Field Testing comments
The MFJ-1026 Phasing Unit

By Mark Connelly
July 28, 1997

This is a quick "heads up" on field testing of the modified MFJ-1026 phasing unit done on Saturday, July 26, from the Robbins Road - Holmes Field beach-DXpedition site located off Route 3A in Plymouth, MA (approx. GC= 70.68 W / 41.98 N).
Equipment: I used the Drake R8A receiver. Both the R8A and the MFJ-1026 were powered from the car battery.
Antennas: Two 90 ft. / 27 m wires lying on the ground were used. The "main" antenna for the MFJ-1026 ran on a slight downslope along the side of Robbins Road straight towards the sea at a bearing of about 70 degrees. The "auxiliary" wire ran out at a right angle into an open field of grass at an approximate 160 degree bearing.

I was on site at about 7 p.m. local / 2300 UTC. This is about an hour before sunset. The two wires were laid out, narrowly missing a patch of poison ivy and some "dog dirt": just a couple of "the hazards of DXpeditioning".
I felt that one of the big challenges would be to null WPLM-1390, located less than 2 miles / 3 km from the site. Luckily, its very large signal did less overloading damage than WRKO-680 does back at home near the Shawsheen River marsh.
Nulling WPLM a good 50 dB was easy! It wasn't too long before evidence of co-channel skip stations from ME, NY, and VT started bubbling in behind the nulled WPLM audio. Better yet was rather good audio from Netherlands on 1395 heard somewhat later !

The stations that the MFJ-1026 had the most trouble nulling were those with high-angle skip, especially if some groundwave was blended in.
The stations on the top end of the dial, such as WNRB-1510, were particularly troublesome in this regard. Null control settings required constant adjustment, especially in the period from an hour before sunset to an hour after.
The best sustained null depth I could manage on stations such as WNRB, WDCD, and WQEW was about 15 dB (although momentarily-deeper nulls popped in and out).
Shortwave DXers will probably experience similarly "jumpy" results above 2 MHz. Pure "groundwavers" like WPLM and longer-skip / lower-frequency stations such as WLW-700 nulled more deeply and for greater time intervals between required control re-adjustments.
These results are consistent with those found for any previous-used phasing scheme, whether delay-line, tuned L-C, or other. Physically-large antennas often help to reduce some of the problems with jumpiness and shallow null depth on short-skip and higher-frequency propagation. There's only so much you can get out of physically-small antennas, regardless of the phasing method employed. Theoretically, a computer-controlled system could accurately "chase" the optimum null in real time, but no one has made this idea a reality yet.

As the evening progressed, the MFJ-1026 / phased wires set-up proved its value as numerous Trans-Atlantic stations were logged.
Some of these came in fine on the 70-degree "Euro-wire" without the need for phasing, but, in a number of instances, phasing the two wires made the difference between a slop-plagued DX signal and crystal clarity. The two Croatia stations (1125 and 1134) come to mind. WBBR-1130 New York has a very strong signal at night here in eastern Massachusetts. Indeed, outside the immediate groundwave zones of locals, it's one of the five strongest stations night after night. WQEW-1560 and WDCD-1540 would also be on the "King Kongs of medium-wave skip" list.
When I was tuned to 1134, Croatia was running a good S9+20, but it was still trashed by WBBR slop at times - even on the "Euro-wire". With a few quick twists of the controls on the MFJ-1026, WBBR was reduced by better than 20 dB and Croatia-1134 roared in with absolutely beautiful audio. On peaks, it was stronger than what was left of WBBR. Not only did the phasing accomplish a nice clean-up on 1134, but also the much-weaker Croatian on 1125 was brought into the clear with just a bit of co-channel flak from Spain. Prior to phasing, it didn't have a ghost of a chance against the barrage of WBBR slop.

Earlier on July 26, I had done a few daytime DX tests of the MFJ-1026 from Harwich, MA on Cape Cod.
The first battery of tests involved feeding a Quantum Loop into the MFJ-1026 "main" input and using the 1026's built-in broadband active whip as the "auxiliary". With the loop at normal (i.e. high) Q, audio null depths only reached about 20 dB (versus better than 40 dB for carrier).
This is consistent with previous nulling scenarios where a high-Q tuned source is phased against a broadband one. You get what sounds like a double-sideband suppressed carrier signal. If the desired DX is more than 20 dB below the dominant, you probably won't hear it even during stable midday conditions.
Q-spoiling the Quantum Loop (15K resistor shunting the L-C tank) increases nullability of "pests" maybe to 30 dB, but the loop's usable sensitivity is compromised. At night, this is probably a non-issue (except in aurora), but during the day you need every bit of signal you can squeeze out of the small loop.
The MFJ-1026's active whip, by the way, had good sensitivity and low amplifier noise: above 800 kHz its usable signal capture was comparable to that of the Quantum Loop at normal Q. Even at 530 kHz, it provided threshold daytime audibility of Turks & Caicos on 530 kHz from Harwich: that's just about as good as the loop. On an outdoor sloper to the top of a 20 m / 66 ft. pitch pine tree, Turks & Caicos groundwave runs about S5 to S6 on the R8A.

A second battery of tests at Harwich used two wires at a right angle (similar to the set-up employed at Plymouth). Daytime nulls were smooth ("like butter" some would say).
WGAN-560 was easily dumped to reveal WHYN, near-equal WPRO and CFCY on 630 could each be brought up alone, much the same on 740 with WJIB and WGSM, WJTO on 730 was nulled a good 30 dB to pull out WACE over CKAC, strong WCLZ-900 was phased under the co-channel CKDH/WMVU mix, WWNH-930 easily surrendered to CFBC, and so forth. Nulling with two wires was decidedly better than any loop-versus-whip or loop- versus-wire scheme.

Nulling with the MFJ-1026 was easier than with a JPS ANC-4 and most of the homebrew units I've used over the last 30 years. It must be stated that some rather simple, but totally necessary, modifications had to be made to the stock MFJ-1026 to get it to up to "world class" medium-wave performance.

One opinion on the modifcations Before completing all the mods Mark suggests, Europeans might wait with removing the 100 ohm resistors till after testing. All the other mods are fine, fire ahead 100 percent with those, perhaps replace the resistors with something in the order of 220-390 ohms or even take out the T/R delay switch and fit a two-pole rotary switch with facility to switch in either 100, 220 or 390 ohms as well as the open circuit position (i.e. zero resistance)... Then, depending on the amount of local RF, a suitable value could be chosen to prevent spurs appearing... On the other hand you can always do what Mark suggests and that is to wind back the Gain controls a little. Paul Ormandy, New Zealand One more opinion These modifications are trivial to make --- rather than desolder the surface mount components and risk heat damage to surrounding components, the offending filters were "crushed in place" by squeezing them with needlenose pliers to to remove them. Loads easier than attempting to desolder. Should you ever wish to revert, it is easy to desolder the remaining end of each component, and to resolder in new surface mount components. Don, Oregon, on hcdx list, August 27, 2000

The first modification is to correct for excessive signal loss at frequencies below 2 MHz. The unit is advertised as covering down to VLF, but its stock version has insertion losses ranging from 5 dB at the top of the medium-wave band (1700 kHz) to more than 30 dB of loss down at 530 kHz. By the time you get down to the European longwave broadcast frequencies, there is so much loss that you might as well be receiving on a dummy load. Removing six high- pass filtering components - L5, L6, R27, L3, L4, R26 - totally corrected this problem.
Apparently the engineering people decided that this was going to be a shortwave-only unit when the advertising department thought that writing it up as having coverage down to VLF sounded nice ... clearly, somewhere along the line, there was a failure to communicate.

The second modification corrects for an inability, in some cases, to obtain enough phase shift to produce nulls. If the phase adjustment covered 0 to 180 degrees, then setting the reverse switch gives us 180 to 360 degrees (also expressable as -180 to 0 degrees).
But what if the phase adjustment range at some frequencies is only 0 to 120 degrees?
The inverted setting will be 180 to 300 degrees. If we have the need for a shift of 150 degrees to obtain a given null, this condition will not be producible. Judging by some of the medium-wave nulling tests, lack of sufficient phase-shift range does appear to be a problem.
Luckily, there is a relatively simple solution:
Reversing the antenna inputs so that what had been the "main" antenna during a non-working null scenario becomes the "auxiliary" and what had been the "auxiliary" becomes the "main". In the example above, we'd change the +150 required shift to -150 degrees, which is equal to +210 degrees (-150+360). This is now within the window of possible adjustments. As long as the adjustment range is better than 90 degrees, it will be possible to cover all required nulls if the channels are made "swappable". Tests performed still show a few cases where a null occurs at either the extreme right or left setting of the phase shift control, but at least a null can be produced with the antennas connected one way or the other.
A double-pole / double-throw (DPDT) "swap switch" was added to facilitate full null coverage. I mounted it on the rear of the unit because of limited space on the front panel. Two lines were cut to allow swap switch use. The first of these was the lead from the SW4B arm to the R20 pot (auxiliary level). The second cut was in the line from C8 to the R9 pot (main level). With the DPDT switch set to "normal", the switch completes the previously-wired paths: SW4B arm to R20; C8 to R9. In its "swapped" position, the SW4B arm gets connected to R9 and C8 gets connected to R20.
A couple of minutes spent studying the schematic and board layout should make it obvious how to install this modification.
Conceivably increasing the value of the phase shift circuit component C12 could also increase the phase-shifting range. I'm thinking of putting a varactor circuit in the box in place of SW2A, C12, and C13. The potentiometer installed (in lieu of the SW2 high / low frequency range switch) to adjust the varactor capacitance could act as a "vernier" for nailing down deeper nulls.

Once the MFJ-1026 is modified, it makes a very competent phasing unit that will undoubtedly bring the technology into the hands of many DXers who have not previously experienced its value in bringing new stations out of "the mud".