Adam, VA7OJ/AB4OJ's IC-756Pro III User Review

Initial observations (July 2005)

On June 30, I picked up my new IC-756Pro3 (S/N 32025XX) at a local dealer. That evening, I installed the Pro3 in my station. Preliminary findings are that it pulls weak, noisy signals out of the noise somewhat better than the Pro2. I was able to work a number of SSB stations on 40m which I would not have heard on the Pro2. The NR and NB appear to have been significantly improved. Overall, I would judge that the S/N ratio is better than on the Pro2, even in the presence of power-line noise. I find myself routinely operating with the preamp off; on the Pro2, I invariably had Preamp 1 on.

The next day, I was able to sustain the sked with my friend Matt KK5DR somewhat longer, as the Pro3 "hears" signals buried in the noise somewhat better than the Pro2 did. I am able (by using a combination of NR, NB, Twin PBT and Manual Notch) to squeeze intelligible copy out of the assorted "garbage" on the band more effectively than on the Pro2.  In addition, the Pro3's strong-signal behaviour appeared to be excellent. S9 + 60 dB signals 1 to 2 kHz above and below our channel edges did not produce any perceptible artefacts in the received audio.

When switched in, Preamps 1 and 2 do not raise the audible noise level in the receiver as much as they did in the Pro2. In the Pro3, the preamps have been designed for highly linear large-signal handling, consistent with a low noise figure. Preamp 1 has negative feedback to improve linearity without compromising NF. There has also been a significant gain redistribution. The first mixer has lower gain, with the result that the scope module in the Pro3 has its own RF amplifier to bring up the vertical sensitivity of the scope.

Notes on ALC

The ALC is a little less sensitive; I need to set the Quadra ALC menu to 5 bars for 1kW output, vs. 4 bars for the Pro2.

As is the case for all other Icom HF transceivers, the rear-panel ALC jack is compatible with a negative-going ALC voltage in the range 0 to -4V. Positive voltage should never be applied to the ALC jack.

There have been some improvements in operational convenience. The mini-scope is great - you can now display 3 windows; frequency/mode etc., mini-scope and bar-graph meter, filter menu or any other lower window that used to suppress the scope on the Pro2. There are two clock displays - UTC and local - and a simple screen-saver which displays the owner's callsign moving about on a black background. By being able to leave the bar-graph meter on-screen along with the scope, I learned that compression peaks at 6 to 7 dB when it only shows 2 to 3 dB on the much slower moving-coil meter.

Display quality

I have the impression that the Pro3's TFT display is superior to that of the Pro2.  The screen  image is brighter and a little sharper, the contrast is much better and the colours are more saturated.

A few measurements

As the ARRL, RSGB and others have already subjected the Pro3 to exhaustive test suites, I felt that any such effort on my part would be redundant (apart from the fact that my second signal generator is much too "dirty" for meaningful RF 2-tone measurements!)

To satisfy my own curiosity, though, I ran a few quick receiver tests: MDS, reciprocal mixing noise and filter shape factors. The test setup was as follows:

HP8640B signal generator, calibrated against R&S URV4-03 RF power meter with 0 dB probe; Sinadder 3 configured as true RMS audio voltmeter connected to [SPKR] jack. NB off, NR off, ATT = 0 dB, AGC MID. Offset tone pitch = 1 kHz (SSB), 600 Hz (CW). SSB and CW "Sharp" filter shapes selected.

1: MDS (Minimum Discernible Signal). In this test, MDS is defined as the RF input power which yields 3 dB unweighted S/N at the output. The results obtained are very close to those in Ref.2.

Table 1: MDS (Min. Discernible Signal)
MDS at 14.1 MHz (dBm)
Preamp SSB 2.4 kHz CW 500 Hz BPF
Off -126 -131
1 -133 -139
2 -135.5 -142

2: Reciprocal Mixing Noise. Reciprocal mixing occurs in a superheterodyne receiver when the noise sidebands of the local oscillator (LO) mix with strong signals close in frequency to the wanted signal, producing unwanted noise products at the intermediate frequency and degrading the receiver sensitivity. Reciprocal mixing noise is a measure of LO spectral purity.

In this test, a strong "undesired" signal is injected into the receiver's RF input at a fixed offset from the operating frequency. The RF input power is increased until the receiver noise floor increases by 3 dB. The reciprocal mixing noise parameter, expressed as a figure of merit, is the difference between this RF input power and measured MDS. The test is run with preamp off. The higher the value, the better.

Table 2: Reciprocal Mixing Noise
  Recip. Mixing (dB)
Offset kHz 3.6 MHz LSB 14.1 MHz USB
2 80.5 79
3 83.5 81
5 88 86
10 95.3 91.5

3. DSP-IF filter shape factors. A quick check of -6/-60 dB shape factors was performed for the 2.4 kHz SSB, 500 Hz CW and 250 Hz CW  IF filter settings. (Note: BPF Indicator was on for CW filters.) Shape factors for all 3 filter settings met or exceeded Icom's advertised specifications. The measured -6/-60 dB shape factor for the 2.4 kHz SSB filter was 1.24; the spec is 1.5.

Tip: To achieve optimum S/N ratio, the selected IF-filter bandwidth should not exceed the occupied bandwidth of the received signal.

Programmable SSB transmitted occupied bandwidth (TBW)

As in the IC-7800, the upper and lower -6 dB points of the transmitter audio-frequency response are independently programmable. This capability, along with the bass and treble equalisation menu, offers many intriguing possibilities. Thus far, I have been using default settings, but plan to do more experimentation.

The QSK "dit-clipping" issue

As discussed in the ARRL Test Report, CW operators will be very relieved to hear that the "dit-clipping" problem which plagued the 756Pro2 in full-break-in mode has been completely eliminated from the 756Pro3. A quick check has confirmed this.

Noise Reduction (NR)

It has been reported that the Pro2 NR's heuristic correlation-discrimination process (correlated signals vs. non-correlated noise) becomes "acclimatised" to the prevailing SNR and becomes less effective over time. Keying the transmitter reinitialises the NR process, creating a perception of improved NR efficacy. George W5YR has contributed an article on this topic to my IC-756Pro/Pro2 User Review.

I have a sneaking suspicion that the NR process in the Pro3 reinitialises itself automatically every 15 sec. or so. This, added to the improvements in the receiver's RF chain, may explain the improvement in weak-signal handling with NR on (as noted above.)

Ref.1, Section 5-2-1, describes NR in greater detail.

More on NR: Normally, I leave the NR control at 50 to 60%  in SSB mode, with AGC = MID. It knocks down the band noise by about 20 dB, but does cut the treble response somewhat as is to be expected. It increases the S/N ratio of the received signal noticeably, and very often makes the difference between copying and not copying a weak SSB signal. There is a slight, but perceptible decrease in noise level between 50% and 60%. Increasing the setting above 60% does not reduce the noise level further, but does not cause distortion. (In the 756Pro2, 50% was the limit of noise reduction.)

In CW mode, the NR will be less effective, as the noise bandwidth is much less than in SSB, and the NR algorithm has less noise power to work with. I find that with NR at 50%, there is a noticeable noise reduction with the 1.2 kHz filter, less with the 500 Hz filter and none to speak of with the 250 Hz filter. Again, this is to be expected (N = kTB). During the test I conducted, I selected AGC = FAST.

I did not observe distortion of the CW signal at any NR setting, even with the IF bandwidth reduced to 50 Hz via Twin PBT.  (At the time, I was listening to CW signals varying in level between S0 and S5.) When I operate CW, the NR control stays at 50%. The noise reduction obtained with BW = 500 Hz and above is still useful.

Improved CW reception

The received audio in CW mode is much more pleasant to listen to, and less fatiguing, than on the 756Pro2 or 756Pro. This is particularly noticeable when using the narrow IF filters (500 Hz and less).

The "background" is much quieter, and the "ringing" effect previously reported in the 756Pro and 756Pro2 (especially with CW Pitch < 600 Hz)  is almost inaudible in the Pro3. I listened to an S0 to S1 CW signal with BW = 50 Hz (SHARP shape factor, BPF on) and obtained very pleasant single-signal copy, even at that narrow bandwidth setting. The low-frequency rumble reported in the 756Pro has been completely eliminated in the 756Pro3.

Microphones for the IC-756Pro3

Let me start by remarking that my background in telecommunications engineering strongly influences my view of amateur radio.

From my perspective, the sole mission of an SSB radiotelephone communications system is to "get the message through" with the highest possible articulation index at the receiving station, under a wide variety of propagation conditions. This objective should be achieved in a manner that conserves transmitter power, and ensures that the available power is concentrated in that portion of the human speech spectrum which most influences articulation - typically about 350 ~ 2700 Hz at the -6 dB points.

Excessive transmitted energy below approx. 250 Hz robs transmitter power (remembering that an SSB transmitter is peak-power-limited) without contributing to the articulation at the receiving station. In addition, the practice of "force-feeding" - deliberately boosting the low-frequency portion of the baseband - can drive the speech amplifier into non-linearity, generating harmonics which mix with the mid-band speech components to yield a rich harvest of intermodulation products. These products "muddy" the transmitted audio.

It should also be noted that the 15 to 18 kHz high-end response of some "broadcast-type" mics is wasted on a system which is 6 dB down at 2900 Hz and 13 dB down at 2950 Hz. In the IC-756Pro series, a steep analogue LPF in the baseband signal path before the ADC input sharply attenuates frequencies above 3 kHz, to prevent ADC overload at frequencies above the programmed DSP cutoff.

The Heil HC-4 and HC-5 elements are excellent examples of transducers specifically designed for the transmission of human speech with optimum articulation. The frequency response of the HC-5 is 300 Hz to 6 kHz at -6 dB, with 6 dB rise at 2 kHz. The HC-4 frequency response is 600 Hz to 6 kHz at -6 dB, with 10 dB rise at 2 kHz. It will be seen that both these elements are peaky around 2 kHz. This is, in effect, a form of pre-emphasis which if accompanied by a judicious amount of compression, can increase intelligibility at the distant station even under unfavourable band conditions. Their output is significantly lower than that of an electret element, but will drive the Pro-series radios to full output with Mic Gain at 50%, and about 6 dB of compression. Refer to "DSP Audio Settings for Heil Microphones".

I have found that the following settings on the IC-756Pro3, Pro2 and Pro work well with the Heil elements. These settings are also applicable to the IC-756Pro2, IC-756Pro and IC-7800. Note that the TBW settings are individually programmable only on the IC-7800 and the IC-756Pro3.

Table 3: IC-756Pro3 Settings for Heil Elements
Heil Element Mic Gain TBW COMP* Bass EQ dB Treble EQ dB
HC-5 12 o'clock MID 12 o'clock -2 +3
HC-4 1 o'clock NAR 1 o'clock -2 +5
Wide Range 10 o'clock WIDE 9 o'clock -3 +5
Electret 9 o'clock MID 9 o'clock -3 +3
*Set COMP for 6 dB compression on voice peaks (on bar-graph meter).

The IC-756Pro3 transmitter offers a combination of bass/treble equalisation and AF response adjustment (selectable low and high -6 dB points.)  With this much flexibility, it should be possible to configure the radio to work well with almost any microphone. Given the communications mission referred to above, though, the best choice will always be a microphone optimised for radiotelephone communications, as opposed to a microphone more suited to broadcast or recording applications.

Spectrum Scope Vertical Sensitivity (August 2005)

With Preamp 2 enabled,  0 dB attenuation (RF and scope) and autotuner disengaged , the Pro III spectrum scope will display a visible spike with 0.06 µV (-131 dBm) at the antenna socket. (This was measured at 14.1 MHz with the HP8640B signal generator, calibrated against the R&S URV4-03 RF power meter with 0 dB probe.) The radio was also in CW mode, with the 500 Hz filter selected.

The corresponding figure for the Pro II was 0.12 µV (-125 dBm).

Dual Watch/Spectrum Scope interaction (August 2005)

I have noticed a slight interaction between Spectrum Scope vertical amplitude and Dual Watch BAL control setting on the IC-756Pro3.

Dual Watch was enabled, with 14316 kHz on MAIN, and 14186 kHz on SUB. When I rotated the BAL control fully clockwise to cut off the Main channel, I observed that the scope grass level rose approx. 6 dB.

Per the block diagram, the scope pickoff point is at the output of the MAIN (A) 1st mixer, which also feeds the MAIN Balance PIN attenuator. I believe that the following is occurring: When the PIN attenuator is completely cut off, it breaks the signal path to the next stage (the IF preamplifier feeding the 1st IF combiner). This partially unterminates the MAIN 1st mixer output, causing the RF voltage at that point (and consequently at the input to the scope RF amplifier) to rise by about 6 dB. There is no other apparent change in receiver behaviour (e.g. instability etc.)

Comparing the 756Pro3 to the 756Pro and 756Pro2 yielded these results:

This is what I believe is happening: The Pro 1st mixer is followed by an LPF. When the PIN diode attenuator is cut off, the LPF is mis-terminated (looks into an open.) This apparently reflects a lower impedance back towards the 1st mixer, loading it down and reducing the signal amplitude at the scope input.

In the Pro2, there is no LPF following the 1st mixer. The mixer output is fairly low-Z, so the mis-termination when the PIN diode attenuator is cut off will not cause much change in signal level at the scope pickoff point.

In the Pro3, the 1st mixer has a lower IF output than in the Pro or Pro2 (hence the need for an RF amplifier in the Pro3 scope input line). It is likely that the 1st mixer output is noise-matched to the PIN attenuator, in which case it will have a higher source impedance to match the load conjugately. The rise in load impedance when the attenuator goes to cut-off will cause approx. 6 dB increase in output voltage. The scope input voltage will rise by the same amount.

This is surmise on my part; I have not opened up the radio to take measurements. This phenomenon is interesting, but not a major issue as it does not affect receiver performance at all. In practice, with Dual Watch on, the BAL control is normally set between 40 and 60%, and the interaction does not occur.

Comment on the IC-756Pro3 Autotuner (December 2005)

Matt KK5DR has found that the IC-756Pro3 autotuner tunes a little faster than that of the IC-756Pro2.  A quick check on my Pro3 confirmed this.

AM and FM Operation with the IC-756Pro3 (January 2006)

There has been no change from the IC-756Pro2 in this area. Please refer to the following links:

Q: Does the IC-756Pro3 have a synchronous AM detector?

A: None of the Icom DSP-based HF transceivers has an AM synchronous detector in the sense of a phase-lock demodulator which regenerates the incident carrier, demodulates the sidebands against the local carrier and "votes" the sideband with the best S/N ratio.

All the Icom DSP-based radios (including the IC-746Pro, the IC-7800, the IC-756Pro series and the IC-7000) employ the same DSP algorithm. This models a phasing-type modulator and demodulator for SSB and AM. The difference between SSB and AM is that in SSB, the unwanted sideband is cancelled, whereas in AM the upper and lower sidebands are added in phase to produce the composite baseband.

Unlike a true synchronous detector, the AM demodulation process in the Icom DSP design does not "vote" the sideband with the best S/N ratio.

AM broadcast reception is very effective on the IC-756Pro series, but is subject to selective fading. One can use ECSS (by receiving AM in SSB mode) to reduce this, and select the quieter sideband manually.

Q: Can you explain the amplitude-sensitive digital artefacts which appear on the transmitted AM signal?

A: With any system involving A/D and D/A conversion, quantizing distortion will occur at high signal levels. This will manifest itself as noise on modulation peaks. As the ADC and DAC used in the Icom radios are 24-bit devices originally intended for high-grade audio applications, quantizing distortion will be insignificant as long as the devices are not overdriven.

Note that as the analogue input level to the  ADC approaches the "all 1's" point, quantizing distortion will increase sharply. If [MIC GAIN] is set for 90% modulation, using the Spectrum Scope as described here, the ADC will remain well within its linear range.

Also as discussed above, Po must not exceed 25W resting carrier.

The spectrum-analysis test (described here) yielded a clean display, without visible artefacts.

In January 2007, I ran an AM listening test on my IC-756Pro3 at 25W resting carrier and 90% modulation , using a simple tuneable demodulator consisting of a military  ME-61/GRC field-strength meter connected to a good-quality headset . The recovered audio sounded smooth and clean, without audible artefacts. This tended to confirm the spectrum-analysis test.

Note on the RX-ANT Input (February 2006)

The RX-ANT input socket is intended for the connection of a separate receiving antenna such as a Beverage, loop or long-wire. Such antennas, if co-sited with the transmitting antenna, can induce extremely high voltages at the RX-ANT input during transmission. This is particularly true on 160m and 80m. Recently, damage to the front end of an IC-756Pro2 fed from a Beverage was reported on some forums.

On researching this topic a little further, I noted that the 756Pro and 756Pro2 RF Unit boards have a varistor (R111) to absorb excess RF voltage, and a fuse bulb (DS111) in the RF signal path from the RX-ANT jack to the ANT/ANT-R selection relay (RL101). Excessive RF power applied to the RX-ANT jack will blow DS111.

On the 756Pro3 RF-B Unit board, varistor R101 is still seen, but DS111 has been omitted. Interestingly enough, though, the 756Pro3 RX-ANT circuit incorporates an RF sensor (D101, Q103, Q104) which operates the selection relay (RL-101) when RF is applied to the RX-ANT socket, thereby breaking the RX-ANT signal path. This feature is lacking in the 756Pro2.

View partial schematics of the IC-756Pro3, IC-756Pro2 and IC-7800 RX-ANT input circuits.

It is strongly recommended that an external protective relay circuit such as the KD9SV Front End Saver be connected between the receive antenna feedline and the RX-ANT input.

Note on the ANT1 and ANT2 Inputs (January 2007)

Q: Are the ANT1 and ANT2 inputs both grounded when power is off?

A: No, this is not the case. If we follow the RF signal path on the CTRL-A Unit from ANT1 and ANT2 to the RFRX input (RF-B Unit), we see that when RL1, RL2, RL3 and RL4 are released, the signal from ANT1 is routed via RL3/1 NC, RL2 NC, RL1 NC, RL4 NC. L27 and L28 to J2 (RFRX). ANT2 is earthed to chassis via RL3/2 NC. (IC-756Pro3 Service Manual, p. 10-5).

All relays are released when power is off.  View partial schematic.

Note on the Calibration Marker (December 2006)

Q. When I turn on the Calibration Marker, the S-meter reads S9 + 14dB with the antenna disconnected and S9 + 11dB with the 40m antenna connected. Does this 3 dB difference mean that half the Marker power is going out the antenna?  Also, can I use the Marker to check S-meter calibration?

A. The marker amplitude as displayed on the spectrum scope will vary slightly between individual radios. One can use it as a rough check to see if the S-meter is reading correctly.

The marker signal is applied at the input of the splitter which feeds the incoming RF signal to the Main and Sub 1st mixers. The load impedance presented to the marker source changes slightly if the antenna is connected with no preamp in the signal path; this may account for the drop in amplitude with the antenna connected.

With Preamp 1 or 2 selected, there is no change, as the preamp buffers the marker injection point from the antenna input.

I measured the marker signal leakage from the antenna socket with a Millivac MV-723B broadband RF millivoltmeter terminated in 50Ω. The leakage is negligible with either preamp in-line. With  preamp off, marker on and f = 7.4 MHz, I read -48 dBm at ANT2. The readings at 3.0 and 30.0 MHz were both -52 dBm. This is the composite signal power of all the 100 kHz harmonic spikes passing through the BPF for the selected frequency range.

If you are concerned about this leakage, you can activate Preamp 1 or 2 whilst the marker is on.

References

  1. "IC-756Pro II Technical Report", Icom Inc., 2003. (PDF) mirror site  erratum

  2. ARRL IC-756Pro3 Product Review (QST, March 2005 - members only)

Links

To be continued as I learn more about the Pro3...

Copyright � 2005 A. Farson VA7OJ/AB4OJ. All rights reserved. 
(Copyright in contributed items reposes with the respective contributors.)
Last revised: October 17, 2007

Microsoft FrontPage