Hot-Rod Your ICOM IC-725-Series Transceiver by Jukka Vermasvuori

Hot-Rod Your ICOM IC-725-Series Transceiver Part 1  by Jukka Vermasvuori  OH2GF

Receiver Issues

The receiver in my IC-725 exhibited these drawbacks: Its audio frequency response was too narrow, dropping 6 dB from its maximum by about 2 kHz. High-end rolloff that is this profound filters out speech energy necessary for good SSB intelligibility and severely muffles AM broadcast signals, which carry audio up to at least 4.5 kHz. 4 The AM-mode audio had a “mega bass” quality. The level of in-band intermodulation products seemed to be quite high. 5 On CW, I could hear weak phantom beat notes from out-of-passband signals. The pitch change of these spurs was the opposite of that of desired signals during tuning. In effect, a strong signal would come back into the receiver passband if I tuned in the right direction. Although these spurious signals were weak, they sometimes disrupted communication as I worked weak DX stations in crowded bands. During CW reception with the optional 500-Hz filter, the beat note of weak signals close to background level was not pure. This phenomenon disappeared at signal levels above which the IC-725’s automatic gain control (AGC) took hold. This effect was not disturbing with the SSB-width intermediate-frequency (IF) filter selected.

The IC-725’s AGC response was such that the detected audio of SSB signals was compressed. Even short syllabic pauses allowed noise and splatter from adjacent channels to immediately rise to considerable levels. This made listening annoying, and caused me to tire prematurely during contest operation. Strong signals seemed to cover an overly wide bandwidth because the IC-725’s IF filtering had been chosen to allow FM transmission and reception as an option. The resultant wide-bandwidth filtering early in the receive chain gave no protection against adjacent strong SSB and CW signals. The IC-725’s synthesizer phase noise was regrettably high, despite the use of direct digital synthesis (DDS) in the radio’s frequency-generation circuitry. (In the IC-725 and some other radios, embedding a DDS within a noisy phase-locked loop throws away the phase-quietness theoretically achievable by DDS.) Most of the IC-725’s audio-output variation occurs between minimum and 9 o’clock on the radio’s AF GAIN control; turning the control beyond this point has little effect. The control is therefore unnecessarily critical to adjust.

Transmitter Issues

The IC-725’s transmitter suffered most from the lack of any kind of speech processing. Its “talk power” was low compared to its rated 100-W output. Transmitting into a dummy antenna to observe the quality of “hello test” experiments with a spectrum analyzer, I saw some relatively wideband spurious or IMD signals about 350 kHz away from my IC-725’s output carrier frequency. The radio’s automatic level control (ALC) decay time, as indicated by the behavior of the TX LED, was rather long. Circuit Analysis and Modifications Receiver Audio Stages The IC-725’s audio amplification chain ( Figure 1 ) consists of quad audio switch IC8 (a CD4066B CMOS bilateral switch), a two-stage audio amplifier consisting of Q35 and Q36 (both 2SC2458s), active low-pass filter Q37 (a 2SC2458) and audio power amplifier IC9 (a µPD1241H). The high-frequency response was designed to fall very early.

The Q37 low-pass stage does most of the rolloff, but capacitors C300, C297, C224 and C216 also play a role. To widen the IC-725’s audio passband, I doubled the low-pass filter’s cutoff frequency by halving the value of resistors R152 and R153. I did this by soldering additional 10-k resistors in parallel with the existing R152 and R153 parts. In addition, I removed C297 by cutting it off the circuit board, and replaced C300 (originally 0.047 µF) with a 0.025-µF part. After I did these modifications, the reason for the audio rolloff became obvious: A 7-kHz signal from the synthesizer board was leaking into the audio preamplifiers. (I determined the presence of this signal by connecting a sensitive oscilloscope to the collector of Q35.)

By inspection, I determined that this 7-kHz spurious signal was being summed with the desired signal in the IC-725’s product detector. Noting Q35’s low-value, unbypassed emitter resistor (R161, 22 ), I realized that I’d found the phantom beat-note generator. 7 The constant 7-kHz “local oscillator” works with nonlinearity in Q35 to form an audio mixer that produces unwanted beat notes when received-signal audio is present. To linearize Q35, I replaced R161 with a 100- part. Doing so decreases Q35’s gain, and has the indirect effect of expanding the AF GAIN control’s limited range because the control must be turned up higher to compensate. Receiver Detectors The IC-725’s SSB/CW detector, IC5 (a µPC1037HA) is a Gilbert cell device. 8 C125 (0.047 µF), between IC5’s output line and ground, bypasses higher audio frequencies to ground. I removed this part by cutting it off the PC board. (IC5’s output pin is 3, not 2 as indicated in the schematic.) Doing so attenuated 7-kHz leakage considerably. The final cure for the leakage was to solder a new ground wire from the ground side of R132 to the emitter (grounded) lead of Q32. This apparently eliminates a ground loop for the 7-kHz energy. The IC-725’s AM detector ( Figure 2 ) is really something special.

It gives you the impression of “mega bass” fidelity and is designed against all diode-detector rules given in a classical reference. 9 The standard design advice is to make the diode’s audio (ac) load impedance at least 10 times higher than its dc load. This keeps the diode from clipping the audio at high percentages of modulation. In the IC-725’s circuit (which is the same as that used in ICOM’s IC-R70 receiver), the diode’s ac load is 10 times smaller than its dc load.

The IC-725’s AM detector circuit is essentially the IC-735’s voltage doubler circuit with a 1-k resistor (R260) substituted for the ungrounded doubler diode. This resistor forms, through C119, an additional load for emitter follower Q29. The IC-735’s AM detector operates at 455 kHz; the IC-725’s, at 9 MHz. The stock value of C121 is therefore unnecessarily high and could be reduced for better audio response. To avoid complete redesign of this detector, as a “first aid” measure I soldered a 3.3-k resistor across C118 (from the junction of R126-C293 to ground). Figure 3 shows the IC-725’s AM-wide frequency response, measured across R125, before and after my detector modifications.

Receiver In-Band Linearity All of the IC-725’s detectors AM, AGC and SSB are driven by emitter follower Q29. The source impedance of Q29 is rather high (the impedance of a parallel-tuned circuit [the last one in the IF amplifier chain] that consists of L77 and its resonating capacitor, paralleled by the drain impedance of Q28, a 2SK241 FET), which leads to Q29’s output impedance being only moderately low. The highly nonlinear loading contributed by the AM and AGC diode detectors therefore causes significant IMD at this point, destroying the quality of the IF signal supplied to product detector IC5. To solve this, I rewired the product-detector’s IF drive circuit so IC5 gets its IF drive from the L77 circuit, as shown in Figure 4 .

I did this by cutting R261’s Q29 end free from the PC board and connecting a 5.6-pF capacitor between R261’s free end and the hot (C123) end of L77. Improved Beat-Note Purity with the Narrow CW Filter When most of a receiver’s amplification is done at one IF, and especially when that IF is as high as 9 MHz, there’s always the risk that uncontrollable feedback paths will exist. Feedback that’s insufficient for full-blown oscillation can nonetheless lead to Q-multiplier (regenerative) effects that can increase noise level, make a symmetrical IF filter seem asymmetrical, and cause IMD.

Although the IC-725’s mechanical layout is well-designed, such feedback seemed to be present in my IC-725. At maximum IF gain, the radio’s noise floor modulates weak CW signals, giving them a rough beat note. This is noticeable only with a narrow CW filter selected. Slightly detuning L77 reduces this intermodulation somewhat. Although this retuning somewhat reduces the IC-725’s overall gain, it probably reduces IMD more as a result of changed phase shift than IF-gain reduction. Therefore, the direction in which L77 is detuned is important. Experiment with turning L77’s slug clockwise and counterclockwise around resonance to get the best in-band IMD reduction. An emitter follower driven from a high-impedance tuned circuit may oscillate under some conditions. To lessen the chances of oscillation in Q29, I replaced C123 (220 pF), Q29’s base coupling capacitor, with a series-connected 0.001-µF capacitor and 470- resistor. 10 This did not totally solve the beat-note-purity problem, however, so some detuning of L77 was still necessary for best results.

Like all high-gain circuitry, the IC-725’s 9-MHz IF stages are stability-critical. Any changes in them must be done with the shortest possible wiring and suitable miniature components. AGC System AGC is a form of feedback a closed control loop that acts to keep a radio’s audio output constant over a wide range of received signal strengths. For CW and SSB operation, “fast attack, slow decay” AGC has long been the norm. We have come to expect that an SSB or CW receiver will reduce its gain rapidly (the fast attack) when a strong signal appears, and relax its gain control gradually and relatively slowly (the slow decay) after the signal weakens or disappears. Pappenfus, Bruene and Schoenike 11 suggest guidelines of 2 to 10 milliseconds for attack and up to 1 second for decay.

Achieving good fast-attack, slow-decay performance in a multiconversion superhet without popping and audible distortion requires critical circuit design because of the timing and gain-distribution issues involved. Attack timing is critical because we don’t want very short noise pulses to activate the AGC and momentarily deafen our receivers, and because highly selective IF filters behave as delay lines that slow a signal’s rise and fall times (making a sharp noise pulse broader, for instance) and delay its passage.

A strong RF signal that suddenly appears at the receiver’s input reaches the AGC detector after one or more frequency conversions only after the delays and rise-time modifications contributed by the radio’s IF filters. Generally, the narrower the filter, the more it delays signals and slows their rise time. (The delay can be as great as several tens of milliseconds in narrow filters.) Of the IC-725’s possible IF-filter suite, the narrow CW filter therefore introduces the longest delay, and slows signal rise and fall times the most. In the time a reasonably strong signal takes to move from a receiver’s antenna terminals to the receiver’s AGC detector, the receiver operates at maximum gain and its later IF stages may limit or overload. As soon as the AGC detector produces output, AGC action begins.

Although AGC is applied to stages preceding and following the IC-725’s narrowest IF filters, the filter delay causes only the AGCed stages after the filter to initially contribute to gain reduction. (The filter puts out unreduced signal until the reduced signal makes it through the filter.) As a result, the AGC detector overshoots rapidly because of its fast attack, putting out too much control voltage until the reduced-level signal makes it through the filter and IF stages to the AGC detector. When the delayed, somewhat controlled signal makes it through the filter, the AGC adjusts itself (slowly, because of its slow decay) to the new, lower value and turns the receiver gain back up a bit. Now the AGC undershoots because of the added signal reduction! Once again, this new, lower RF level affects AGC-detector output only after filter delay, and so stabilizing the system requires considerably more time with filter delay than without it. It’s easy to see that any additional delay in applying control voltages to gain-controlled stages only makes loop stabilization longer and more complex. An ill-designed AGC system can even oscillate under the right conditions!

The IC-725 has only one AGCed stage (Q27, a 3SK74 dual-gate MOSFET) after its main IF filter. Because of the MOSFETs’ limited gain-control range, overshoot recovery becomes additionally longer. AGC voltage is fed to gate 2 of the IC-725’s earlier dual-gate AGCed stages (Q15 and Q21, both 3SK74s) via RC decoupling (filter) networks. These networks introduce more delay. See Figure 5 .

The last AGCed stage, Q27, responds fastest, having a rise time of 2.2 470 (R143) 0.0047 µF (C136), or 4.9 µs. The other stages are far too slow: Q21’s response time is 2.2 470 k (R33) 0.1 µF (C42), or 103 ms; Q15’s, 2.2 1 M (R78) 0.1 µF (C91), or 220 ms. This is particularly ironic in that 0.1 µF is not an optimum value for bypassing at the frequencies involved (70.45 MHz at Q15 and 9 MHz at Q21), and because the wire-lead capacitors used are inductive at these frequencies anyway. I recommend changing C91 to 0.001 µF and C42 to 0.01 µF. To realize faster attack response, bridge the Gate 2 decoupling resistors R33 (470 k ) and R78 (1 M ) with 1-k resistors. Figure 6 shows the IC-725’s AGC detector/amplifier.

With the IC-725’s fast AGC selected (S5, the AGC button, pushed in), transistor Q30 charges C113 (1 µF, in series with R120 [100 ]) toward the negative rail voltage (-5.2 V) through R123 (470 ). This rise time is about 2.2 570 1 µF (1.3 ms), setting the IC-725’s fastest possible AGC attack. Selecting slow AGC ( AGC button out, closing S5) adds, in parallel with the C113-R120 combination, C112 (10 µF) in series with R119 (10 k ). C112 can neither charge nor discharge rapidly with R119 in series, so switching in C112-R119 probably makes a difference only when receiving constant-carrier signals (unmodulated carriers, full-carrier AM, radioteletype, and so on). On SSB, the influence of C112-R119 is minimal. 12 The decay time of the IC-725’s AGC voltage depends mainly on the discharging of C113 (fast) or C113 and C112 (slow) through R118 (3.3 M ) and R114 (1 M ), the input resistor of an op-amp dc amplifier that feeds AGC voltage to the IC-725’s squelch and S-meter circuitry.

Notes

(1) Kirk Kleinschmidt, NT0Z, “ICOM IC-725 MF/HF Transceiver,” Product Review, QST , Mar 1990, pp 38-41.

(2) Peter Hart, G3SJX, “ICOM IC-725 HF Transceiver,” Radio Communication , Sep 1989, pp 56-58. (3) OH2GF has tested these modifications only for the IC-721, which is electrically identical to the IC-725 and IC-726 transceivers in the areas addressed by this article. Throughout, this article therefore uses “IC-725” to mean “IC-721, ’725 and ’726.” Because the IC-725 component designators and values involved in the modifications presented in this part also apply to the IC-728 (the IC-725’s current US-market successor) even though the ’728’s frequency conversion scheme differs from the ’725’s, the circuit changes presented can likely improve the IC-728’s performance as well. The modifications presented are not ICOM approved and may void your transceiver’s warranty. Ed.

(4) See the article cited in Note 1: “Receiver audio + IF response...430-2090 Hz” (from the test-results table) and (from the review text) “the IC-725’s AM filter is far too wide for good AM reception in Amateur Radio and shortwave-broadcast bands and the ’725 sounds as if audio rolloff has been built in to compensate for the filter’s wideness.”

(5) See the article cited in Note 2: “In-band linearity measured with 200 Hz tone spacing was poor.”

(6) See the article cited in Note 1: “The rig’s audio-gain control is quite touchy. Rarely did I have to turn the knob past 9 o’clock (that means 3/4 of the knob’s range is never used!).”

(7) W. Cocking, “Linearity of the Transistor Amplifier,” Wireless World , May 1972, pp 210-211.

(8) See Robert Schetgen, KU7G, Editor, The ARRL Handbook for Radio Amateurs , 1995 edition (Newington: ARRL, 1994), pages 15.13-15.14.

(9) F. E. Terman, Electronic and Radio Engineering , 4th edition (New York: McGraw-Hill, 1955), pp 547-557.

(10) M. Chessman and N. Sokal, “Prevent Emitter-Follower Oscillation,” Electronic Design , Vol. 13, Jun 21, 1976, pp 110-113. (11) E. W. Pappenfus, Warren B. Bruene and E. O. Schoenike, Single Sideband Principles and Circuits (New York: McGraw-Hill, 1964), p 284. (12) From QST ’s IC-725 review (Note 1): “I did not notice a great deal of difference between the fast and slow settings.” Coming in Part 2: “Hang” AGC for the IC-725 receiver, and RF speech processing and better ALC for the IC-725 transmitter.

Figure 1 The IC-725’s audio amplification chain includes significant AF rolloff to tame IF hiss and 7-kHz leakage from the radio’s digital circuitry. Modifying these stages for better performance involves changes to the asterisked components as described in the text. (Except as indicated, in this and all other schematics in this article, decimal values of capacitance are in microfarads [µF]; others are in picofarads [pF]; resistances are in ohms; k=1000; and M=1,000,000.)

Figure 2 The IC-725’s AM detector is nearly identical to the IC-735’s voltage doubler AM detector with one diode replaced by 1-k resistor R260. For better-sounding AM, bridge a resistor across the asterisked capacitor (C118) as explained in the text.

Figure 3 The IC-725’s AM-wide frequency response before and after detector modifications.

Figure 4 Rewiring the product detector to get its IF drive from a point before buffer amplifier Q29 keeps the IMD generated by the IC-725’s AM and AGC detectors from affecting the sound of received SSB and CW signals.

Figure 5 The time constants of the RC decoupling networks in the AGC lines to the IC-725’s three AGCed IF amplifier differ greatly. The time constants of the Q15 and Q21 feeds are so long that the IC-725 cannot respond rapidly enough to SSB speech transients and the onset of Morse code dots and dashes. The response time can be improved by adjustments to the asterisked components as described in the text.

Figure 6 Simplified schematic of the IC-725’s AGC detector and amplifier. The high value of C112’s series resistor (R119, 10 k ) keeps C112 from charging rapidly and explains why switching the IC-725’s AGC decay to slow makes little difference during CW and SSB reception. For Safer, Easier Modifications... In developing and testing these modifications, I worked as noninvasively as possible by adding resistors in parallel with existing resistors on the component side of the IC-725’s Main Unit circuit board. For other changes, I just cut off capacitors with small, sharp pliers.

Some of the modifications require access to the foil side of the board anyway (done by carefully disconnecting all of the board’s connectors and taking out mounting screws). If you’re going to do them, you might as well save time by removing the Main Unit board and replacing any values that need to be changed with parts of the correct values rather than just paralleling parts on top of the board. Some of these modifications involve FET circuitry that operates at high impedances. Such circuitry may be damaged by static electricity or ac-line leakage from a soldering iron tip, so be sure to ground your iron’s tip to the ground foil of the circuit under modification (use a thin cable) so no potential difference can exist between the tip and the circuitry to be soldered. OH2GF

Hot-Rod Your ICOM IC-725-Series Transceiver Part 2

Concluding his two-parter on improvements for these popular transceivers, OH2GF adds “hang” AGC, reduces SSB-transmit IMD and builds in a simple RF speech processor. By Jukka Vermasvuori , OH2GF Viputie 3 FIN-01640 Vantaa Finland Photos by the author Part 1 of this article appeared on pages 42-45 of September 1995 QST . OH2GF has tested these modifications only for the IC-721, which is electrically identical to the IC-725 and IC-726 transceivers in the areas addressed by this article.

The IC-725 component designators and values involved in the hang-AGC, AM-wide bandwidth and ALC modifications presented in this part also apply to the IC-725’s current US-market successor the IC-728 even though the ’728’s frequency conversion scheme differs from the 725’s, so the circuit changes presented can likely improve the IC-728’s performance as well. (The IC-728’s transmit circuitry differs enough from the IC-725 that some reengineering may be necessary to make the RF speech clipping modification work in the IC-728.) The modifications presented are not ICOM-approved and may void your transceiver’s warranty. Ed .

To make the IC-725’s automatic gain control (AGC) work even better, I developed a simple “hang” AGC system as shown in Figure 7 . In a standard AGC system, the control voltage returns gradually to its no-signal value after a signal disappears. In a hang AGC system, the control voltage stays at the level set by the incoming signal for a predetermined time, and then rapidly returns to its no-signal value. 13 To do this, remove C112 (10 µF) and C113 (1 µF) from the board, reinstall C112 in C113’s PC-board holes we’ll call this capacitor “C113” from now on and install a 1-M resistor between Q30’s collector and the PC-board hole that took the old C112’s negative (-) lead. Remove the original discharge resistor (R118, 3.3 M ) and replace it with a BS170 MOSFET (drain toward +8 V, source to the AGC line) in series with a new discharge resistor (2.2 M ). Then complete the rest of the changes shown in Figure 7 .

At signal onset, AGC amplifier transistor Q30 charges C113 and the BS170’s gate capacitor. When the signal disappears, C113 discharges very slowly through R114, but the BS170’s gate capacitor begins to discharge at once through its associated 3.3-M resistor. When the BS170’s gate-source voltage increases enough, the BS170 turns on, connecting the 2.2-M resistor across C113-R120. C113 then begins to discharge. As it does, the AGC voltage becomes more positive and the receiver’s gain increases. The BS170’s delay in connecting 2.2 M across C113 produces the system’s hang action. The hang time is determined by the FET’s gate time constant, which the AGC switch now selects. The FET’s channel starts to conduct when its gate-source voltage rises to about +2.5 V. When this happens depends on the AGC transistor’s charging capability with increasing signal strength, and on the slow AGC discharge time constant of 4.7 µF in parallel with 3.3 M .

Setting the AGC switch to fast, parallels the 3.3-M gate resistor with the series 1-M and 10-k (originally R119) resistors through 470 . This hang-AGC modification therefore reverses the AGC switch’s fast and slow positions: Pushing the AGC button in (opening it) now selects slow AGC, and leaving the button out selects fast AGC. Reducing the Wide Apparent Bandwidth of Strong Signals The IC-725’s first 9-MHz filter FI2 (a “9M15” filter [FL-23 in ICOM nomenclature]) does double duty, setting the radio’s FM selectivity to 15 kHz and band-limiting the signals fed to the radio’s noise-blanker circuitry. Because I didn’t intend to operate FM with my IC-725, I decided to replace the 9M15 with a narrower filter. I removed the 9M15 filter and replaced it with a “9M6” filter (ICOM FL-116) the same type used for FI4, the IC-725’s 9-MHz AM filter which I bought from ICOM as a replacement part. 14

The 9M15 and 9M6 are mechanically identical (the size of a small crystal), so the work was easy to do. This modification disallows using the IC-725 on FM, but gives better protection for the radio’s later stages against strong out-of-channel signals in non-FM modes. The improvement isn’t perfect: L21 and L22, the resonators in tuned circuits at the new filter’s input and output, must be retuned after the new filter is installed, and the IC-725’s IF response is asymmetrical with the 9M6 installed in place of the 9M15. For serious AM reception, an altogether different (better) filter would be necessary. An optional ICOM filter (FL-112) with good selectivity at the right center frequency has been available in the past, but it is mechanically bigger than the 9M6 and 9M15 and therefore could not be directly installed at FI2’s location. Transmitter Improvements ALC Automatic level control is transmit-mode AGC, so the dynamics of an ALC system are similar to those of an AGC system. Figure 8A shows part of the IC-725’s ALC circuitry.

IC11A, one half of an M5218L dual op amp, acts as a comparator, comparing the dc voltage set by the IC-725’s RF PWR control with RF voltage peaks rectified by the IC-725’s forward-power transmit-output detector (D7, a 1K60 diode in the PA Unit). When detected power peaks exceed the reference value, IC11A generates audio-frequency-range output, which is rectified by D76 (a 1SS133 diode). The resulting negative voltage charges C264 (1 µF), the ALC memory capacitor. As shown in Figure 8B , the ALC voltage is fed to Q22 and Q7 (2SK241 MOSFET IF amplifiers used only in transmit mode) via low-pass RC networks: R87 (1.5 M )/C101 (0.1 µF) and R22 (150 )/C28 (0.0047 µF).

As in the radio’s AGC circuitry, two different time constants are used (1.5 M 0.1 µF and 150 0.0047 µF) in control-voltage feeds to the system’s two controlled stages. At this point, it’s instructional to keep in mind that an AGC or ALC-controlled stage is really a form of mixer or modulator in which the desired signal is mixed with or modulated by an extremely low-frequency signal (the AGC or ALC voltage). Since an AGC or ALC system is intended only to control relatively slow changes in the desired signal’s level, it must not be capable of following the signal so closely that it tries to “iron out” the rapid signal variations that convey the information we want to receive or transmit. Otherwise, distortion will result.

The IC-725’s ALC system is flawed in this sense. A low-value resistor (R215, 150 ) in series with the ALC memory capacitor C264 (1 µF) works against C264’s filtering action. As a result, low-frequency ac is left riding on the ALC dc. This ac amplitude-modulates the envelope of the transmitter signal at IF, creating spurious signals and splatter. (C264 discharges through R198 [1 M ] the lower end of which is connected to IC11’s input [pin 6], virtually 0 . 15 ) The ALC loop design is somewhat less complex that the receiver’s AGC because there are no narrow filters in the loop to cause time-delay problems. The use of two different RC low-pass time constants seems to be harmful, however. The stage with the longer delay can be ignored for the moment, since the faster stage stabilizes the ALC loop to generate a predetermined ALC dc level. This dc level is “memorized” by the loop’s memory capacitor, C264, which discharges slowly. When the second (slower) stage’s delay is over, it attenuates the signal based on the ALC value determined by loop stabilization with the faster stage. Now the system knocks down signals too much ! During the momentary time the slower stage takes to achieve loop equilibrium, the radio’s output power is lower than it should be. The time constants of the low-pass networks that feed ALC voltage to these stages can be carefully chosen to minimize this undershoot. This is probably why R215, no analogy to which was used in the IC-735, was introduced in the IC-725 design. It allows (at first) faster ALC loop decay, but supports the generation of spurious AM. 16 The audio signal components present in IC11A’s output are distorted because of the chip’s comparator function. Yet, for proper ALC action, no audio-range phase shifts or delays should be added before ALC detection; only RF components can be filtered out. Also, as in an AGC system, the ALC detector should be driven from a low-impedance source. But this is not the case with the IC-725, in which audio-range ALC-voltage components are filtered with R214 (470 ), C262 (0.047 µF) and C263 (0.001 µF) an arrangement that causes unwanted phase shift while driving the ALC detector (D76) from a relatively high impedance. (In this respect, the IC-735 uses more solid detection principles, although it also suffers from the two-time-constants problem.)

I achieved a cleaner transmitted SSB signal by short-circuiting R215, bridging R87 with a 150- resistor, cutting C262 off the PC board, and adding a 100-k resistor in parallel with C264 for shorter ALC decay. RF Speech Processing With average, unprocessed speech, the ratio of peak envelope power (PEP) to average power on SSB is about 15 dB. Operating the IC-725 at 100 W PEP on SSB therefore results in an average power output of 3.2 W! By using an RF clipper with 20 dB of clipping, the ratio of PEP to average power is 9 dB. 17 The average power increases by 6 dB, meaning four times the power from 3.2 W to 12.8 W. This is a considerable gain in talk power. With this in mind, I undertook the most ambitious aspect of this modification suite: I designed and installed RF speech clipping! All I needed was a tuned IF amplifier with 20 dB of gain, limiting diodes, and an SSB-width IF filter to strip away the adjacent-channel splatter added by the clipping process.

The processor can be fitted in the transmitter path either (a) following the balanced modulator and before the IC-725’s SSB filter or (b) after the original filter, where the transmitter path has been separated from the receiving chain. Because of the instability risks I saw in my investigations of the IC-725’s receiver, option b seemed better because it promised to allow the shortest leads to the necessary external PC board. Figure 9 shows the circuit. I originally obtained the required gain with a single dual-gate FET, but later replaced it with a single 2SK241 FET, which offers similar very low feedback capacitance (0.035 pF) and forward transfer admittance of 10 mA/V. Doing this reduced the amplifier’s component count by five.

I used two antiparallel, diode-connected 2N3904 transistors as limiter diodes because they limit more sharply than 1N4148-class switching diodes. C1 matches the high impedance of the tuned circuit to the termination resistor of 1 k in parallel with the filter’s input impedance. R85 (10 k ) at Q22’s drain is adjusted so that with hard limiting in the clipper, the IC-725’s ALC indicator just lights on the band in which the transceiver has the lowest transmit gain. If necessary, you can use C1 for output level adjustment if R85 doesn’t have enough control range. 18 I built the clipper using ground-plane construction 19 according to the layout shown in Figure 10 .

Figure 11 shows the installed assembly.

I recommend running Q22 at close to maximum gain. Because there’s no gain to spare, every adjustment of C1 requires readjustment of the clipper’s sharply reacting tuned circuit to get maximum output. Adjusting the IC-725’s MIC control adjusts the clipping level, maximum MIC gain giving maximum clipping. Field tests with the RF clipper and faster ALC have been most satisfying. Your Turn Regardless of their manufacture, many of today’s shortwave transceivers use similar circuitry, especially in low-level stages. Most use AGCed dual-gate MOSFETs as IF amplifiers in receive and transmit; most use similar ALC and AGC circuitry. As basic even uninteresting as these time-tested subsystems may appear to be, they and their dynamics play a large part in determining how our radios sound and feel on the air.

With so much attention paid to computer control, memories and exotic anti-interference features, “design drift” sometimes causes the fundamental radio performance of even our best transceivers to be poorer than it should be. Armed with practical knowledge of how our radios should work and sound, we can do much to improve already-solid commercial designs by analyzing and adjusting the performance of their low-level IF and AF subsystems. That done, we can enjoy radios that work as well as their many bells and whistles suggest they should!

Notes

(13) The ARRL Handbook for Radio Amateurs , 1995 edition, p 17.27.

(14) With the IC-725 having been discontinued, this filter (ICOM 955-07272) is in short supply.

(15) Paul Horowitz and Winfield Hill, The Art of Electronics , 2nd edition (Cambridge: Cambridge University Press, 1989), p 178: “What is the input impedance? Simple. Point A [the input pin] is always at zero volts (it’s called a virtual ground ).”

(16) In the receive side, the resistors (R119, 10 k , and R120, 100 ) in series with the stock IC-725’s AGC memory capacitors (C112 and C113 in Figure 6 of Part 1) cause AGC-line ripple that adds similar spurious AM to signals passing through the radio’s AGCed stages. This spurious AM may be a major factor in poor in-band IMD receive performance reported for the IC-725 and apparent in more than a few transceivers with similar AGC systems. The hang-AGC modification presented earlier in this article minimizes this effect through its hang action, and by keeping the series resistance at 100 regardless of the AGC speed chosen. See the “AGC Loop Problems” portion of the 1995 ARRL Handbook ’s Receivers, Transmitters, Transceivers and Projects chapter, which includes this passage: “Audio frequency components on the AGC bus can cause problems because the amplifier gains are modulated by the audio and distort the desired signal. A hang AGC circuit can reduce or eliminate this problem.” Ed.

(17) W. Sabin and E. Schoenike, Single-Sideband Systems & Circuits , 2nd ed (New York: McGraw-Hill, 1995), p 284: “Measurements verify that with 20 dB of clipping, the peak-to-average ratio of speech...is about 9 dB.”

(18) Once the RF clipper is installed, it’s difficult to adjust the tuned circuit coil. So when the added circuit board is finished, you can pre-tune it outside the transceiver by soldering temporary 50- coaxial cable via 1-k resistors to the filter’s output and connecting the other end into the transceiver antenna connector. Tune the transceiver’s general-coverage receiver to the filter center frequency. Apply dc to the clipper board and adjust the clipper’s tuned circuit for maximum noise output from the clipper amplifier. The noise level will be low, but recognizable with full receiver sensitivity.

(19) The ARRL Handbook for Radio Amateurs , 1995 edition, p 25.8.

Jukka Vermasvuori has been an active ham ever since receiving his license in 1956 at age 15. His hobby within the hobby is designing HF receivers and transmitters: He received “First 1970 Award” at an RSGB Radio Engineering and Communications Exhibit for a receiver that used dual-gate FETs and a double-balanced mixer. Jukka has spent 30 years with the Finnish Broadcasting Company purchasing new AM, FM and TV transmitter equipment. Highlights of his career have included the purchase and commissioning of three 1-megawatt-PEP SSB transmitters for shortwave broadcasting, and recently dealing with IOT common amplification UHF-TV transmitters. He has been actively participating in the preparation of “CCIR Report to World Administrative Radio Conference (HFBC-93) Dealing With Matters Connected With The HF Broadcasting Service” (part: SSB System for HF Broadcasting). He is also the author of several technical articles published in Finland and elsewhere.

Figure 7 A simple hang AGC system for the IC-725. The BS170 FET switches a 2.2-M resistor in parallel with the system’s AGC memory capacitor (C112, 10 µF, moved into C113’s location on the IC-725’s Main Unit circuit board and henceforth referred to as C113) when its gate voltage rises high enough, as determined by the time constant of the parallel RC network at its gate. The IC-725’s AGC switch now changes the BS170’s gate time constant rather than changing the value of AGC memory capacitance. As in Figure 6 , D58 is biased to act as a voltage clamp that keeps the AGC line from going more positive than +2.2 V during no-signal periods.

Figure 8A In the IC-725’s ALC detector circuitry (A), the voltage drop across R215 works against C264’s filtering action, allowing voice-frequency ac components to modulate the ALC voltage. This adds spurious AM sidebands to the IC-725’s transmitted SSB signal and causes adjacent-channel interference.

Figure 8B B shows the two time constants involved in the control-voltage feed to the IC-725’s ALCed stages. The asterisks indicate components affected by modifications described in the text.

Figure 9 A simple RF speech processor for the IC-725. This circuit amplifies, clips and filters the IC-725’s SSB signal at IF for much greater talk power than that achievable with audio processing. T1 is a Toko RLC-20241 10.7-MHz IF can, which can be tuned down to 9 MHz by adding 33 pF across its tuned winding. (Toko’s 154AC-T1038Z and KACS-K586HM parts, both of which are available from Digi-Key Electronics, are possible substitutes.) The crystal filter is ICOM’s FL-103 “wide SSB” filter, sold as an option for the IC-780/781 transceiver.

Figure 10 How the RF-processor subunit fits into the IC-725.

Figure 11 The installed RF-processor subunit: (A) underside; (B) top; and (C) ready to go.