The distortion artifacts were subtle but noticeable. They sounded like a slight "crunching" or "crackling" sound (very noticeable when someone says "ahhh."), and could be made to stand-out (when looking for its presence) by rotating the "Shift" knob to accentuate high frequencies. With the shift in its normal position, the distortion was still present, but it tended to be masked (in most cases) by the higher level of the voice signal.
(One quick test I use for locating the cause of audio distortion is to reduce a receiver's RF Gain. If the audio sounds clearer with less RF gain, then, in my experience, there's a very good chance that AGC action is creating the distortion.)
I reduced the FT-1000D's RF Gain. The the audio sounded clearer. Ah ha! There was a very good chance that the distortion, therefore, was related to AGC action, and so I started experimenting...
My main concern was SSB operation (for which I usually use either SLOW or MEDIUM AGC rates). Looking at the schematic, I noticed that there was a 10K resistor (R2140) in series with the 2.2 uF cap used for Slow AGC. Shorting-out this resistor reduces SSB distortion. But it will distort the CW envelope. My feeling was that I never use Slow AGC for CW, so this was an acceptable compromise to make.
Even the 4066 analog switch introduced some distortion artifacts (per its datasheet, its resistance is about 500 ohms for a 10V supply voltage). I replaced this analog switch with a relay, which has a much lower "ON" resistance.
I also replaced the analog switch used for switching in the MEDIUM AGC circuit (the 0.47 uF cap) with a relay. Experiments for FAST and MEDIUM AGC settings revealed that, for CW, a 30K resistor actually worked better than the original 10K ohm resistors used for these time-constants. So I added a 30K resistor (replacing the two 10K resistors) that is common to both the FAST and MEDIUM caps (0.22 and 0.47 uF).
However, the 30K does produce distortion artifacts on SSB. So I decided to compromise: only for strong signals is the 30K ohm resistance switched in -- for normal or low-level signals, the resistance is very low. (For low-level signals, the AGC line sits high. This turns on a 2N2222 which in turn shorts out the 30K resistor. For strong signals, the AGC line is driven lower, until, for very strong signals, it drives the 2N2222 into cut-off (threshold set using the two series-diodes attached to the base), which then places the 30K back into the circuit.)
I found that the 30K also increases distortion on AM signals, so I also short it out for AM operation. (Therefore, it is only present for strong CW, SSB, and, I suppose, FM (it has no effect on FM)).
Here's my original markup of the FT-1000D schematic:
Here's a drawing that might be a bit clearer...
1. If one would like to keep the changes simple and not incorporate all of the mods that I made, I would recommend the following two modifications:
- Short out R2140 (10K ohms: Slow AGC.)
- If you use Medium AGC for SSB, then also short out R2141 (10K ohms: Medium AGC. But note that this may create some fuzziness on CW signals.)
3. I used Clare DSS41A05 Relays.
4. Audio envelope testing was performed by modulating the RF signal from an HP 8640B (in AM mode) with a pulse generator (100ms pulse, rep rate of 1 second, 2 ms rise/fall, 1.4v peak, -1.1v DC offset). The AGC waveform was monitored @ TP2005, and the audio envelope monitored at the headphone jack.
5. Diodes are in series with the base of the transistors to ensure that they fully turn off when the control signals go low.
I made these modifications to suit the style of operating that I prefer (SSB ragchewing). These modifications may not be suitable for your style of operation. So, if you're also experiencing annoying distortion, please consider these mods to be a starting-point for your own experiments in improving the performance of the 1000D.