1. Introduction

As the main problem in dense areas of many well equipped VHF contest teams appears to be TX noise originating in HF RIG a possible solution to this problem is TX IF preselector.

Although the main motivation for this project was HF RIG TX noise reduction, the IF preselector brings tremendous benefit on RX as well. Just imagine the "crazy" situation when the XVRT TX driving signal going out of HF RIG can be 20 dB lower than the RX IF signal going from XVRT into HF RIG receiver (50 km LOS, 2x 1kW and 17 dBi antennas pointed one to other ==> 10 mW signal level that enters HF RIG receiver, while XVRT drive signal is in the 0,1 mW range).

The preselector also has a major role in setting up an efficient remote RX system (RX system 50 ... 200 meters away from main TX antennas that can be used during transmit time). 

More about this was said at regular S5 VHF&up meeting in Nemčavci 2011 (also available in Czech - tnx to OK1UBO) and by other hams many many years ago (i.e.DL7QY in 1987). 

2. Block Scheme

Block sheme is shown on the diagram below and is very similar to the older project by Vlada/OK1VPZ.

Block Scheme

2. Results

After more than 10 units tuned it can be said that the filters are of a very high quality and very repeatable - i.e. insertion loss of different filters and between different units is below +/- 0,2 dB.


Photo of a prototype unit (as tested in AA VHF 2012 and IARU VHF 2012 by S59DEM).

Photo of a final unit - front view (Pout potentiometeer on the left, positions marked with star are pluggable, filter selection buttons on the right).
front view

Photo of a final unit - rear view (DC, remote buttons and PTT connectors on the left).
rear view

Photo of a final unit - top view (two pluggable filters on the right are in the "parking" positions).
top view

3. Specifications





4. Construction

Design is pretty straightforward and in accordance with KISS principle.

Photo of a pluggable filter PCB.
pluggable filter

Crystal filter datasheet (made by http://www.krystaly.cz/en/): SSB, CW
PIC program listing and HEX file.

Last minute schematic correction: the gate of both PTT BS170 FETs should be connected to the ground via 100k resistor (can be easily soldered on top of 1n capacitor).

5. Description and Operation

Preselector has 13 filter selectrions: 8 for SSB filters, 4 for CW filters and one bypass. It is used on TX as well as on RX.

Insertion loss is less than 2 dB so it does not have significant impact on receiver sensitivity. Beacuse filters protect HF receiver from very strong signals the preamplifier on the HF RIG can freely be swithed on (AIP OFF on FT-1000mp). Preamplifier has to be switched on in case we add additional losses on the IF RX line (i.e. antenna IF switch matrix) to maintain required sensitivity of the complete receive system.

PIN attenuator and high dynamic range IF amplifier are placed in front of filters in the TX path. PIN attenuator functions as main output power setting control. TX amplifier replaces the IF TX AMP in the Javornik-144/14 XVRT. This means that IF TX AMP in XVRT needs to be switched off (position of jumpers J2, J3 and J4 need to be changed). By doing so about 6 dB lower TX noise of the XVRT can be achieved. Nominal TX level out of preselector is +1 dBm. This level is high enough to drive two XVRTs in parallel up to about 30 W (XVRT Pout potentiometer fully open!). In case output power from XVRT is not high enough the missing gain should be obtained by changing PI attenuator between BFG196 driver and power module (default 11 dB for 60W version) and NOT by switching-in IF TX AMP.

Two PTT signals can be connected to the preselector, »main« and »sub«. »PTT main« is normal PTT function – it switches both RX/TX relais. »PTT sub« function is supporting more advanced contest setups with two HF RIGs, where the second RIG (»sub«) is used for S&P. The »sub« RIG will be mainly outside the »main« RIG filter passband so »PTT sub« also switches »bypass relay« beside RX/TX relais. This way any transmission with »sub« RIG will always bypass filters.

Filters can be selected by two push buttons on front panel (  and ). Remote push buttons can be connected to the 3.5 mm stereo jack on the back panel. It makes sense to assemble 3 push buttons on a small PCB with long enough cable so that such a small control PCB can be freely positioned on the desktop area. Third, the middle, button should connect both signals to the ground in parallel - pictures below are self explanatory. Pressing middle button selects bypass position - pressing it again returns to the last selected position. It turned out that the middle button is being the most pressed button during the contest operation.

remote           remote sch

                                                                                                                                                                                        Remote filter selection buttons 

HF TX ATT (see block scheme) value depends on the HF RIG in use. Install 0 dB for FT-1000mp. Nominal TX input level is between -9 dBm and -11 dBm. In case HF RIG provides higher TX level, the difference needs to be leveled out by HF TX ATT. Example: TS-850 XVRT TX level is +10 dBm --> HF TX ATT value needs to be 20 dB.

How to set TX chain levels:

6. "Measurement day" results

On 15th December 2012 we organized a filter tuning and TX signal quality measurement day in Ljubljana. From far it looked like a "FT1000mp fan club" meeting :) Some photos can be seen here.
We had 15 HF RIGs on a test bench, about the same amount of Javornik-144/14 XVRT and XTAL-BOX-144M and three solid state 2m kW PAs. Event was visited by 23 S5 VHF contest enhusiasts.
The test results are summarized in the tables below. Please note that the goal was not to perform very accurate measurements - it was more on comparison and finding "hidden" issues (precision of the results is probably not better than +/- 2 dB). All the measurements were made with QS1r as a measurement receiver (SDRMAX v5.0.0.8, 4096 FFT, 64 AVG, 0.125 Msps and 1.25 Msps); external step atenuator and low noise preamplifier were used to bring the test signal 1 ... 2 dB below QS1r clipping level to use its full dynamic range. QS1r modification enabled measurements directly at 144 MHz (external 10 MHz wide bandpass filter was used).

6.1 Phase noise test results

TX phase noise - HF RIG @ 14200 kHz TX sideband noise level dBc/Hz
  20 kHz offset 50 kHz offset 200 kHz offset
FT1000mp (s53ww; 6K150120)  -132,0 -134,0 -143,0
FT1000mp (s53d; 3F480049)  -130,0 -132,0 -135,0
FT1000mp (s57q; 9H460108) -131,0 -133,0 -143,0
FT1000mp (s57c; 0X123456)  -132,0 -134,0 -142,0
FT1000mp (s55m; 0E540077) -134,0 -136,0 -145,0
FT1000mp (s50c; 2H070329)  -134,0 -136,0 -142,0
FT1000mp (s50c; 2H070329) @ ANT connector 5 W -123,0 -123,0 -123,0
FT1000mp (s50c; 2H070329) @ ANT connector 100 W 130,0 132,0 -142,0
FT1000mp (s53v; 7I200025) -134,0 -136,0 -143,0
FT1000mp MV (s59a; 41630092) -128,0 -130,0 -145,0
FT1000mp MV (s57m; 1E270014) -134,0 -136,0 -143,0
FT1000mp MV (s53d; 4L260088)  -135,0 -138,0 -144,0
FT1000mp MV (s59r; 5L360040)  -132,0 -134,0 -143,0
TS590 (s52ei; B2700070) -132,0 -134,0 -138,0
TS590 (s59p; B1400177)  -132,0 -134,0 -138,0
IC7600 (s50c; 0402264) -130,0 -136,0 -143,0
IC7600 (s50c; 0402264) @ ANT connector 5 W -121,0 -121,0 -125,0
IC7600 (s50c; 0402264)  @ ANT connector 100 W -131,0 -136,0 -143,0
FT2000 (s50l; 7E190058)  -127,0 -130,0 -133,0

I yellowed the values that are below 120/125/135 dBc/Hz @ 20/50/200 kHz respectively (my arbitrary set limits).  From the test results it can be seen that FT1000mp and MV behave well and there is no much difference between different radios. It can also be noted that driving XVRT with TX signal from regular antenna connector with reduced output power (to let say 5 W) should never be used as the phase noise floor is degraded by about 13 dB (for 100 W to 5 W reduction as an example)!

TX phase noise - HF RIG + XVRT @ 144200 kHz TX sideband noise level in dBc/Hz  
  20 kHz offset 50 kHz offset 200 kHz offset w/ XFLT
FT1000mp (s53ww; 6K150120) + XVRT(s50k/v1)@25W -132,0 -134,0 -138,0 -140,0
FT1000mp (s53ww; 6K150120) + XVRT(s53rm/v2)@25W -131,0 -133,0 -137,0 -139,0
FT1000mp (s53ww; 6K150120) + XVRT(s53rm/v2)@5W + HLV1000@750W -130,0 -132,0 -137,0 -138,0
FT1000mp (s53d; 3F480049) + XVRT(s53d/v1)@40W -128,0 -130,0 -131,0 -138,0
FT1000mp (s57c; 0X123456) + XVRT(s57q/v2)@60W -132,0 -134,0 -138,0 -138,0
FT1000mp (s53v; 7I200025)+ XVRT(s53v/v2)@50W -131,0 -133,0 -136,0 -138,0
FT1000mp (s55m; 0E540077)+ XVRT(s57c/v2)@10W + ITB#1@700W -131,0 -131,0 -137,0 -138,0
FT1000mp (s55m; 0E540077)+ XVRT(s57c/v2)@10W + ITB#2@700W * * * *
FT1000mp MV (s59a; 41630092)+ XVRT(s59a/v2)@30W -130,0 -131,0 -138,0 -138,0
FT1000mp MV (s57m; 1E270014)+ XVRT(s57m/25W)@25W -133,0 -135,0 -138,0 -138,0
FT1000mpMV (s59r; 5L360040) + XVRT(s59r/v1)@90W -132,0 -134,0 -137,0 -138,0
TS590 (s52ei; B2700070)+ XVRT(s52ei/v2)@25W -131,0 -132,0 -136,0 -138,0
TS590 (s59p; B1400177) + XVRT(s59p/v1)@50W -130,0 -131,0 n/a -138,0
FT2000 (s50l; 7E190058) + XVRT(s50l/v2)@50W -128,0 -130,0 -135,0 -138,0
FT847(s57m; 01410695)@50W  -118,0 -124,0 -134,0 n/a

The sideband noise level was estimated from averaged (64x) spectrum plots with SDRMAX version which has +3 dB error in displaying noise level (with version one can disable additional spectrum filtering and the reading is then equal to S-meter). The levels in the above table were corrected for this 3 dB error on 12th Jan 2013.

The phase noise floor for 2m test results at around -138 dBc/Hz @ 200 kHz offset is not comming from the XVRT 130 MHz LO as initially speculated. Later I repeated the test (using 14 MHz XO + XFLT) with one more XVRT as test downconverter, so that measurements taken with QS1R were always at 14 MHz. The XVRT LO phase noise at 50 kHz offset is around -144 dBc/Hz; phase noise at larger offsets could not be measured as it was below receiver noise floor. The -138 dBc/Hz phase noise floor is comming from QS1R internal sampling clock when using it in undersampling mode; my measurements show that its 125 MHz XO has phase noise of around -128/-137/-139 dBc/Hz at 1/5/200 kHz respectively. Those values are reduced by 19 dB when using it at 14 MHz resulting in the phase noise of -147/-156/-158 dBc/Hz at 1/5/200 kHz which are all below its blocking dynamic range  - so no limitation to the receiver performance.

6.1 CW key clicks test results

The test was done with the help of  internal keyer at maximum speed. I choose arbitrary signal width values at approximately 105 and 125 dBc/Hz below PEP for comparisson. I yellowed the values that are wider than 10/25 kHz @ 105/125 dBc/Hz  respectively.

CW width - HF RIG @ 14055 kHz kHz for dBc/Hz from peak power
  105 dBc/Hz 125 dBc/Hz 125 dBc/Hz w/ XFLT
FT1000mp (s53ww; 6K150120)  4,0 20,0 9,0
FT1000mp (s53d; 3F480049)  5,0 25,0 9,0
FT1000mp (s57q; 9H460108) 3,0 15,0 8,0
FT1000mp (s57c; 0X123456)  4,0 22,0 10,0
FT1000mp (s55m; 0E540077) 5,0 18,0 n/a
FT1000mp (s50c; 2H070329)  5,0 20,0 10,0
FT1000mp (s53v; 7I200025) 4,0 20,0 9,0
FT1000mp MV (s59a; 41630092) 4,0 25,0 9,0
FT1000mp MV (s57m; 1E270014) 3,0 16,0 8,0
FT1000mpMV (s53d; 4L260088)  4,0 18,0 n/a
FT1000mpMV (s59r; 5L360040)  4,0 20,0 9,0
TS590 (s52ei; B2700070) 2,0 21,0 8,0
TS590 (s59p; B1400177)  2,0 24,0 8,0
IC7600 (s50c; 0402264) 5,0 28,0 n/a
FT2000 (s50l; 7E190058)  12,0 40,0 13,0

CW width - HF RIG + XVRT @ 144055 kHz  kHz for dBc/Hz from peak power
  105 dBc/Hz 125 dBc/Hz 125 dBc/Hz w/ XFLT
FT1000mp (s53ww; 6K150120) + XVRT(s50k/v1)@25W 7,0 28,0 12,0
FT1000mp (s53ww; 6K150120) + XVRT(s53rm/v2)@25W n/a n/a n/a
FT1000mp (s53ww; 6K150120) + XVRT(s53rm/v2)@5W + HLV1000@750W n/a n/a n/a
FT1000mp (s53ww; 6K150120) + XVRT(s53rm/v2)@5W + HLV1000@500W 5,0 30,0 10,0
FT1000mp (s53ww; 6K150120) + XVRT(s53rm/v2)@5W + HLV1000@1000W 5,0 30,0 10,0
FT1000mp (s53ww; 6K150120) + XVRT(s53rm/v2)@5W  5,0 25,0 10,0
FT1000mp (s53d; 3F480049) + XVRT(s53d/v1)@40W 4,0 50,0 10,0
FT1000mp (s57c; 0X123456) + XVRT(s57q/v2)@60W 7,0 35,0 13,0
FT1000mp (s57c; 0X123456) + XVRT(s57q/v2)@50W n/a n/a n/a
FT1000mp (s53v; 7I200025)+ XVRT(s53v/v2)@50W 4,0 22,0 10,0
FT1000mp (s55m; 0E540077)+ XVRT(s57c/v2)@10W + ITB#1@700W n/a n/a n/a
FT1000mp (s55m; 0E540077)+ XVRT(s57c/v2)@10W + ITB#2@700W n/a n/a n/a
FT1000mp MV (s59a; 41630092)+ XVRT(s59a/v2)@30W 4,0 24,0 10,0
FT1000mp MV (s57m; 1E270014)+ XVRT(s57m/25W)@25W 4,0 25,0 10,0
FT1000mpMV (s59r; 5L360040) + XVRT(s59r/v1)@90W 4,0 22,0 10,0
TS590 (s52ei; B2700070)+ XVRT(s52ei/v2)@25W 2,0 23,0 8,0
TS590 (s59p; B1400177) + XVRT(s59p/v1)@50W n/a n/a n/a
FT2000 (s50l; 7E190058) + XVRT(s50l/v2)@50W 12,0 45,0 13,0
FT847(s57m; 01410695)@50W  15,0 >100 n/a
FT847(s57m; 01410695)@25W  n/a n/a n/a

As can be seen FT847 should never be used as a main TX in serious contest setup.

6.1 SSB spaltter test results

This test was done with processor ON and at around 10 dB compression level. The effect of compresson level on SSB splatter width for FT1000mp can be seen here and in practice it should be set to around 5 dB. It was obseved that compression setting widely varied between same type of radios (on some one could not set level above 5 dB at all) while on TS590 I could not see any difference in setting processor level at any range so I used factory default 50. The input level to QS1r was set to be below clipping while talking into the mike. As the peak to average ratio varies on different radios due to different compression settings the accuracy of the results are probably more dispersed but can still be used for relative comparison. I yellowed the values that are wider than 25/40 kHz @ 105/125 dBc/Hz.

SSB splatter width - HF RIG @ 14200 kHz kHz for dBc/Hz from peak power
  105 dBc/Hz 125 dBc/Hz 125 dBc/Hz w/ XFLT
FT1000mp (s53ww; 6K150120) @ 10 dB processor gain  25,0 60,0 38,0
FT1000mp (s53d; 3F480049) @ 10 dB processor gain 20,0 50,0 34,0
FT1000mp (s57q; 9H460108) @ 5 dB processor gain 28,0 65,0 40,0
FT1000mp (s57c; 0X123456) @ 5 dB processor gain 18,0 38,0 30,0
FT1000mp (s55m; 0E540077) @ 5 dB processor gain 23,0 55,0 n/a
FT1000mp (s50c; 2H070329) @ 8 dB processor gain 25,0 60,0 38,0
FT1000mp (s53v; 7I200025) @ 10 dB processor gain 35,0 80,0 37,0
FT1000mp MV (s59a; 41630092) @ 5 dB processor gain 28,0 70,0 34,0
FT1000mp MV (s57m; 1E270014) @ 5 dB processor gain 15,0 33,0 25,0
FT1000mpMV (s53d; 4L260088)  n/a n/a n/a
FT1000mpMV (s59r; 5L360040) @ 10 dB processor gain 20,0 55,0 35,0
TS590 (s52ei; B2700070) @ processor level 50 7,0 20,0 17,0
TS590 (s59p; B1400177) @ processor level 50 7,0 24,0 15,0
IC7600 (s50c; 0402264) n/a n/a n/a
FT2000 (s50l; 7E190058) @ max. processor gain 8,0 20,0 20,0

SSB splatter width - HF RIG + XVRT @ 144200 kHz  kHz for dBc/Hz from peak power
- same processor setttings as above 105 dBc/Hz 125 dBc/Hz 125 dBc/Hz w/ XFLT
FT1000mp (s53ww; 6K150120) + XVRT(s50k/v1)@25W 28,0 65,0 38,0
FT1000mp (s53ww; 6K150120) + XVRT(s53rm/v2)@25W 20,0 55,0 32,0
FT1000mp (s53ww; 6K150120) + XVRT(s53rm/v2)@5W + HLV1000@750W 20,0 52,0 40,0
FT1000mp (s53ww; 6K150120) + XVRT(s53rm/v2)@5W + HLV1000@500W 20,0 60,0 38,0
FT1000mp (s53ww; 6K150120) + XVRT(s53rm/v2)@5W + HLV1000@1000W 27,0 75,0 45,0
FT1000mp (s53d; 3F480049) + XVRT(s53d/v1)@40W 18,0 >100 33,0
FT1000mp (s57c; 0X123456) + XVRT(s57q/v2)@60W 65,0 >100 >100
FT1000mp (s57c; 0X123456) + XVRT(s57q/v2)@50W 35,0 >100 80,0
FT1000mp (s53v; 7I200025)+ XVRT(s53v/v2)@50W 30,0 80,0 45,0
FT1000mp (s55m; 0E540077)+ XVRT(s57c/v2)@10W + ITB#1@700W 28,0 65,0 40,0
FT1000mp (s55m; 0E540077)+ XVRT(s57c/v2)@10W + ITB#2@700W 25,0 80,0 80,0
FT1000mp MV (s59a; 41630092)+ XVRT(s59a/v2)@30W 18,0 45,0 37,0
FT1000mp MV (s57m; 1E270014)+ XVRT(s57m/25W)@25W 18,0 35,0 33,0
FT1000mpMV (s59r; 5L360040) + XVRT(s59r/v1)@90W 20,0 65,0 45,0
TS590 (s52ei; B2700070)+ XVRT(s52ei/v2)@25W 13,0 25,0 23,0
TS590 (s59p; B1400177) + XVRT(s59p/v1)@50W 17,0 35,0 30,0
FT2000 (s50l; 7E190058) + XVRT(s50l/v2)@50W 23,0 60,0 45,0
FT847(s57m; 01410695)@50W  25,0 100,0 n/a
FT847(s57m; 01410695)@25W  25,0 100,0 n/a

The test results revealed some differences in linearity of XVRT PA at close to maximum output level.While some XVRTs showed still acceptable performence at 50 W other were not OK already at 40 W. But setting XVRT power to 30 W and less always brought the SSB width to acceptable value. It was observed that HLV-1000 performs extremely well even at the power levels that constantly lights it's IMD LED (this PA was recently at service and upgrade by BEKO). One of the ITB PA (uses same power FET as HLV) showed some serious noise-like spurious signals at 35 kHz and 75 kHz (probably comming from its DC/DC switching power supply). It was also noted that TS-590 has extremely low splatter levels!

Some typical spectrum plots.

In summary, the use of  30 kHz preselector at IF TX reduces the width of SSB transmission by a factor of 2 ... 3 (depending on the HF RIG and XVRT output power). In other words, for the same amount of splattes on the band 2 ... 3 more stations operate!

Date published: Sept. 2012
Change Log:
13.01. 2013 (adjustment of QS1r noise level reading, phase noise)
01.03. 2013 (90/110 dBc/Hz corrected to 105/125 dBc/Hz (30 Hz RBW not taken into account - tnx to s53rm)