Deze website maakt gebruik van zogeheten cookies. Klik op [OK] om deze melding te verbergen. Klik hier om meer informatie te lezen over de gebruikte cookies.
P O L Y T E C H . N U
English

BG7TBL 10 MHz bandpass filter

introduction
Time related measurement equipment use generally a local oscillator signal as a system 'clock'. Stable oscillators are rather expensive and require periodically adjustment due to natural ageing of oscillator crystals. Instead of using a expensive high precision (OCVCXO) oscillator in every device, it's generally possible to use an external oscillator signal. The internal oscillator is then disabled or bypassed. In this situation only one (expensive) high precision oscillator is required to feed more equipment with a high precision oscillator signal. Usually the external clock input is 10 MHz, but 1 MHz and 5 MHz is known as an external input frequency as reference. A high precision oven controlled (OCXO) can be used, but these crystals age naturally and adjustment is therefore required. Due to the rather low price of a GPSDO, this is a common used 'master clock'. This 10 MHz output signal is distributed (and amplified) to feed other devices which require an external oscillator input.

10 MHz filtering
The 'cleaner' the master signal is, the better the performance. Harmonics of the 10 MHz signal (20 MHz, 30 MHz and so on) or other high frequency components reduce the signal quality. By using a 10 MHz low pass filter, higher frequency components can be reduced. (Near) DC signals are not blocked/reduced. By using a 10 MHz band pass filter, the desired 10 MHz signal can be 'cleaned' from high and low frequency components. A band pass filter is the best way to get the desired 10 MHz signal without other unwanted signals.

BG7TBL band pass filter
BG7TBL designed and produced a rather small 10 MHz band pass filter. The board (as shown below) is equipped with to SMA connectors. The quality is rather nice and the board has a descent ground plane. Even the components are marked with silkscreen. My board is marked with 2017-02-09 which can be assumed to be the version number. The schematic is rather simple. There's a 3-pole C-L-C high pass filter at the input side consisting of C1, L3 and C2. The high pass filter blocks DC signals and reduces lower frequency components. After the high pass filter is a 5-pole C-L-C-L-C low pass filter placed. This low pass filter reduces the higher frequency components. The construction is nice and small. Due to the price wouldn't spend time to recreated something myself but just buy one.

image

measurements
I measured the signal behaviour on a DSA815-TG spectrum analyser with tracking generator. The measurement is normalised using a SMA-SMA adapter to require the best results. The raw measurement is shown below.

image

Due to the limitations of 'just' four markers, I exported the measurement information and created the graph below. The signal reduction at 10 MHz is 'just' 1,42 dB. That seems very reasonable. The first harmonic (f1) at 20 MHz is almost 28 dB reduced. The reduction of higher frequency components is rather good. Definitely good enough for it's purpose.

image

harmonics filtering
Sometimes the master oscillator signal is a 10 MHz square wave. By feeding the signal through this band pass filter, the harmonics are suppressed. The result is that a 10 MHz square wave will output as a (near perfect) sine wave. To show this, I made the measurements shown below. The oscilloscope signal of a 10 MHz square wave input is shown on the image below. The filtered (sine wave) output is visible. This signal is considered as a good quality sine wave. The band pass filter can therefore be used to 'smooth' a 10 MHz square wave to a 10 MHz sine wave.

image

I made also a 20 MHz square wave measurement. The output of the filter is shown below. The 20 MHz square wave is fed into the low pass filter and the output is shown below. The (linear) vertical scale is the same as the image above. Shown is that the signal strength of harmonics are substantially reduced.

image