+++++++++++++++++++ Date: Sat, 02 Mar 2002 10:06:41 +0100 From: Roland CORDESSES To: elecraft at mailman.qth.net Subject: [Elecraft] Sall bandwidth of the OP1 SSB filter My K2 #2379 is now completed as well as the KSB2 option board. Thanks to the fine construction notice (Heathkit style) and the numerous informations available on the Elecraft WEB site, I had no troubles to adjust it and to properly tune the BFO frequencies. While I still have not made any QSO, I have carried out many measurements which show the performances of my K2 (TX and RX) are always as good and often better that the Elecraft published data. As a designer and builder of many homebrew high performance ham receivers, I have been specially impressed by the measured IP3 and dynamic range reached in such a small package. I am now concerned with a problem related to the IF bandwidth of the SSB filter used in the KSB2. This BW(-6dB), measured with Spectrogram or in the transmit mode, is only about 1.75kHz, a small value compared to the 2.2kHz published BW. In order to confirm my home measurements I conducted some tests on the filter alone with an HP Vector Network Analyzer and obtained the same results. Before building the KSB2, I measured the crystals parameters and obtained the following data (they are the averaged values of the 7 crystals but the dispersion is rather small): Series resonance frequency : Fs=4.913866 MHz Motional inductance # 80.8 mH Motional capacitance # 0.013 pF Unloaded Q : Qu # 130000 Parallel capacity : Cp # 4 pF In the past few years I have build many ladder crystal filters (5 to 12 poles CW, SSB and noise filter) using the tools and methods provided by Wes Hayward W7ZOI and others. Thus I simulated the KSB2 crystal filter using these tools and the ANSOFT Harmonica software (the "big brother" of the ARRL Radio Designer program) : I obtained a 1.8kHz BW with the above crystal parameters. I then computed the capacitors needed for a 0.5dB 2kHz Chebycheff ladder filter (1500 ohms I/O impedances)and get values different from those of the KSB2: moreover, as the 7 crystal frequencies have very small dispersion (delta F max=10Hz), I find that I need more serial capacitors in the meshes than the KSB2 PCB can accept. I am now thinking about building a piggyback PCB for an eventual new filter but before, I would be glad to receive comments about my problem. A feedback from the Elecraft K2 designers would be particularly appreciated. 73's Roland Cordesses/ F2DC K2 # 2379 +++++++++++++++++++ Date: Sat, 02 Mar 2002 10:13:31 -0800 From: Wayne Burdick Organization: Elecraft To: Roland CORDESSES Cc: elecraft at mailman.qth.net, KI6WX at pacbell.net, Eric Swartz Subject: [Elecraft] Re: bandwidth of OP1 Hi Roland, Our models and measurements of the KSB2 filter came in at an average of 2.2 kHz, with some small variation from one crystal lot to the next. This result has been duplicated by others, most recently John Grebenkemper, KI6WX, who wrote an application note regarding improving the ultimate attenuation of the filter in receive mode. John has more recently modeled filters with wider bandwidths and in some cases less ripple, using the same topology represented by the KSB2 PC board. Of course termination impedance of the filter is critical for obtaining the correct bandwidth. Eric and I will re-model the filter using the crystal parameters you measured. Thanks for this information. You might also want to compare notes with John, at KI6WX at pacbell.net. I copied him on this e-mail. 73, Wayne N6KR ++++++++++++++++++++ Date: Sat, 02 Mar 2002 11:40:06 -0800 From: "John Grebenkemper, KI6WX" Subject: Re: [Elecraft] Sall bandwidth of the OP1 SSB filter To: Elecraft Mailing List Roland; For the last month, I have been working with Eric and Wayne to arrive at a wider SSB Xtal filter without causing any change in the KSB2 circuit board. While theory provides some straight-forward ladder filter capacitor sizes to achieve a wide range of response, practice turns out to be much more difficult. You have to pick capacitor sizes that are standard values (for monolithic ceramic capacitors used in the KSB2 this is 10% values at a 5% tolerance) and you have to analyze the circuit for capacitor and Xtal variation. These restrictions end up eliminating most bandwidth choices. I have some tentative choices, but I am still analyzing this to make sure they will work over a wide set of component variations. When we finish this, the data will be published on my web site and I would assume it will also be available on the Elecraft web site. I hope to be able to put out several sets of capacitor values so the builder can decide what bandwith SSB filter they desire. On the receive side, you also have to take into account the response of the second Xtal filter. This filter effects the SSB response of the K2, and its current response tends to narrow the receive bandwidth of the SSB filter. Elecraft sorts the crystals by resonant frequency and bins them in 10 Hz increments. There is no control on the motional inductance and capacitance. Your resonant frequency and motional inductance values are somewhat higher than my measurements. I would appreciate it if you would email me directly your measurements on all of the individual crystals. Also, did you directly measure the Rs of the crystals? I will add this to my data on the K2 crystals to help determine a reasonable estimate on crystal variation. For the filter design impedance, you assumed a value of 1500 ohms. In the transmit mode, the load impedance on the filter varies from about 1600 ohms to 900 ohms depending on the setting of the D1/D2 attenuator in the KSB2. Since generally the attenuator is reducing the output, the impedance tends to fall on the lower side of these values. I've also made some measurements on the NE602 that indicate that its impedance is somewhat less than 1500 ohms. These variations will effect the filter response. -John KI6WX +++++++++++++++