Noise Phasing and MFJ-1025/1026 Data
The original concept of the 1025 is based on a bridge phasing system. Many phasing boxes, such as the QRN Squasher, do not rotate phase. They are based on the concept that a 180-degree inversion allows rotation of phase when mixed through a potentiometer, but that does not occur. The only thing mixing a 180-degree shifted phase signal with another base phase signal does is vary the amplitude, and produce an abrupt 180-degree flip at the zero-crossing point. Some people have copied the flawed QRN-Squasher concept, and offer similar products.
I recently reviewed another phasing system design (on paper) which is implied to be the "ultimate". It uses a simple R/C phase advance circuit, not a bridge. The maximum phase rotation of such a system is limited to less than 90 degrees. With 180-degree or zero-degree inversion in a later stage, that allows only a 150-160 degree slice of phase choice. Even transposing inputs would only bring that to 320 degrees. Not only that, dynamic range is limited by a very poor choice of operating parameters for the FET driving the phasing system.
To have a perfect product you need 360-degree rotation with no level change. Even with perfect design, phasing systems are tedious to adjust. Other systems rotate phase over a wide range, but have severe level change as phase is rotated. The effect of a level change is you have to "walk" back and forth with phase an level every time you adjust phase! When level changes with phase adjustment, phasing units are even more unpleasant to adjust than a proper system is.
With a *well designed* bridge system, amplitude shift is minimal. The MFJ1025 series, despite being manufactured as cheaply as possible, outperforms any other unit I have tested (by a large margin) over its intended frequency range of 1.8-20 MHz.
The phasing system at W8JI, used to pinpoint signal directions of transmitters or jammers more than to actually cancel noise, is a more sophisticated version of the 1025 with higher dynamic range. It uses tandem bridges, and is calibrated in degrees shift. Phasing is spread over a wide, linear, control range. Phase rotates a full 360-degrees with less than 1dB amplitude shift. It is possible to resolve as little as 1- degree directional difference in arriving signals.
A Word About Dynamic Range and Noise Floor
160-meter DX'ers have the most demanding noise floor and dynamic range requirements of anyone using a phasing unit. 160 ops not only must hear exceptionally weak DX stations using small inefficient antennas and low power, they must contend with multiple 1500-watt signals (often with excellent antennas) parked just hundreds of Hz away! CW operators often listen to CW signals with 250Hz BW systems, pushing noise floors up to 50 times lower than 10kHz filters would produce at any location. Amplifiers and other components must also handle the large powerful window of AM BC stations just below 160, as well as the SW BC stations above. At my house the accumulation of signals from stations *outside* Amateur bands is enough to light a small 12v light bulb when using a 15dB amplifier!
Obviously what works well at MW for sorting 10kHz spaced wide bandwidth signals apart is entirely different. Here we have the strong window of stations, but we are looking in that strong window for other stations that often are strong (compared to 100-watt transmitters with poor antennas). Noise floor is also much higher, and so dynamic range and noise floor are of almost no concern. The critical performance parameter is mostly adjustment range, and not dynamics.
SW BC listeners need something a bit different yet.
Everyone should be cautious assuming a device optimized for one application somehow makes it fit the other.
The only drawback of the higher noise floor of the MFJ-1025/1026 is you must place it after any amplifiers used with low-level low-noise antennas. The MFJ-1025/1026 noise floor is typically around 14dB.
The dynamic range and noise figure of my personal unit is excellent. Unfortunately such a system is impractical for amateurs and SWL's because of cost. Each unit has about $300 in raw material cost, plus the assembly time. The FET's alone are 28-volt devices capable of several watts RF power while providing a 1.5dB noise figure. But for typical operation the 1025 is almost as good. If you want to spend several hundred dollars, I can sell you a system that works almost perfect in almost any application!
Even with MFJ's well-known QC issues and affinity for building things cheap (not good) I still recommend the MFJ-1025 over all other mainstream units. It has the potential to be better than anything else with only minor changes, and is certainly betetr than all other mainstream units like the ANC-4 or QRN Squasher just as it is shipped. If you want a custom modified unit, and can not make the changes yourself, I would be happy to help.
The following table are gain measurements for the MFJ-1025/1026. Wiltron Network Analyzer (50 ohm detectors), transmission loss mode.
Stock Unit and Filters 50-ohm input and output
Aux Port Gain
Frequency Gain (dB)
|2.5MHz and higher
From the above losses, deduct -2dB (or add 2dB!) if the input protection light bulb is bypassed or replaced with a jumper wire. I.e. 1.8 MHz gain becomes -3.4dB with the lamp removed.
Main Port Gain
Frequency Gain (dB)
|2.5MHz and higher
Bypass is essentially lossless.
Gain Compression and IM Dynamic Range
With gain controls wide open and pre-amp disabled, blocking begins at about -3dBm. Third order intercept is about -21dBm. This is generally well beyond what most receivers are capable of obtaining. (Click here to see the receivers page on this site.) The drawback of the 1025/1026 is the lack of any form of filtering, other than the high pass filter. This means the MFJ-1015/1026 sees a very wide window of signals.
With that in mind, it is questionable if removing diodes actually is worth the risk of RF or ESD damage. It would be much better to add a simple bandpass filter system for the band you are operating. However, there is some improvement offered by removing diodes. IM and blocking also can be improved by increasing quiescent current in Q5 and 6, as well as Q4. The bias modification is generally too complex for most people to do, and adding a bandpass filter in from of the unit would be much more effective anyway.
Removing Diodes and Lamps
IM dynamic range will improve about 6dB with this modification. The lamp will improve gain and noise figure of the AUX port (reduce loss) by about 2dB.
If you never transmit through the unit:
1.) Remove C5
2.) Remove D1,D5 at the output, remove D12 and 13 (or D10 and 11 in 1025) and D6 and 9 at the inputs.
3.) Replace Lamp1 with a jumper wire.
NOTE: If you transmit through the unit do NOT remove C5 and D1 and D5.
Sensitivity to Load Impedance
Like any typical amplifier system, this unit is sensitive to load impedance. Since it has an emitter follower at the output, it likes to see a low load impedance. Gain will increase about 5dB if you add a 1:4 step up transformer at the emitter of Q4.
The following are modifications or changes to the MFJ-1025/1026:
Gain Improvement Mod.
Increases gain and dynamic range by approximately 5dB (assumes your receiver has a 25-ohm or higher input Z):
1.) Cut the trace from C6 to D1/D5 near the small dip relay.
2.) Add 1:4 step-up transformer, low impedance side towards C6 and high impedance side to load.
Transformers are available, including US shipping, for $6.
The MFJ-1025/1026 contains BCB filters that roll off at about 1.9MHz. To extend operation lower in frequency:
Remove L3, L4, L5, L6.
Short C8 and 16
Increase the value of C12 and 13 by the same factor as the decrease in frequency. For half the frequency, double C12 and 13.
Caution: DO NOT remove filters if you intend to use this unit for weak-signal reception on Amateur bands above 1.8 MHz, unless you are sure you do not have strong BC signals.
C12 and C13 mods will reduce upper frequency limit proportionally to the change.
Extending Phase Range
Phase range of the MFJ-1025/1026 is limited by the ratio of R16 to reactance of C12 or C13. Normally phase will rotate through at least 130-degrees. SW3B inverts phase 180-degrees, and this moves the phase shift to the other side of the circle. Typically the MFJ-1025/1026 has about 280-degrees of phase rotation.
The remaining 80-degrees can be covered a number of ways. You can swap inputs (only when using it entirely passive), or modify the phasing network. I also have successfully added a relay that swaps the outputs of the low-pass input filters. A second method is to add an additional capacitor in series with the ground lead of the bottom of R16 (the wiper remains grounded) back to the junction of C14 and R17.
I'd recommend swapping the inputs, it is easier and does not degrade amplitude response or bandwidth.