Part 5 of the W4KAZ filter project series discusses filter losses, an idea for getting a very rough S-meter calibration, and trying to estimate the out of pass band attenuation provided by the filters.
The Losses:
The filters do have losses in the pass band. This is known as the insertion loss, and is reported in db. When discussing the pass band, we want the losses to be as low as possible, or approaching 0.0db of loss. The old rule of thumb is for every 3db of loss you are losing about half of your power. So, 100 watts of RF transmitted through a 3db loss component means there is only 50w coming out the other end.
Run that through the loss formula…. db loss = 10*[log(100/50)] = 10*log(2) = 10*.30103 = 3.01db of loss.
Since loss is defined as a ratio of the actual power levels, a simple watt meter and dummy load can be used to measure the losses of a component in db. That gives a nice yardstick for comparision to known commercial filters. The accuracy of the wattmeter is an issue, but part of the game is to compare the values I come up with against values measured with better test equipment. If I ever manage to hook up with one of the guys who are willing to help with that.
The set up to measure the loss in the pass band is simple.
Transmitter–> filter –> watt meter –> dummy load
By replacing the filter with a barrel connector, you can get the baseline power. The watt meter on hand here is not sensitive or accurate enough to use the same technique for measuring signals outside the passband. Not if the filters are working. 😮
Aside: This is also a good way to test a piece of unknown coaxial cable. Rather than rely on an estimate of what the loss should be for a known length of similar cable, it is pretty easy to measure the loss. A quick computation of the loss into db gives you a yardstick on the quality of the cable by comparing it to known losses specified by cable manufacturers.
The Guess-Estimate:
Anyway…. My set of NVARC filters came in measuring actual losses between 0.6db to 0.8db, and about 1.0db when installed in the integrated switch box. The set of K4VX filters came in at 0.3db to 0.6db.
The problem is that the insertion losses in the pass band tell little about their effectiveness on the 2N or N/2 harmonics.
The only tool available in the KazShack for measuring this type of loss turns out to be the S-meter on the receivers. Receivers are quite good at hearing RF. Kinda their whole purpose in life, right? The new problem is the unknown scale of the S-meter. Is it telling us anything useful?
So: How to calibrate the S-meter?
Okay. I couldn’t solve that one. Is there a possible work-around, or a way to determine the existing calibration of the S-meter?
This puts me off into an area that may eventually turn out to have little real-world validity, but here’s what I came up with. The FT-920 has a three step attenuator pad which is a known quantity. Assuming a simple resistance pad can be easily calculated and implmented by engineers capable of designing such an otherwise comlpetely slick gizmo. For some reason the pad is coincidentally in 6db steps, giving 6,12, and 18db. How convenient. 😮
The unofficial rule of thumb is that an S-unit is supposed to be 6db, with S-9 the 50 microvolt level. So with 18db attenuation, an S-9 signal should be knocked down to S-6. I don’t have a 50Ö¶ standard, one of countless other things I don’t have, but I am able to generate a signal at various levels. So I decided to use the attenuator pad to calibrate the S-meter markings. Although I may have no idea what level actually causes the meter to read S-9, I CAN use the known values to figure out the values from S-9 down, or S-9 up. I still don’t know the actual signal levels or what signal level corresponds to an S-9 meter reading, but the scale allows measurement of the relative differences in known quantities. In this case, that is exactly what I need.
What this gives me is a round-about way to guess-estimate the effectiveness of the filters where it counts, on the sub harmonics or harmonics. If I know the value of attenuation causing a signal drop from between S-9 to S-2, inserting a filter that causes that same drop will have that amount of attenuation on that frequency.
Nothing is ever THAT easy. S-meters are known far and wide for non-linear behavior, right? Sheesh. But life is full of surprises.
Previous in series: Band Pass Filter Fever – The Kludgy Switch Box – Part 4
Next in Series: BPFF – The Guess-timated Scale and actual Guess-timates – Part 6
Leave a Reply