[Elecraft] K3 - FT8

[Alan Ibbetson]

Thanks for this, Gordon. I failed to search properly and missed that 
previous thread. Elecraft ought to at least acknowledge and document the 
issue even if they don't fix it. I did some googling and I now see that 
this is a well known and widespread problem. It seems to have a well 
known set of solutions, discussed for example here

https://en.wikipedia.org/wiki/Dither

So it does seem fixable in principle but it will involve engineering and 
business trade-offs which only Elecraft themselves can address.

73, Alan G3XAQ

On 14/05/2018 12:52, Gordon LaPoint wrote:
> The problem was discussed in this thread  last year.
> http://elecraft.365791.n2.nabble.com/K3-Odd-ALC-behavior-td7633052i20.html
> 
> Gordon - N1MGO
> On 05/14/2018 03:05 AM, Alan Ibbetson wrote:
>> Don,
>>
>> I'm puzzled you when you say
>>
>> "If it were a K3 problem, that aliasing would also be happening with
>> other  data modes as well. Other soundcard data modes do not have a
>> problem with a TUNE at 1000 Hz"
>>
>> I can reproduce this problem with my K3 with MMVARI on other data
>> modes. I have tried it with another PC and sound card and again I can
>> still see the problem. So I suspect that feeding a pure sine wave from
>> any source into LINE IN can cause ALC bargraph artefacts at certain
>> specific audio frequencies.
>>
>> This is not FT8 specific, nor is it audio harmonics or IMD. It is
>> simply an unfortunate choice of audio tone frequency interacting with
>> the K3 internals. Or at least that is my best guess from the evidence
>> I see here. The only constant throughout these tests is my K3. If it
>> is faulty then OK, no big deal.
>>
>> I hoped by reporting this effect on this mailing list I might help the
>> others who have also commented about it. If there is a consensus that
>> it's specific to my setup then OK. I can reliably avoid my problem by
>> changing the audio tone frequency just a few Hz so I will just do that
>> and get on with my life.
>>
>> 73, Alan G3XAQ
>>
>> On 13/05/2018 23:20, Don Wilhelm wrote:
>>> Alan,
>>>
>>> That is getting into issues associated with WSJT-X and FT8.
>>> My website article is written to apply to all soundcard data modes and
>>> not FT8 specifically.  There was no intention for addressing all the
>>> various data modes that could be used.  All of them require the right
>>> amount of audio drive.
>>>
>>> You should look to the FT8 forums for questions about that aliasing.  If
>>> it were a K3 problem, that aliasing would also be happening with other
>>> data modes as well.
>>> Other soundcard data modes do not have a problem with a TUNE at 1000 Hz,
>>> so why is FT8 any different?
>>>
>>> Of course, if your soundcard is generating harmonics or significant IMD
>>> before it reaches the K3, there is not much one can do other than
>>> getting a different soundcard.
>>>
>>> I don't know about the "moving 10kHz" thing, but I have heard reports of
>>> similar symptoms.
>>>
>>>
>>> 73,
>>> Don W3FPR
>>>
>>> On 5/13/2018 4:22 PM, G3XAQ wrote:
>>>> Don,
>>>>
>>>> I suggest you need to draw attention in your web page and general
>>>> advice on
>>>> here to the requirement to choose the AUDIO FREQUENCY with some care
>>>> when
>>>> setting up ALC for FT8 via the TUNE button in WSJT-X.
>>>>
>>>> The problem seems to be that the DSP sampling rate inside the K3
>>>> must not be
>>>> related to the TUNE frequency. Ideally I expect they should be
>>>> co-prime, or
>>>> something like that, to avoid aliasing artifacts in the ALC bargraph
>>>> readout. The actual frequency you should choose is probably affected
>>>> by PC
>>>> and K3 clock frequency accuracy becasue a change of just a few Hertz
>>>> can
>>>> have a dramatic effect.
>>>>
>>>> In my case, the default WSJT-X tune frequency of 1000Hz is about as
>>>> bad as
>>>> you can get. My ALC bargraph oscillates with a beat frequency of not
>>>> much
>>>> more than 1Hz. And the ALC goes from zero to 6 bars! What is really
>>>> confusing is that WSJT-X leaves the tune frequency at whatever it
>>>> was when
>>>> you last transmitted, so if you tune around, click on some signals,
>>>> have a
>>>> listen and then try using TUNE the audio frequency could be anywhere
>>>> unless
>>>> you take specific action.
>>>>
>>>
>>
> 

-- 

Alan Ibbetson
alan at g3xaq.net