THE HALLICRAFTERS SX-117 AND PERFORMANCE EXPECTATIONS
The SX-117 is probably one of the most misunderstood radios that came off the drawing boards at Hallicrafters. There are as many opinions about its performance as there are people that use them. Great audio, second rate audio, quiet, noisy --- pops and crackles, Notch filter works great, Notch doesn’t appear to work at all, Bandwidth is never right, best radio I have ever owned, one of the worst.
The performance of a radio is not a democratic exercise. There are reasons for every like or dislike. I hope to provide the information needed to set realistic expectations and help to understand the eccentricities of the SX-117. Hope you find it interesting.
The SX-117 includes six distinct engineering levels, “A” thru “F”. The schematics for first five can be found on the www.K9axn.com web site under schematics. I cannot find any information that the sixth schematic ever existed. This note and any that follow regarding the SX-117 will be added to the SX-117 service note folder on the site for reference.
I believe the level of your radio can be determined by the last digit of the first group of numbers in the serial number. The upgrades were not mandatory but installed for individuals to solve specific problems.
Of the six levels, only the “B” to “C” change was profound. It corrected some problems but changed the personality of the radio from what most are accustomed, to something quite different.
The following information will focus on the most likely cause of differing opinions, and choices that you can select to make the radio fit your preferences.
The 50Kc I.F. system, including the Notch filter and the AVC function are the primary cause of differing opinions. Download the “B” and “C” level schematics for reference. They will be key to understanding the major evolution of the SX-117.
There are several Hallicrafters radios that used the 50Kc 3rd I.F. system. The Hallicrafters design team knew from the beginning that the intricacies of the design were great, but so were the problems that accompanied it. They struggled with nagging technical problems until the last one left the plant. The following should help to demystify the intricacies of the radio, the changes that were attempted, and degree of success each change provided.
In order to understand the SX-117 character, start with a review of the SX-100 MARK1. It used the 50.5Kc I.F. system and behaved like most radios of the era that did not have a product detector. To use SSB or CW, you had to reduce the RF gain and use the AF gain near full and vary the RF gain just to the point where it was intelligible but not distorted, and the S-Meter was useless for SSB or CW reception.
The SX-100 Mark2 followed, and behaved like a significantly different radio. For SSB and CW you could run with the R.F. gain at full and vary the audio gain for comfort. It behaved as though it had a product detector and the S- Meter behaved normally. What changed? Simply better control of the circuitry mimicked the performance of a product detector. The BFO injection was increased --- the AGC delay was extended --- the 4 volt positive bias on the cathode of the AVC detector was reduced to ground enabling AVC action at a much lower signal level --- and the 50.5Kc system was changed to 50.75Kc. The AVC engaged earlier and held longer, stabilizing the signal level so that it would not exceed the BFO injection level. Increasing the BFO injection level put the finishing touches on the fix. Together, they mimic a product detector. There were a few other changes but they are not relevant to this note.
Have you ever wondered why the 50.5Kc was changed to 50.75 in the SX-100 MARK2? There were two reasons. The MARK1 had a problem with the BFO feeding backward to the AVC diode. When the BFO was enabled, it generated AVC voltage and the S-Meter indicated signal when there was none. The path was from the BFO to the detector diode V7, to T6, to T5, then to the AVC detector diode in V7. The BFO frequency, 50Kc, was within the band pass of T6 and T7. It passed backward through the transformers. Yes, T5 and T6 will pass a signal in both directions equally. Adjusting the transformer band pass from 50.50Kc to 50.75, 250 cycles further away from the BFO reduced the effect but was not a complete fix. That was yet to come in the SX-117.
Now to the SX-117:
The SX-117 uses the same 50.75 I.F. system. They added a product detector and amplified AVC, both combined, isolate the BFO from the AVC circuits.
The “A” and “B” level radios behave close to what’s considered normal. The S-meter sees what you can hear and the Notch filter effect can be seen on the S-Meter and behaves as it should.
The “C” thru “F” radios behave unusually. The S-meter will incorrectly indicate a much wider band pass and presence of a signal. Also, the Notch filter effect does not appear on the S-Meter even though it is working seemingly properly. I had to say seemingly correct because it is working properly but other circuits prevent it from doing what most are accustomed to. This will be explained later in detail.
Note: In the “A” and “B” level radios the amplified AVC uses voltage divider R64 1Meg and R63 220K to create a +24 volt bias on the cathode of V10B. This is depicted on the voltage charts and is correct for the A and B level radios. Your first thought is how can this radio develop any AVC? Remember the SX-100 simply fed the I.F. signal directly to the AVC rectifier whereas the SX-117 amplifies the signal before sending it to the AVC diode in V10. The 24 volt threshold was required. Note, the plate resistor in the AVC amplifier is 22K so a meager 1ma will generate 22 volts and the bias to overcome is 24 volts. Yep it works. Even though the A and B radios have some problems, they behave in a way that most understand and are accustomed
Note: The voltage charts for V10BP2 were not updated in the later schematics. They should have been changed from 24volts to 4.2volts for the C, D, and E radios.
THE “B” TO “C” upgrade
First: They reduced the 5Kc bandwidth to 4.2Kc, almost unacceptable for AM ops. This was done by changing R85 and R87 from 470 ohms to 390 ohms increasing the Q, and removing C65 which moved the band center toward the carrier shrinking the band pass. You will see C65 on the “B” level schematic. It won’t be there beginning with the “C” level. I believe the reason for this change is the Notch filter did not have enough swing to remove it from the band pass causing distortion when the 5Kc width was selected. I would add a switch to the Notch dial to totally disable the Notch filter and restore the band pass to 5Kc.
Note: The 4700pf and .01uf capacitors on the band switch are correctly positioned on the schematics. They are inserted in series with the 390pf capacitors in both transformers to increase the center frequency (Resonant point) of the band pass. The 4700pf cap in series will be used for 5Kc, the widest band width and highest shift of the resonant frequency of the band pass to accomodate the widened pass band. The .01uf cap is used for 2.5Kc, the midrange band pass and band center. And a direct short around the capacitors for the .5Kc bandwidth and anchor resonant frequency.
Second: They moved the AVC sense from the output of the 50Kc i.f. amp to the input of the Notch filter. This change resulted in some strange behavior. The Notch filter seems to have little if any effect on the S-Meter and the S-Meter incorrectly indicates that the bandwidth is far wider than it truly is. This a result of placing the AVC detect before the Notch filter and final I.F. stage.
Have to admit, I have no idea why they moved the AVC sense. The BFO feedback problem in the SX-100 was solved with the AVC amplifier and 50.50kc to 50.75Kc change, isolating the BFO.
Anyone have a comment here? My vote is it’s a mistake.
I have become accustomed to the following Notch character:
I hear a weak signal and a stronger one appears say 800 cycles away. This generates AVC that de-senses the weak signal and is also annoying. Now the weak signal is below the threshold because of the increased AVC. I enable the Notch and target the offending signal. As the strong signal is diminished the AVC decreases and the weak signal reappears.
That is the way the “A” and “B” level radios work --- properly
The “C” thru “E” radios are different. The Notch being after the AVC sense does not allow the AVC to diminish as the offending signal is notched. The AVC level remains the same even though you notched the strong signal. The weak signal does not reappear.
Third: This change required the bias on the AVC detector to be reduced because the new AVC amplifier sense point is earlier and much lower. You will find a 6.8M resistor, R63 on the cathode bias of V10b. The “B” level used a 1M resistor. The new bias level will be 4.2 volts.
Fourth: They also changed C107, the AVC timer from .47uf to .22uf. I would go back and retry the .47uf cap.
The “C”, “D”, and “E” level radios:
Consequences follow. With the AVC tap earlier in the I.F. chain two distinct things change. Both very unsettling for old timers like me.
First: The S-Meter sees things that you cannot hear. You are tuning along and the S-Meter goes to S9 but you hear nothing until you tune further. That’s because the AVC is looking at a wider band pass than you are.
Second: You are in QSO and see your contacts signal at S7 and all of a sudden he is S9+10. Both of you collaborate and believe that you have just discovered one way QSB??? Nope, the AVC amp just sees a wider band pass --- it's some other signal that you can't hear.
Third: You have an interfering carrier and enable the Notch filter. You tune to quench it and it seems to work but the S-Meter says “No, it did nothing”. Wrong, it is working as designed. You believe your next project will be to fix a poorly functioning Notch filter and waste hours because it is normal operation.
When the Notch filter is ahead of the AVC detection as in the “A” and “B” level radios, notching a carrier not only reduces the audio but also blocks the signal to the AVC circuits. This reduction in AVC allows weaker signals to seemingly bloom to an audible range. The SR-400 Notch filter behaves this way. That is the behavior of the “A” and “B” level SX-117s.
When the Notch filter follows the AVC detection as in the “C”, “D”, and “E” radios, it has no effect on the AVC or S-meter, and does not reduce the AVC so the weak signal blooming effect is absent. To me that’s not proper notch operation.
There are a few other changes from the B to C level that I would recommend even if you don’t do the full B to C upgrade. It will improve sensitivity and strong signal performance.
1. Check for a 4.7uuf capacitor between V8BP9 and V8AP7. If there is one there, change the V8 circuitry to match the “C” level schematic. It’s a very simple change but improves third mixer performance significantly.
2. Check to see if T4P1 is grounded. If it is rewire T4P1 to the C level schematic. Significant improvement in AVC performance and SSB popping.
3. Check for a 470 ohm resistor between V2P3 and V2P6. If not rewire V2 pin 3 and 6 to the C level.
NOT CHANGE RELATED:
The Noise limiter: The noise limiter in the SX-117 and SX-115 are are as effective as most modern noise blankers, if it has not been injured. BTW, the HD6225 diodes can be replaced with 1N459A diodes. There is no cross reference for the HD6225 but the 1N459A tests to be right on and works well. This noise limiter is able to mitigate a neighbor’s security system that uses spread spectrum sweeping from 7.1Mc to 8.3Mc. Turn the limiter on and it’s gone --- similar to the woodpecker. Most noise blankers can't do that.
You will find what look like paper caps on the band pass filter switch, that is, if someone hasn’t changed them already. THEY ARE NOT PAPER regardless of the parts description. They never were at any production level. They are General Instrument’s 5% Polyester film caps and will be as good today as they were 60 years ago. If you find Ceramic disc caps on the band switch, remove them and replace them with Polypropylene film caps (No, they are not inductive). Do not use a class 2 or 3 ceramic disc cap. If a class 1 NPO ceramic capacitor was available in a size that would fit, that would be fine. A ceramic disc cap i.e. class 3 Z5U will cause the center frequency of the band pass to move upward as much as 20% and the pass band to widen up to 100% as the temperature changes to operating mode. See the sweep results with the use of ceramic caps in the SX-115 filter compared to film caps.
No designer uses ceramic class 3 i.e. Z5U in a tuned circuit. Use Polypropylene film.
A detailed explanation of the Hallicrafters 50Kc filter will be described in the SX-115 section.
Hope you found this interesting and if you find any part that needs clarification don’t hesitate to correct or help make it more useful.
Kindest regards Jim K9AXN