50 MHz Noise Canceling Receiver Front End
Noise has long been a problem for communication. Typically the lower the frequency the more the noise and the less the required receiver sensitivity. Man made noise is probably the worst of all noises that we have to contend with today. After many years of trying to minimize it I am convinced there is no one magical solution for the elimination of man made noise. It's a combination of many types of noise reduction: noise blankers, antenna pattern, DSP and phasing cancellation.
Refer to the block diagram of a receiver front end. A frequency (any) is converted to the First Intermediate Frequency (1st I.F.) by means of the mixer and local oscillator. In the newer HF rigs the 1st I.F. Is around 70 MHz. In many 2 meter rigs the 1st I.F. Is usually 10 to 11 MHz (10.7 or 10.8 etc).
The 1st I.F. crystal filter is sometimes called a "roofing filter" and it sets the maximum bandwidth into the rest of the receiver. The bandwidth of this filter is typically 12 to 15 KHz in most radios. This is not just to pass an FM signal but it is also a good compromise on blanking bandwidth. If this filter was made too narrow then the noise blanker wouldn't function at all.
Noise pulses (ignition and power line) pass through the mixer and 1st I.F. filter and are split into two channels: one to the noise amplifier and the other to the delay line and the noise gate (an on & off switch). The output of the switch goes to the rest of the receiver. Noise pulses are amplified greatly by the noise amplifier and are then used to turn the noise gate on and off. This is the same as turning the receiver on & off but very fast so you never hear it. When a noise pulse enters the receiver the noise gate "opens up" and "blanks" or turns the receiver off until the pulse has passed. This gate turns on & off in a few dozen nano seconds.
If a signal is "outside" the pass band of the 1st I.F. Filter it will never get to the noise gate. If however a signal is within the 12 KHz "window" of this filter it will get to the noise gate and get turned on & off by the gate. This re-modulates the signal you want with noise gate pulses. The presence of this other signal has essentially rendered your noise blanker useless and you might as well turn it off.
If stations will spread out to 20 KHz or more then most noise blankers will continue to operate effectively. QSYing up the band only 5 or 10 KHz keeps you within someone's blanking band width.
By careful design of your antenna you can effectively reduce many sources of man made noise. If you are not pointing directly at a noise source why should you have to hear it? It has been proven using today's antenna software that very clean antennas can be built where the side lobes (any lobe which is not the main lobe) can be as much as 22 to 25 db suppressed. All that at a cost of only 0.25 db in forward gain. That means you can replace your gain maximized older antenna with 15 db down lobes and reduce your off axis noise level by 6 to 10 db. This is one of the best improvements you can make to your noise problem.
Digital signal processing has come a long way and has given us some real improvements in noise reduction. It still isn't the "cure all" we have hoped for.
The basic operation of the noise canceling receiver front end is shown in the block diagram below. Noise and the desired signal are received by the main 50 MHz antenna. Noise and hopefully not too much of the desired signal are received by the noise antenna. The output of the noise antenna is filtered and amplified then it's phase and amplitude are adjusted to give cancellation of the noise at the "sum" point. Equal amplitude and opposite phase will give perfect cancellation.
In reality I have been able to get better than 50 db of cancellation of one signal using this technique on 50 MHz. One signal means it can be any signal, noise, carriers and even local stations and beacons. We must also have the capability of 360 degrees of phase control. One must be careful however since improper adjustment means you can cancel out the desired signal as well. It is not an automatic system and it requires a lot of "tweaking". . . . That's what amateur radio is about isn't it ?. . . . . Tweaking?
Let's start with the front end .
Below is a super clean noise antenna which has virtually no side lobes. It is designed using "YO", Yagi Optimizer by Brian Beezley K6STI. It has low gain since gain is not needed to hear the offending noise and a clean pattern to minimize picking up unwanted signals. It would be ideal to have the exact same beam width as the main antenna but that would mean we need the same boom length as the main antenna. What we really need is an antenna with a 1 degree beam width and no gain all on a one foot boom for 50 MHz! This is a compromise of course.
Filtering and gain are required to have enough "noise" signal to control and thus cancel. Here are two combinations of amplifiers and filters which I have tried with success. The FET preamplifiers are very narrow in bandwidth and have proven reliable in stability and noise figure. They have been used as the main receiver front end as well.
The bipolar is a new design primarily since I have started monitoring for six meter MUF indicators as low as 45 MHz. I needed a high dynamic range, low noise figure preamplifier which would cover the entire band (43 thru 51 Mhz) and the FET preamplifiers won't do that. They both provide about the same amount of gain however.
We need to filter the noise channel since we don't want to bring in any extra signals. After all we are amplifying and injecting them into the receiver so filtering is important. Below is a 3 pole band pass filter which covers 45 to 51 MHz. It is used not only as a noise front end filter but for MUF and main receiver filtering as well.
This filter is much narrower (1 MHz) but has more insertion loss
Either one will work depending on your application. I have been using many of them as a "bread board" front ends which I am always changing.
We need to control the phase with very fine resolution while maintaining a constant 50 Ohm impedance in order to get good cancellation. The following circuit is an adaptation of a microwave idea which has been around for a long time. A 90 degree quadrature hybrid has some interesting properties: as long as the two original outputs (B and C) have the same load impedance then the input (A) and output (D) impedances remain good. That means we can place two identical (tracking) loads on B and C and control the phase without hurting the impedance of the hybrid. It also has very low insertion loss. It is essentially a lumped component version of quarter wavelength transmission lines.
This is the schematic of the phase shifter control. It uses Variable capacitance diodes as the tuning elements in order to get a "remote" tuning capability.
Since the 0 to 100 degree phase control isn't adequate to give complete phase coverage, (360 degrees needed) a two stage 90 degree and 180 degree fixed shift was also used. These are essentially 50 Ohm quarter wave length lines simulated with lumped components again and are switched in and out with relays.
The variable gain adjustment is provided by a PI type PIN diode attenuator. While more precision resistors are called for this version works quite well. It also have very high dynamic range so it can handle very strong signals without intermodulation products. It has slightly less than 2 db of insertion loss and has better than 35 db of gain control while maintaining close to a 50 Ohm impedance.
We can't tolerate much insertion loss at the summing point since the main 50 MHz signal is present and we don't want to attenuate it very much. A directional coupler was chosen since it can very low loss to the main thru put. The insertion loss of the coupled port 10 to 20 db can then be compensated for by the gain of the noise amplifier gain. A reasonable compromise can be achieved with a 12 db coupler. It is fabricated using high permeability toroids and is simple to fabricate.
One other approach is to put a filter preamplifier in front of the sum point but it isn't really necessary as the loss of the directional coupler is under one decibel. That won't hurt your sensitivity at 50 MHz.
The main 50 MHz signal is injected into the "In" port and the processed noise signal is injected into the "F" port. They are essentially added at this point and the difference (for nulling) is taken from the "out" port.
The combined signal then goes to the 50 MHz receiver or transverter. Care must be taken to not run too much gain from the noise channel as it is very easy to inject unwanted signals and noise. When first using this device it is wise to practice on carriers as they are usually stable. Trying to null a ssb signal is virtually impossible. To disable operation merely remove DC power to the noise channel amplifier. Changing the phase from "null" to "addition" (180 degrees) will add noise to the main receiver and the gain will have to be reduced. It's easiest to remove DC power. Insertion loss to the noise channel is about 2 db (attenuator) + 12 db (directional coupler). The 50 main attenuation is about 1 db.
This plot is "null mode" using the FET amp string:
Since the null is critical, minor changes in amplitude or phase will produce fluctuations in the null, like wind moving your antenna. Anytime you change anything you will have to re-null the interfering signal.
This plot is of the noise channel only. In this case it's a BPF- FET amplifier BPF-FET amplifier. This string does a very good job of filtering most additional potential interference that may be added by the noise channel.
Switching to the filter-bipolar-filter combination in "null mode". Note the additional gain at unwanted frequencies. This hasn't been a problem in real time use. The dynamic range of the preamplifier (+35dbm IP3) has helped prevent any noticeable intermodulation from occurring. The trade off is well worth it for the extended bandwidth MUF capability of this front end.
Now the filter-bipolar-filter noise channel only. There is still adequate filtering out of band to prevent problems. Even the bottom end of channel 2 has 30 db of attenuation. This same front end is now in use for the front end of the new six meter transverter. There has been no interference with it.