Wenn man ein Filter beliebiger Art messen möchte, muss man Eingangs- und Ausgangs- Impedanz des Filters an die 50 Ohm des NWT anpassen, da man sonst sehr stark verfälschte Kurven bekommt. Als ich letztlich vor dem Problem stand, viel mir ein Artikel aus der QRP Quarterly von 1997 ein. Ich habe nachgefragt und der Author hat mir freundlicherweise sein Manuskript geschickt. Vielleicht findet ja noch ein Forum Mitglied die Zeit, den Text zu übersetzen.
Kar Larsen, K5DI hat ein Excel Blatt geschrieben, mit dem man die nötigen Anpassglieder berechnen kann. Für die Praktiker: lasst Euch nicht durch die errechneten Widerstände mit völlig krummen Werten verwirren, der Rechner weiss es nicht besser. Es reicht, wenn man in die Nähe der Werte kommt.
From October 1997 QRP Quarterly, Idea Exchange column, WA8MCQ
CHANGING ATTENUATOR IMPEDANCES
From me, WA8MCQ--I always see attenuators for sale at hamfests, sometimes at very good prices, but there's a little "gotcha" that'll bite you if you're not careful--they may not be for fifty ohms. Before buying one, look carefully at it. It could be 75, 93, 125, 170 or 600; I've seen all of those and there may be more lurking out there. Assuming you're working only with 50 ohms, does that mean you can't use them at all? Not necessarily.
A couple years ago W6TOY bought a really nice Kay Electric step attenuator and discovered too late that it was for 75 ohms--probably why the price was so good. I told him I could sell it and get his money back and I did. A quick ad on QRP-L produced a buyer. I could have sold it as-is and someone would probably have had a use for a 75 ohm device, but I tossed in a free bonus from my junk box: a pair of "minimum loss pads" to convert it to 50 ohms. They had male and female BNC connectors on the ends, so they plugged right onto the attenuator.
A minimum loss pad is a simple resistive device consisting of a couple of resistors, used to convert between two different impedances. Connect it to anything, then plug your cable into it and instead of 75 ohms you'll see 50 (or 75 instead of 170, etc, depending on the value of the pad). It does have a drawback--being a resistive device, it introduces additional loss. But that's not a problem if whatever you're using can tolerate that, and the loss is a known value--just add that to the attenuation of the attenuator itself. The loss of a 50/75 ohm pad is 5.7 dB, and a pair of them on both sides of a 75 ohm attenuator adds 11.4 dB to the loss. That turns a 0-80 dB 75 ohm attenuator into an 11.4-91.4 dB 50 ohm unit. You still have the full range available; you've just shifted it a bit and converted it to the impedance you need. While you do lose a bit of flexibility on the low end--you're stuck with a minimum of 11.4 dB in this case--that's not always a problem since you won't always need those lower values of attenuation. And to use the unit, simple read off the attenuation from the switches or dials and add the 11.4.
I occasionally see these devices at hamfests, usually for 50/75 ohms. They usually look like attenuators, small cylindrical devices with connectors on each end, although some are in little rectangular metal boxes. Figure XX shows a typical BNC fixed attenuator; many of the pads look the same. How do you tell them apart? If it's an attenuator, there will be an attenuation value on it, such as 3 dB or 10 dB. If a minimum loss pad, each end will be marked with a different impedance, such as 50 and 75.
The sex of the connectors is critical. An example: If you have a 75 ohm attenuator with female connectors on both ends (the most common by far), you'll need pads which have males on the 75 ohm side and females on the 50 ohm side; otherwise you'll have to add a pair of adapters. (Unfortunately, the freebies I included with the W6TOY attenuator were backward to begin with--the female was on the 75 ohm side, male on the 50 ohm side, which would have required a pair of BNC male to male adapters to plug them onto the attenuator. But modification was easy enough. It wasn't too hard to unscrew the pads--although not shown in the figure, there are several flats on the units to grip with wrenches. (It is necessary to remove the label before unscrewing.) After it was open, I flipped the innards 180 degrees to get the proper orientation. The soldering work was easy enough, and working under a 7X stereo microscope--while not truly necessary--made it more enjoyable.)
You can sometimes find these things at hamfests; bargains exist, but they are often several dollars. Be sure that they are for the pair of impedances you need, and it's best if the sexes are properly oriented (see above; modification may be possible). Most that I've seen are 50/75 ohms, but I've also seen 50/600 pads (which introduce a little over 16 dB loss each). If you can't buy what you need, you can also make your own. One way is to buy some dirt cheap coaxial attenuators--maybe something of an oddball impedance that no one wants--and open them up, rip out the innards and replace with your own resistors. You could also get some small Pomona die cast aluminum boxes with connectors on the ends (preferably male and female; see Figure X) and build into that. You can often find these boxes at hamfests at good prices, either unused or with something already built into them. I don't like to pay more than a buck or two, though $3 is still a good price. They come in several sizes, and the best one is about an inch and a half long.
Here are the resistor values and the losses that these add. Don't forget that this is for one pad; if you use two of them, on both ends of an attenuator, you'll get twice as much additional loss (such as 11.4 dB for the 75/50 ohm conversion). Resistor values are given for both 1% and 5% parts. These were calculated using a freeware program from Teledyne Microelectronics called RF Toolbox. It's been around for quite a few years now and is fairly simple and limited but does have some useful features. It can be found at ftp.lehigh.edu, under pub/listserv/qrp-l/tools, filename XXXX. Although it shows a pi configuration with a third resistor of 90K or more, in the real world you could probably eliminate it with no significant effects. The pads that I modified did not have that resistor.
Figures X and X show two different attenuators with and without pads. These are two commonly seen types. Rotary units are more convenient since you turn the knob to whatever value you want and you read it directly, rather than adding up the numbers on the switches that are flipped up on the other type--a possible source of error. But the drawback is that rotary attenuators come in steps of 1 or 10 dB, and sometimes 0.1 as well. To get a range of, say, 0 to 70 dB you'd have to have two rotary attenuators in series, with steps of one and ten dB each. (Some contain two attenuators in one housing, with concentric knobs; they are longer and more expensive than their single counterparts but more convenient.) One of my Kay Electric units has switches for 20, 20, 20, 20, 10, 5, 3, 2 and 1 dB, allowing selection of any value from 0 to 101 dB in 1 dB steps (but be sure to add all the numbers correctly!).
Note that the resulting attenuator and pad combo is starting to get a bit large and clunky. There's not much you can do in the case of the rotary attenuator, but if you have one with individual switches you have the option of putting the pads inside the case and eliminating the add-on units. You could open the unit and wire in the resistors between the connector center pins and the original wiring. Be sure to put a label on the outside indicating the new impedance and the new minimum insertion loss of 11.4 dB (or appropriate value if the impedance transformation is something other than 75 down to 50 ohms).