Home.
January 2025 - Accurate Attenuators & Calibration Standards
A few years ago I made some attenuators that would (among other things) do a simple job of protecting the front end of my spectrum analyzer. At the time I was using the very useful SDR Play receiver in conjunction with the free software that enables it to also be used as a spectrum analyzer.
I felt that device would benefit by protecting from overload too.
I called them 10dB (ish) attenuators because they weren't quite 10dB. The reason being that I had used preferred value resistors which were closest to the calculated necessary values. They did the job in a rough and ready way.. Moving forward in time we can now buy a great hobby spectrum analyzer for way under £100 and if you insert a 10dB attenuator - and then want to accurately measure, you would like the confidence of knowing that it really is providing 10dB attenuation.
Last month, I had the chance to join in a "limited numbers group kit offer " provided by a QRP group in California. These units use precise values of surface mount resistors with a small printed circuit board cleverly engineered to provide excellent response all the way past 1 GHz.
In the above example you can see how Doug KI6DS finished off his attenuators.
Doug has advised me "A set of boards is available, all the builder has to supply are the parts.
They get 7 attenuator boards and the 2 calibration standards boards for $8 shipped in the USA.
If you want a set of boards and live in the USA, send $8 via PayPal to KI6DS1@gmail.com.
If you live outside the USA, contact Doug at KI6DS1@gmail.com and we will work out the details.
More info on all of our projects at CalQrp groups.io group".
In the photo above is the supplied printed circuit board. I used a blob of Bluetak (underneath) to hold the PCB and the SMA connector still while soldering.
As mentioned in a previous note , I really enjoy the challenge of building with surface mount devices and find it isn't a chore as long as I have good magnification and a gadget to hold the blighters still while soldering.
Now if you were to try duplicating this kit you wouldn't have the luxury of the super little printed circuit boards provided with the kit. But with something like "perfboard" you could fashion something close utilising 1206 size SMD resistors. These are closer to pure resistance than the axial lead alternatives that I previously used with the "10dB ish" units.
Short connecting wire will substitute for the tracks on the CalQRP PCB.
Of course, at some point I was bound to sneeze and I managed to lose one of the 866 ohm resistors in the 1dB unit. Fortunately, I have a good supply of surface mount resistors, but only preferred values - and probably not all of them!.
The ideal computed resistor values were supplied with the kit instruction (as shown below:)
The value actually supplied for R1&R2 was quite close (to the calculated value) at 866 ohms.
The resistor was marked 8660. The convention is that if there are four digits the first three digits are the value and the forth is a (x10) multiplier. If there are just three digits, then the first two are value and the third a multiplier.
So if I was going to replace that 866 ohm resistor it was going to be with two resistors in either a series or parallel combination. Bless the people who create this handy web site, but no doubt you can find other places that will compute the values needed.
The "plus" in the second column indicates series combinations. Those with two vertical lines are parallel.
You can see below that I managed to squeeze two resistors together on the one set of pads. I chose a parallel combination as they fitted the pads better. I went with 1000 ohms and 6800 ohms with the resulting resistance of almost 872 ohms. Just 6 ohms short of the target.
The resulting fix produced performance just like the others.
This is useful information for those who would want to build from scratch when the precise calculated values are not at hand.
Once built I thought that these units needed some kind of protection and some screening might be useful.
The fiscally prudent QRPer will look in the junk box for inspiration and I had a few Bic pen tubes squirreled away for that one day when they might become useful.
A Dremmel cutting disc was used to make a slot.
.. and the attenuator board slid inside.
I wound copper "slug" tape around the pen tube then soldered a wire between the two connectors while also soldering the tape. A flux pen is useful when soldering to tape.
Finally I used heat shrink sleeving to cover the copper tape.
The NanoVNA showed the frequency response over 1 to 50MHz for the 10dB attenuator.
The finished project is shown below and here you also see the two VNA calibration standards from the California Club. One is terminated in sockets, and the other plugs.
All three standards - Open, Short and Load are on one board.
The problem with the individual calibration standards that come with the NanoVNA is their propensity to roll off onto the floor (where they go to hide). It is far more convenient (as in the example below) to construct something that brings the short, open and load standard together in one unit.
With surface mount resistors you can easily do that and it will work pretty well over the usual ham frequencies.
I wouldn't make any claims for the following unit in the GHz range - but then, I do not build things - or make antennas way up there.
.
Above on the left are three SMA sockets soldered together in a triangle shape. On the right is one of the sockets with a crude illustration showing a 50 ohm SM resistor soldered to the centre pin and the ground of the socket. You can use two 100 ohm resistors in parallel instead - if that's what you have.
One of the other sockets has a short circuit and the other no connection at all.