A further study of the spurious emissions from an FT-847 on 70MHz
by Nige Coleman G7CNF
As the Four Metre Band is gaining in popularity, with new EU allocations surfacing from time to time, the Yaesu FT-847 is becoming a much sought-after set, due to its ability to operate multimode on 4m.
With this thought in mind, I felt it pertinent to share my recent research on my FT847 with the 4m community.
It has long been known that there are problems with using modified FT847s on 70MHz, but the extent to which their spurious emissions may cause a nuisance is not so obvious. These findings came about when I bought a new "UK spec" FT-847 (with 4m included as standard) from a well-known emporium. Naturally, as I had bought a commercial amateur-band radio, I only undertook very basic spurious emissions testing with an absorption wavemeter (after all, it was a UK spec radio, so it must be OK, surely??).
The results were more-or-less satisfactory, but left some unanswered questions, so in recent weeks, I have thoroughly re-tested my FT847 with a spectrum analyser.
What I am about to reveal came as quite a shock and has really worried me, as there are quite a number of these radios on the air. I have checked and double-checked these results, and my
test gear, and would issue the warning:
use an unmodified 847 on 4m at your own risk: remember your licence conditions, you may well be in breach of them!
Details of the tests
All the tests were performed on a "UK spec" Yaesu FT-847, manufactured in December 2004 (serial 4N, batch 96). The initial readings were taken with the radio "fully widebanded" (as per the internet mods, to allow transmission on the 60m band). (Details of the test equipment used can be found at the bottom of this page.)
It has been thought that the 847's thirst for current on 4m has been entirely down to poor PA efficiency (i.e. converting the surplus DC energy into heat), but these test results are much more revealing.
The first chart shows how the current drawn by the radio when transmitting at 10W varies as the frequency is increased from 37MHz to 72MHz.
Next follows a chart which shows how the PA current varies as a function of the final (broadband) output power: note how steeply the 4m plot rises. (The current values in the chart have taken into account the ambient current drawn by the other radio functions, so the values plotted here have had 1.2A subtracted from the actual readings.)
The pink 70.2MHz plot stops at 30w: this is the maximum output power obtained from this particular radio on 4m during the tests, without the inductor mod.
The PA efficiency can be calculated from measurements of the RF output and DC input powers.
When transmitting 100W on 6m, the efficiency is 45%.
Reducing to 30W output on 6m, the efficiency falls to 23.5%.
However, for the same 30W output on 4m, the efficiency is a mere 10.3%.
Why is this? The spectrum analyser may provide some answers.
(All the measurements were taken with an output power of 10W, with the radio terminated directly into a 50 ohm load. A variable attenuation coupler samples the signal for the spectrum analyser.)
The first plot is a 'control' reading at 50.2MHz: the 20MHz span shows a nice clean signal with a couple of sidebands which are around -70dB relative to the fundamental. No cause for concern here, as to be expected.
The next trace has a 500MHz span, so that the fundamental is just visible to the left of the left-most graticule and the 2nd to 5th harmonics are clearly visible: the 3rd harmonic is the worst at -55dBc.
The next trace, for comparison, has the same sweep, but this time the radio is TX'ing on 70.200MHz.
See the difference? Pay particular attention to the 'mess' on the far left of the display.
The tallest line represents 70.2MHz, and each harmonic has a companion, very close to it in terms of frequency.
Taking a closer look at the lower end of this spectrum:
This represents the spectrum from DC to 100MHz. Shocking isn't it?
The most prominent spurious emission is at 45.6MHz, and is only 3dB down on the
Other spurii can be observed at the following frequencies:
3.5, 7.4, 17.6, 21.0, 24.5, 27.9, 42.0, 49.0, 52.4, 66.6, 73.5, 94.6 and 115.6 MHz
And that's all before the 2nd harmonic!
In fact, each sprog is accompanied by its own pair of sidebands. I do not believe I have seen anything like this in my >20 years in the hobby, or professionally.
At this point I recalled Keith G4FUF's previous comments about the boot condition of the CPU and the shift registers.
Since the radio under test (although a UK spec. model) had been modifed for widebanded transmit, to allow transmission on the 60m band, I wondered if the wrong filters were being selected.
I spent the next day or so altering and reconfiguring the jumpers and recording the results, both in terms of band limits, and and effects these might have on the spurious emissions.
As there are 6 jumpers, there are 64 possible combinations, and I wanted to know what all of these did.
I can confirm that on a December 2004 production FT-847, the position of the jumpers makes absolutely no difference to the band-pass filter selection (each alteration of the shift register was
accompanied by a full CPU reset, as on this radio, the alteration of the registers had no effect until a reset was performed).
Now read on: there are more surprises to come in
part two, plus a solution to the spurious emissions problems.
Test equipment used
Daiwa broadband power meter (calibrated against Bird Thruline at 50/70MHz)
Anritsu spectrum analyser
Agilent sampling coupler
Rymsa 50 ohm load #937
Metex M-3650B DVM.
Nissei DPS-300GL 30A variable voltage PSU, set to 13.800V.
(All traceable calibration)
Transmissions were in FM mode at 70.2 and 50.2MHz, with the microphone removed