Elecraft › [K3]

K3 Wide AGC threshold

Classic

List

Threaded

5 messages Options

Bob Henderson-2

K3 Wide AGC threshold

As a new K3 owner, it is entirely possible I am treading ground already well
trodden. My apologies should this be so.

While waiting for delivery of an Inrad 500Hz roofing filter, I have been
accumulating experience with S/N 4904. I am enjoying my new rig greatly,
although I confess to finding an odd thing or two puzzling. One of these
relates to the level at which wide AGC has been set.

My primary shack transceiver is a home built, G3XJP designed Picastar, which
has a similar architecture to the K3 but is based upon the much older Analog
Devices ADSP2181 and AD1885 AC'97 codec. Unlike the K3 with its multiple
choice roofing filter bandwidths, my Picastar uses a single 4kHz wide roofer
in all modes. The thing puzzling me is that I find de-sensing by strong
signals within roofer b/w but outside DSP b/w far more noticeable on the K3
than on Picastar. I expected the opposite to be true given the Picastar
codec has only a 16 bit ADC whereas the K3 has 24 bit.

A quick measurement reveals K3 wide AGC threshold around 70dB above MDS
whereas on Picastar it is closer to 80dB. With Star's 16 bit ADC this
equates to around 10dB headroom whereas the K3's 24 bit ADC implies headroom
in excess of 30dB. On the face of things it would appear some 20dB of extra
defence against strong signals in the no man's land between roofer b/w and
dsp b/w has been sacrificed. I am wondering why? Perhaps I am missing
something?

Thanks.

Bob, 5B4AGN
______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

Guy, K2AV

Re: K3 Wide AGC threshold

That's not the "wide" AGC. It's the defensive hardware AGC that is in
the actual IF string after the roofing filter before the analog to
digital converter (ADC). It's function is to limit the level
presented to the ADC as you surmise. It is also primitive, easily
seen in the schematics. Handling strong signals without any AGC
pumping, using an SSB roofer ahead of a CW DSP bandwidth, I doubt was
a design goal of the RX. What you experience is common for those
without a CW roofer.

There was some variation in how the K3 handled this which was up in
beta firmware very early on, and various settings were evaluated by
the community. I'm sure you can find the blow-by-blow back in the
archives.

73, Guy.

On Fri, Nov 19, 2010 at 10:17 AM, Bob Henderson <[hidden email]> wrote:

> As a new K3 owner, it is entirely possible I am treading ground already well
> trodden. My apologies should this be so.
>
> While waiting for delivery of an Inrad 500Hz roofing filter, I have been
> accumulating experience with S/N 4904. I am enjoying my new rig greatly,
> although I confess to finding an odd thing or two puzzling. One of these
> relates to the level at which wide AGC has been set.
>
> My primary shack transceiver is a home built, G3XJP designed Picastar, which
> has a similar architecture to the K3 but is based upon the much older Analog
> Devices ADSP2181 and AD1885 AC'97 codec. Unlike the K3 with its multiple
> choice roofing filter bandwidths, my Picastar uses a single 4kHz wide roofer
> in all modes. The thing puzzling me is that I find de-sensing by strong
> signals within roofer b/w but outside DSP b/w far more noticeable on the K3
> than on Picastar. I expected the opposite to be true given the Picastar
> codec has only a 16 bit ADC whereas the K3 has 24 bit.
>
> A quick measurement reveals K3 wide AGC threshold around 70dB above MDS
> whereas on Picastar it is closer to 80dB. With Star's 16 bit ADC this
> equates to around 10dB headroom whereas the K3's 24 bit ADC implies headroom
> in excess of 30dB. On the face of things it would appear some 20dB of extra
> defence against strong signals in the no man's land between roofer b/w and
> dsp b/w has been sacrificed. I am wondering why? Perhaps I am missing
> something?
>
> Thanks.
>
> Bob, 5B4AGN
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:[hidden email]
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
>

______________________________________________________________
Elecraft mailing list
Home: http://mailman.qth.net/mailman/listinfo/elecraft
Help: http://mailman.qth.net/mmfaq.htm
Post: mailto:[hidden email]

This list hosted by: http://www.qsl.net
Please help support this email list: http://www.qsl.net/donate.html

Bob Henderson-2

Fwd: K3 Wide AGC threshold

Guy

Thank you for putting me straight on the terminology. Perhaps I can
rephrase my question.........

The hardware AGC appears to be set at a level which precludes effective
utilisation of the 100dB+ DR of the DSP system. This seems to suggest the
existence of something between roofing filter and ADC which needs protecting
due to it having rather less than 100dB+ DR?

73 Bob, 5B4AGN

On 19 November 2010 15:44, Guy Olinger K2AV <[hidden email]> wrote:

> That's not the "wide" AGC. It's the defensive hardware AGC that is in
> the actual IF string after the roofing filter before the analog to
> digital converter (ADC). It's function is to limit the level
> presented to the ADC as you surmise. It is also primitive, easily
> seen in the schematics. Handling strong signals without any AGC
> pumping, using an SSB roofer ahead of a CW DSP bandwidth, I doubt was
> a design goal of the RX. What you experience is common for those
> without a CW roofer.
>
> There was some variation in how the K3 handled this which was up in
> beta firmware very early on, and various settings were evaluated by
> the community. I'm sure you can find the blow-by-blow back in the
> archives.
>
> 73, Guy.
>
> On Fri, Nov 19, 2010 at 10:17 AM, Bob Henderson <[hidden email]> wrote:
> > As a new K3 owner, it is entirely possible I am treading ground already
> well
> > trodden. My apologies should this be so.
> >
> > While waiting for delivery of an Inrad 500Hz roofing filter, I have been
> > accumulating experience with S/N 4904. I am enjoying my new rig greatly,
> > although I confess to finding an odd thing or two puzzling. One of these
> > relates to the level at which wide AGC has been set.
> >
> > My primary shack transceiver is a home built, G3XJP designed Picastar,
> which
> > has a similar architecture to the K3 but is based upon the much older
> Analog
> > Devices ADSP2181 and AD1885 AC'97 codec. Unlike the K3 with its multiple
> > choice roofing filter bandwidths, my Picastar uses a single 4kHz wide
> roofer
> > in all modes. The thing puzzling me is that I find de-sensing by strong
> > signals within roofer b/w but outside DSP b/w far more noticeable on the
> K3
> > than on Picastar. I expected the opposite to be true given the Picastar
> > codec has only a 16 bit ADC whereas the K3 has 24 bit.
> >
> > A quick measurement reveals K3 wide AGC threshold around 70dB above MDS
> > whereas on Picastar it is closer to 80dB. With Star's 16 bit ADC this
> > equates to around 10dB headroom whereas the K3's 24 bit ADC implies
> headroom
> > in excess of 30dB. On the face of things it would appear some 20dB of
> extra
> > defence against strong signals in the no man's land between roofer b/w
> and
> > dsp b/w has been sacrificed. I am wondering why? Perhaps I am missing
> > something?
> >
> > Thanks.
> >
> > Bob, 5B4AGN
> > ______________________________________________________________
> > Elecraft mailing list
> > Home: http://mailman.qth.net/mailman/listinfo/elecraft
> > Help: http://mailman.qth.net/mmfaq.htm
> > Post: mailto:[hidden email]
> >
> > This list hosted by: http://www.qsl.net
> > Please help support this email list: http://www.qsl.net/donate.html
> >
>
>

Bill W4ZV

Re: Fwd: K3 Wide AGC threshold

Bob Henderson-2 wrote

The hardware AGC appears to be set at a level which precludes effective
utilisation of the 100dB+ DR of the DSP system.

Hi Bob...spoken like a former Orion user! You're absolutely correct...but it's actually much better than it was originally. Why? I have no clue.

73, Bill

Hardware AGC threshold change (HAGC Mod Kit)

http://www.elecraft.com/K3/k3_app_notes.htm (scroll nearly to the bottom)

SMD HAGCMDKT instructions

Both through hole and SMD parts are included. Users can install either set of parts.

Some users prefer tuning the band with a wide filter bandwidth selected (for example, using a 2.7 or 2.8 kHz SSB filter in CW mode rather than a 500 Hz or narrower filter). Strong off-frequency signals (above about S9+5) that are inside the crystal filter passband can then activate the hardware AGC circuit, reducing the level of desired signals.

This modification raises the hardware AGC threshold from about S9+5 to about S9+25 to +30. The new threshold still protects the DSP's analog-to-digital converter, while providing better receive performance when wider crystal filters are used. This is not as significant an issue when using narrow CW filters (500 Hz and below).

Guy, K2AV

Re: Fwd: K3 Wide AGC threshold

There was some writeup I seem to remember that there were some
conditions where the lower threshold worked better. If one does NOT
PREFER using wide bandwidth to tune CW, the current stock setting IS
better for some reason that I can't give you.

My CW contesting use makes wide filters simply useless for CW. There
is ALWAYS a 35 over S9 signal up 350 Hz from me, no doubt due to some
natural law related to entropy. To get him out of my bandpass and
copy the calling S minus 3 QRP basement noodle antenna station that
accidentally disconnected his noodle and doesn't need it yet, I need
BOTH the DSP and the narrow filter. Your current filter situation
would be simply hopeless. You will love it when you finally get it.

73, Guy.

On Fri, Nov 19, 2010 at 4:50 PM, Bill W4ZV <[hidden email]> wrote:

>
>
> Bob Henderson-2 wrote:
>>
>> The hardware AGC appears to be set at a level which precludes effective
>> utilisation of the 100dB+ DR of the DSP system.
>>
>
> Hi Bob...spoken like a former Orion user! You're absolutely correct...but
> it's actually much better than it was originally. Why? I have no clue.
>
> 73, Bill
>
> Hardware AGC threshold change (HAGC Mod Kit)
>
> http://www.elecraft.com/K3/k3_app_notes.htm (scroll nearly to the bottom)
>
> SMD HAGCMDKT instructions
>
> Both through hole and SMD parts are included. Users can install either set
> of parts.
>
> Some users prefer tuning the band with a wide filter bandwidth selected
> (for example, using a 2.7 or 2.8 kHz SSB filter in CW mode rather than a 500
> Hz or narrower filter). Strong off-frequency signals (above about S9+5) that
> are inside the crystal filter passband can then activate the hardware AGC
> circuit, reducing the level of desired signals.
>
> This modification raises the hardware AGC threshold from about S9+5 to about
> S9+25 to +30. The new threshold still protects the DSP's analog-to-digital
> converter, while providing better receive performance when wider crystal
> filters are used. This is not as significant an issue when using narrow CW
> filters (500 Hz and below).
>
> --
> View this message in context: http://elecraft.365791.n2.nabble.com/K3-Wide-AGC-threshold-tp5755595p5756731.html
> Sent from the Elecraft mailing list archive at Nabble.com.
> ______________________________________________________________
> Elecraft mailing list
> Home: http://mailman.qth.net/mailman/listinfo/elecraft
> Help: http://mailman.qth.net/mmfaq.htm
> Post: mailto:[hidden email]
>
> This list hosted by: http://www.qsl.net
> Please help support this email list: http://www.qsl.net/donate.html
>