"At 5, 10 and 20 kHz, there is no significant (IMD) difference between the
stock 1.0 kHz Ten-Tec filter and the Inrad filter.
At 2 kHz spacing, the improvement is on the order of 6 to 7 dB.
At 1 kHz spacing the improvement is closer to 10 dB."
Robert Sherwood NC0B, Sherwood Engineering
In October 2003 I realized some strange things were going on with gain in the optional 500 Hz filter position in Orion. In December, ARRL published their Expanded Test Report results which indicated that the standard 1000 Hz roofing filter had much better IMD performance than the 500 Hz. Apparently this is because Ten-Tec opted to use the older 6/8 pole 250/500 Hz filters which work in their Omni VI and previous models. These filters have higher insertion loss than the 4-pole filters used in the 1000 and above positions, so an additional amplification stage for the 250/500 was added to compensate for the increased insertion loss of the narrow filters, and this extra amplification stage apparently causes IMD degradation. Further complicating the issue, the 250/500 Hz filters use a slightly different IF frequency (9.000750) than the 1000 and higher filters (9.001500), so you cannot simply move the 500 Hz filter to the 1000 position.
After thinking about this I felt it might be possible to use a filter like Inrad's #350 600 Hz "Filter Kit for Experimenters". The center frequency was about right (9.000000 MHz); it was a 4-pole just like Ten-Tec's 1000/2400/6000 filters so it should have similar insertion loss as the 1000 Hz filter; and its input/output impedance was 200 ohms which is the same as all Ten-Tec filters. Shape factor beyond about 30 dB down on the skirts is not a critical issue for roofing filters in a DSP radio like Orion because tighter filtering is done by the DSP at the 3rd IF stage.
In January 2004 I gathered up Inrad's required minimum order (10 units) and placed a special order for the #352, which was basically a #350 but centered on 9.001500. We finally received our kits on April 1. After construction, the #352 appeared to work well, with a 6 dB bandwidth of 640 Hz and insertion loss which appeared to be similar to the 1000 Hz. However, I really could not tell about any IMD improvement since I do not have equipment to properly test IMD, and we were out of the heavy contest season which is the best on-the-air test for IMD. Fortunately, Rob Sherwood NC0B of Sherwood Engineering recently tested a few samples of the #352 and reported the following results, which were actually better than I expected when starting this project:
> At 5, 10 and 20 kHz, there is no significant difference between the
> stock 1.0 kHz Ten-Tec filter and the Inrad filter.
> At 2 kHz spacing, the improvement is on the order of 6 to 7 dB.
> At 1 kHz spacing the improvement is closer to 10 dB.
> There is some strange hysteresis of a couple dB, as you change the level
> around. The IMD does not act exactly as one would expect.
> I am guessing the digital AGC is doing strange things, causing the
> The insertion loss of filters #2 and #3 are identical to the Ten-Tec 1.0
> kHz filter.
Needless to say I am very pleased with the further improvement that this simple kit can make in Orion's already outstanding front-end! With this filter, Orion maintains its mid-90's dB IMD performance for interfering signal spacings all the way down to 1 kHz (but we can even improve on that for certain cases by using Orion's PBT...see below).
UPDATE: Inrad has now sold out of the #352 kits but is offering the same filter as a completed unit (#762) for $80 plus shipping.
UNIQUE AND POWERFUL CAPABILITY OF ORION'S PBT COMBINED WITH THE INRAD #762
Combined with the Inrad #762, Orion has a very unique and powerful capability that does not exist in any other current receivers that I am aware of. Orion's PBT is unique in that it controls not only the DSP at the 3rd IF, but also the roofing filter at the 1st IF. Below is a plot of RF through Orion's #762 filter at the 1st IF . I've set the 3rd IF DSP BW = 6000 Hz in order to show only the effects of the #762 roofing filter. Because an RF Noise Generator is used as the signal source, the resulting Spectrogram (tm) plot appears somewhat noisy. Inrad's plot of the full #762 response curve can be seen by clicking here, but the Spectrogram plot below will suffice for our example. The black cross is at the 800 Hz center frequency and the vertical lines are at the -6dB points of 500 Hz and 1100 Hz respectively. (NOTE: The absolute value of the vertical axis is meaningless...we are only interested in the relative shape and position of the resulting plot.)
Imagine that you are tuned to a signal at 800 Hz above (black cross). I used an 800 Hz Pitch in this example to provide a complete spectrum display but normally listen with a much lower Pitch. Now let's assume you begin experiencing an IMD product (2f1-f2) being generated by a signal at 500 Hz (f1) combined with another strong signal at 200 Hz (f2). Let's see what happens when we simply adjust PBT by +250 Hz.
Both interfering signals have now been reduced about 20 dB, which should reduce 3rd order IMD products even more (theoretically 3 times as much), while the desired signal at 800 Hz is unaffected. Obviously this will not work if you have interference from both sides, but Orion's basic IMD performance even with only the stock 1000 Hz filter is already the best available at close spacings (see W8JI, Sherwood and ARRL comparative measurements).
PBT combined with the #762 is a very powerful combination uniquely available in Orion. No current receivers I am aware of (Omni family, FT-1000 family, IC-756/7800 or FT DX 9000) can offer this capability since it requires BOTH a narrow (600 Hz) roofing filter AND control of the 1st IF via PBT. No other receiver has either capability.
Do we have any idea how good the Inrad 762 might be at very close spacings (i.e. <1 kHz)? Let's take another look at the Inrad plot. I've marked points below which show a total bandwidth of about 1300 Hz at 30 dB attenuation on the skirts. Assuming the #762 is correctly centered, this means signals at a spacing of 650 Hz or more should be attenuated by at least 30 dB, which means that their IMD products will be attenuated by at least 90 dB. If we further assume interference is coming primarily from one side, we can use PBT to shift 250 Hz which should effectively yield ~90 dB IMD rejection at 400 Hz spacings on one side of the passband (of course the other side of the passband would also shift 250 Hz thus resulting in ~90 dB IMD rejection for signals at 900 Hz spacings and above...still better than any other receiver I know of). My estimates have not been verified by actual measurements yet but I am hoping Sherwood or someone will be able to confirm them in the near future and will post any results here.