Purpose: Eliminate the well known parasitic and adjacent frequency artifacts observed on the 80 meter band during alignment procedures that have plagued the Collins KWS-1 since its first production in 1955.
Summary Findings: A 250 kHz coax line layout problem is causing RFI feedback (3-4 MHz) that disrupts the 250 kHz signal driving the first mixer. The disturbance is caused by electrical coupling of RFI due to the placement of the coax lines that travel between the 250 kHz crystal oscillator, Carrier Level potentiometer, Emission switch, and the output of the 250 kHz mechanical filter. The magnitude and effects of RFI feedback is dependent on SWR, type of antenna system used, and the proximity of the KWS-1.
Failure analysis observation: The offending electrical coupling acts like an antenna and receiver that mixes RFI signals with the 250 kHz carrier signal, especially at lower carrier settings used on CW, AM, and during tune-up. SSB is unaffected because the carrier signal path is electrically opened and one side of the mechanical filter’s output signal is grounded by the Emission switch through a 1000 pf capacitor. RFI feedback is caused by coupling capacitance, mutual inductance, and weak shielding of the 250 kHz coax lines mounted next to other wires, cables, and switches. The Emission switch does not make a good RF band switch for the 250 kHz carrier signal, but works ok for switching 120 VAC, Interlock, PTT, CW Key, ALC, and grounds for SBB, AM, and Calibration circuits. The 250 kHz coax lines travel along parallel paths mounted on top of other wires and cables which are bound together with lacing string. The slightest disturbance caused by electrical coupling will disrupt the carrier signal, cause parasitic and adjacent frequency artifacts, and result in the KWS-1 to see random spikes of excessive plate current that may blow the High Voltage fuse during tune-up.
Resolution & Modification Steps (Ref: Figure 1)
Carrier Level position on SSB: The Carrier Level position must be turned all the way CCW to zero when operating SSB. This will ensure that a signal ground is placed on one side of the 250 kHz Mechanical filter’s output through a series 1000 pf capacitor. Doing this will prevent the 250 kHz crystal oscillator from inserting a carrier signal into the First Mixer during SSB operation. The Carrier Level position and its use on AM, CW, and CAL remains the same.
If you want to preserve original functionality of the Emission switch to provide a redundant signal ground and unnecessarily opening the 250 kHz signal path to the Mechanical Filter, please consider adding a SPDT relay mounted close to the Carrier Level potentiometer terminals. You can energize the SPDT relay by tapping off the 6 VAC panel lamp power and switching it through the original Emission switch wafer terminals located on the front panel. However, maintaining this functionality using a SPDT relay is not necessary and not tested.
Once the modification steps are complete, it is recommended that you follow the excellent Collins KWS-1 Care & Feeding document along with the Collins KWS-1 manual and perform the alignment procedure to bring the drive levels back to specification. Please note; ignore any reference to detuning L207 to eliminate parasitic signals across the 80 meter band as this modification fixes that problem and makes it unnecessary. Therefore, adjust L207 for peak signal as designed.
Benefits of this modification:
All peer reviews are welcome!
|Collins KWS-1 Service Bulletin #7 by WA6JKN|
|Collins KWS-1 Service Bulletin #7 - Detailed Analysis and Technical Interchange Review.|
The primary question that started the journey behind my investigation and failure analysis review into the parasitic and adjacent frequency artifacts showing up when aligning the 80 meter RF stages in the KWS-1 are as follows:
This review will examine why adjusting L207 as the only corrective action recommended by the care and feeding document of the Collins KWS-1 that appears to be in direct conflict with the official Collins KWS-1 design requirement with regard to the official alignment procedures IAW in the KWS-1 manual from it's first publication through revision 7.
Specifically, my detailed Analysis and Technical Interchange Review behind the discovery of the KWS-1 design defect and the effects of adjusting L207 that only seemed to mask the root cause of the problem that lead my investigation into the discovery of the real root cause of the problem almost 70 years later.
All responsible peer-reviews and comments are welcomed.
1. What are the parasitic artifacts that you are trying to eliminate? Are they the results of oscillations or internal mixing products? Do you have spectrum analyzer photos?
First, let me introduce a little history about the Collins KWS-1 transmitter used in my research. My KWS-1 was restored by Howard Mills and sold to Wayne Spring. I bought the rig from Wayne Spring many years ago. During my KWS-1 "As built" / "As design" configuration audit to verify what service bulletins were installed, I was able to confirm that they were all installed.
During this audit, I examined all grounds, cleaned, and tighten as necessary to include cabinet, chassis, terminals, lugs, and tube sockets and performed an overall visual inspection. All tube voltages and resistance measurements were made and corrected as necessary. I also decided to replace most of the silver mica capacitors and a few out of tolerance resistors. All these steps had to be completed before investigating why it was necessary to detune L207.
To get my KWS-1 ready for "The L207 Investigation", it took about 5 years to replace the old parts in an effort to eliminate any possibility of a component and/or ground failure as a contributing factor. Special tools were purchased and/or made to work in tight spaces.
After studying the design of the KWS-1, I was able to conclude that Collins engineers attempted to solve incidences of parasitic and adjacent frequency artifacts in its design approach and through multiple service bulletins.
At a top level, these documents were used in some of my early discoveries and data points that I captured in my Fault Tree Analysis. Just to give you a flavor of some of my findings, please see below.
· Was it caused by plate, cathode, and/or grid neutralization issues regarding the 6CL6 driver tubes? No
· Was it caused by RFI within the B+, bias, and/or keying circuits? No
· Was it caused by bad grounds? No.
· Was it caused by RFI (RF Feedback) on 80 meters? Yes
· Was it caused by faulty components? No
· Can you see the adjacent frequency artifacts during TX with a Spectrum Scope? Yes
2. Is their 3-4 mHz RF energy is coupling back to the 250kHz section of the circuitry, causing the artifacts? If so, how does that work?
Yes, the 3-4 mHz RFI is disturbing the 250 kHz signal "path" driving the 1st mixer (V201) as a direct result of RFI producing RF Feedback in the circuit.
What does "The RF Feedback Footprint" look like? It looks like adjacent frequency artifacts surrounding the VFO's true "transmit frequency".
How does this work? Through capacitive and/or inductive coupling!
It did not take much to disturb the poorly laid out 250 kHz signal "path" as it carries a very small signal by design as it travels to a high impedance input of the 1st mixer (V201).
In addition, there is "design evidence" where Collins engineers were aware of this defect and incorporated traditional RF bypass components to ground, series chokes, and use of coax cables, faraday compartments, voltage dividers, and 47 ohm termination tricks to overcome the effects of RFI (RF Feedback), especially in the 3-4 mHz range which was most susceptible.
In order to fully understand the failure mechanism (root cause), it cannot be simply understood by measurement via an oscilloscope or spectrum scope because the KWS-1 TX signal also interferes with the measuring equipment and the measuring equipment also interferes with the KWS-1's circuits during transmit. To overcome this, I used an external RF energy source (Icom IC-7610) where I was able to loosely couple an RF field to pinpoint the area susceptible to disturbance inside the KWS-1 while terminated to a 50 ohm dummy load and with L207 adjusted for peak value IAW the Collins KWS-1 manual.
Most importantly, the fact that detuning L207 during normal TX was the "dead giveaway" indicator that RFI was triggering the RF Feedback disturbance with many adjacent frequency artifacts, a.k.a "The RF Feedback Footprint" showing up along with the VFO's "true" frequency as seen on the spectrum scope.
It is my professional judgment that decreasing the gain (lowering the Q) of the last RF stage (80 meters) was the wrong approach because it treated the symptom, not the root cause, and subjects the KWS-1's design to many disadvantages.
Was the RF disturbance mixing or sharing because of the long 250 kHz coax cable run as it traveled along its poorly laid out path? Most likely both!
My solution was to shorten the overall length of the 250 khz coax cable, minimize the number and length of coax center conductor pigtails, and avoid long parallel paths and coupling opportunities with a goal of using point-to-point connections as a better design approach to eliminate any ground loop concerns. My SB#7 explains this process.
The route and length of the 250 kHz cable was the major contributing factor of electrical coupling. The coax cable (antenna) traveled somewhere between 2 and 2.5 feet from the 250 kHz local oscillator, then over to the carrier level potentiometer and ground lug, then to the emission switch and ground lug, than back to the carrier level potentiometer and ground lug, then finally to the 1st mixer (V201) and chassis ground. There were many long "center conductor pigtails" at each termination point exposing the 250 kHz coax cable to RFI along its path.
As explained in SB#7, it was unnecessary to make these long runs to the emission switch and back to the carrier level potentiometer to ensure the "customer/user/operator" did not forget to turn the carrier level potentiometer fully CCW during SSB operation, unless you wanted to operate SSB with a carrier to produce AM. As pointed out, the long coax configuration runs required breaking up the 250 kHz signal path many times by creating multiple center conductor pigtails and grounds at each termination point and bundling the coax cables on top of other wires along its path.
It is my professional judgment that this was not the best engineering practice to use. However, if you can get away with it, why not? As discovered in this case, it was not a good idea. It became clear that improving the 250 kHz coax cable layout was the answer!
The obvious solution was to shorten the coax cable using point-to-point connections to avoid electrical coupling opportunities as much as possible. By doing so yielded significant improvements in the purity of the VFO's signal during TX with absolutely No detectable signs of adjacent frequency artifacts across the 3-4 mHz band. Most importantly, it allowed you to adjust L207 and C221 to its peak value (max Q) IAW the Collins KWS-1 manual.
3. Is the problem you are seeing solved by simply de-tuning L207 which lowers the gain, as described in "The Care and Feeding of the KWS-1"? Have you tried all the usual grounding improvements that are often the cause of instabilities?
Your question is a Bellringer!!! The operative question is; why on earth would you want and/or need to detune L207 in a circuit designed for high gain and stability? I asked myself; would detuning L207 create any performance issues? What actually happens when adjusting L207 or C221 for that matter? Does the adjustment alter the VFO's signal in anyway? As you can imagine, I had many questions that needed to be investigated first before accepting a departure from the Collins KWS-1 design and alignment procedure.
During my first observation, I did not like the fact that while tuning L207 or C221 would trigger the adjacent frequency artifacts to appear when making the peak adjustments IAW the Collins KWS-1 manual. Realizing this adjustment (L207 & C221) was the last (2nd RF amplifier(V203)) stage, I was not fooled because the 1st RF amplifier, 1st mixer adjustments as well as the 1st mixer plate trap null adjustment C229 all behaved in the same manner. These circuits could easily trigger the unwanted adjacent frequency artifacts too while making their adjustments IAW the Collins KWS-1 manual.
I also noticed that while turning L207 through max Q, it would push and/or pull the VFO off its "true" frequency at lower carrier drive levels. The method of detuning L207 was not a good idea because all I was doing was masking the real problem by treating the symptom, not the root cause and causing other less than desirable performance issues. I immediately thought; "How did this get overlooked for so many years?" It became imperative that I continued to investigate this issue until I discovered the root cause.
Most importantly; The Care and Feeding of the KWS-1 document recognized this too by saying; "the Collins KWS-1 "manual" procedure will result in a oscillating or marginally stable" condition. Leveraging my work experience with designs exposed to environmental qualification test levels at 4.5 db above maximum predicted environment and being involved with thousands of failure reviews over a span of 40 years on Complex Major Weapon System platforms in the operation areas of air, land, sea, and space, I could only conclude that this was an unwise engineering decision to lower the gain of a circuit designed to be high Q to ensure maximum gain was achieved in order to maintain full performance and "stability" across the 3-4 mHz band. I do not think lowering gain would be advisable even at normal qualification levels unless it was design that way in a commercial product with no military standards being specified.
I was able to confirm that detuning L207 "masks" the root cause and tricks the KWS-1 into working, but not as designed by Collins and certainly without achieving preferred design performance levels! Obviously, the problem was somewhere before L207. As I said, adjacent frequency artifacts and having to detune L207 was the dead giveaway that RFI was causing the RF Feedback somewhere in the circuit.
I also performed the Care and Feeding of the KWS-1 document suggestion to check all grounds, cleaned all mechanical termination points, and tighten as necessary to include the cabinet, chassis, terminals, lugs, and tube sockets.
The proof is in the pudding! My Collins KWS-1's technical performance significantly improved with SB#7.
As we now know, the root cause was due to a poorly laid out 250 kHz coax cable which became solely responsible for causing RFI to produce RF Feedback which disturb the purity of the VFO's signal coming out of the 1st mixer (V201).
My KWS-1 is now more robust on 80 meters and has more drive gain and stability as required by the Collins design. With the 1st mixer and 1st & 2nd RF amplifier stages tuned to max Q, it now delivers more drive to the 2nd mixer and up the chain until it finally reaches 10 meters.
Were there other benefits to fixing the 250 kHz coax cable lay out problem? Yes
Most have heard about that KWS-1 owners complaining about the drive level being too low on LSB. Did the LSB crystal or maybe the 250 kHz mechanical filter shifting with age cause it? These two issues have plagued the KWS-1 for years and were mostly caused by the lack of gain, and were improved by SB #7 too.
It is all about maintaining proper gain and Q in the 1st mixer and 1st & 2nd RF amplifier stages.
How about the Carrier Level potentiometer having to be readjusted during subsequent transmissions when operating AM or CW? That too was significantly improved by SB#7 by requiring little to no adjustment after warm up. Collins' Service Bulletin to replace the Carrier Level potentiometer may have been unnecessary to perform. I was not able to confirm this as I have the newer pot installed.
There were many other noticeable improvements gained by SB#7, as seen by a spectrum scope showing the KWS-1 being clean and clear of parasitic and adjacent frequency artifacts, full rated RF output power, improved behavior during tune up and operating the KWS-1 on all modes, especially AM and CW.
Please Note that L206 and L204 are drawn incorrectly (reversed) on Figure 8 in the Care and Feeding of the KWS-1 document.
4. How many other KWS-1 units have displayed the artifacts you describe? Are there other KWS-1 owners who have solved the problem you describe by installing your modification?
This is a very good question! I found the answer in the excellent Care and Feeding of the KWS-1 document. According to a caution note, it says; "Caution: There is one Strong caution that must be heeded. On almost 50 % of the KWS-1s that l've aligned, the manual procedure will result in a oscillating or marginally stable 2nd IF stage (V203). This will result in, at the best, spurious emissions and at worse, full output with no modulation on SSB. This will appear on a frequency very close to the units true "transmit frequency". Solution When aligning L207 and the corresponding 3. 1 MHz trimmer, monitor the IF waveform at the 6CL6 driver plate with a scope or spectrum analyzer. If any sign of instability (a beat frequency on a scope or spurs on the spectrum analyzer) is present, de-tune L207 until the instability disappears. After tuning (or de-tuning), the Stability should be checked CW and under modulation across the full frequency range of the PTO, not just the alignment points. As a final check stability, remove the PTO oscillator tube and monitoring both the scope and final grid current, there should be no output across the full PTO/slug rack tuning range. Ample drive will still exist after any required detuning."
This Caution Note was a good attempt to lessen the problem, but it did not go far enough to identify the root cause of the problem. Remember, the Collins KWS-1 manual did not instruct you to detune L207 during alignment! In addition, what was seen on the spectrum scope was The RF Feedback Footprint showing up at the output of the 6CL6 plates!
Since the Care and Feeding of the KWS-1 document suggested that the problem was within the 80 band, this narrowed down my search. I also confirmed the problem as a practical matter. Since the Collins KWS-1 alignment procedure did not say that L207 and/or C221 detuning was required or necessary for either the 1st RF amplifier stage (V202) or the 2nd RF amplifier stage (V203), my focus immediately went looking at the 250 kHz local oscillator, 250 kHz mechanical filter, and first mixer (V201) circuits more closely.
5. How is the type of antenna and its proximity to the KWS-1 important as far as the effect on the problem you describe? How have you demonstrated that radiated energy is getting back into the 250kHz balanced modulator and IF circuitry?
I found no disturbance with the 250 kHz balanced modulator. Given it is very low impedance (Cathode follower); I would not expect it to have any concerns with RFI (RF Feedback) unless it was being picked up somewhere in the audio circuits.
Antenna used, RFI and RF Feedback (cause & effects) are as old as Radio itself.
--- Common Mode
--- Electrical Coupling
--- Near field (location, wavelengths, & distance)
--- RF Power Levels Used
--- Coax, connectors, and/or bad connections
In closing, it is my professional judgment that Service Bulletin #7 is without a doubt one of the most important Service Bulletins ever to come out for the Collins KWS-1 since 1955, even though it took almost 70 years to materialize.
I hope this helps with explaining the approach I used, the discoveries I made along the way, and the final solution behind Service Bulletin#7. I know Art Collins must be smiling now, even though it took almost 70 years to solve this problem and making the KWS-1 as robust as its big brother the Mighty KW-1.
All responsible peer-reviews and comments are welcomed.