I'm using the RF SpeechProcessor of Joachim Munch, DF4ZS, on my Yaesu FT-817 since about more than 1 1/2 year. The circuit is build into the MH-31 handmike and consists of a real 455 kHz SSB RF SpeechProcessor (RF Clipper), which has much less audio distortions than other circuits which clip directly on the AF side. Just see his homepage on http://www.jwm.de/afu/index0.htm for more informations.
I only had some first RF hum problems as I was on summer holidays this year and worked with an ATX-1080 or a MiracleWhip from the hotelroom. So here I like to tell you my totally and finally modifications for that great external RF SpeechProcessor.
See http://www.jwm.de/afu/0ft817eng.htm for details.
I fixed a wire to each inside of the MH-31 plastic case. The wires were stripped of their insulation on one end and these ends are only fixed with two dots of superfast-glue. Then the inside of the front and back cases are painted with silverfluid (from the model railway or electronic shop). Let the silverfluid dry for 3 - 5 hours. I fastened that up by laying both case sides into the summer sun on a hot sunny day. Finally the other ends of the front/back wires are soldered to the GND connection on the PCB.
Now we have a great and RFI stable metal case in our MH-31 handmike, even if the outsides are still plastic, of course, hihi.
To be sure not to produce an electronic shortening, I cutted all peaked ends of the solderings on the PCB and fixed some additionally isolating tape to some (maybe critital) areas.
On the DF4ZS PCB you can locate two pairs of antiparallel diodes (D1-2 and D3-4). One pair is on the parts side, the other pair is on the soldering side. These 4 diodes are 1N4148 types which produce a strong self-noisefloor. I replaced those 4 diodes with 1N5711 (or HP5082-2800), which have the same forward voltage of 0,7V but a much less noisefloor. Now you can hear the clipping effect by using the DF4ZS clipper without having a high background noise. The S/N ratio is much, much better now.
C3 (100nF) was replaced by 22µF ("+" to R1/R2, "-" to IC1-Pin1)
C4 (100nF) was replaced by 22µF ("+" to IC1-Pin2, "-" to Gnd)
C17 (2,2µF) was replaced by 10µF
This loweres the lower side of the AF highpass significant and let the basses come through. The sound is much fuller and more understandable without loosing the sharp "clipper" punch.
The alignment of my DF4ZS module:
P1 (Clipping degree) = 60% clockwise, shown from top
P2 (BFO) = 50% clockwise, shown from top
P3 (AF Output level) = 20% anti-clockwise, shown from side (like on his homepage)
Normally the module comes in a pre-aligned mode, equal to the FT-817 needs. If you don't like that and if you don't have the special measurement equipment, you can do the alignment like I did by using a second RX with headphones.
Reducing P3 to a real minimum AF output to prevent AF distortions.
P2 to middle.
Re-align P1. Turning clockwise will raise up the clipping degree. With my 1N5711 there would be distortions on a level more than 60%. Find out the edge of the highest clipping level without having too much distortion and where the audio signal is still clear and understandable. If this level is too low there wouldn't be a significant clipping and you loose the motive of this clipper. So you always should go on the highest possible and well-sounding edge.
Change P2 to have a darker or brighter sound characteristik. On the darkest end (100%) the heights were attenuated on my module as I still was not on the optimum bandpass area of the 455 kHz filter. So the middle (50%) was the optimum choice for me.
Raise up P2 until the final audio level is OK for your rig. If this level gets too high the FT-817 produces distortions on its own mic-amp, so please take care on that level. Too much is REALLY too much and won't sound good, hihi. On mine about 20% is really the upper end.
As shown in the block diagram I'm using the electret capsule on the "B"-switch position of the backward switch too to have a non-amplified or non-clipped option (for FM). But I removed the 680 ohms resistor.
You have 2 choices:
Still using the originally 680 ohms resistor. Then the electrolyt should have a value of 1µF or more. The originally "tone-switch" cap of 0,33µF is not a good choice cause it would produce an AF highpass on 700 Hz and above. The result would be a thin, high pitched modulation without any basses.
On this version a possible loud background would be reduced significant by the 680 ohms resistor and you have to speak close to the mike all the time. So this would be the better choice for high-noise environments.
Remove the 680 ohms resistor to get a real powerful and sensitive behaviour of the electret capsule. Then you should use an electrolyt of 0,1µF - 0,33µF instead to prevent a dark modulation. The mouth-to-mike distance can vary widely without loosing too much modulation output.
...so now you know M Y choice, hihi.... I'm using a 0,22µF coupling electrolyt.
...Many words, long description. But the best thing is to hear it "in action".
So now it's time and here's the recorded WAV file. The file was recorded on my FT-847 while listening on a background QSO with about S5 to S7. The FT-817 was connected to the dummyload and had produced nearly the same signal level on the FT-847, so both audio signals are like a multi-station listening.
So let's hear how the DF4ZS breaks through the noise. Download the WAV file.