Some time ago I bought a couple of low-pass / TVI filters from a ham-fest. It is mandatory to ensure that you are not producing out-of-band or otherwise harmful interference and so having such a filter assists in ensuring that you are in compliance with the regulations. I picked up a Bencher YA-1 52Ω 1.8-30MHz 1500W low-pass filter for about $20 and a B&W FL-10/1500 filter for around $10.
Bencher and B&W low-pass filters
DSA815-TG being prepared to test a Bencher YA-1 filter
Having somewhat recently purchased a spectrum analyzer with tracking generator, I was naturally curious to see how the two filters performed. I started by “normalizing” the test system to account for any impedance mismatch and transmission line losses inherent in the system prior to the addition of the DUT (device under test).
Normalizing the DSA815-TG for filter measurements
To normalize the test system, you must make a connection from the tracking generator output to the spectrum analyzer input using the cables which will be part of your test-bed. I used a BNC-BNC barrel connector in the photo but changed to a SO-259barrel for the actual measurements.
The advantages of the BNC connectors are as follows: they are rated for higher frequency than SO-259/PL-259 connectors and they are also “quick connect.” My rationale for using a SO-259 barrel was to allow the inclusion of the BNC-to-PL-259 adapters which are necessary for the test (since the filters are equipped with SO-259 connectors.)
Normalized trace for filter measurement
The normalization procedure was done from 9kHz to 500MHz at a 1MHz RBW (receiver bandwidth) with no input attenuation and the tracking generator set to output at 0dB. The spectrograph shows a nice flat curve which is what you should expect as the spectrum analyzer is mathematically accounting for any impedance mismatches, etc.
Bencher YA-1 filter being “swept” by the DSA815-TG
Upon inserting the DUT into the system, it is immediately apparent as to what is going on.
The tracking generator’s signal sweeps from 9kHz to 500MHz and shows a very characteristic low-pass response curve. The device under test shows very little attenuation (approx. -0.6dB) and then a nice steep roll-off starting around 40MHz.
Bencher YA-1 9kHz-500MHz 1MHz RBW
If we examine the spectrograph, we see that there are a few points of interest:
- insertion loss
- -3dB point (the bandwidth point)
- -23dB point
- any “spurs” or peaks
It should be noted that this spectrograph was taken with an RBW of 1Mhz which is not very precise.
Bencher YA-1 9kHz-500MHz 10kHz RBW showing increased detail
As we see with an RBW of 10kHZ we can see some more detail, as well as lowering the noise floor of the test-system.
The effect of lowering the noise floor as well as increasing our relative sensitivity shows us some minor oscillations in the filter where there is a 10dB difference in filtering ability in the 100-500MHz range. This is completely normal.
By further decreasing the RBW to 1kHz, we can see even more detail as seen in the accompanying spectrograph below.
Bencher YA-1 9kHz-500MHz 1kHz RBW
The system floor now exists around the -85dB point where we can clearly see the periodic increase and decrease in signal passing ability. We are still talking about attenuation of 75dB at worst. Recall from decibel math that 10dB means 10x and 20dB means 100x .. so -70dB means that the signal is 10000000x less powerful coming out of the filter in this range.
Notice also the peak at 425MHz where the attenuation is -53dB – this is still very good but it is interesting to note all the same. This indicates that the filter is less able to attenuate signals around 425MHz, however, it is still an effective filter all the same.
Bencher YA-1 9kHz-100MHz 1kHz RBW
If we examine the filter a little more closely, limiting the frequency range to between 9kHz and 100MHz, we can see that the filter’s “roll-off” is nice and flat and that the insertion loss is approximately 0.6dB and that the -3dB point is a little more accurately shown as ~40MHz.
Bencher YA-1 35-45MHz 300Hz RBW
By further decreasing the window in which we are looking (ie. 35MHz to 45MHz at RBW 300Hz) we can quite accurately determine the parameters of interest.
We can see now that the insertion loss through the filter is 0.65dB and that -3dB point is at 39.8MHz and the -23dB point is 43.5MHz.
From the investigation of the Bencher YA-1 filter, it is clear that the unit performs well under the test conditions and should be an effective filter to prevent TVI / harmonic emissions.
Bencher YA-1 9kHz-1500MHz 10kHz RBW
For the sake of completeness, I’ve attached a spectrograph of the filter’s response curve over the range of 9kHz to 1.5GHz. Notice the filter is effectively unable to attenuate any signals around the 1.05GHz range.
The purpose of the filter is to knock out harmonics from the HF amateur band and so for example, at 20m (14MHz) we might expect to see harmonics being generated in the following series: 42, 70, 98, 126, 154, 182, 210, 238MHz as 3rd, 5th, 7th, 9th, 11th, 13th, 15th and 17th harmonics. In order for a transmission on 20m (14Mhz) to create interference on the 1.05Ghz band, a transmitting system would have to generate 75th harmonics! (rest assured the power content of 75th harmonics would be .. well .. purely academic in importance.
