Prikaz objav z oznako IMD2. Pokaži vse objave
Prikaz objav z oznako IMD2. Pokaži vse objave

četrtek, 24. julij 2025

"Are laboratory measurements a valid criterion for selecting a transceiver?"

"Among radio amateurs, it's worth asking whether the laboratory measurements provided by ARRL, Sherwood, Adam Farson, or Peter Hard - RadCom are truly meaningful criteria for selecting a transceiver."







How to Choose a Radio for Your Location — A Slightly Different Approach

Introduction

Not long ago, I started thinking about upgrading some of my radio equipment. My trusty HF transceiver,  FT 920 which had served me well for many years as a backup rig, was slowly making its way toward retirement. Although the radio itself still met all my basic needs, its protocol and software support were clearly outdated.

These days, the market is flooded with high-end amateur transceivers, with price tags reaching astronomical levels. Their technical specs are indeed impressive, at least according to Sherwood and ARRL test results. One of the latest examples is the Yaesu FT-101D MP, which boasts some stunning figures:

  • MDS = –136 dBm (Minimum Discernible Signal, or sensitivity)

  • BDR = 150 dB (Blocking Dynamic Range at 100 kHz spacing)

  • RMDR = 120 dB (Reciprocal Mixing Dynamic Range, reflecting phase noise performance)

  • IMD3 @ 2 kHz = 110 dB (Third-order intermodulation products at 2 kHz spacing)

But the real question is: Do we actually need radios like this?

Personally, I’m not a fan of choosing radios solely based on Sherwood (5) or ARRL (6) receiver test rankings, where the primary focus is on third-order intermodulation distortion (IMD3). Last spring, I came across an article in Funkamateur, written by Werner DK1CU (1), which takes a different approach to radio selection.

In his article, Werner analyzes environmental noise, receiver sensitivity, and dynamic range—and makes the case that before choosing a radio, it makes sense to take a few measurements at your specific location. A similar line of reasoning can be found in an ARRL QEX publication (2).

Receiver sensitivity (MDS) is often highlighted as one of the most important specifications. But sensitivity is closely tied to dynamic range, susceptibility to third-order IMD, and reciprocal mixing caused by oscillator phase noise (RMDR).

Interestingly, neither Sherwood (5) nor ARRL (6) currently test for second-order intermodulation products (IMD2, where f1 + f2), which tend to be generated by strong shortwave broadcast stations. These signals are much more of a problem in Europe than in the U.S. — especially in the 6 to 13 MHz range — where the signal levels of broadcast stations are often significantly higher than those of amateur signals (see Figure 2).

These second-order products can appear as annoying “birdies and spurs” or phantom signals on the 14 and 21 MHz bands, and in some cases, strong 40-meter broadcast signals can even block your receiver entirely on 7 MHz. See Figures 1 and 2 for examples.

They also dont,t test  ergonomics and audio  receiver quality which is very important in a pile up.!



  
  Figura 1: Signal strenght of broadcast stations in the 40m band





Figure 2  Signal strenght of broadcast stations between 9 and 14MHz

How Much Sensitivity Do We Really Need?

In real-world environments, there is always some level of ambient noise present. This noise level depends on several factors, including frequency, time of day, and—most importantly—your location. There’s a big difference between operating in an urban setting versus a quiet rural area.

Environmental noise adds directly to the receiver’s internal noise floor, which in turn limits both sensitivity and the dynamic range of the receiver.

(See Figure 3)




Figure 3: Statistical Distribution of External Noise vs. Frequency

As we can see in Figure 3, external noise is highly frequency-dependent. For example, if we take a mid-range noise level between the "B city" and "B quiet" curves from the chart, we get approximately NF = 65 dB at 1.8 MHz and NF = 40 dB at 14 MHz. This environmental noise adds directly to the receiver’s own noise floor—especially at lower frequencies—thus limiting both sensitivity and dynamic range.

Let’s calculate the total noise floor of a receiver at two different frequencies, assuming a CW filter bandwidth of 500 Hz (which corresponds to +27 dB), and using the thermal noise floor of a 50-ohm resistor at 20°C, which is –174 dBm/Hz.

  • NFrx (1.8 MHz) = –174 dBm + 27 dB + 65 dB = –82 dBm
    → Total receiver noise at 1.8 MHz

  • NFrx (14 MHz) = –174 dBm + 27 dB + 40 dB = –107 dBm
    → Total receiver noise at 14 MHz

After doing these theoretical calculations, I wanted to see how much noise is actually present at my own receiver , using my real antennas and location. So, over the course of more than two weeks, I used a spectrum analyzer to measure noise levels in various bands—morning, midday, and evening.

The result was an averaged external noise profile for my location JN76DA, shown in Figure 4:

Frequency (MHz)Noise Level (dBm)
1.8–84
3.5–73
7–99
14–108
21–125
28–122









Figure 4: Measured Noise Levels at My Location (JN76DA) Using My Antennas



From the measurements shown above, we can see that the theoretical noise floor I calculated earlier differs from the actual measured values on my antennas by only about 3 dB—a relatively small difference. The only exception is the 3.5 MHz band, where the noise level is at least 10 dB higher, likely due to some local interference.
After these findings, I became curious about another important question:
What actual signal levels are present on my antennas?


Are we perhaps overestimating the strength of received signals when designing or testing receivers?


To answer this, I decided to measure real-world signal levels during two of the most extreme contest scenarios:
  • CQWW CW 2019
  • CQ 160 CW 2020
Once again, I used a spectrum analyzer connected directly to my antennas, performing measurements across all bands from 1.8 MHz to 21 MHz. Unfortunately, there was no activity on 28 MHz during the test period.
I repeated the measurements at hourly intervals throughout each contest: in the morning, afternoon, and night. Using the maximum values observed during these sessions, I compiled signal snapshots—shown in Figures 5, 6, and 7.






Figure 5:  The strongest signals in the CQWW160m contest measured on my full size inverted L antenna 




Figure 6:  the strongest signals in the CQWW  contest measured on my 7MHz  rotary dipole



  Figure 7 : the strongest signals in the CQWW contest measured on my 3el. rotary beam. 

Receiver Selection Criteria


Based on the above measurements, I can now roughly define the required characteristics of my new receiver according to the following criteria:

MDS (Minimum Discernible Signal): According to recommendations [1,2], the noise level present at our location should be increased by 15 dB to obtain the actual receiver sensitivity. Increasing sensitivity beyond this point makes no sense and in fact degrades the dynamic range.

BDR (Blocking Dynamic Range): From the signal strength diagrams of 40 m broadcast stations, we can see that signals spaced 100 kHz away are nowhere stronger than –20 dBm (see Figures 1 and 2).

RMDR (Reciprocal Mixing Dynamic Range): On the amateur bands, I have never received stronger signals than –28 dBm (local "big guns") during CQWW and CQWW160 contests. From this data, I calculate the required RMDR (see Figure 6).

IMD3 (Third-order Intermodulation Distortion): The "big gun" signals have never been closer than 5 kHz to each other at a level of –33 dBm. It,s also not logical for two big guns to get so close ( 2kHz) to each other  calling CQ in pile up.. Based on this data, I calculate the minimum IMD3 and IP3 requirements  at 5kHz spacing.

Blocking dB (100 kHz): The strongest measured signal from broadcasting stations in the 6–14 MHz range was at a level of –20 dBm.




Conclusion:

  1. MDS must vary with frequency band, ranging from –132 to –88 dBm. Increasing sensitivity beyond this range will degrade the dynamic range. From the results in Figure 4, with an added 15 dB margin, we get Table 1, which shows that the MDS should range from –132 dBm (at 28 MHz) to –88 dBm (at 3.5 MHz). To adjust sensitivity appropriately for each band, multiple attenuator stages are needed, providing at least 20 dB of total attenuation. For frequencies above 20 MHz, at least one preamplifier is required.

  2. RMDR (2 kHz): The strongest signal received during contests on my antennas was a local "big gun" at –28 dBm, which can increase phase noise. Therefore, RMDR (2 kHz) should be at least 85–90 dB. Sensitivity to phase noise is also the most important parameter for receiver quality.

  3. IMD3: Since "big gun" signals never appeared closer than 5–7 kHz to each other in all measurements, there is no need to measure IMD3 at 2 kHz, as Rob Sherwood does. Instead, I take IMD3 at 5 kHz as the reference.
    As a sample, I took the worst-case measurement at 7 MHz. From this data, I then defined the following requirements:


Example Calculation of Required Receiver Parameters for 7 MHz (see Table 1):


  • MDS (sensitivity) = NF_rx + (–15 dB) = –99 dBm + (–15 dB) = –114 dBm

  • RMDR (reciprocal mixing at 2 kHz) = MDS – strongest signal on 7 MHz = –114 dBm – (–28 dBm) = 86 dB

  • IMD3 (third-order products at 5 kHz spacing) = MDS – strongest pair of signals at 5 kHz spacing = –114 dBm – (–28 dBm) = 86 dB

  • IP3 (5 kHz) = 1.5 × IMD3 – MDS = +15 dBm

  • Blocking dB (100 kHz) = MDS – strongest broadcast signal in the 6–14 MHz band = –114 dBm – (–20 dBm) = 94 dB



Based on the noise measurements and the above parameters, I obtained the results shown in Table 1. For RMDR calculation, I use the highest signal level on each band, while for IMD3 at 5 kHz I consistently use an input signal level of –33 dBm.

FrequencyNoise Floor CW 500 HzMax Signal LevelDesired MDSMin RMDR 2 kHzMin IMD3 5 kHz
1.8 MHz–84 dBm–28 dBm–99 dBm71 dB71 dB
3.5 MHz–73 dBm–33 dBm–88 dBm55 dB60 dB
7 MHz–99 dBm–28 dBm–114 dBm86 dB86 dB
14 MHz–108 dBm–38 dBm–123 dBm85 dB95 dB
21 MHz–125 dBm–38 dBm–130 dBm92 dB102 dB
28 MHz *–122 dBm–35 dBm–132 dBm97 dB104 dB

Table 1: Required receiver parameters for individual bands.

From Table 1, I take the maximum values for the receiver parameters, which should be:

  • MDS: –132 dBm

  • RMDR (2 kHz): 97 dB

  • IMD3 (5 kHz): 104 dB

  • Blocking (100 kHz): 112 dB

  • Multi-stage attenuators with at least 20 dB total attenuation

  • Preamplifier needed only for frequencies above 21 MHz

From this analysis, it is clear that at my location, I do not need a transceiver with 150 dB blocking and 130 dB RMDR. A good mid-range transceiver is perfectly sufficient. The numbers clearly show that a mid-class receiver from the middle section of Rob Sherwood’s list is completely adequate for my location. There is no need to buy expensive models such as the FT-101D MP, TS-890, Flex 6700, or Hilberling 8000A, priced at € 4000 or more.

I'd rather invest the price difference in better antennas—and still have money left over for a DXpedition to the Maldives (HI).


Since the market is full of used transceivers and prices are falling sharply, I’ve decided to wait for a good opportunity—a well-preserved modern transceiver that allows external RX antenna input, has software support and compatibility with my existing setup, and is ergonomic and suitable for small modifications (e.g., adding a panadapter or roofing filter). With a bit of patience and luck, such a radio can be bought second-hand for around € 600.


Conclusion

Radio amateurs often go for expensive rigs, influenced by Close-In, RMDR, and NPR test results published by Rob Sherwood, Adam Farson, or Peter Hart (RadCom).
But in real-world band conditions, even in the most demanding contests like CQWW, CQWPX, or CQWW CW 160m, older and cheaper radios—with minimal upgrades—can perform just fine in environments with strong signals.

A similar conclusion was reached in a study for the Bavarian Contest Club by DK4YJ.
It showed that RMDR/phase noise sensitivity is much more critical than third-order products (IMD3).
Sometimes it’s better to choose an ergonomic mid-range radio than to overspend on a high-end rig that performs slightly better on lab tests.

This, of course, applies to HF receivers. On VHF (2 m) during contests, the situation is entirely different when you are using transverter. 
Unfortunately, I didn’t conduct similar measurements last year during the VHF Marconi contest, where the signals are truly extreme.

These measurements apply only to my location and antennas. Results may differ significantly if using long Beverage RX antennas.


Most often, the real source of unbearable interference is the correspondent's transmitter, which may key-click several kilohertz wide.
It’s a shame that some operators don’t know—or don’t bother—to fix their CW parameters (such as rise time) to reduce splatter.

For example, reducing CW rise time from 10 ms to 3 ms increases key-clicks by 20 dB at 750 Hz offset from the carrier.
Jeff AC0C describes this issue clearly on his website (ref. 3).

Similar problems occur in SSB contests, with overdriven compressors, overdriven final stages, and resulting broadband noise across the band.
On the receiving end, there’s no remedy, not even with a €4000 FT-101D MP or TS-890S.
In such cases, we can only rely on the technical culture of our fellow operators.

This exact situation was described years ago by Robi S53WW (ref. 7).
Figure 8 shows an example of such a poor signal recorded during the CQWW SSB contest.
Figure 9 shows, for comparison, a clean signal of the same strength.




Figure  8 : A snapshot of wide SSB signal during the CQ WW  SSB contest



 Figure  9 : A snapshot of two equal-strength signals in the CQ WW contest ,but with a significant difference in bandwith.


It’s also unfortunate that, due to the aforementioned tests, manufacturers tend to focus primarily on producing top-tier receivers, while the final stages of transmitters do not receive the same level of attention.


                                    Golden Rules for Choosing a  Radio Station


  • Don’t buy a station based on the brochure!

  • Consider Sherwood, ARRl,  Radcomm, ..... test,s  with caution.

  • "Instead of relying solely on lab specs, try evaluating the transceiver by comparing it side by side with a reference station, as I described in my blog post 'Comparison of Yaesu FT-2000 with build -in roofing filter with Kenwood TS 890S in ARRL 2025 CW competition."

  • Avoid buying a transceiver  that’s just hit the market!

  • Don’t test the transceiver at the dealer's shop.

  • Don’t be fooled by "bells and whistles." The panorama should be displayed on a PC. Even a "touch screen" can cause issues if you have wet fingers Avoid buying a station from a smoker.

  • If possible, borrow a transceiver  you're interested in and test it at home with your own antennas, over the weekend or during a WW contest.

  • Preferably, the transceiver should have a built-in power supply.

  • It's crucial to understand the protocol and what software (SW) options are available.

  • If you need to purchase a power supply opt for a linear one with transformer , not a "switcher."

  • The station should have an integrated tuner, dual antenna connectors,separate receiver connector  and, ideally, an RF preselector.

  • If you’re buying from a dealer, does the transceiver come with service support?

  • The best time to test the receiver is in the evening on 7 MHz, not in the morning.

  • The ultimate test of a transceiver  is the CQWW 160m contest or the UKV Marconi CW contest , ideally in a contest environment.

  • For your first transceiver , consider buying a second-hand one. The prices of used equipment tend to drop over time, similar to how used car values depreciate. Later, once you will have  more experience and if you participate in contests, invest in a new  transceiver according to reason.

  • Don’t overestimate technical requirements. If you’re not a contester  a mid-range transceiver on Sherwood scale priced around $1,200 is more than adequate. Ergonomics are far more important.



References:

  1. Werner Schnorrenberg DC4KU, Antenna Noise in the Shortwave Range – Funkamateur FA 12/14, pages 1290–1291.

  2. QEX May/June 2002 by G3RZP: HF Receiver Dynamic Range: How Much Do We Need?, pages 36–41.

  3. Jeff Blaine AC0C: https://ac0c.com/main/page_gear_mods_filter_failure.html

  4. BCC – Bavarian Contest Club: Dynamic Range or: How Much Roofing Does the Contester Need?
    http://www.bavarian-contest-club.de/projects/hardware/Close-In-Dynamic-Range-oder-Wie-viel-Roofing-braucht-der-Contester-;art376,1772

  5. http://www.sherweng.com/table.html

  6. https://www.remeeus.eu/hamradio/pa1hr/productreview.pdf

  7. Robi Vilhar S53WW: The Final Milliwatts or “Splitting Firewood at Two Meters”, CQ ZRS 6/1995
    http://slovhf.net/poslednji-milivati-ali-seku-drva-na-dva-metra/

* Unfortunately, there was no activity on 28 MHz, so the signal level data is estimated at –35 dBm.






torek, 29. december 2020

Some tips how to make good roofing filter for Up conversion receiver ( Slovenian language -English abstract)

Abstract 

Despite the great trend to switch to a down- conversion receiver or hybrid receiver design we have with up-conversion receiver design much better suppression of image interference.( Dr. Ulrich L.Rohde N1UL,DJ2LR-  QEX), that the importance of good second -order IMD performance can be in Europe more important than the third-order IMD product.!!! (  A typical 49m broadcast signals mix with signals near 8MHz generating ghost broadcast signals)

The down-conversion receiver may not perform on 14,10 and 7MHz as well as on other bands because of images.

The up-conversion receiver moves the image range into the VHF which a 30MHz low pass front end filter easily removes ! 

In the past in up-conversion receiver design VHF crystal filters weren,t available in the narrow widths and with the steep skirts that enabled the high dynamic range up-converting transceivers. The big change is the use of narrower roofing filter to bring the up-conversion rig to current standars for top IMD3 performing transceiver. Such narrow filter are now possible in the 50 to 75MHz range to fit the up-converting architecture

Most mid range radios have roofing filters that are to wide for serious contest. My  narrow 1,4kHz CW  and 2,4kHz SSB Roofing  50 Ohm  Filter design  can be  with minor changes used for up conversion receivers  with first  intermediate frequency (IF)  between 50 in 75 MHz  like :

          ICOM IC : 756, 765, 775,7200,7400, 7410, 7600,7700, 7850, 7851, 9100

          Kenwood  TS: 450, 480, 850, 870, 950, 2000

          Yaesu  FT :  450,847, 897, 920,950,991, 1000D, 1000MP,  2000, 9000


Some tips,how to make your own Roofing filter ( Use Google translate )

I am not responsible for any hardware failure. You made this mods on your own risk.! 




Nekaj nasvetov , kako narediti dober  roofing filter za Up conversion sprejemnike.


Down conversion sprejemniki so imeli pri produktih 3. reda IMD3 in recipročnem mešanju RMDR prednost pred up conversion sprejemniki , ker se nedavno na VHF frekvencah še ni dalo narediti ožjih filtrov od 3kHz.  Inradovi roofing filtri širine 5-6kHz  so bili za CW teknovanja preširoki , ožji NS 2,4kHz filtri, ki jih je za FT 2000 in FT 950 prodajal  Jeff AC0C pa so bili dražji od 300€.. 

Up conversion sprejemniki pa imajo zlasti v Evropi prednost , ker so bolj odporni na produkte 2. reda IMD2 ( f1+f2 , 2x f1- f2,..... na primer 7,1MHz +7,2MHz =14,3MHz  ali pa  2 x 11,95MHz - 9,7MHz = 14,2MHz ) -  črički ( birdies and spurs) zaradi broadcasting postaj , ki jih pa Rob Sherwood ne testira.
V spodnjem diagramu je jasno vidno, da so broadcasting signali,ki so vzrok motenj  bistveno močnejši od amaterskih postaj. Zato so down conversion in SDR sprejemniki na frekvencah med 7 in 15 Mhz lahko bolj občutljivi na motnje  od sprejemnikov s prvo medfrekvenco nekje med 60 in 70Mhz.





Slika 1: Jakost broadcasting postaj v dBm med 7 in 15MHz merjena na mojih antenah 

 Zato sem se odločil, da poiščem kvalitetnega  proizvajalca kristalov ,da izboljšam IMD3 ter RMDR in  sam naredim poceni  CW roofing filter 1,4 kHz,ki je primerljiv z filtrom proizvajalca  ICOM za IC 7851.  https://www.strictlyham.com.au/icom-ic-7851. 

Za izdelavo kvalitetnega roofing filtra za VHF področje  podajam nekaj nasvetov.

Poiščite kvalitetnega proizvajalca kristalov ter pri njem  naročite kristale z pribljižno sledečimi karakteristikami.;

Kristal za 3. overtonsko frekvenco npr 70 455 kHz ( primer FT 1000MP)

Kristal naj ima za SSB širino filtra 2,4kHz serijsko resonanco okrog 1000 Hz višjo Fs = 70 456 kHz

Kristal naj ima za CW širino filtra 1,4kHz serijsko resonanco okrog  1400 Hz višjo  Fs = 70 456,4kHz

kapacitvnost C0 manjša od 4pF
Upornost Rs manjša od 16 Ohm
kapacitivnost Cm= manjše od 1,5 If
toleranca manjša od +/-5 ppm
Priporočam ohišje HC 49/u
Za izdelavo roofing filtra ne potrebujete več kot 4  kose  kristalov.

Nekaj EU dobaviteljev kristalov :
www.omig.com
www.andyquarz.de
www.pupin.rs
www.krystaly.cz


Slika 2:  Primer izbire kristala za  CW filter FT 1000MP  frekvence  70.455  ( 70456,4) kHz proizvajalca Krystaly 

Program za izračun : Crystal Ladder Filter Calculator " DISHAL" by Horst  DJ6EV http://warc.org.uk/wp-content/uploads/2014/01/eDishalHelp-1.pdf

Za izračun CW filtra uporabite  tip  filtra  Butterworth za izračun SSB filtra pa  tip Chebyshev.

S spreminjajem B3dB parametra in PB ripple parametra  v programu Dishal se poskušajte čimbolj približati impedanci filtra 50 Ohm. V primeru,da konstruirate filter za drugačno impedanco ga morate
impedančno prilagoditi!

Na vhodu filtra obvezno predvidite  diplexer,ki prilagodi impedanco filtra v širokem območju na  izhod mešalnika. Uporabite diplexer calculator ( Bridged Tee) 
Kvaliteta Q na frekvencah 50 -70 MHz naj bo  1-3
https://www.changpuak.ch/electronics/calc_16a.php

V postaje vgrajeni originalni filtri so običajno prilagojeni na impedanco 50 Ohmov zato je primerno izdelati 50 Ohmski roofing filter in ga z obstoječim filtrom vezati v kaskado. Tako pridobimo tudi večje končno dušenje celotne verige filtrov.Prvemu kristalu je paralelno vezana induktivnost s katero kompenziramo Cp in s tem izboljšamo simetrijo filtra.

Zaradi dušenja filtra okrog 8dB v prepustnem pasu sem na izhodu dodal nizkošumni ojačevalnik z veliko dinamiko PGA103+, ki se mu z attenuatorjem prilagodi ojačenje tako, da celotna veriga nima večjega ojačenja od 1, sicer lahko pokvarimo IMD3 !!!





Slika 3: Primer - Shema roofing filtra za frekvenco 68,330 kHz (  FT 897D, FT 857.. )










  Slika 4: Primer PCB tiskanine


Po vgradnji boste v primeru, da lahko preklopite med različnimi roofing  filtri,kot je to npr . primer v postaji FT 2000( 15 kHz, 6 kHz, 3 kHz)  pri vključenem novem roofing filtru zaznali bistveno manjši šum,
kar je normalno !!!!

Karakteristiko filtra najlaže izmerite z VNA mostičem. Na spodnjih slikah   podajam primer izmerjenega  2,1 kHz SSB filtra. za FT 2000   izmerjeno z NanoVNA .



  Slika 5: Karakteristika 2,1kHz roofing filtra ( Logmax, SWR, Smith )
  span 15kHz 




 Slika 6:  2,1KHz  SSB filter  prenosna karakteristika na - 6 dB / 2,1kHz 
(   NanoVNA WebUSB Client - v1,4 android  10  )







Slika 7 : 2, 1kHz SSB filter prenosna karakteristika na - 65 dB / 10,5 kHz 
(  NanoVNA WebUSB Client - v1.4 android  10 ) 

Nekaj nasvetov,kako z NanoVNA merimo kristalni filter : 
https://www.youtube.com/watch?v=3uRA1p6OotA

Kako lahko sami izmerimo parametre kristalov :
https://www.youtube.com/watch?v=G9zZRNzhsEE
https://www.youtube.com/watch?v=rVIHVi-7brs

Kako simuliramo filter :
https://www.youtube.com/watch?v=Y0_H8e4QTy8

Načrt celotnega vezja je objavljen na mojem blogu :  YAESU FT 950 SSB 2,4kHz roofing filter.

Opomba:  Vgradnja v postajo na lastno odgovornost. Priporočam le izkušenim z tehniškim znanjem. Za eventuelne napake ne odgovarjam. 




nedelja, 5. januar 2020

YAESU FT 950 SSB 2,4Khz roofing filter


As a backup to my FT 2000 radio I am using a FT 950, which has the same 69,450MHz first IF frequency. The  first IF stage features three roofing filters (15 kHz, 6 kHz, and 3 kHz) automatically selected by mode. Each roofing filter is a four-pole, fundamental-mode monolithic crystal filter design. Unfortunately, some choices made in the first IF design of the rig left is with quite a poor performance score in one area - close-in high-strength signal handling capability. The original Yaesu 3 kHz filter measures only about 7 kHz wide at -6dB.  After Jeff AC0C wrote that  FT2K/FT950 excellent NS roofing filter in no longer in a production I have decided to make my own roofing filter. Roofing filter attenuates and reduce IMD3 products and DSP has less bandwith to process.


After I successfully installed in the second IF 450kHZ a selection of  2 (two)  Murata +/- 2kHz CFWLA450KJFA ceramic filters placed in a cascade and  therefore reduced IF noise and IF image(see picture 1) ,  I also decided  to improve the FT 950  input IMD performance with my own SSB roofing filter.  The same roofing filter can also be used for FT 2000 receiver.




Picture :1  New 450Khz  IF

A homemade SSB the 4 pole 50 Ohm Cohn crystal filters of 2,7 KHz bandwidth at  6dB has been modeled with a AADE Filter design V4,5 program and is built from discrete 3rd overtone crystals. Crystal parameters:  3rd overtone  FL 69,450 Mhz , RR = 20 oHm, C0 = 4pF, FL= 5ppm, package UM1  (see picture 2 and 3)





Picture:2 Cohn  4 pole filter  modeled with an AADE Filter design





Picture:3  Input impedance modeled with an AADE Filter design

To improve the impedance mismatch between 1st mixer and roofing filter I added a diplexer to properly terminate the roofing filter. Diplexer terminates the impedance of the roofing filter to the first  mixer in a wide frequency range. To compensate filter loss I have added ultra-low noise (0,55dB) and  high performance MMIC LNA- SPF 5189Z (or MMIC PGA 103+)with output third order intercept point typ +38.5 dBm. To lower the MMIC  gain a 4dB T attenuator has been built at the output.(see picture 4)





Picture: 4  The 69,450Mhz roofing filter circuit


Unlike Jeff AC0C, I decided to improve the ultimate attenuation and   installed my new 2,7 kHz roofing filter before the original YAESU 3kHz XF1004 filter.( see picture 5)








Picture: 5 


Before the 2,7 kHz roofing filter has been built in the Yaesu 3 kHz filter measures about 6-7 kHz wide at -6dB. The spectrum is taken from the 2nd IF TP 1056 and shows  the filter shape of the original FT filter.( see picture 6)





Picture 6 : Spectrum taken before 2,7kHz  roofing filter has been built



Preparing the 2,7 kHz roofing filter for installation

1.       Locate the working area on the PC board ( see picture 7)





Picture 7: Working area



2.       Unsolder SMD capacitor C 1583 (see picture 8)





Picture: 8


3.  Cut the PCB trace between T1026 and original Yaesu XF 1004 3kHz filter (see picture 8)

4. Solder the input coax center conductor to the center pads of T1026 (see picture 8)

5. Solder the input coax shield to the transformer T1026 casing.(see picture 8,9)



Picture: 9

6. Solder the output coax center conductor to the input pad of original XF1004   Yaesu 3 kHz filter (see pictures 7,8,9)
7. Solder the output coax shield to the ground near filter XF1004 (see pictures 7,8,9)
8. Solder back SMD the capacitor C 1583

9. Connect 9V DC to the new filter from pin 14 - J 1005 ( see pictures 10,11,12)



Picture:10





Picture: 11 






Picture:12


9. Trim T1026 and T1033 to maximum signal level


After the 2,7 kHz roofing filter has been built in ,(see picture 13)  the front end is far more selective and a new filter measures about 2,7 kHz wide at -6dB and 8kHz at -50dB. The spectrum is taken again from the 2nd IF TP 1056 and it shows a new filter shape .(see picture 14)






Picture: 13 Roofing filter  build in



Picture :14  Spectrum taken after 2,7kHz roofing filter has been built in




More about the built in procedure can be seen from the pictures above.




After this modification and  adding a selection of  2 (two)  4kHz Ceramic filter in second IF 450kHz ,   I measured IMD and RMDR dynamic range at offsets of 2kHz  and 5kHz  with   2 x -6dBm  DC4KU   HF- Zweitongenerator FA-2-HF.  Two-tone third order dynamic range (IMD DR) is the diference between MDS and the levels of two interfering signals causing IMD products 3dB over the noise floor. RMDR is Reciprocial mixing dinamic range , measured as a 3dB increase in noise floor.    See Picture 15 




Picture 15: DC4KU  Two- tone RF generator -6dBm



Results of two tone IMD testing - 1.st preamp OFF ; 500Hz Bandwith; Roofing 2,7KHz; AGC Off , Test band is 7MHz ;

MDS = - 128 dBm, 
Blocking above noise floor @ 10 kHz ,measured as 3dB increase in noise floor  107 dB
IMD2 Dinamic range second order on 14 Mhz    = 76 dB

Spacing offset:                                              IMD DR3:                    RMDR              

2  kHz                                                               86dB                       90 dB           
5 kHz                                                                88dB                       98 dB              


According to the experts’ opinion (e.g. Rob NC0B and Tom W8JI), the IMD DR3 of 85 dB is enough for CW. Thus my 2 kHz IMD DR3 86*dB results are very good.


I have been testing the new roofing in CQWW_2019 CW/LP   contest with more than 2010 QSOs  and I was really satisfied with a new installation. I used the radio on the air  several days before  CQ  WW Contest and great difference was noted  between the 6kHz filter and the new 2,7 kHz filter at dynamic range







With a little changes the same filter can be build also in FT2K


Warning !!! Changing any hardware inside FT950 will void your warranty.I am not responsible for any hardware failure. You made this mods on your own risk.! This modification requires a high level of soldering skill, possibly beyond that normally possessed by the average HAM. Professional assistance is advised if you are not confident that you have  this ability.

Note for simplified installation  10.08.2020

I found that is much easier  to  unsolder (zero) oHm  SMD resistor R 1393 to insert  roofing  filter in cascade with original Yaesu 3kHz roofing filter. See pictures  below.
















Some Tips for Contest Operators Using Yaesu FT -2000 or FT- 950 with N1MM Logger

  During contests, operators using the FT-2000 or FT-950 sometimes struggle to locate small buttons on the radio, such as ATT, IPO, RX, or...