++++++++++++++++++++++++++++++ Date: Wed, 06 Jun 2001 22:38:32 -0700 From: "John Grebenkemper, KI6WX" Subject: [Elecraft] Mod to K2 CW Filter To Improve Rejection A month or two ago, KO0B sent out a modification to the CW filter that improved its sound quality, but there were no measurements of this improvement. There were some folks who made the change that couldn't hear any difference before and after. I don't trust my ears unless I can compare audio a few seconds apart, so I set out to measure the improvement if possible. One thing led to another, and I have come up with a somewhat different modification that definitely makes a substantial improvement to the CW filter response. For instance, at the 100 Hz bandwidth, the filter bandwidth at 60 dB down is 1210 Hz before the modification but is reduced to 710 Hz after the change. The filter ultimate rejection is now better than 100 dB. The basic modification improves the crystal grounding with a change that should take 5-10 minutes. A second change improves the bypassing to the varactor bias circuit using chip caps, but there is no proof that this change makes any further improvement and most people should skip doing it. The changes are given in a pdf file that can be downloaded from http://home.pacbell.net/johngreb/KI6WX.html. Let me know if you have any trouble reaching this page as I just set it up today. - -John KI6WX ++++++++++++++++++++++++++++++++++++++ Date: Thu, 07 Jun 2001 07:25:29 -0700 From: Wayne Burdick Subject: [Elecraft] KO0B/KI6WX mod to K2 CW filter I wanted to thank John for doing such a comprehensive job in updating KO0B's original mods. The plots show the mod to be working well, and the extra insertion loss is not likely to be an issue for most ops. John will be visiting me this weekend so we can compare the KO0B/KI6WX crystal filter mod to the N6KR audio filter ;) In case you didn't recognize his call or name, John is the designer of the "Tandem Match," an accurate directional wattmeter, that has been featured in the ARRL Handbook. In the 1996 Handbook it can be found on page 22.36. 73, Wayne N6KR "John Grebenkemper, KI6WX" wrote: > > A month or two ago, KO0B sent out a modification to the CW filter that > improved its sound quality, but there were no measurements... > > One thing led to another, and I have come up with a somewhat different > modification... ++++++++++++++++++++++++++++++++++++++++++ Date: Thu, 07 Jun 2001 07:16:45 -0700 From: Louis Hlousek Subject: Re: [Elecraft] Mod to K2 CW Filter To Improve Rejection >From what I understand the danger in soldering the crystals near their bases is the possibility of damaging the can seal. I don't know how easy it is to damage the xtals with heat but using low temp indium solder might be a way to avoid potential damage. Alloys are available with melting temperatures raging from below room temp to over 250 C. Indium Corp is one manufacture. www.indium.com/ . I used this stuff in graduate school to solder directly to metalized piezoelectric materials. I don't know how readily available this stuff is in small quantities but it might be useful for soldering the grounding leads to the xtal cans if they can indeed be easily damaged by normal soldering temps. Anybody else have experience with low temp solders? Lou W7DZN ++++++++++++++++++++++++++++++++++++++++++++ Date: Thu, 07 Jun 2001 18:29:00 -0700 From: "John Grebenkemper, KI6WX" Subject: Re: [Elecraft] Mod to K2 CW Filter To Improve Rejection Tim; The grounding of both sides of the case is what causes the increased insertion loss. Before the change, the insertion loss of all three bandwidths that I measured was roughly the same. After grounding both sides, the insertion loss increases by 4-5 dB for the 100 Hz bandwidth compared to the 700 Hz bandwidth. Unfortunately, I was not looking carefully at insertion loss when I did these changes and I had to go back and infer the number from the data I had taken. If the measurements are done more carefully, one should be able to precisely measure the change in the insertion loss. The absolute insertion loss in the K2 circuit is more difficult to determine, but could also be done. This brings me to a point I didn't make in my original message. This change has only been done on one K2. It would be nice if some of the folks modifying other K2's take before and after measurements with Spectrogram following the procedure in Appendix A. These measurements could be sent to me and we could see if this difference holds across multiple K2's. Please note whether you have bypassed the second crystal filter, and I would prefer to receive Spectrogram Log files because they provide more accurate information than the plots. - -John KI6WX > John, > > I have one question after reading your documentation. Is the changing of > the ground leads on the cans of the filter what increases the insertion > loss, or is it adding the bypass chip capacitors? > > Tim Raymer > > At 22:38 06/06/01 -0700, John Grebenkemper, KI6WX wrote: > >A month or two ago, KO0B sent out a modification to the CW filter that +++++++++++++++++++++++++++++++++++++++++++++++++ Date: Thu, 07 Jun 2001 18:51:37 -0700 From: "John Grebenkemper, KI6WX" Subject: Re: [Elecraft] Mod to K2 CW Filter To Improve Rejection Steve; The majority of the benefit comes from grounding the Xtal cases. This provides a significant benefit to a rejection of at least 75 dB, and the chip caps made no additional improvement to this level. I made the much more sensitive measurement that measured down to a rejection of 110 dB with both the Xtal case grounding and the chip caps installed. I did not want to publish data that did not reflect exactly what I had done, so I did include the chip cap addition. In my opinion, I don't think you will be able to hear any difference with or without the chip caps. You might be able to measure some difference, but it will be completely swamped out by oscillator phase noise. So the chip cap change is described for completeness, but most people shouldn't bother to add them unless we get some additional data that says they really do help at rejections greater than 80 dB. - -John KI6WX > Ditto... > Am I reading this correctly??? Simply grounding the Xtal cans from the > side improves the filters? Is this a step in the right direction without > adding the chip caps? How much improvement is assignable to the ground > changes, and how much additional improvement is had with the cap addition? > > Any benefit to doing a similar ground change to the xtals on the SSB > adapter? > > Steve > aa8af +++++++++++++++++++++++++++++++++++++++++++++ Date: Thu, 07 Jun 2001 20:14:30 -0600 From: Larry East Subject: Re: [Elecraft] Mod to K2 CW Filter To Improve Rejection At 10:38 PM 6/6/01 -0700, John Grebenkemper wrote: ...snip... >The changes are given in a pdf file that can be downloaded from >http://home.pacbell.net/johngreb/KI6WX.html. Nice work and a very nice write-up, John! My observations lead me to believe that better crystal can grounding gives the greatest gain in filter performance. When I built my K2, I grounded the crystal cans very close to the bottom edges (you can view a photo at http://www.ida.net/users/w1hue/k2_xtals.gif ). I recently made filter response measurements using a marker generator as a signal source and an HP 403B wideband AC voltmeter on the audio output. The results that I obtained for my 150 Hz and 400 Hz filters are as follows: 150 Hz 400 Hz Filter Filter --------------- 6dB Bandwidth 110 Hz 200 Hz 20dB Bandwidth 260 Hz 340 Hz 60dB Bandwidth 700 Hz 870 Hz These values are similar to yours, although my filters appear to be a bit narrower, particularly at the -6dB level. Unfortunately, my filters show considerably more insertion loss at the narrower settings than do yours. The following table shows my filter losses relative to the 1000 Hz filter: Filter Loss --------------- 1000Hz - 0.0dB (reference) 600Hz - -2.5dB 400Hz - -4.5dB 150Hz - -9.0dB I tried the mode by Roy, AB7CE, intended to reduce insertion loss at narrow sittings with very poor results -- it did not significantly reduce the loss and produced "double peaked" shapes for filter widths greater than 500Hz. The rather narrow -6dB bandwidths, high loss at narrow settings and failure of Roy's mod to work in my rig (plus that fact that my K2 is an early unit [serial #491]) led Gary S. (and others) to suggest that my crystals might not be well matched and that I should try replacing them. Perhaps I will ... but since the filter appears to have excellent rejection characteristics, I'm tempted to just live with the loss at narrow bandwidths. I have Spectrogram running on my Toshiba laptop, but I am unable to get a dynamic range greater than about 40dB. Since you and others appear to be able to get a much greater dynamic range, I suspect that something is amiss with the "sound card". I have recently installed a sound card in my "desk top" PC (although a rather inexpensive one...) but I have not yet tried Spectrogram with it. Has anyone else tried Spectrogram on a Toshiba laptop? 72/73 Larry W1HUE/7 ++++++++++++++++++++++++++++++++++++++ Date: Thu, 07 Jun 2001 23:30:08 -0700 From: "John Grebenkemper, KI6WX" Subject: Re: [Elecraft] Mod to K2 CW Filter To Improve Rejection I used both versions 6.0.8 and 5.1.6 and found no difference in their performance and results. I'm not convinced that it makes any difference which version you use. If someone has a definitive comparison of both versions getting different results with the same input, I'd be very interested in hearing about it. - -John KI6WX > The maximum display range for SpectroGram is 90 db. The newer releases, > especially 6.0.8, give much better control over displayed range than the > older releases. > > Vic Rosenthal wrote: > > > > Larry East wrote: > > > > > Has anyone else tried Spectrogram on a Toshiba laptop? > > > > Yes, my old 100 MHz Tecra. It also does not do much better than 40 db dynamic > > range regardless of how I fiddle with the level. ++++++++++++++++++++++++++++++++++++++++++ Date: Fri, 08 Jun 2001 12:10:29 -0500 From: "George, W5YR" Subject: Re: [Elecraft] Mod to K2 CW Filter To Improve Rejection Depends upon what you are doing with the program, etc., John. For example, 6.0.8 has Peak Hold which is invaluable for some applications and unneeded for others. There are a number of other differences - but again their merit depends upon what you are using the program for and how you are using it. Used properly, you should NOT get different results with the various program releases for the same input data. What you *do* get is an increased feature set with added capabilities. SpectroGram now compares, within its capabilities, very favorably with SpectraPLUS which is a multi-hundred$$$ commercial program. Previous to 6.0.8. this was not the case. 72/73, George W5YR - the Yellow Rose of Texas QRP-L 1373 NETXQRP 6 Fairview, TX 30 mi NE of Dallas in Collin county EM13qe Amateur Radio W5YR, in the 55th year and it just keeps getting better! Icom IC-756PRO #02121 Kachina #91900556 IC-765 #02437 "John Grebenkemper, KI6WX" wrote: > > I used both versions 6.0.8 and 5.1.6 and found no difference in their > performance and results. I'm not convinced that it makes any difference > which version you use. If someone has a definitive comparison of both > versions getting different results with the same input, I'd be very > interested in hearing about it. > -John > KI6WX +++++++++++++++++++++++++++++++++++++++++++++++++ Date: Fri, 08 Jun 2001 22:13:03 -0700 From: "John Grebenkemper, KI6WX" Subject: Re: [Elecraft] Elecraft List Ok, I put up an HTML version of the mod on the web page. Because of server limitations, I needed to post it as a single file (mht) version, but most new web browsers should be able to read this. I don't know about Web TV. You can access the HTML version at http://home.pacbell.net/johngreb/KI6WX.html. - -John KI6WX > Well........... I know this is my problem but with my webtv box I can't > download anything! As in pdf format. Reading between the lines am I to > believe that I can get the improved cw filter skirts by shortening the > ground leads? Where do the bypass caps go? You guys are going to shame > me into buying a computer! These K2's are expensive! > > 73, Kurt > W7QHD (since 1951) > web page: http://W7QHD.tripod.com > +++++++++++++++++++++++++++++++++++++++++++++++++ Date: Sun, 10 Jun 2001 11:20:02 -0700 From: "John Grebenkemper, KI6WX" Subject: [Elecraft] Update on CW Filter Improvements Wayne and I met yesterday and went over the changes I had made in my K2. One of the things of which we are still not sure, is why grounding the other side of the crystal cans improves the filter response. However, we both had independently come up with a theory to explain this result. The crystal cans are very close to the SSB board; we might be coupling energy from the outside of the cans into the SSB board and bypassing some of the crystal filter rejection. This can be easily checked by someone who has an unmodified K2 and access to Spectrogram and PC sound card that can obtain the full dynamic range (many older PC's can't). Measure the filter response with the KSB2 board plugged in. Then remove the KSB2 board and insert jumpers between pins 1 and 3 on J9 and J10. Repeat the measurement with Spectrogram. If there is a significant improvement with the KSB2 board not installed, then coupling to the SSB board is the likely cause of the filter degradation. If anyone does this, please let the list know your results. - -John KI6WX +++++++++++++++++++++++++++++++++++++++++ Date: Wed, 13 Jun 2001 20:36:16 -0700 From: "John Grebenkemper, KI6WX" Subject: Re: [Elecraft] (K2) Insertion loss and CW filter mod. Kevin; > My thesis, then, is that the leakage path for in-band signals is much more efficient than > it is for out-of-band stuff. (nearly as efficient as the intended signal path, if I am > interpreting John's data correctly) This leakage lowers the insertion loss of the filter. When the > leakage of the filter is essentially eliminated, the insertion loss increases to it's > "real" value. Now don't misunderstand me...This leaked signal is just as good as what passed > through _all_ the rocks, but I'm willing to give it up in order to hear signals "fade-to-nothing" as I tune > across the band. > One of my early speculations was that the crystal cases were coupling to other parts of the circuit. However, it is pretty hard to explain the change in insertion loss in the middle of the passband when the skirts don't show any effect until 50 dB down. One of the measurements that I made that was not in the note was a network analyzer measurement of the filter before and after the change. The network analyzer directly drives the CW filter through J9 and receives it at J10. None of the other K2 circuitry is involved. These measurements show that there is no difference in the filter response before and after the changes. This means that whatever is causing the increase in filter bandwidth and change in insertion loss, it is not due to coupling between the filter crystals, but coupling between the crystals and other parts of the K2 circuitry. I borrowed a set of K2 crystals for the CW filter from Wayne. Today I measured them in two different impedance analyzers and came up with the equivalent electrical model of the crystals. Interestingly, the crystal response varied only slightly between case grounded and case floating. I will now be able to run a simulation in SPICE and see what the theoretical response of the filter should be including insertion loss variation with filter bandwidth. - -John KI6WX ++++++++++++++++++++++++++++++++++++++ Date: Thu, 14 Jun 2001 22:27:47 -0700 From: "John Grebenkemper, KI6WX" Subject: [Elecraft] K2 CW Filter Crystal Equivalent Circuit Values I have received several requests about the measured values I took on a sample of crystals used in the K2 CW filter. All measurements were taken with a HP-4195A Impedance Analyzer and spot checked on a HP-4192 Impedance Analyzer. For those interested, the geometric mean of these values is: Rs=18.1 ohms Ls=71.4395 mH Cs=0.0146858 pF Cp=4.09736 pF fs=4.913631 MHz Q=122000 The data is repeatable to the first two digits and usually only varied by 1 unit at the third digit. The L's and C's are shown to the full resolution of the Impedance Analyzer calculation in case you want to use these to calculate resonant frequencies. Remember, these crystals have been binned by Elecraft so that fs is in 20 Hz bins. - -John KI6WX ++++++++++++++++++++++++++++++++++++++++ Date: Sat, 15 Sep 2001 13:02:55 -1000 From: John Buck Subject: [Elecraft] KAF2 comments and Test of CW Filter Rejection Mod. per KO0B/KI8WX Finished the KAF2. It tuned as advertised and really does a good job eliminating residual noise outside the desired passband. This filter is a must have for both CW and SSB modes. The listeners noise fatigue is greatly reduced. While I had the system open I decided to incorporate the KO0B/KI8WX shielding suggestions to improve the basic CW filter. K2 #1242 has all currently available options and significant mods, such as XFIL, Sine sidetone and PSK 31 installed. I made the best measurements I could using Spectrogram 6.2.3 The measurement system had a 70db below signal capability using the 90 db range. No Spectrogram or receiver setting changes (except for filter selection) were made during the modification and test process. I used a broadband noise generator as the signal source. Careful measurements were made three times: 1. Before modification; 2. After the crystal ground wires were modified by soldering to the side of the cans; 3. After the addition of the .1uf surface mount capacitors per KI6WX with the Gull Wing shield wires per KO0B. I printed the Display plots for each of the 3 measurements for each of the four filter settings. OP1, 700Hz, 400Hz and 100Hz. FL2, 3, and 4 are centered on 750Hz sidetone choice. When the plots were overlayed using a light box so I could see any differences, the following the conclusions were reached. A. There was no harm of any type due the the changes. I did not see any evidence of the slight broadening of the top of the filter response curve reported by one of the authors. In all cases, each sucessive plot coincided with or fit within the previous plot. B. As expected, the FL1 OP1 filter showed no significant changes. (KSB2 installed) C. The FL2 700 Hz filter shows 2 to 3 db improvement in the 1500 to 2000Hz area. Almost all of the improvement is from the crystal can grounding change. No effect from the gull wing and capacitor additions is seen. D. The FL3 400 Hz filter also shows 2 to 3 db improvement in the 2000Hz area with noticable improvement starting in the 1000Hz region. The gull wing plot is almost identical to the grounding plot but there is a db or so in the 2000Hz area. E. The FL4 100 Hz filter shows about 5 db in the 1800/2000Hz region. About 2 db is due to the gull wing mod. About 1 db overall improvement starts as close as 1000 Hz. Was it worth it? Probably. My K2 now has even better close in rejection of strong signals than before. Knocking down a signal 400 Hz above the center of my filter by another db or two or a signal 1000 Hz higher by 5db could make a real difference. Then I can turn on the KAF2 audio filter for even more help! Who said that K2 does not have bells and whistles? I have .jpg files of the plots on my system if someone really wants to analyze my data. I did not make the .txt spreadsheet readable files however. I freshly downloaded Spectrogram for this effort and did not discover that feature until I was almost done. Aloha, John KH7T +++++++++++++++++++ Reply-To: "Ron D'Eau Claire" From: "Ron D'Eau Claire" To: "Elecraft-list" Subject: Re: [Elecraft] RFI: KI6WX K2 XTAL filter mod Date: Sat, 8 Jun 2002 18:14:50 -0700 > I can't seem to find this on the Elecraft site lately. Has it fallen > from grace? > > 73, Rod N0RC You can find the mod on the Elecraft WEB site under Builder's Resources, Modifications, "2nd Stage Xfil Mod for a Flatter SSB Response." As for its value, I questioned Wayne about that recently and here's what he replied: "Let me be unequivocal on this, as I always have: the 2nd XFIL mod makes a HUGE improvement in USB/LSB matching, and simultaneously allows L34 to be set close to its optimal CW setting. "I strongly recommend that everyone make the mod. It is so important that I will be adding it to the K2 RF board the next time it is revised.." Ron AC7AC K2 # 1289 +++++++++++++++++++++ From: "Rod N0RC" To: "Elecraft-list" Subject: Re: [Elecraft] RFI: KI6WX K2 XTAL filter mod Date: Sun, 9 Jun 2002 06:08:05 -0600 Thanks to all for help and assistance. I have located the article. Turns out it was NOT on Elecraft's site, it WAS on John's site: http://home.pacbell.net/johngreb/KI6WX.html 73, Rod N0RC ----- Original Message ----- From: "Rod N0RC" To: "Elecraft-list" Sent: Saturday, June 08, 2002 6:12 PM Subject: [Elecraft] RFI: KI6WX K2 XTAL filter mod > Folks, > > I can't seem to find this on the Elecraft site lately. Has it fallen > from grace? +++++++++++++++++++