The cause of (and cure for) the spurious emissions from an FT847

by Nige Coleman G7CNF

The strongest of the spuruous signals previously described is at 45.580 MHz, and does not change in frequency or amplitude as the radio is tuned. This in fact a transmit intermediate frequency, which is leaking through the buffer and bandpass stages into the PA, and comes out only 3dB down on the wanted signal.
In this radio, the transmit signal is fed via one of three bandpass filters to a buffer before the PA. These bandpass filters are for HF, 50 and 54-76MHz. The 6m filter comprises a series of four cans plus varicap diodes, but The 54-76MHz filter is much simpler, and appears to have a low Q.

Information subsequently received suggests that the UK "C2" modified sets are fitted with a different filter to reduce the problems described below.
See this note for further details.

Herein lies the secret of the 847!
The next spectrum was taken with the radio (fully widebanded) TX'ing on 53.999.999MHz.

It is similar to that of the earlier 50.2MHz sweep, but if the dial is moved up in frequency just 1Hz to 54.000MHz, the broadband 54-76MHz BPF is switched in. Now look at the output spectrum:

Bingo, there is the offender!
The 54-76MHz broadband BPF appears to be leaking other signals through, which then intermodulate in the amplifiers, producing spurii at regular intervals.

Improving the bandpass filter

I mounted two 20pF trimmers in parallel with L3024 and L3029, to inject a little variable reactance into the circuit.

Within moments of trimming, I had the radio set on 70.2 putting out 100W with a PA current of 16A! Just the ticket I thought, but that was not the end of the story.
The trouble was that all the RF output was on 45.580MHz! I had successfully attenuated the 70MHz output, leaving only the original LO injection component!
This illustrates the importance of watching the output spectrum in combination with the current and RF power when making any adjustments, to see exactly what is going on.

Setting the rig to 10W output on the broadband power meter, I could tune the sprogs in and out, the wanted 70MHz component in and out, and watch the PA current vary from 6 to 13 amps (since the PA was amplifying the spurious signals along with the wanted output). Irrespective of the dominant output frequency, the power output remained more-or-less constant at 10W on the broadband meter.

By now it is pretty clear that the cause of the FT-847's spurii is the inadequate filtering of the signal being injected into the PA. I made some effort to 'shunt' the unwanted frequencies away, as follows:
C3082 and C3097 were temporarily replaced with the same 20pF trimmers that were used in the previous section, with a view to crudely altering the bandpass characteristics as widely as possible with the least invasion; I wanted an operational radio at the end of this investigation!

The effect of adding the trimmers at this point was to empower me with almost full control over the spurious emissions, to such an extent that I was able to eliminate all except the TXLO sprog, however even this was attenuated to a more realistic 30dB below the wanted output. Unfortunately this also caused excessive attenuation to the desired 70MHz signal: the output was somewhere less than ½ watt!

These two little trimmers were able to alter the PA current (with RF Power fully clockwise) from 3.5A to the nominal >22A that is often seen. Over that range, it is painfully clear from the spectrum analyser that all this energy is being used to amplify the dozens of spurii which the 54-76MHz BPF stage fails to remove. These are then either transmitted out of the back of the radio or lost in the low-pass filter, as heat.

The logical conclusion was to break the RF path at L3024/L3029 and install a more selective bandpass stage, perhaps with a small buffer amplifier. However, I was conscious that fitting some fiddly surface-mount components may not easy for many amateurs, so with this thought in mind, I set out to find a solution that almost anyone can replicate.
And here it is....

Cleaning up the FT847: the modification

Get a 2.2nF capacitor, preferably one with a close tolerance, or if possible have someone measure the value for you.
Cut its legs to exactly 8mm: this is fairly critical, to achieve the correct self-resonance of the capacitor.
Carefully scrape away a small area of resist on the ground plane of the PCB, just above L3024, as pictured, and apply a reasonable amount of solder to it, being careful not to damage the plated-though holes.
Now, with the radio dial facing you, lay the capacitor onto the PCB, with the capacitor disc on the left and the legs pointing right.
Solder one leg to the PCB ground, and the other leg to the left leg of L3024.
Note that in my efforts to produce a 'tidier' job by situating the capacitor elsewhere, I was unable to come close to the results obtained from seating the capacitor exactly as shown below!

That's all there is to it! All the spurii were removed or considerably reduced to a satisfactory level. This, in combination with Keith G4FUF's inductor mod has resulted in a clean 20W from the back of the radio, with a PA current of the order of 16A (which is far from ideal, but is a universe away from the utter rubbish that the radio was emitting prior to this mod with worse efficiency).
The TXLO output on 45.580MHz which was only -3dBc is now reduced to -55dBc, and all the other sprogs are even lower.

Happy 4 metre-ing to all FT847 owners, be good - BE CLEAN!