[Elecraft] FTDX5000 Design Flaw

[Wayne Burdick]

Jim,

Thanks for your analysis of transmitted phase and keying noise of various transceivers. I'd like to provide a couple of technical details, and to give credit where it is due.

To achieve the excellent results shown in your paper for the K3, we used a combination of four techniques:

1. A hybrid PLL/DDS synthesizer with a very high C/L ratio VCO: Many of the radios shown in your plots use unfiltered DDS for their VFO, or in the case of PLL designs, VCOs with a much higher C/L ratio than the K3. Both of these design decisions can increase phase noise and dynamic artifacts. To preserve a consistently high C/L ratio, we use up to 128 different C/L combinations as the VCO is band-switched. (This is in contrast to the usual 1, 2, or 3 VCO C/L ranges used in other rigs.) We also used very narrow-band crystal filtering of the DDS output to completely remove any of the usual spurs due to quantization, etc. John Grebenkemper assisted greatly with the synthesizer design.

2. Transmit ALC with long time-constant, pre-calculated power calibration, and virtually no dynamic artifacts: The K3's transmit ALC is, in effect, open-loop in relation to keying waveform rise/fall timing. In other words, we do not apply power corrections over short periods, since this can distort the keying envelope. This is especially noticeable in CW mode but applies to T/R switching in other modes as well.

3. Conversion to a low I.F., with both RX and TX signals running through a narrow crystal filter: This band-limits I.F. noise from early stages (DSP and D-to-A converter) and helps establish a very low transmit noise floor ahead of the main mixer. Even in speech modes, all fast ALC is applied ahead of the crystal filter, which I believe is unique to the K3.

4. Sigmoidal keying waveform applied at the DSP: Our DSP engineer (Lyle Johnson, KK7P) studied the sidebands resulting from various sigmoidal and raise-cosine modulation envelopes. He selected the one that provided the smallest keying bandwidth consistent with rise and fall times of approximately 4 ms.

One clarification. The reason we created the K3's "QRQ mode" was to provide faster full break-in at very high code speeds (up to 100 WPM). The keying dynamics and phase noise are the same with QRQ mode either on or off; they are not "soft" with QRQ mode on. The KPA500 amplifier also switches very fast, so it is fully compatible with either mode.

73,
Wayne
N6KR


On Aug 31, 2014, at 3:40 PM, Jim Brown <jim at audiosystemsgroup.com> wrote:

> On Sat,11/20/2010 12:48 PM, John, KI6WX wrote:
>> The ARRL is aware of this issue.  This problem was discussed in my 1988
>> articles on phase noise in QST (March & April issues).  The article notes
>> that a high phase noise transmitter degrades the performance of a low phase
>> noise receiver to that of the transmitter.  This article started the ARRL
>> making composite transmit noise measurements.
> 
> I've recently added fuel to the fire with an online document plotting ARRL TX data for multiple radios on the same graphs, making comparison's easier. I did that by taking data point by point off of ARRL's published plots, then putting them in my own spreadsheet. I did that for both keying spectra and phase noise.
> 
> Seeing that document, Bob Allison sent me the raw data for the CW keying spectra for those radios, and I've been processing and plotting that data. He was not able to provide the phase noise data in electronic form, so we're stuck with what I am able to eyeball from the published plots.
> 
> You can see an "in-progress" version at k9yc.com/TXNoise.pdf
> 
> 73, Jim K9YC