this device is to think of the RF passing through the device, but a tiny portion of it gets passed to the third port. For it to work you need 50 Ohm loads everywhere - a high power dummy load on the power out port and, usually, an instrument  with a 50 Ohm input impedance on the sample out port. If 10mW is too high for it you can insert an attenuator between the sample out port and the instrument.

To understand the theory of the sampler you need to know about potential dividers.

As the text in Figure 1 above says you need to take into account the load resistance and the series internal impedance of any voltage source connected to the input. Normally 50 Ohms, Figure 2 below has the circuit of Wes's power sampler redrawn to take these into account. Note that R1 will be three 820 half Watt resistors connected in series. And R2 should be 51 Ohms (gives a slightly more accurate value than 50 Ohms.

I did a spreadsheet to calculate the exact values, these will need confirmed by measurement and calibration. The resistors are 5% and I don't have 51 Ohm resistors at the moment, I can either parallel two 100 Ohm resistors or use a 47 Ohm one. This gives 40.12 dB or 40.39 dB

The spreadsheet is in Figure 3 below and stored on my google drive at Blogfiles folder /Resistive_tap.ods

Fig 3: Screenshot of spreadsheet

It is easy to design a low frequency power sampler, at higher frequencies the inductance of even a straight bit of wire, and the capacitance from the resistors matters, the physical layout is critical. To avoid spending ages experimenting I tried to imitate Wes's layout exactly. Though his design is good to 500MHZ, he did use N-type connectors, I choose to use SO237 sockets as I was happy to work with lower frequencies. Wes linked the Power in and Power out with a flat piece of Brass sheet to reduce inductance, a 1 inch by 1.5 inch piece just fits a Hammond 1590A die cast box (the box is fairly small and available from Mouser.com at under £7 or from Amazon at £13 with free postage.). I had some one inch wide thick copper band which would do instead of the brass and I had to purchase half watt, 820 Ohm Carbon Film resistors. (if you want some copper and some resistors free, just ask as I ended up buying 20 for £3!)

Here is Wes's schematic and a drawing of his assembly

Note the flat plate forms L1 and a tiny piece of insulated wire forms C1, it is probably about a third of a picofarad, 

C1 extends just over half an inch (0.6")  from the plate and is kept parallel  and touching R1A and a bit of R1b

Use #22 AWG insulated hookup wire.

Fig 4: Schematic of constructed Power Sampler  

(From June 2001 QST © ARRL)

Fig 5: Drawing of the 40dB power Tap assembly. A 1x1.5 inch piece of brass links Power in and out

(From June 2001 QST © ARRL)

Here is a photograph of my case and the tool I used to make the holes for the UHF SO237 and BNC connectors. 

 Fig 6, the parts of my resistive power tap

The eagle eyed among you may see two 100 Ohm resistors in parallel instead of a single 51 Ohm value - I have none at present so I am making a power tap of -40.12 dB instead of -40.04dB - close enough...I made up the unit with the compensating "capacitor" that the original article describes (a 0.6 inch of 22AWG wire conneced physically close to the first 820 Ohm resistor. I then connected my NanoVNA to the Power in port and the BNC sample port. Most importantly I connected a dummy load to the power out port. I swept the NanoVNA from 3.5MHz to 30MHz for my first test and observed S21. The readings on a poorly calibrated NanoVNA were -40.35dB but more importantly they were flat across the entire HF band. A second sweep to 500MHz showed a 2dB flucuation.

Fig 7: Without a compenstaing capacitor a flat sweep from 3.5MHz to 30MHz 

Fig 8: Sweeping from 3.5MHz to 500MHz showed some fluctuations, (this is with no Capacitor)

I then added the capacitor, you can see it below in Figure 9.

Fig 9: A view inside the completed unit, complete with Wes's compensation capacitor (yellow wire)

With the wire added I repeated the measurements, shown in Figure 10 and 11 below.

Fig 10: 3.5MHZ to 54MHz with Capacitor in circuit, output has risen a bit but is still flat.

Fig 11 3.5 to 500MHz with capacitor added, the fluctuations have increased a tiny amount.

I also tested the return loss by plotting S11, it was fairly poor with an SWR of 1.3 to 1 for the HF bands. I then bypassed the unit and tested my dummy load directly and discovered it has an SWR of 1.3 to 1 all on its own, obviously you need to be careful with dummy loads! mine was made by Bird and should be good. I did feed the dummy load through a length of coax that might have been near 1/4 wavelength at UHF, or perhaps I need to more carefully calibrate my nanoVNA. I usually use the NanoVNASaver PC software and Calibrate there but my windows 11 upgrade has created a USB driver problem so for now I have to use a magnifying glass and photograph my nanoVNA screen.

In summary, this resistive power tap will work very well on HF and 6M. Still useful on UHF up to 500MHz, but I'd need to plot a calibration/correction chart if I wanted sub-decibel accuracy. However as a way to get my spectrum analyser to view my linear amplifier outputs, it will function very well.