K3S (SDR/superhet hybrid) vs. entry-level direct-sampling radios

[wayne burdick]

Hi all,

Over the past two days I’ve been deluged with requests for our K3S vs. IC7300 comparison chart. This stems from a posting that mentioned the latter rig, and my subsequent offer to send out the chart, which we normally use for in-house training purposes.

What’s become clear from all the questions and comments is that there’s a need for clarification on receiver architecture.

The K3S, an SDR/superhet hybrid, includes narrowband protection of its A-to-D converters in the form of roofing filters (crystal filters). This is fundamentally different from the approach taken by direct-sampling radios, which have only very broadband filtering ahead of their ADCs. Typically, front-end band-pass filters are 0.5 to 4 MHz wide. Their ADCs will be impacted by all strong signals in this range, alone and in summation.

When the ADC over-ranges in this “pure” SDR architecture, the radio usually reduces its gain automatically by either turning off preamps or adding attenuation. This increases the noise figure, often resulting in the loss of weak signals, both audibly and in the panadapter. Another word for it is “desense.” You might hear a pulsing sound as the noise floor goes up and down in response to a particular signal, or the sensitivity may degrade for many seconds at a time while the firmware waits for levels to drop at the input to the ADC.

Such over-ranging is almost unheard of in the K3S, for multiple reasons. First, we use very strong mixers and gain stages. Second, in many cases there’s no need to turn the preamp on at all. (Example: the K3S’s preamp-off sensitivity is typically -135 dB, while the preamp-off sensitivity on a typical direct-sampling SDR is -115 to -125 dB.) The third reason for the K3S’s strong-signal performance is its crystal roofing filters. These protect the ADC from not only wide-spaced signals, but also from signals very close by. As K3 and K3S owners will attest, you can have a huge signal just a few hundred Hz away and not even know it’s there -- unless that station’s transmit phase noise is blanketing the band anyway. (There’s no defense against an unclean or clicky transmit signal.)

I’ve updated the referenced comparison chart to clarify this important difference. Our webmaster will put it up next week.

Meanwhile, I’ll continue to send it on request (email me directly). Feel free to share the information with anyone interested in the topic.

73,
Wayne
N6KR




______________________________________________________________
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
Message delivered to [hidden email]

[Gary K9GS]

Thanks for sending Wayne.   To take this discussion a step further,  is the Flex series of radios in the same category as the 7300?


73,
Gary K9GS
-------- Original message --------From: Wayne Burdick <[hidden email]> Date: 7/30/17  11:14 AM  (GMT-06:00) To: Elecraft Reflector <[hidden email]> Subject: [Elecraft] K3S (SDR/superhet hybrid) vs. entry-level
  direct-sampling radios
Hi all,

Over the past two days I’ve been deluged with requests for our K3S vs. IC7300 comparison chart. This stems from a posting that mentioned the latter rig, and my subsequent offer to send out the chart, which we normally use for in-house training purposes.

What’s become clear from all the questions and comments is that there’s a need for clarification on receiver architecture.

The K3S, an SDR/superhet hybrid, includes narrowband protection of its A-to-D converters in the form of roofing filters (crystal filters). This is fundamentally different from the approach taken by direct-sampling radios, which have only very broadband filtering ahead of their ADCs. Typically, front-end band-pass filters are 0.5 to 4 MHz wide. Their ADCs will be impacted by all strong signals in this range, alone and in summation.

When the ADC over-ranges in this “pure” SDR architecture, the radio usually reduces its gain automatically by either turning off preamps or adding attenuation. This increases the noise figure, often resulting in the loss of weak signals, both audibly and in the panadapter. Another word for it is “desense.” You might hear a pulsing sound as the noise floor goes up and down in response to a particular signal, or the sensitivity may degrade for many seconds at a time while the firmware waits for levels to drop at the input to the ADC.

Such over-ranging is almost unheard of in the K3S, for multiple reasons. First, we use very strong mixers and gain stages. Second, in many cases there’s no need to turn the preamp on at all. (Example: the K3S’s preamp-off sensitivity is typically -135 dB, while the preamp-off sensitivity on a typical direct-sampling SDR is -115 to -125 dB.) The third reason for the K3S’s strong-signal performance is its crystal roofing filters. These protect the ADC from not only wide-spaced signals, but also from signals very close by. As K3 and K3S owners will attest, you can have a huge signal just a few hundred Hz away and not even know it’s there -- unless that station’s transmit phase noise is blanketing the band anyway. (There’s no defense against an unclean or clicky transmit signal.)

I’ve updated the referenced comparison chart to clarify this important difference. Our webmaster will put it up next week.

Meanwhile, I’ll continue to send it on request (email me directly). Feel free to share the information with anyone interested in the topic.

73,
Wayne
N6KR




______________________________________________________________
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
Message delivered to [hidden email]
______________________________________________________________
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
Message delivered to [hidden email]

[wayne burdick]

> GaryK9GS <[hidden email]> wrote:
>
> Thanks for sending Wayne.   To take this discussion a step further,  is the Flex series of radios in the same category as the 7300?


Flex radios do use a direct-sampling architecture (with no roofing filters). The degree to which they’re subject to ADC over-range, and what they do about it in firmware, has been the subject of ongoing debate. It may be the case that, generally, the more you pay for a direct-sampling radio, the better the A-to-D converters, and the higher the ADC overrange threshold.

That said, when measured with preamp off, the Flex 6700 has about 7 dB less IMD dynamic range than the K3S (see note Y in Sherwood’s chart), and 20 dB less blocking dynamic range. Both are direct consequences of not having narrowband roofing filters.

On the other hand, the Flex radios have a very broad panadapter display for those who require it.

73,
Wayne
N6KR

>
>
> 73,
> Gary K9GS
> -------- Original message --------From: Wayne Burdick <[hidden email]> Date: 7/30/17  11:14 AM  (GMT-06:00) To: Elecraft Reflector <[hidden email]> Subject: [Elecraft] K3S (SDR/superhet hybrid) vs. entry-level
> direct-sampling radios
> Hi all,
>
> Over the past two days I’ve been deluged with requests for our K3S vs. IC7300 comparison chart. This stems from a posting that mentioned the latter rig, and my subsequent offer to send out the chart, which we normally use for in-house training purposes.
>
> What’s become clear from all the questions and comments is that there’s a need for clarification on receiver architecture.
>
> The K3S, an SDR/superhet hybrid, includes narrowband protection of its A-to-D converters in the form of roofing filters (crystal filters). This is fundamentally different from the approach taken by direct-sampling radios, which have only very broadband filtering ahead of their ADCs. Typically, front-end band-pass filters are 0.5 to 4 MHz wide. Their ADCs will be impacted by all strong signals in this range, alone and in summation.
>
> When the ADC over-ranges in this “pure” SDR architecture, the radio usually reduces its gain automatically by either turning off preamps or adding attenuation. This increases the noise figure, often resulting in the loss of weak signals, both audibly and in the panadapter. Another word for it is “desense.” You might hear a pulsing sound as the noise floor goes up and down in response to a particular signal, or the sensitivity may degrade for many seconds at a time while the firmware waits for levels to drop at the input to the ADC.
>
> Such over-ranging is almost unheard of in the K3S, for multiple reasons. First, we use very strong mixers and gain stages. Second, in many cases there’s no need to turn the preamp on at all. (Example: the K3S’s preamp-off sensitivity is typically -135 dB, while the preamp-off sensitivity on a typical direct-sampling SDR is -115 to -125 dB.) The third reason for the K3S’s strong-signal performance is its crystal roofing filters. These protect the ADC from not only wide-spaced signals, but also from signals very close by. As K3 and K3S owners will attest, you can have a huge signal just a few hundred Hz away and not even know it’s there -- unless that station’s transmit phase noise is blanketing the band anyway. (There’s no defense against an unclean or clicky transmit signal.)
>
> I’ve updated the referenced comparison chart to clarify this important difference. Our webmaster will put it up next week.
>
> Meanwhile, I’ll continue to send it on request (email me directly). Feel free to share the information with anyone interested in the topic.
>
> 73,
> Wayne
> N6KR
>
>
>
>
> ______________________________________________________________
> 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
> Message delivered to [hidden email]
> ______________________________________________________________
> 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
> Message delivered to [hidden email]

______________________________________________________________
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
Message delivered to [hidden email]

[k4vd]

Concerning Flex, one of the engineers posted a pretty detailed explanation
of how they handle strong signals in a wide band environment. This stuff is
still a bit over my head but owning both the KX3 and the Flex-6500 I can
say, I am never disappointed with either receiver in any environment. I
kind of feel like the luckiest dog in the world to have both.

I hope it is OK to post the link:

https://community.flexradio.com/flexradio/topics/adc-overload-myths-debunked

73,
Kev K4VD



On Sun, Jul 30, 2017 at 12:27 PM, GaryK9GS <[hidden email]> wrote:

> Thanks for sending Wayne.   To take this discussion a step further,  is
> the Flex series of radios in the same category as the 7300?
>
>
> 73,
> Gary K9GS
> -------- Original message --------From: Wayne Burdick <[hidden email]>
> Date: 7/30/17  11:14 AM  (GMT-06:00) To: Elecraft Reflector <
> [hidden email]> Subject: [Elecraft] K3S (SDR/superhet hybrid)
> vs. entry-level
>         direct-sampling radios
> Hi all,
>
> Over the past two days I’ve been deluged with requests for our K3S vs.
> IC7300 comparison chart. This stems from a posting that mentioned the
> latter rig, and my subsequent offer to send out the chart, which we
> normally use for in-house training purposes.
>
> What’s become clear from all the questions and comments is that there’s a
> need for clarification on receiver architecture.
>
> The K3S, an SDR/superhet hybrid, includes narrowband protection of its
> A-to-D converters in the form of roofing filters (crystal filters). This is
> fundamentally different from the approach taken by direct-sampling radios,
> which have only very broadband filtering ahead of their ADCs. Typically,
> front-end band-pass filters are 0.5 to 4 MHz wide. Their ADCs will be
> impacted by all strong signals in this range, alone and in summation.
>
> When the ADC over-ranges in this “pure” SDR architecture, the radio
> usually reduces its gain automatically by either turning off preamps or
> adding attenuation. This increases the noise figure, often resulting in the
> loss of weak signals, both audibly and in the panadapter. Another word for
> it is “desense.” You might hear a pulsing sound as the noise floor goes up
> and down in response to a particular signal, or the sensitivity may degrade
> for many seconds at a time while the firmware waits for levels to drop at
> the input to the ADC.
>
> Such over-ranging is almost unheard of in the K3S, for multiple reasons.
> First, we use very strong mixers and gain stages. Second, in many cases
> there’s no need to turn the preamp on at all. (Example: the K3S’s
> preamp-off sensitivity is typically -135 dB, while the preamp-off
> sensitivity on a typical direct-sampling SDR is -115 to -125 dB.) The third
> reason for the K3S’s strong-signal performance is its crystal roofing
> filters. These protect the ADC from not only wide-spaced signals, but also
> from signals very close by. As K3 and K3S owners will attest, you can have
> a huge signal just a few hundred Hz away and not even know it’s there --
> unless that station’s transmit phase noise is blanketing the band anyway.
> (There’s no defense against an unclean or clicky transmit signal.)
>
> I’ve updated the referenced comparison chart to clarify this important
> difference. Our webmaster will put it up next week.
>
> Meanwhile, I’ll continue to send it on request (email me directly). Feel
> free to share the information with anyone interested in the topic.
>
> 73,
> Wayne
> N6KR
>
>
>
>
> ______________________________________________________________
> 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
> Message delivered to [hidden email]
> ______________________________________________________________
> 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
> Message delivered to [hidden email]
>

______________________________________________________________
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
Message delivered to [hidden email]

[Lynn W. Taylor, WB6UUT-2]

Wayne,

How much resolution is available (how many bits per sample) in the
typical modern A/D converter used in an IC-7300 or Flex?

73 -- Lynn

On 7/30/2017 9:14 AM, Wayne Burdick wrote:
> The K3S, an SDR/superhet hybrid, includes narrowband protection of its A-to-D converters in the form of roofing filters (crystal filters). This is fundamentally different from the approach taken by direct-sampling radios, which have only very broadband filtering ahead of their ADCs. Typically, front-end band-pass filters are 0.5 to 4 MHz wide. Their ADCs will be impacted by all strong signals in this range, alone and in summation.
______________________________________________________________
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
Message delivered to [hidden email]

[wayne burdick]

We’re quite familiar with this document. But blocking is blocking: a single strong signal in the passband of the RF band-pass filter can overrange the ADC, causing desense or spurious responses, thus requiring that gain be reduced ahead of it. Clever firmware can help, but it can’t  overcome the basic limitation on ADC input voltage spec.

73,
Wayne
N6KR

> On Jul 30, 2017, at 9:47 AM, Kevin der Kinderen <[hidden email]> wrote:
>
> Concerning Flex, one of the engineers posted a pretty detailed explanation
> of how they handle strong signals in a wide band environment. This stuff is
> still a bit over my head but owning both the KX3 and the Flex-6500 I can
> say, I am never disappointed with either receiver in any environment. I
> kind of feel like the luckiest dog in the world to have both.
>
> I hope it is OK to post the link:
>
> https://community.flexradio.com/flexradio/topics/adc-overload-myths-debunked
>
> 73,
> Kev K4VD
>
>
>
> On Sun, Jul 30, 2017 at 12:27 PM, GaryK9GS <[hidden email]> wrote:
>
>> Thanks for sending Wayne.   To take this discussion a step further,  is
>> the Flex series of radios in the same category as the 7300?
>>
>>
>> 73,
>> Gary K9GS
>> -------- Original message --------From: Wayne Burdick <[hidden email]>
>> Date: 7/30/17  11:14 AM  (GMT-06:00) To: Elecraft Reflector <
>> [hidden email]> Subject: [Elecraft] K3S (SDR/superhet hybrid)
>> vs. entry-level
>>        direct-sampling radios
>> Hi all,
>>
>> Over the past two days I’ve been deluged with requests for our K3S vs.
>> IC7300 comparison chart. This stems from a posting that mentioned the
>> latter rig, and my subsequent offer to send out the chart, which we
>> normally use for in-house training purposes.
>>
>> What’s become clear from all the questions and comments is that there’s a
>> need for clarification on receiver architecture.
>>
>> The K3S, an SDR/superhet hybrid, includes narrowband protection of its
>> A-to-D converters in the form of roofing filters (crystal filters). This is
>> fundamentally different from the approach taken by direct-sampling radios,
>> which have only very broadband filtering ahead of their ADCs. Typically,
>> front-end band-pass filters are 0.5 to 4 MHz wide. Their ADCs will be
>> impacted by all strong signals in this range, alone and in summation.
>>
>> When the ADC over-ranges in this “pure” SDR architecture, the radio
>> usually reduces its gain automatically by either turning off preamps or
>> adding attenuation. This increases the noise figure, often resulting in the
>> loss of weak signals, both audibly and in the panadapter. Another word for
>> it is “desense.” You might hear a pulsing sound as the noise floor goes up
>> and down in response to a particular signal, or the sensitivity may degrade
>> for many seconds at a time while the firmware waits for levels to drop at
>> the input to the ADC.
>>
>> Such over-ranging is almost unheard of in the K3S, for multiple reasons.
>> First, we use very strong mixers and gain stages. Second, in many cases
>> there’s no need to turn the preamp on at all. (Example: the K3S’s
>> preamp-off sensitivity is typically -135 dB, while the preamp-off
>> sensitivity on a typical direct-sampling SDR is -115 to -125 dB.) The third
>> reason for the K3S’s strong-signal performance is its crystal roofing
>> filters. These protect the ADC from not only wide-spaced signals, but also
>> from signals very close by. As K3 and K3S owners will attest, you can have
>> a huge signal just a few hundred Hz away and not even know it’s there --
>> unless that station’s transmit phase noise is blanketing the band anyway.
>> (There’s no defense against an unclean or clicky transmit signal.)
>>
>> I’ve updated the referenced comparison chart to clarify this important
>> difference. Our webmaster will put it up next week.
>>
>> Meanwhile, I’ll continue to send it on request (email me directly). Feel
>> free to share the information with anyone interested in the topic.
>>
>> 73,
>> Wayne
>> N6KR
>>
>>
>>
>>
>> ______________________________________________________________
>> 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
>> Message delivered to [hidden email]
>> ______________________________________________________________
>> 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
>> Message delivered to [hidden email]
>>
> ______________________________________________________________
> 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
> Message delivered to [hidden email]

______________________________________________________________
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
Message delivered to [hidden email]

[wayne burdick]

I believe the Flex to be 16 bits and the Icom to be 14 bits, but if anyone has data to the the contrary, feel free to post it.

Even the most expensive 16-bit A-to-D converters, operating at RF sampling rates, can exhibit nonlinearity or even non-monotonicity in their lowest-order bits. The usual work-around is to dither (spread) the clock signal, either on the chip itself or by injecting a separate tone. This can help, but the results can be uneven. The result is that two direct-sampling radios of exactly the same model can vary considerably in IMD performance.

In the K3S, the ADC is operated at much lower sampling rates (48 kHz, as required for the 15 kHz second IF). Since it is also fully protected by crystal roofing filters, an entire class of wideband sampling spurious responses is eliminated, and the much higher “NBSFDR” spec applies (narrowband spurious-free dynamic range). No dithering is required, and receiver sensitivity never has to be reduced due to out-of-band signals. That and the strong front end are why so many Field Day, multi-op contest, and DXpeditions stations use K3S’s and K3’s.

73,
Wayne
N6KR

> On Jul 30, 2017, at 9:49 AM, Lynn W. Taylor, WB6UUT <[hidden email]> wrote:
>
> Wayne,
>
> How much resolution is available (how many bits per sample) in the typical modern A/D converter used in an IC-7300 or Flex?
>
> 73 -- Lynn
>
> On 7/30/2017 9:14 AM, Wayne Burdick wrote:
>> The K3S, an SDR/superhet hybrid, includes narrowband protection of its A-to-D converters in the form of roofing filters (crystal filters). This is fundamentally different from the approach taken by direct-sampling radios, which have only very broadband filtering ahead of their ADCs. Typically, front-end band-pass filters are 0.5 to 4 MHz wide. Their ADCs will be impacted by all strong signals in this range, alone and in summation.

______________________________________________________________
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
Message delivered to [hidden email]

[Joe Subich, W4TV-4]

On 7/30/2017 12:47 PM, Kevin der Kinderen wrote:

 > I hope it is OK to post the link:
 > >
https://community.flexradio.com/flexradio/topics/adc-overload-myths-debunked
Welcome to Flex's own myth.  If the information posted in that paper
were true, OFDM and/or CDMA base stations could use class C (or D or E)
amplifiers rated for average power.  That is demonstrably not true as
many signals of that type require class A or AB amplifiers with pre-
correction and peak power capacity of 6 to 20 dB greater than average
power output to prevent significant IMD issues.

73,

    ... Joe, W4TV


On 7/30/2017 12:47 PM, Kevin der Kinderen wrote:

> Concerning Flex, one of the engineers posted a pretty detailed explanation
> of how they handle strong signals in a wide band environment. This stuff is
> still a bit over my head but owning both the KX3 and the Flex-6500 I can
> say, I am never disappointed with either receiver in any environment. I
> kind of feel like the luckiest dog in the world to have both.
>
> I hope it is OK to post the link:
>
> https://community.flexradio.com/flexradio/topics/adc-overload-myths-debunked
>
> 73,
> Kev K4VD
>
>
>
> On Sun, Jul 30, 2017 at 12:27 PM, GaryK9GS <[hidden email]> wrote:
>
>> Thanks for sending Wayne.   To take this discussion a step further,  is
>> the Flex series of radios in the same category as the 7300?
>>
>>
>> 73,
>> Gary K9GS
>> -------- Original message --------From: Wayne Burdick <[hidden email]>
>> Date: 7/30/17  11:14 AM  (GMT-06:00) To: Elecraft Reflector <
>> [hidden email]> Subject: [Elecraft] K3S (SDR/superhet hybrid)
>> vs. entry-level
>>          direct-sampling radios
>> Hi all,
>>
>> Over the past two days I’ve been deluged with requests for our K3S vs.
>> IC7300 comparison chart. This stems from a posting that mentioned the
>> latter rig, and my subsequent offer to send out the chart, which we
>> normally use for in-house training purposes.
>>
>> What’s become clear from all the questions and comments is that there’s a
>> need for clarification on receiver architecture.
>>
>> The K3S, an SDR/superhet hybrid, includes narrowband protection of its
>> A-to-D converters in the form of roofing filters (crystal filters). This is
>> fundamentally different from the approach taken by direct-sampling radios,
>> which have only very broadband filtering ahead of their ADCs. Typically,
>> front-end band-pass filters are 0.5 to 4 MHz wide. Their ADCs will be
>> impacted by all strong signals in this range, alone and in summation.
>>
>> When the ADC over-ranges in this “pure” SDR architecture, the radio
>> usually reduces its gain automatically by either turning off preamps or
>> adding attenuation. This increases the noise figure, often resulting in the
>> loss of weak signals, both audibly and in the panadapter. Another word for
>> it is “desense.” You might hear a pulsing sound as the noise floor goes up
>> and down in response to a particular signal, or the sensitivity may degrade
>> for many seconds at a time while the firmware waits for levels to drop at
>> the input to the ADC.
>>
>> Such over-ranging is almost unheard of in the K3S, for multiple reasons.
>> First, we use very strong mixers and gain stages. Second, in many cases
>> there’s no need to turn the preamp on at all. (Example: the K3S’s
>> preamp-off sensitivity is typically -135 dB, while the preamp-off
>> sensitivity on a typical direct-sampling SDR is -115 to -125 dB.) The third
>> reason for the K3S’s strong-signal performance is its crystal roofing
>> filters. These protect the ADC from not only wide-spaced signals, but also
>> from signals very close by. As K3 and K3S owners will attest, you can have
>> a huge signal just a few hundred Hz away and not even know it’s there --
>> unless that station’s transmit phase noise is blanketing the band anyway.
>> (There’s no defense against an unclean or clicky transmit signal.)
>>
>> I’ve updated the referenced comparison chart to clarify this important
>> difference. Our webmaster will put it up next week.
>>
>> Meanwhile, I’ll continue to send it on request (email me directly). Feel
>> free to share the information with anyone interested in the topic.
>>
>> 73,
>> Wayne
>> N6KR
>>
>>
>>
>>
>> ______________________________________________________________
>> 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
>> Message delivered to [hidden email]
>> ______________________________________________________________
>> 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
>> Message delivered to [hidden email]
>>
> ______________________________________________________________
> 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
> Message delivered to [hidden email]
>

______________________________________________________________
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
Message delivered to [hidden email]

[Igor Sokolov-2]

May I add that another good feature of a direct sampling SDR is low IMD
in the passband. That is due to lack of crystal filter in the passband.
The result is that that output audio is cleaner especially when there
are several stations calling on the same frequency.

73, Igor UA9CDC

30.07.2017 21:46, Wayne Burdick пишет:

>> GaryK9GS <[hidden email]> wrote:
>>
>> Thanks for sending Wayne.   To take this discussion a step further,  is the Flex series of radios in the same category as the 7300?
>
> Flex radios do use a direct-sampling architecture (with no roofing filters). The degree to which they’re subject to ADC over-range, and what they do about it in firmware, has been the subject of ongoing debate. It may be the case that, generally, the more you pay for a direct-sampling radio, the better the A-to-D converters, and the higher the ADC overrange threshold.
>
> That said, when measured with preamp off, the Flex 6700 has about 7 dB less IMD dynamic range than the K3S (see note Y in Sherwood’s chart), and 20 dB less blocking dynamic range. Both are direct consequences of not having narrowband roofing filters.
>
> On the other hand, the Flex radios have a very broad panadapter display for those who require it.
>
> 73,
> Wayne
> N6KR
>
>
>>
>> 73,
>> Gary K9GS
>> -------- Original message --------From: Wayne Burdick <[hidden email]> Date: 7/30/17  11:14 AM  (GMT-06:00) To: Elecraft Reflector <[hidden email]> Subject: [Elecraft] K3S (SDR/superhet hybrid) vs. entry-level
>> direct-sampling radios
>> Hi all,
>>
>> Over the past two days I’ve been deluged with requests for our K3S vs. IC7300 comparison chart. This stems from a posting that mentioned the latter rig, and my subsequent offer to send out the chart, which we normally use for in-house training purposes.
>>
>> What’s become clear from all the questions and comments is that there’s a need for clarification on receiver architecture.
>>
>> The K3S, an SDR/superhet hybrid, includes narrowband protection of its A-to-D converters in the form of roofing filters (crystal filters). This is fundamentally different from the approach taken by direct-sampling radios, which have only very broadband filtering ahead of their ADCs. Typically, front-end band-pass filters are 0.5 to 4 MHz wide. Their ADCs will be impacted by all strong signals in this range, alone and in summation.
>>
>> When the ADC over-ranges in this “pure” SDR architecture, the radio usually reduces its gain automatically by either turning off preamps or adding attenuation. This increases the noise figure, often resulting in the loss of weak signals, both audibly and in the panadapter. Another word for it is “desense.” You might hear a pulsing sound as the noise floor goes up and down in response to a particular signal, or the sensitivity may degrade for many seconds at a time while the firmware waits for levels to drop at the input to the ADC.
>>
>> Such over-ranging is almost unheard of in the K3S, for multiple reasons. First, we use very strong mixers and gain stages. Second, in many cases there’s no need to turn the preamp on at all. (Example: the K3S’s preamp-off sensitivity is typically -135 dB, while the preamp-off sensitivity on a typical direct-sampling SDR is -115 to -125 dB.) The third reason for the K3S’s strong-signal performance is its crystal roofing filters. These protect the ADC from not only wide-spaced signals, but also from signals very close by. As K3 and K3S owners will attest, you can have a huge signal just a few hundred Hz away and not even know it’s there -- unless that station’s transmit phase noise is blanketing the band anyway. (There’s no defense against an unclean or clicky transmit signal.)
>>
>> I’ve updated the referenced comparison chart to clarify this important difference. Our webmaster will put it up next week.
>>
>> Meanwhile, I’ll continue to send it on request (email me directly). Feel free to share the information with anyone interested in the topic.
>>
>> 73,
>> Wayne
>> N6KR
>>
>>
>>
>>
>> ______________________________________________________________
>> 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
>> Message delivered to [hidden email]
>> ______________________________________________________________
>> 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
>> Message delivered to [hidden email]
> ______________________________________________________________
> 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
> Message delivered to [hidden email]

______________________________________________________________
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
Message delivered to [hidden email]

[wayne burdick]

Igor,

The original DSP code had a math error in the AGC algorithm that caused some nonlinearity at certain signal levels. This was corrected. The crystal filters have low group delay and low ripple, so the only remaining factor is ADC quantization noise. Our ADCs are 16 bits and sampling at 48 kHz (at the 2nd IF), with performance as good or better than RF sampling ADCs. Additional ADC reference filtering was added to the K3S RF board, and the DSP audio path was improved. Now, in practice, there is almost no difference in in-band demodulation sound or IMD.

73,
Wayne
N6KR

> On Jul 30, 2017, at 10:46 PM, Igor Sokolov <[hidden email]> wrote:
>
>
> May I add that another good feature of a direct sampling SDR is low IMD in the passband. That is due to lack of crystal filter in the passband. The result is that that output audio is cleaner especially when there are several stations calling on the same frequency.
>
> 73, Igor UA9CDC
>
> 30.07.2017 21:46, Wayne Burdick пишет:
>>> GaryK9GS <[hidden email]> wrote:
>>>
>>> Thanks for sending Wayne.   To take this discussion a step further,  is the Flex series of radios in the same category as the 7300?
>>
>> Flex radios do use a direct-sampling architecture (with no roofing filters). The degree to which they’re subject to ADC over-range, and what they do about it in firmware, has been the subject of ongoing debate. It may be the case that, generally, the more you pay for a direct-sampling radio, the better the A-to-D converters, and the higher the ADC overrange threshold.
>>
>> That said, when measured with preamp off, the Flex 6700 has about 7 dB less IMD dynamic range than the K3S (see note Y in Sherwood’s chart), and 20 dB less blocking dynamic range. Both are direct consequences of not having narrowband roofing filters.
>>
>> On the other hand, the Flex radios have a very broad panadapter display for those who require it.
>>
>> 73,
>> Wayne
>> N6KR
>>
>>
>>>
>>> 73,
>>> Gary K9GS
>>> -------- Original message --------From: Wayne Burdick <[hidden email]> Date: 7/30/17  11:14 AM  (GMT-06:00) To: Elecraft Reflector <[hidden email]> Subject: [Elecraft] K3S (SDR/superhet hybrid) vs. entry-level
>>> direct-sampling radios
>>> Hi all,
>>>
>>> Over the past two days I’ve been deluged with requests for our K3S vs. IC7300 comparison chart. This stems from a posting that mentioned the latter rig, and my subsequent offer to send out the chart, which we normally use for in-house training purposes.
>>>
>>> What’s become clear from all the questions and comments is that there’s a need for clarification on receiver architecture.
>>>
>>> The K3S, an SDR/superhet hybrid, includes narrowband protection of its A-to-D converters in the form of roofing filters (crystal filters). This is fundamentally different from the approach taken by direct-sampling radios, which have only very broadband filtering ahead of their ADCs. Typically, front-end band-pass filters are 0.5 to 4 MHz wide. Their ADCs will be impacted by all strong signals in this range, alone and in summation.
>>>
>>> When the ADC over-ranges in this “pure” SDR architecture, the radio usually reduces its gain automatically by either turning off preamps or adding attenuation. This increases the noise figure, often resulting in the loss of weak signals, both audibly and in the panadapter. Another word for it is “desense.” You might hear a pulsing sound as the noise floor goes up and down in response to a particular signal, or the sensitivity may degrade for many seconds at a time while the firmware waits for levels to drop at the input to the ADC.
>>>
>>> Such over-ranging is almost unheard of in the K3S, for multiple reasons. First, we use very strong mixers and gain stages. Second, in many cases there’s no need to turn the preamp on at all. (Example: the K3S’s preamp-off sensitivity is typically -135 dB, while the preamp-off sensitivity on a typical direct-sampling SDR is -115 to -125 dB.) The third reason for the K3S’s strong-signal performance is its crystal roofing filters. These protect the ADC from not only wide-spaced signals, but also from signals very close by. As K3 and K3S owners will attest, you can have a huge signal just a few hundred Hz away and not even know it’s there -- unless that station’s transmit phase noise is blanketing the band anyway. (There’s no defense against an unclean or clicky transmit signal.)
>>>
>>> I’ve updated the referenced comparison chart to clarify this important difference. Our webmaster will put it up next week.
>>>
>>> Meanwhile, I’ll continue to send it on request (email me directly). Feel free to share the information with anyone interested in the topic.
>>>
>>> 73,
>>> Wayne
>>> N6KR
>>>
>>>
>>>
>>>
>>> ______________________________________________________________
>>> 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
>>> Message delivered to [hidden email]
>>> ______________________________________________________________
>>> 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
>>> Message delivered to [hidden email]
>> ______________________________________________________________
>> 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
>> Message delivered to [hidden email]
>
> ______________________________________________________________
> 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
> Message delivered to [hidden email]

______________________________________________________________
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
Message delivered to [hidden email]

[wayne burdick]

Gary, K7EMF, caught an error in my paragraph below. I had used ‘dB’ when I meant ‘dBm’ (dB relative to 1 milliwatt). Now corrected.

73,
Wayne
N6KR


> Such over-ranging is almost unheard of in the K3S, for multiple reasons. First, we use very strong mixers and gain stages. Second, in many cases there’s no need to turn the preamp on at all. (Example: the K3S’s preamp-off sensitivity is typically -135 dBm, while the preamp-off sensitivity on a typical direct-sampling SDR is -115 to -125 dBm.) The third reason for the K3S’s strong-signal performance is its crystal roofing filters. These protect the ADC from not only wide-spaced signals, but also from signals very close by. As K3 and K3S owners will attest, you can have a huge signal just a few hundred Hz away and not even know it’s there -- unless that station’s transmit phase noise is blanketing the band anyway. (There’s no defense against an unclean or clicky transmit signal.)




______________________________________________________________
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
Message delivered to [hidden email]