+++++++++++++++++++++ Date: Wed, 26 Sep 2001 20:37:55 -0700 From: "John Grebenkemper, KI6WX" Subject: [Elecraft] Frequency Stability Improvements to the K2 I've been working on a set of modifications to the K2 that greatly reduce frequency drift. I've applied these changes to my K2 and measurements show that they definitely work. Before I publish this set of changes, I'd like to find out if other K2's have a drift rate similar to mine. The link listed below will provide a set of instructions on how to measure the frequency drift of a K2 and separate that drift into the VFO and BFO components using just WWV and the normal daily temperature variations. I'd appreciate receiving email on the results of any K2 owners who wish to undertake this measurement. If all of the K2's have a similar underlying drift rate, then I can publish the changes as a modification that can be done the same on all K2's. If the drift rate differs by K2, then each K2 will have to be tuned to minimize drift. The circuit mod allows a single resistor to control the correction of each oscillator, so this can be done but will require each K2 to be individually measured for frequency drift. Further details can be found at: http://home.pacbell.net/johngreb/measuring_and_improving_the_frequency_stabi lity.pdf - -John KI6WX +++++++++++++++++++++++++ Date: Thu, 27 Sep 2001 00:18:25 -0700 From: "John Grebenkemper, KI6WX" Subject: [Elecraft] Frequency Stability Improvements to the K2 Since the pdf address hiccuped in the email, you can also get to this pdf file at: http://home.pacbell.net/johngreb/KI6WX.html > I've been working on a set of modifications to the K2 that greatly reduce > frequency drift. I've applied these changes to my K2 and measurements show > that they definitely work. --- snip --- ++++++++++++++++ Date: Thu, 27 Sep 2001 19:06:03 -0700 From: "John Grebenkemper, KI6WX" Subject: Re: [Elecraft] Frequency Stability Improvements to the K2 Once you feel confident that you can make the measurements with one or two minutes, I would start with the K2 cold. Its internal temperature takes nearly an hour to stabilize, so there is very little change in frequency in the first minute or two. On my K2 with the LCD in nightime mode and the mixer set for performance (this gives maximum internal heat generation), my K2 would have an internal temperature that was 10F above ambient. - -John KI6WX > "John Grebenkemper, KI6WX" wrote: > > > > I've been working on a set of modifications to the K2 that greatly reduce > > frequency drift. > > Assuming that I want to make a test run, should the K2 be turned on > > a) Immediately before the first measurement > > b) Half an hour before the first measurement > > Should it be turned off at any time between the measurements? > > 73, > Vic, K2VCO > Fresno CA ++++++++++++++++++ Date: Tue, 20 Nov 2001 21:53:18 -0800 From: "John Grebenkemper, KI6WX" Subject: [Elecraft] Reducing K2 Frequency Drift For the last few months, I have been working on several changes to my K2 to reduce frequency drift in the VFO and BFO. Several versions of a document describing these changes have been posted on my home page at http://home.pacbell.net/johngreb/KI6WX.html since September. After several reviews, the document is now ready for general distribution. In a simulated PSK31 QSO, these changes reduced the drift of my K2 from 155 Hz to 65 Hz. Electrically, the changes are pretty easy to implement requiring only 3 resistors, an LM334, and cutting one trace for each oscillator. However, each K2 drifts a different amount and the value of 1 resistor will have to be selected to minimize the drift of each oscillator. A method is described within the note of a way to measure the drift of each oscillator using WWV as a frequency standard. If you have any questions about this document, please email them to me. Based on any feedback I receive, I may make some changes in the description of the modification to further clarify the content. The modification works very well in my K2, and I don't expect there will be any changes to the circuit. This modification has not been reviewed or approved by Elecraft. - -John KI6WX +++++++++++++++++++ Date: Wed, 21 Nov 2001 18:54:14 -0800 From: "John Grebenkemper, KI6WX" Subject: [Elecraft] Reducing K2 Frequency Drift (Update) In my original message, I stated that my original K2 drifted by 155 Hz in a simulated PSK31 QSO. It should be noted that this is a worst case number measuring the K2 frequency drift from a cold start and running a digital mode in a 50% duty cycle. In more normal operation with a K2 that has warmed up, I would expect the drift to be less than 1/4 of the above number. I've updated the document on my web site to more accurately reflect this measurement. - -John KI6WX +++++++++++++++++ Date: Thu, 07 Feb 2002 00:14:29 -0800 From: "John Grebenkemper, KI6WX" Subject: Re: [Elecraft] K2 Frequency Stability To: Elecraft Mailing List Roy; In a normal home indoor environment with the temperature changes you describe, I would expect a K2 that is left on would drift less than 50 Hz over your time period. If you are operating outdoors with much larger temperature excursions, the drift will be greater. The results of the frequency stability measurements and changes I made to my K2 are located on my web site, http://home.pacbell.net/johngreb/KI6WX.html . In about the worst case torture test I could design, I was able to make my unmodified K2 drift about 150 Hz in one hour. As Wayne pointed out to me, my test method is not reflective of the operating methods of most K2 owners, and the real values that they see will be considerably better. The note on my web site shows a method that was used to reduce the drift to about 60 Hz using the same torture test. These changes are not recommended for most K2 owners because of their complexity. -John KI6WX ----- Original Message ----- From: To: Sent: Wednesday, February 06, 2002 1:13 PM Subject: [Elecraft] K2 Frequency Stability > John Grebenkemper, KI6WX, has recognized the need for better frequency > stability and has made modifications to reduce frequency drift. He has asked > others about their K2 drift rates. I have not heard about his findings. -- snip -- > Roy Morris W4WFB +++++++++++++++++++ Date: Thu, 07 Feb 2002 20:03:51 -0800 From: "John Grebenkemper, KI6WX" Subject: Re: [Elecraft] K2 Frequency Stability To: Elecraft Mailing List Dave; The data I have from 6 K2's on frequency stability shows that the K2 VFO drifts with temperature by between -0.4 and -1.1 ppm/C with an average drift of -0.8 ppm/C. Ten meters is clearly the worst band since this involves the VFO tuning to 25 MHz. At this frequency, the average drift would be -20 Hz/C. If your VFO drifts significantly worse than this, try changing X1 and X2 on the RF board. To get the best absolute frequency accuracy, the 4 MHz oscillator on the control board must be precisely adjusted. There have been several ways described to this on the Elecraft site and in the mail archives. Because of the interactions between the 4 MHz oscillator and CAL PLL and CAL FIL, a error in the 4 MHz oscillator can translate into different errors on different bands. No matter what you do, I doubt whether you will be able to get the absolute accuracy much better than 50 Hz across all bands. -John KI6WX ++++++++++++++++++++ Reply-To: From: "Ron D' Eau Claire" To: Subject: RE: [Elecraft] K2 Frequency Stability Date: Fri, 8 Feb 2002 15:24:37 -0800 > I note that measurements are being conducted on 10 MHz against WWV, the > worst band on my K2 is 28 MHz. The drift is of the order of 100 or more Hz > 73's Dave, G4AON First, I stand corrected. As a couple of sharp readers including Lou, W7DZN, pointed out, the amount of drift in the PLL output is increased when it is running at a frequency above the reference oscillator frequency. In the K2, that means that worst case drift will be observed on 10 meters as Dave noted above. I ran a measurement on my K2 at 29 MHz today. From a "cold start" (the K2 had been powered Off for 12 hours) the frequency drifted downward at total of 40 over the first 90 minutes. After that it has slowly wandered over a range of 10 to 12 Hz of the median frequency (total drift of 20 to 25 Hz). It seems to cycle over a period of a couple of hours although the movement is somewhat erratic. Over a period of 5 hours the most rapid movement was 20 Hz over a period of 70 minutes. Other than that one 20 Hz shift, the rate of change is closer to 10 Hz per hour. With an outside air temperature of 40 degrees F, the forced air heating is cycling on and off, shifting the room temperature probably 5 degrees each time the furnace runs. That is probably the greatest cause of the on-going frequency changes after the initial warm-up. Ron AC7AC K2 # 1289 +++++++++++++++++ Reply-To: "Bob Lewis (AA4PB)" From: "Bob Lewis (AA4PB)" To: Subject: Re: [Elecraft] K2 frequency error/drift figures Date: Sat, 9 Feb 2002 09:57:05 -0500 Guys, the K2 does not use TCXOs and I would not expect it to have the same stability as a rig that does. I think Elecraft has done a great job of providing a good amount of stability (and other features) while keeping the power consumption in line with battery operation and the design something that can be built by most individuals. +++++++++++++++++ Date: Sat, 09 Feb 2002 12:13:28 -0600 To: elecraft at mailman.qth.net, joe levy From: Tom Hammond =?iso-8859-1?Q?N=D8SS?= Subject: [Elecraft] K2 Case-distortion-induced Freq INstability Folks: This note recently received from Joe Levy (unfortunately, callsign unknown). Joe had come up with a cheap and simple fix to a complaint that several users have written about... that being that the K2 shifts frequency just a bit when the bottom of the case is squeezed, or when the bail (flip-down stand) is lowered. Here's Joe's posting on this subject. Hello Tom As an owner of a recently built K2 (2315) and a devoted reader of the reflector for the last 2(?) years I came across a lot of "complaints", some that were eventually fixed (either by Elecraft or by some users ingenuity). One of the minor complaints was that the bottom plate of the k2 is too compliant and when the folding legs are opened the frequency moved, and the same frequency shift happened when the bottom of the case was squeezed during handling mor moving of the case. The solution to this problem is extremely simple as you'll see in the attached photo. (I'm omitting Joe's photo because ATTACHments can't be posted to the reflector. - Tom) Just added 2-3 plastic spacers of the exact height (secured with hot glue) so the bottom panel does not change it's spacing from the printed board when the case is handled. The length of the spacers has been 'adjusted' a bit, so they just touch the insied of the bottom cover when it is installed, thus preventing it from being distorted during handling. These spacers are situated in the so the movement near L33 and L31 is minimized. Their location is also determined by available 'empty' PC board space, so they don't interfere with with already-installed components. Joe Levy, in Israel _______________ Tom's few comments: I do have a picture of the bottom of Joe's K2, showing where he situated the added plastic spacers. If you cannot visualize what he's describing, drop me a note and I'll e-mail the picture to you, but I doubt it should be required... the explanation seems pretty self-evident. Joe's solution should go a LONG way to curing any pesky problems you might have with your K2 shifting frequency as a result of minor movement in the bottom (front) cover. Thanks for such an easy fix, Joe. 73, Tom Hammond N0SS ++++++++++++++++++ Date: Sat, 09 Feb 2002 14:59:18 -0800 From: lhlousek Subject: Re: [Elecraft] K2 Case-distortion-induced Freq INstability To: elecraft at mailman.qth.net, joe levy Early on I installed another 1/4" threaded spacer in the unused mounting hole in the RF board and a matching hole that I drilled in the bottom cover. This adds one more spacer between the bottom cover and the RF board and reduces the botom cover frequency tweak a great deal. Lou W7DZN ++++++++++++++++++ Date: Fri, 22 Feb 2002 20:39:09 -0800 From: "John Grebenkemper, KI6WX" Subject: Re: [Elecraft] K2 Frequency Accuracy To: Elecraft Mailing List From: "Sherman Banks (W4ATL)" > If I measure my frequency error on 7000 KHz and make the proper calculation > of my frequency error, then Cal PLL and Cal Fil (moving the BFO one count > and then back) I get pretty good accuracy on 80 and 40 M throughput the > band. But the other bands seem to be off. 20 M is 50 Hz low. 15M is about > 80 Hz low. 10M error increases to about 120 Hz. 80M is right on. Ver 2 > firmware. Is this normal? Should I try to measure the error on the highest > band for more accuracy? > The absolute frequency error in the K2 will vary significantly by band if the 4 MHz oscillator is not precisely adjusted to 4 MHz. This error occurs because the BFO and the VFO frequencies are determined by the 4 MHz oscillator during CAL FIL and CAL PLL. A second uncertainty is created because the BFO step size is in the range of 20 to 30 Hz. If you measure the frequency error in both LSB and USB (or CW and CWR), you can compute the error for both the VFO and BFO which is very useful to determine why the error is occuring. Assuming that your data is for LSB on 80 and 40 meters and USB on 20, 15, and 10 meters, it fits quite well for the 4 MHz oscillator to be 17 Hz high (this offset predicts -59 Hz on 20 meters, -89 Hz on 15 meters, and -119 Hz on 10 meters). For those who are interested in the expected frequency error, I've computed the following table. There is roughly a 20 Hz uncertainty due to the finite step size of the BFO and VFO. The frequency error table assumes that the 4 MHz oscillator is 10 Hz high in frequency. Freq Mode Error (Hz) 3500 LSB +09 7000 LSB +18 14000 USB -35 21000 USB -53 28000 USB -70 It is very difficult to set the 4 MHz oscillator to a better accuracy than ~10 Hz, plus this oscillator does drift at a measured rate in my K2 of -1 Hz per degree F increase in temperature. The K2 internal temperature increases by about 10F over a 1 hour warmup period in receive mode which will shift this oscillator by -10 Hz as the K2 warms up. The most accurate way to set this oscillator is to drive the K2's internal counter with a known accurate frequency source in the 20 to 30 MHz range. You may also try to set the fifth harmonic (20 MHz) of this oscillator so that it is at exactly the same frequency as the 20 MHz WWV signal. You have to remove the K2's antenna to hear the fifth harmonic. Finally, I think it is unreasonable to expect a K2 to have an absolute frequency error better than the above table. My K2 can regularly achieve an absolute frequency accuracy of 40 Hz on 15 MHz WWV. The K2 was not designed with a temperature compensated master oscillator and a 1 Hz frequency synthesizer. While there are radios that have this level of performance, they tend to be expensive, heavy, and power hungry. Very few communication modes require an absolute frequency accuracy of a few Hz. -John KI6WX ++++++++++++++++ From: James Hammons To: elecraft at mailman.qth.net Date: Wed, 11 Sep 2002 16:52:09 -0400 Reply-To: james-e-hammons at worldnet.att.net Subject: [Elecraft] Freq dirft in crystals explanation. Years ago I worked for a company that made crystals and filters. We made the filters that went into the telemetry that was on the Lunar Lander for example. It's been a long time but here goes. The K2 uses voltage controlled crystal oscillators for the synthesizer and BFO. The varactor diodes,the inductors, the voltage source and the crystals all change with temperature to some extent. KI6WX and others measured the frequency drift of 8 K2's and found them all to drop in frequency with temperature. The frequency change he found is generally within the range of the specs I found on the internet for other radios without TCX0's, Meaning the vast majority. So it satisfies most users. The crystals used are no doubt AT cut crystals which can have a + or - change with temperature depending on the angle they are cut from the crystal with respect to the molecular structure. For example a 35degree21minute cut gives 0 change in frequency over a wide range each side of room temperature. changing the angle by 7 minutes less produces a + change of about .0015% in 10 degree centigrade. Increasing the angle produces a - change. A minute of arc is about equal to one inch at 100 yards. So control of this angle is difficult, which means more expensive. But a crystal supplier can do it. By chance your crystals may tend to correct more or less the drift of the other components in the circuit. KI6WX and I ended up with one of the worse examples, drifting about 150hz at 14mhz on warmup. =20 KI6WX improved his K2 to about 65hz with his compensation mod. Compensation could be done by selecting a matching crystal, not practical on the individual level. Or by having the CPU or another CPU measure the temperature and compute the compensation needed. =20 How much would that be worth as an add on?=20 James Hammons +++++++++++++++++++ From: James Hammons To: elecraft at mailman.qth.net Subject: [Elecraft] Rig drift Date: Sat, 28 Sep 2002 22:06:38 -0400 Reply-To: james-e-hammons at worldnet.att.net An easy accurate way to evauate dirft and frequency accuracy is to use the free program for PC's, BeaconSee. All it requires is a connection from the headphone jack to the line in of your soundcard. If you are set up for soundcard digital you are ready. This method just about eliminates errors in tuning and zero beating WWV, and it should work anywhere in the world. =20 Reading the setup instructions tell you to use lower sideband and tune 1kc high. With the K-2 it's easier to tune to the beacon using cw mode and set the center equal to your sideband frequency eg 600cps. Drift on SSB is not likely to be a problem anyway. If your frequency calibration is accurate you will see a line trace the beacons you can hear and see. Use the plug only partway in to see and hear. If the calibration is off the line will be above or below the center frequency. If there is drift the line will slope up or down as the frequency changes. These beacons are accurate. Often I can hear all of the beacons on 14,100 kc, on my 20m dipole. After the K2 warms up there will be a continuous line from all the beacons. =20 Most likely the line will not be exactly centered at your sidetone frequency. You can center it up with tuning knob and then you can try cw reverse and various filters and modes to see how they affect the frequency. You will be able to see the bandwidth of the various modes and see if the signal remains in the passband. The advantage of BeaconSee is the trace is over a long period of time. It is more than worth the effort to set up BeaconSee because it can be used to evaluate band conditions from you to the whole world. It gives good indication that your receive is up to snuff also. You can simulate normal operation by attaching your dummy load and going into tune for 10 minutes and then receive the beacon for 10 minutes. Don't use more power than you would for digital mode eq 5watts for the base K-2. CW and SSB do not use 100% power on average. You will overheat the finals if you use full power and it won't simulate CW or SSB power anyway. If you have the 160 meter option with the second receive antenna connector you won't have to change the output to the dummy load by hand, but otherwise you will have to switch to the second output or physically connect and disconnect the dummy load. Or you can punch in the frequency and pop over and see how normal operation has affected operation anytime. This is easiest if you mark the AF gain setting with a bit of masking tape. You will likely want to use Pre all the time so you can hear the weak ones. James ++++++++++++++++ From: "Don Brown" To: "Elecraft" Subject: Re: [Elecraft] Rig drift Date: Sat, 28 Sep 2002 21:46:23 -0500 Hi Hey guys if we are all patient the drift problem may be fixed. There is a= fix in the works. As soon as Elecraft is sure it works we will get the d= etails. One part of it is the removing of X1 or X2. If you can still get = enough shift in the reference oscillator with only one crystal then this = will improve the stability of the VFO. Eric told me most of the crystals = will do this. At least you have 2 of them to play with. I am going to try= this on my 2 K2's and see how it works. The temperature compensation of = the BFO info will come as soon as it is proved out. It is a very simple m= od something like the 2nd filter flattening mod. Don Brown KD5NDB +++++++++++++++++ From: "Don Brown" To: , "Elecraft" Subject: Re: [Elecraft] K2 freq stability... Date: Wed, 9 Oct 2002 08:08:42 -0500 Hi I asked the same question a couple of weeks ago when talking to Eric. The= 4 mhz oscillator is only required to be accurate while CAL PLL is runnin= g so the temperature stability is only required for the few minutes while= the calibration is running. Once all of the calibration information is s= tored into the EEPROM the 4 mhz clock is out of the picture as far as any= drift is concerned. The drift is in the reference oscillator and the BFO= oscillator. The BFO drift causes the bandpass to drift and if you have s= etup very narrow filters (less than around 150 HZ) then the bandbass may = drift away from your selected sidetone frequency causing a drop in signal= . Don Brown KD5NDB =20 ----- Original Message ----- From: John Clifford Sent: Wednesday, October 09, 2002 6:35 AM To: elecraft at mailman.qth.net Subject: [Elecraft] K2 freq stability... Could a custom TXCO be substituted for the 4 Mhz crystal, in order to reduce/eliminate the K2's temperature-resultant drift? - jgc John Clifford KD7KGX Heathkit HW-9 WARC/HFT-9/HM-9 Elecraft K2 #1678 /KSB2/KIO2/KBT2/KAT2/KNB2/KAF2/KPA100 Ten-Tec Omni VI/Opt1 email: kd7kgx at arrl.net ++++++++++++++++ From: "Don Brown" To: "g3pjt" , "Elecraft" Subject: Re: [Elecraft] K2 drift Date: Wed, 23 Oct 2002 07:52:23 -0500 Hi Yes you can remove X2 and improve the PLL drift. I have done this on seve= ral K2's and it works. Check that you can get 10 KHz shift in the referen= ce oscillator and then redo the PLL calibration. Check the shift by conne= cting the counter probe to TP3 and select FCNTR mode, then tap band+ and = band- to read the upper and lower limits of the reference oscillator. If = by chance the oscillator does not shift 10 KHz then try the other crystal= at X1 (the one you removed from X2). I have not found a crystal that did= not have enough range so you should be OK Don Brown KD5NDB ----- Original Message ----- From: g3pjt Sent: Wednesday, October 23, 2002 7:29 AM To: elecraft at mailman.qth.net Subject: [Elecraft] K2 drift Hi Having analysed ( as KI6WT note) the sources of drift in my K2 I find tha= t any drift is largely in the PLL rather than the BFO Am I right in assuming that this drift can be reduced if I only use one crystal rather than two at 12MHz provided that I can swing the 12MHz oscillator by sufficient kHz so that the 5 kHz interpolation functions correctly. I know this topic has been raised here recently so sorry for the bandwidt= h but I can't find the reference. 73 Bob G3PJT K2# 2642 +++++++++++++++ From: James Hammons To: elecraft at mailman.qth.net Subject: [Elecraft] K2 drift Mod Date: Wed, 23 Oct 2002 17:18:57 -0400 Reply-To: james-e-hammons at worldnet.att.net Drift is not a real operating problem, in the K2, even for digital modes using a computer software using SSB. Nor is the drift unusual if compared to the specs of most transceivers. My K2 measured about 1.2ppm per degree C drift for the PLL before removing the extra crystal. Thus for a 10 degree cold to warm up change at 14mhz it would drift 168hz. This would only change the tone on CW or SSB the same amount. Unless you were using a narrow filter less than what we have found is recommended, eg .1 or .2khz, would you loose the signal. Most likely you would just touch up the tuning before this happened. On digital modes AFC would follow the receive signal or you could touch it up and you might not even be aware of it. The BFO drift is about the same but at 5Mhz it would be a fixed amount, It will reduce or add to the PLL drift depending on whether the VFO frequency is above or below the signal frequency. So the BFO is much less of an influence on drift and will reduce overall drift, at some frequencies. After removing one of the crystals in K2 #2748, the PLL drift was reduced to less than .5 ppm per degree C. This was an improvement to less than half the drift before. One of my crystals would pull 9.95 khz and the other 10.54 khz, which is the one I used. I did the same to the BFO to see what would happen and it didn't have near the range needed. =20 I am among those who like to tinker and improve, Even though this isn't strictly needed, I also have much experience removing parts and doing repair. If you aren't this type I would wait until official word comes out as what can be done. I am going to wait for more information before I do any further mods. With the addition of the KI6WX mod for drift PLL drift could be reduced to near zero. =20 Most people will not have the equipment or interest to measure the drift. If you are among those, I would wait before doing anything. ++++++++++++++++ Date: Thu, 24 Oct 2002 12:43:34 -0700 From: Vic Rosenthal Organization: Transparent Software To: Don Brown Cc: Helmut Usbeck , Elecraft Subject: Re: [Elecraft] K2 drift Mod Don Brown wrote: > > I don't remember what the drift spec is but removing X2 will improve it by a factor of 2. It also will improve the frequency tracking of the PLL because the swing is smaller for the same number of steps. Just make sure it will swing at least 10 KHz I've been told on good authority that Elecraft is working on an 'official' mod to reduce drift in the BFO and the VCO. It might be worthwhile to wait and see this before making any mods for this purpose. Vic K2VCO ++++++++++++++++ Date: Sun, 10 Nov 2002 13:55:45 -0800 From: lhlousek Subject: [Elecraft] To PLL mod or not To: elecraft list , rod at n0rc.us, Ron AC7AC As I understand it, characterizing the drift of a K2 isn’t quite as simple as seeing what happens at a single frequency like 10.000 MHz. The operation frequency of the K2 is determined by a combination of Local Oscillator (VCO/PLL) and BFO frequencies so if one or the other drifts, the operating frequency will drift. If both drift, the total drift can be better or worse than the individual drifts depending on whether they add or subtract. Whether they add or subtract changes when you go from 17 meters and below, where the operating frequency is equal to the LO minus the BFO frequency, to 15 m and above, where the operating frequency is equal to the LO plus the BFO. The magnitude of the contribution to drift due to the BFO is more or less constant from band to band since it is always operated at the nearly the same frequency (around 4913 KHz or ~ 5MHz). However, because of the wide tuning rang of the LO (~ 6.7 to 24 MHz), its contribution depends on the band. This wide range is achieved by using a PLL to multiply a VXCO (running near 12 MHz) so the drift depends on the multiplier. For example, on 40 meters the LO is 7 + 5 = 12 MHz. The PLL multiplier is ~1 so LO drift will be equal to the VXCO drift. At 28 MHz, the LO is 28 - 5 = 23 MHz. Since the VXCO operated at 12 MHz, the PLL multiplier is ~ 1.9 and the contribution to drift from the VXCO will be nearly twice as big. To make things interesting, the amount of temp drift of the LO or BFO is dependent on where you are in the tuning range of their respective VXCOs. Changing filters or sidebands changes the BFO frequency and so can affect its contribution to drift. The LO VXCO is used to tune over successive 5 KHz segments of a band so the contribution to drift due to the LO will depend on where you are in each 5 KHz segment. And, due to the PLL multiplier, the tuning range of the LO VXCO ranges from about 2.6 KHz on 10 meters to about 9 KHz on 180m so the contribution due to where in the 5 KHz segment you are is also band dependent! The end result is that the overall magnitude and the direction of the drift can vary a lot depending on band, side band, and over 5 KHz tuning intervals. Drift in the BFO not only affects operating frequency but also affects where the bandpass of the filter lies with respect to its setpoint. If you are using a narrow 150 Hz CW filter centered on 750 Hz and the BFO drifts 200 Hz, a CW signal zero beat to 750 Hz will no longer pass through the filter. The drift of my K2 without the KPA100 was a non-issue. However, with the KPA running at 100W , I was getting no significant drift on 40 meters but upwards of 400Hz drift during a QSO on 15 meters after a long over. I was one of the guinea pigs for the new L33 toroid and (wound two of them myself) and the single VCO Xtal and these changes together have reduced the drift to acceptable levels (100 to 200 Hz during a QSO) an all bands. Lou W7dzn +++++++++++++++++