[TenTec] IM with Orion's 500 Hz roofing filter

[Sinisa Hristov]

Eric Scace K3NA wrote:

> Your guide also mentions the problem of 500/250 Hz roofing filters coming after an
> amplifier stage.  According to my understanding, when the 500 or 250 Hz roofing filters
> are employed, the signal is routed first through the wide L-C filter (I think it's 12
> kHz?)... then through this amplifier that causes the problems... and then into the
> 500 or 250 Hz filter... and finally through another amp to make up for the filter loss.

I have made a mistake by blaming additional amplification
for degrading IM performance, without any detailed insight.

Just opened schematic and the box itself and verified that signal path is as follows:

    preamp Q27 (if engaged)
    first mixer Q1/Q2/Q3/Q4
    post mixer amp Q24
    2-pole filter X1/X2
    hardware noise blanker gate D1/D2/D3/D4 (if engaged)

=>  one of 1/1.8/2.4/6/15 kHz filters (15 kHz if 500/250 Hz selected)
=>  postfilter amp Q9
=>  through or (250/500 Hz filter + postfilter amp Q10)

    hardware AGC PIN diode attenuator D4/D5/D6/D7
    second mixer Q9/Q10
    cascaded 455 kHz filters FL5/FL1, etc.
  
Additional amplification (Q10) appears AFTER 500/250 Hz filters and
therefore cannot be blamed for degrading IM performance,
nor it can help noise figure/sensitivity, but of course
it can maintain proper signal level at ADC input.

I performed the following test:

  two signals -43 dBm each (S9 + 30 dB) at 7001.0 and 7001.7 kHz to ANT1 input
  preamp off, attenuator off, 1 kHz roofing filter selected
  main RX to ANT1, sub RX to ANT2

At 1 kHz filter input IM level was undetectable.
The 6 kHz filter was then inserted into 1 kHz slot and at RFIC output
IM was -68 dBc, translating into input-referred IP3 of only -9 dBm.

Therefore, your assumption is correct:  Q9 seems to be creating IM
(not claiming it's the transistor itself, but it's clearly that stage).

This is an instance of a long-standing mistake of building weak stages
after the first mixer. Unfortunately, Orion is no exception,
it can mask the problem to some degree by selecting 1-6 kHz filters,
but the emperor appears completely naked with 500/250 Hz filters.


> then the following two potential modification may correct the problem:
 
>    1) Change the routing of the PIN diode control signal, so that the 1 kHz filter
>       position is selected instead of the L-C filter
>       position in the first stage.  This could be done in firmware.
 
>    2) Replace the L-C filter stage with a 600 Hz bandwidth 4-pole filter centered
>       on 9000.75 kHz.  One may surrender the ability to
>       transmit wideband modes such as AM.  For many of us, that is no sacrifice.

None of these can correct the real problem - overload of Q9,
although some IM products may be masked out by using 1 kHz filter
in front of 500/250 Hz, which doesn't appear as a significant
improvement over 1 kHz filter alone.

500/250 Hz filters can be promoted to real IM-fighting devices only
by removing the essential cause, which can be done in two ways:

  a) insert them before Q9, that is where other filters are, or
  b) improve Q9's large signal handling capabilities.


73,

Sinisa  YT1NT, VA3TTN