K3 & ADAT ADT-200A by HB9CBU

[Joe Subich, W4TV-3]

> You are right
> that if you exceed the maximum voltage level into the ADC no
> amount of processing gain will change this fact.

Thank you.  Now, since you have set a hard limit on the signal
to the ADC, any architecture must either allow sufficient
margin to prevent overload or must install some form of hardware
AGC to prevent overload from multiple, simultaneous, strong
signals.  Any headroom comes from the practical dynamic range
(maximum signal level minus absolute MDS (even with processing
gain).  Choose the AGC route and you're back into AGC blocking
where strong unwanted signals can decrease the gain to a point
that a weak signal drops below detection level even with all
the processing gain in the world.  

All SDRs, even the Perseus, attempt to minimize the effects
of strong "unwanted" signals by using some form of front end
filtering.

It is the strong multiple signal condition that is the problem
for contesters/DXers on the lowbands where the goal is to copy
a -140 dBm signal in the middle of 20 or S9+40 dB "locals" all
within 20 or 30 KHz.  This is the same problem on 7 MHz in
Europe where a user might be trying to copy a -140 dBm signal
with three or four 0 to +10 dBm mega-power broadcasters only
100 KHz away.  In either case, ADC overload (overflow) and/or
ACG induced blocking are almost guaranteed with ANY SDR design
that does not utilize very tight front end ("roofing") filters.

> The weakest signal
> that can be resolved plays just as important a part in the
> calculation of the blocking dynamic range of a receiver as
> its strong signal handling capability - and this is true
> whether we are talking about ADCs or analog receivers.

Of course, but there is still a finite range between the
two limits.  The maximum instantaneous peak is driven by
the number and strength of incoming signals and the minimum
discernable signal level is defined not by processing gain
but by "sky noise."  All the processing gain goes only so
far ... the limit is still well above the theoretical ability
to detect one signal in the absence of noise.  Processing
gain may be of value at microwave frequencies with a quiet
sky but it is not going to help below 10 MHz with extreme
signals and high noise levels.

73,

   ... Joe, W4TV
 
 

> -----Original Message-----
> From: [hidden email]
> [mailto:[hidden email]] On Behalf Of Philip
> Covington
> Sent: Monday, December 01, 2008 4:40 PM
> To: [hidden email]
> Cc: [hidden email]; Bill W4ZV
> Subject: Re: [Elecraft] K3 & ADAT ADT-200A by HB9CBU
>
>
> On Mon, Dec 1, 2008 at 3:56 PM, Joe Subich, W4TV
> <[hidden email]> wrote:
> >
> >> This is not correct.  You can't just compare the number of bits
> >> without taking into consideration the ADC sampling rate and signal
> >> bandwidth.  Please see the concept of "signal processing gain" or
> >> "process gain".
> >
> > The concept of processing gain is completely different than
> blocking
> > dynamic range.  The maximum large signal capability of any
> DAC has an
> > absolute limit based on the largest integer the DAC can resolve and
> > the DAC reference voltage. When more than one signal is
> present within
> > the "window" (DC to maximum
> > frequency) of the DAC the DAC needs to be able to handle the sum of
> > the instantaneous peak (vector maximum) voltages not their average
> > levels.
> >
> > No amount of decimation and processing will reduce the
> instantaneous
> > peak voltage that the ADC must handle to prevent overload.  The
> > clipping (overload) level results in distortion (IMD) or imposes an
> > operating condition that must be prevented by the application of
> > hardware AGC (gain
> > reduction) ahead of the ADC (blocking).  Decimation and processing
> > gain are only of value as long as he hardware is operating
> within its
> > linear range.
> >
> > 73,
> >
> >   ... Joe, W4TV
>
> I assume you mean ADC above and not DAC...  You are right
> that if you exceed the maximum voltage level into the ADC no
> amount of processing gain will change this fact.  What the
> processing gain does give you is the ability to hear weak
> signals that are much lower than the 14 or 16 bit ADC without
> decimation would be capable of hearing.  The weakest signal
> that can be resolved plays just as important a part in the
> calculation of the blocking dynamic range of a receiver as
> its strong signal handling capability - and this is true
> whether we are talking about ADCs or analog receivers.
>
> --
> Phil Covington
> Software Radio Laboratory LLC
> Columbus, Ohio
> http://www.srl-llc.com _______________________________________________
> Elecraft mailing list
> Post to: [hidden email]
> You must be a subscriber to post to the list.
> Subscriber Info (Addr. Change, sub, unsub etc.):
>  http://mailman.qth.net/mailman/listinfo/elecraft   
>
> Help: http://mailman.qth.net/subscribers.htm
> Elecraft web page: http://www.elecraft.com

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[dj7mgq]

> has anybody done a serious test of the ADT-200A transceiver yet?

When I asked this question, I sure did not expect the very educational
discussion which has broken out. I'm very thankful, as a photographer
who programs graphic systems for television, many good points were made
which I might not have thought about on my own.

 From a purely amateurish SWL point of view, I have had the chance to
play with a few very expensive high end LW/MW/SW receivers, designed for
use by the government, military, etc. One thing which has struck me, was
the often poor IMD2 behavior of these radios, unless if they were using
good, fairly small-band input filters.

Also a real world test of a radio, would also have to deal with aspects
such as the user interface as well. This is one of the reasons I did not
"upgrade" from the Flex-1000 to the Flex-5000. I did however "upgrade"
K2 to K3, and despite whatever inherent minor weakness it might have,
I'm not sorry that I did.

vy 73 de toby
--
Real radios don't care whether they have knobs or not...

DD5FZ (ex 4n6fz, dj7mgq, dg5mgq, dd5fz)
K2 #885, K2/100 #3248, K3/100 #67
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Bill W4ZV]

Joe Subich, W4TV-3 wrote

Processing
gain may be of value at microwave frequencies with a quiet
sky but it is not going to help below 10 MHz with extreme
signals and high noise levels.

Joe, as near as I can tell, you've hit on the primary cause of arguments between intelligent folks like Leif SM5BSZ and Tom W8JI (both were on ARRL's committee to review their Product Test procedures).  One is talking weak signal EME, galactic noise and extremely wide interference spacings; the other is talking about weak signal 160 meters, atmospheric noise and relatively close interference spacings (e.g. the ARRL 160 or CQ 160 contests).  It may be that both are right for their respective applications, so we may be having a discussion of apples and oranges.  Normal HF applications of the K3 are much closer to Tom's world than Leif's world IMHO.

I'm still waiting for independent measurements of BDR that show other SDR approaches are near the results of the K3.  When I see that, I'll be the first to be a believer.  Maybe the Perseus will do this, but I haven't seen the "corrected" ARRL results yet.  When it comes to receiver performance, I'm from Missouri (the "Show Me" state).

73,  Bill

[Nico Palermo, IV3NWV]

In that case the BDR would be 111 dB, would it not?  I believe the K3 was

measured at BDR of 140 dB at 2 kHz spacing on 14 MHz in the most recent ARRL

test.  Not that we would actually try to operate so close to such a strong

signal on CW due to key clicks, etc, but BDR results at wider spacings are

very important if there are other stations on the same band (e.g. one on SSB

and one on CW in Field Day or during a DXpedition like VP6DX, or if you had

a close neighbor on the same band).

You are not replying my question, Bill.
I've simply asked if you are able to listen to a -105 dBm with a + 7 dBm inteferer placed at
a 2 kHz offset with your receiver.

You have two choices:
1) No, I can't.
2) Yes. I can.

If the reply is #1 you should explain me why if the BDR of your receiver is declared to be 140 dB
at 2 kHz spacing and you are not able to do handle a 111 dB signal dynamic range.
If the reply is #2 I kindly ask you to provide me the proof and I will apologize with you.

BTW, if in a real situation nobody would try to operate 2 kHz apart a strong CW signal, due to the
clicks, what the ARRL Blocking Compression Gain test at 2 kHz is meant for? Is it for dummies?
Don't you think that questioning the utility of the test while being so proud that your receiver is rated with that figure
is a really poor argument?

Can you briefly explain why you feel ARRL's definition of BDR below is not

correct?

...The blocking dynamic range is the difference between the level of the noise floor from the level of undesired

signal that produces a 1 dB decrease in a weak desired signal...

I already know the ARRL definition.
What "weak" means? Is it a signal at  the MDS level, 20 dB stronger, or 50?
Weak is not 1 foot, nor 2 inches/second. It is NaN, not a number.
For DXers weak means S 8, for others it is right the MDS, for other it is the level of the least audible CW signal immersed in the noise, say 10 dB less than its power.
If "weak" is meant to be S 8, be sure your receiver has a 140 BDR, as defined by the ARRL, but I kindly let you note that the difference between 10 dBm (the level of the interfering carrier that produce a 1 dB decrease in the weak signal in a receiver which has a sensitivity of -130 dBm) and -79 dBm (the weak S8 signal) is not 140 dB, but just 89 dB.
Do the same test with a weak desired signal which is S 4 and let me know if you are able to measure the gain compression of your receiver or if instead you simply you are not, just because the phase noise of the LO is 20 dB higher and has already completely desensitized your equipment.
In the case you are not able to do it, why do you claim that your receiver can handle a 140 dB dynamic range if it is not able to handle signal level differences which are much less?

73,
Nico Palermo, IV3NWV

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Philip Covington]

On Mon, Dec 1, 2008 at 8:38 PM, Joe Subich, W4TV <[hidden email]> wrote:

> Of course, but there is still a finite range between the
> two limits.  The maximum instantaneous peak is driven by
> the number and strength of incoming signals and the minimum
> discernable signal level is defined not by processing gain
> but by "sky noise."  All the processing gain goes only so
> far ... the limit is still well above the theoretical ability
> to detect one signal in the absence of noise.  Processing
> gain may be of value at microwave frequencies with a quiet
> sky but it is not going to help below 10 MHz with extreme
> signals and high noise levels.
>
> 73,
>
>   ... Joe, W4TV

You are making the incorrect assumption that we have no control over
the upper or lower limits.  If band or sky noise is the limiting
factor on the low end, then adding attenuation in front of the ADC to
adjust for this has the benefit of increasing the high end limit also.
 This ideally has not changed the dynamic range, but only shifted the
upper and lower limits.

The Perseus or QS1R will easily outperform the K3 in the case of
multiple strong signals in discerning a weak signal as you have
described.  This is WITHOUT AGC in front of the ADC.

--
Phil Covington
Software Radio Laboratory LLC
Columbus, Ohio
http://www.srl-llc.com
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Philip Covington]

On Mon, Dec 1, 2008 at 4:48 PM, Kok Chen <[hidden email]> wrote:

>
> On Monday, December 01, 2008, at 01:16PM, "Bill W4ZV" <[hidden email]> wrote:
>
>>...which is why the hybrid approach of a narrow roofing filter before the
>>ADC works so well in Orion and the K3.
>
> Bill is correct.
>
> Think of this this way: the clipping level of a codec does not change, but the noise floor of the processed passband falls.
>
> You gain nothing anymore only after the Nyquist rate of the signal of interest has exceeded (think energy density) the decimated sampling rate.
>
> 73
> Chen, W7AY
>

Chen,

You must also realize that placing analog components in front of the
ADC introduces non-linearity and distortion.  In the case of DSP IF
radios like the K3, there is *a lot* of analog components in front of
the ADC, each degrading the signal to some extent.  There is also the
problem of the phase noise of the LO.   Even QSD based SDRs which use
a DDS for the LO have worse phase noise (and spurs) than the crystal
oscillator LO in direct sampling receivers like the Perseus or QS1R.
In the DDS, the DAC only approximates an analog signal resulting in
spurs, and depending on whether the internal PLL is used in the DDS,
added phase noise.  In direct sampling receivers such as Perseus or
QS1R, the LO is not converted back to an analog signal as in the DDS
case - the quality of the LO is then pretty much determined by the
quality of the sampling clock (very good) and how many bits you use in
your NCO in the FPGA.  PLL based LOs (like the K3) have much higher
phase noise than this.

--
Phil Covington
Software Radio Laboratory LLC
Columbus, Ohio
http://www.srl-llc.com
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Philip Covington]

On Mon, Dec 1, 2008 at 9:02 PM, Bill W4ZV <[hidden email]> wrote:

>
>
>
> Joe Subich, W4TV-3 wrote:
>>
>> Processing
>> gain may be of value at microwave frequencies with a quiet
>> sky but it is not going to help below 10 MHz with extreme
>> signals and high noise levels.
>>
>
> Joe, as near as I can tell, you've hit on the primary cause of arguments
> between intelligent folks like Leif SM5BSZ and Tom W8JI (both were on ARRL's
> committee to review their Product Test procedures).  One is talking weak
> signal EME, galactic noise and extremely wide interference spacings; the
> other is talking about weak signal 160 meters, atmospheric noise and
> relatively close interference spacings (e.g. the ARRL 160 or CQ 160
> contests).  It may be that both are right for their respective applications,
> so we may be having a discussion of apples and oranges.  Normal HF
> applications of the K3 are much closer to Tom's world than Leif's world
> IMHO.
>
> I'm still waiting for independent measurements of BDR that show other SDR
> approaches are near the results of the K3.  When I see that, I'll be the
> first to be a believer.  Maybe the Perseus will do this, but I haven't seen
> the "corrected" ARRL results yet.  When it comes to receiver performance,
> I'm from Missouri (the "Show Me" state).
>
> 73,  Bill
> --
> View this message in context: http://n2.nabble.com/K3---ADAT-ADT-200A-by-HB9CBU-tp1597981p1601915.html
> Sent from the Elecraft mailing list archive at Nabble.com.
>
> _______________________________________________
> Elecraft mailing list
> Post to: [hidden email]
> You must be a subscriber to post to the list.
> Subscriber Info (Addr. Change, sub, unsub etc.):
>  http://mailman.qth.net/mailman/listinfo/elecraft
>
> Help: http://mailman.qth.net/subscribers.htm
> Elecraft web page: http://www.elecraft.com
>

The difference is that while W8JI may be an "expert" in many other
(analog) areas, his disparaging comments made on various mailing lists
shows an ignorance in the SDR realm (except maybe for a bad experience
with a SDR-1000).  It is not much different than Joe-blow using his
nifty cell phone while having absolutely no comprehension of how his
voice gets from point A to point B.  It is hard on the old "analog
experts" when the new digital technology is quickly leaving them
behind.  Whether Leif's interests lie in weak signal EME or other
areas, it does not negate the facts that he presents concerning SDR -
those facts apply in Tom's world also.

--
Phil Covington
Software Radio Laboratory LLC
Columbus, Ohio
http://www.srl-llc.com
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[P.B. Christensen]

>> Its dynamic range may be OK for IMD but today's SDRs cannot approach the BDR of
>> rigs like the K3 until even higher resolution ADCs become available.

> Sorry to tell you a bad new: you are wrong!

Nico & Phil,

 

When can we expect to see direct R.F.-sampled SDRs with better MDS performance?   When we compare MDS data across QST product reviews in a 500 Hz filter B/W, we see marginal figures of ~ -128 dBm @ 14 MHz.  This is true of the Perseus, SDR-IQ, and Flex 5000A. 

 

By contrast, radios such as the K3 and Orion offer excellent MDS figures approaching -140 dBm while still providing excellent close-spaced blocking and I3 dynamic range.

 

Paul, W9AC

 

 

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Bill W4ZV]

Nico Palermo, IV3NWV wrote:

You are not replying my question, Bill.
I've simply asked if you are able to listen to a -105 dBm with a + 7 dBm
inteferer placed at
a 2 kHz offset with your receiver.

You have two choices:
1) No, I can't.
2) Yes. I can.

If the reply is #1 you should explain me why if the BDR of your receiver is
declared to be 140 dB
at 2 kHz spacing and you are not able to do handle a 111 dB signal dynamic
range.
If the reply is #2 I kindly ask you to provide me the proof and I will
apologize with you.

BTW, if in a real situation nobody would try to operate 2 kHz apart a strong
CW signal, due to the
clicks, what the ARRL Blocking Compression Gain test at 2 kHz is meant for?
Is it for dummies?
Don't you think that questioning the utility of the test while being so
proud that your receiver is rated with that figure
is a really poor argument?

Can you briefly explain why you feel ARRL's definition of BDR below is not
>
correct?


...The blocking dynamic range is the difference between the level of the
> noise floor from the level of undesired
>
signal that produces a 1 dB decrease in a weak desired signal...
>

I already know the ARRL definition.
What "weak" means? Is it a signal at  the MDS level, 20 dB stronger, or 50?
Weak is not 1 foot, nor 2 inches/second. It is NaN, not a number.
For DXers weak means S 8, for others it is right the MDS, for other it is
the level of the least audible CW signal immersed in the noise, say 10 dB
less than its power.
If "weak" is meant to be S 8, be sure your receiver has a 140 BDR, as
defined by the ARRL, but I kindly let you note that the difference between
10 dBm (the level of the interfering carrier that produce a 1 dB decrease in
the weak signal in a receiver which has a sensitivity of -130 dBm) and -79
dBm (the weak S8 signal) is not 140 dB, but just 89 dB.
Do the same test with a weak desired signal which is S 4 and let me know if
you are able to measure the gain compression of your receiver or if instead
you simply you are not, just because the phase noise of the LO is 20 dB
higher and has already completely desensitized your equipment.
In the case you are not able to do it, why do you claim that your receiver
can handle a 140 dB dynamic range if it is not able to handle signal level
differences which are much less?

######################################################

Nico I believe I finally understand your point.  In simple terms, you're saying the K3's phase noise overrides close-spaced IMD and BDR performance.  In this case, I completely agree with you,  This has been a stumbling point that Rob Sherwood NC0B and Peter Hart G3SJX of RSGB have been making for many years now.  Both Peter and Rob have used the term "phase noise limited" to tell us that phase noise is actually overriding close spaced measurements like IMD.  

What is truly important is the "Spurious-Free Dynamic Range" (G3SJX's term) which includes phase noise, BDR and IMD performance.  SFDR is the worst case of all 3 factors, and phase noise will clearly override IMD or BDR results at close spacings in the K3.  ARRL's relatively recent change in test procedure to use a narrowband spectrum analyzer to measure IMD is masking the actual SFDR result because it's using a narrow band filter to remove the phase noise component when making the measurement.  Our brain/ear only has a ~50 Hz filter at best, so we will hear the phase noise in that 50 Hz bandwidth instead of the ~1 Hz (?) BW ARRL's spectrum analyzer uses.  What good does it say we have extremely good IMD or BDR at close spacings if that performance is theoretical and does not represent what our ears hear?  I agree it is misleading.

I believe what you're actually saying is that Perseus has much better close-spaced phase noise than the K3, and ARRL's measurements of close-spaced IMD and BDR are imaginary numbers that do not represent what our ears will actually hear.  In that case, I completely agree with you!   Would you tell us how Perseus compares in the following phase noise performance?

The following table is from a post by Eric WA6HHQ in September 2007.  I've added the Flex 5000 based on measurements published after Eric's note:

http://mailman.qth.net/pipermail/elecraft/2007-September/073931.html

"Here are transmit composite phase noise numbers from the Lab for the K3.
Needless to say, we are pleased with the results. :-) I've also included
some numbers from the ARRL reviews for several late model rigs.

Rig     1kHz    2   10   20   50   100   1M
K3      -110  -119 -136 -140 -143 -144  -150
IC7800  -103  -112 -130 -138 -140 -140  -140
FT2000  -102  -105 -128 -129 -128 -128  -128
ORION 2 -121  -129 -126 -125 -118 -128  -138
OMNI 7  -102  -103 -120 -123 -127 -129  -126"
F5000   -123 (same at all spacings de W4ZV)
Perseus   ?

What I personally wish is that ARRL should do is go to a metric like SFDR which G3SJX has proposed.  I believe this would be the most meaningful.  It would show us the worst case of what our ears will hear from any of the spurious components at various spacings.  

I'm going on a little trip with my XYL for a few days now but will be interested to see your response about Perseus.  Again, I agree with you that the close-spaced IMD and BDR results published by ARRL will be overridden by phase noise.

73,  Bill  W4ZV

[David Woolley (E.L)]

Philip Covington wrote:

> problem of the phase noise of the LO.   Even QSD based SDRs which use
> a DDS for the LO have worse phase noise (and spurs) than the crystal
> oscillator LO in direct sampling receivers like the Perseus or QS1R.

It seems to me that high speed A/D conversion is much more difficult to
do accurately than high speed D/A conversion, so any degradation in
converting the LO to analogue is going to be more than compensated for
by that in converting a wideband signal to digital.

--
David Woolley
"The Elecraft list is a forum for the discussion of topics related to
Elecraft products and more general topics related ham radio"
List Guidelines <http://www.elecraft.com/elecraft_list_guidelines.htm>
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[David Woolley (E.L)]

Nico Palermo wrote:

> You probably are unaware that an ADC dynamic range in a given bandwidth
> does not depend just on its ENOB (effective number of bits) but also on
> the sampling frequency.

Surely, in terms of the interfering signal, that bandwidth is something
like 30MHz, rather than the 2.8kHz in the K3 type of SDR.

--
David Woolley
"The Elecraft list is a forum for the discussion of topics related to
Elecraft products and more general topics related ham radio"
List Guidelines <http://www.elecraft.com/elecraft_list_guidelines.htm>
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Bill W4ZV]

On Tue, Dec 2, 2008 at 6:37 AM, Philip Covington <[hidden email]> wrote:

> The difference is that while W8JI may be an "expert" in many other
> (analog) areas, his disparaging comments made on various mailing lists
> shows an ignorance in the SDR realm (except maybe for a bad experience
> with a SDR-1000).  It is not much different than Joe-blow using his
> nifty cell phone while having absolutely no comprehension of how his
> voice gets from point A to point B.  It is hard on the old "analog
> experts" when the new digital technology is quickly leaving them
> behind.  Whether Leif's interests lie in weak signal EME or other
> areas, it does not negate the facts that he presents concerning SDR -
> those facts apply in Tom's world also.

I suspect you'll find Tom is a very practical performance-oriented
engineer.  He's probably agnostic about what's inside the box as long
as better performance is demonstrated.  The problem with the QSD SDRs
so far is that better performance has NOT been demonstrated (not to
mention IMHO an unfriendly user interface for most contesters...but
that's another discussion).

Perhaps direct-sampling SDRs will be different, but I believe the jury
is still out.  Show us via independent measurements that any radio is
demonstrably better and I bet Tom will become a "believer".

73,  Bill  W4ZV
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Philip Covington]

On Tue, Dec 2, 2008 at 8:05 AM, David Woolley (E.L)
<[hidden email]> wrote:

> Philip Covington wrote:
>
>> problem of the phase noise of the LO.   Even QSD based SDRs which use
>> a DDS for the LO have worse phase noise (and spurs) than the crystal
>> oscillator LO in direct sampling receivers like the Perseus or QS1R.
>
> It seems to me that high speed A/D conversion is much more difficult to do
> accurately than high speed D/A conversion, so any degradation in converting
> the LO to analogue is going to be more than compensated for by that in
> converting a wideband signal to digital.
>
> --
> David Woolley

It may seem that way to you, but in real life it turns out the the DDS
generates spurs due to only approximating a sin function, clock
leakage, number of bits, etc...  There is not the same issue in the
NCO generated LO in the FPGA in direct sampling receivers such as
Perseus and QS1R.

--
Phil Covington
Software Radio Laboratory LLC
Columbus, Ohio
http://www.srl-llc.com
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Philip Covington]

On Tue, Dec 2, 2008 at 6:49 AM, Paul Christensen <[hidden email]> wrote:

>>> Its dynamic range may be OK for IMD but today's SDRs cannot approach the
>>> BDR of
>>> rigs like the K3 until even higher resolution ADCs become available.
>
>> Sorry to tell you a bad new: you are wrong!
> Nico & Phil,
>
> When can we expect to see direct R.F.-sampled SDRs with better MDS
> performance?   When we compare MDS data across QST product reviews in a 500
> Hz filter B/W, we see marginal figures of ~ -128 dBm @ 14 MHz.  This is true
> of the Perseus, SDR-IQ, and Flex 5000A.
>
> By contrast, radios such as the K3 and Orion offer excellent MDS figures
> approaching -140 dBm while still providing excellent close-spaced
> blocking and I3 dynamic range.
>
> Paul, W9AC
>

Hi Paul,

There is no inherent reason why the MDS could not be that low in
direct sampling SDRs.  It is mainly a matter of design
decision/implementation.

--
Phil Covington
Software Radio Laboratory LLC
Columbus, Ohio
http://www.srl-llc.com
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[dj7mgq]

Hi Phil,

> There is no inherent reason why the MDS could not be that low in
> direct sampling SDRs.  It is mainly a matter of design
> decision/implementation.

This may be a stupid (or at least ignorant) question, but if the dynamic
range of the ADC itself is defined by the number of bits it resolves and
a maximum voltage must never (should never) be exceeded, wouldn't
improving MDS by 10 to 15db create major problems with the high end of
the scale? I would tend to think that analog solutions could be more
forgiving or that the K3 path which provides an AGC to protect the ADC
might be (at present) a "better" way to go.

Or are you suggesting using much faster ADCs so that you have a higher
level of over sampling, allowing more process gain during decimation?

vy 73 de toby


_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[David Woolley (E.L)]

Philip Covington wrote:
> On Tue, Dec 2, 2008 at 8:05 AM, David Woolley (E.L)

>
> It may seem that way to you, but in real life it turns out the the DDS
> generates spurs due to only approximating a sin function, clock
> leakage, number of bits, etc...  There is not the same issue in the

There is no fundamental reason why the the DDS sine function should be
any worse than the DSP one, nor for it to have any less bits than that
with which the signal is digitised.

It might be that commonly used DDS chips are rather old technology, and
you are comparing state of the art DSP with ten year old DDS.

> NCO generated LO in the FPGA in direct sampling receivers such as
> Perseus and QS1R.
>


--
David Woolley
"The Elecraft list is a forum for the discussion of topics related to
Elecraft products and more general topics related ham radio"
List Guidelines <http://www.elecraft.com/elecraft_list_guidelines.htm>
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Nico Palermo, IV3NWV]

Bill,
the phase noise of Perseus is as follow:

Offset      (kHz)      2    5     10    20   >50
Phase Noise (dBc/Hz)  -141 -145  -148  -151  <-155

As a picture is worth one thousand words, see here for more:

http://microtelecom.it/perseus/tests/Perseus-phasenoise.jpg

To make the test, the phase noise of the source generator,
a homebrew crystal oscillator, has been neglected and it has been assumed
that all the phase noise was due to the receiver oscillator, which is a crystal
oscillator too.

A more realistic assumption is that the two oscillators, the source generator
and the receiver LO, are contributing equally to the phase noise figure and in this
case the performance of the receiver would be even 3 dB better.
The figures indicated are thus worst case values.

For offsets larger than 50 kHz the phase noise performance cannot be really
measured as the prevailing source of noise is of thermal origin and the
noise induced by the reciprocal mixing is below the receiver thermal noise floor.

Now that you can compare the full table, it would be not difficult for you to conclude
that even at 1 MHz offset, the blocking dynamic range of Perseus is greater than
that of your receiver.


73
Nico Palermo, IV3NWV


>Would you tell us how Perseus compares in the following phase noise
>performance?

>The following table is from a post by Eric WA6HHQ in September 2007.  I've
>added the Flex 5000 based on measurements published after Eric's note:

>http://mailman.qth.net/pipermail/elecraft/2007-September/073931.html

>"Here are transmit composite phase noise numbers from the Lab for the K3.
>Needless to say, we are pleased with the results. :-) I've also included
>some numbers from the ARRL reviews for several late model rigs.

>Rig     1kHz    2   10   20   50   100   1M
>K3      -110  -119 -136 -140 -143 -144  -150
>IC7800  -103  -112 -130 -138 -140 -140  -140
>FT2000  -102  -105 -128 -129 -128 -128  -128
>ORION 2 -121  -129 -126 -125 -118 -128  -138
>OMNI 7  -102  -103 -120 -123 -127 -129  -126"
>F5000   -123 (same at all spacings de W4ZV)
>Perseus   ?

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[Philip Covington]

On Tue, Dec 2, 2008 at 10:27 AM, David Woolley (E.L)
<[hidden email]> wrote:

> Philip Covington wrote:
>>
>> On Tue, Dec 2, 2008 at 8:05 AM, David Woolley (E.L)
>
>>
>> It may seem that way to you, but in real life it turns out the the DDS
>> generates spurs due to only approximating a sin function, clock
>> leakage, number of bits, etc...  There is not the same issue in the
>
> There is no fundamental reason why the the DDS sine function should be any
> worse than the DSP one, nor for it to have any less bits than that with
> which the signal is digitised.
>
> It might be that commonly used DDS chips are rather old technology, and you
> are comparing state of the art DSP with ten year old DDS.

Not really.

While there is no fundamental reason, we have to deal with what
hardware DDS chips are available to designers with their 32/48 bit
accumulators and their 10/12/14 bit DACs.

Even the very latest DDS chips from Analog (ex. AD9910, AD9912), while
better than the older DDS chips by far, still have spurs at certain
programmed frequencies if using a very good low phase noise 1 GHz
clock and if you use the internal PLL to multiply up to the 1 GHz
clock rate still have phase noise issues.

--
Phil Covington
Software Radio Laboratory LLC
Columbus, Ohio
http://www.srl-llc.com
_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[P.B. Christensen]

>Rig     1kHz    2   10   20   50   100   1M
>K3      -110  -119 -136 -140 -143 -144  -150
>IC7800  -103  -112 -130 -138 -140 -140  -140
>FT2000  -102  -105 -128 -129 -128 -128  -128
>ORION 2 -121  -129 -126 -125 -118 -128  -138
>OMNI 7  -102  -103 -120 -123 -127 -129  -126"
>F5000   -123 (same at all spacings de W4ZV)
>Perseus   ?

 

Apologies for being a "stickler" on the details, but I see no units of measure above -- and moreover, QST has failed to show units of measure on the vertical scale of their product review graphs. 

 

Does this mean that the Flex 5000's phase noise power is -123 dBm referenced to some noise B/W and thus -144 dBc/Hz (i.e., -123 dB + (-21 dB))?  Or, is it -123 dBc/Hz? 

 

Paul, W9AC

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com

[John A. McCabe]

Hi Guy's

Thanks for the interesting discussion on the state of the art in SDR
technology. This discussion does raise a few questions in my mind that I
would like to ask. To me it is a question of the theoretical as opposed
to the practical. Does it really matter if a receiver can hold up to a
80db over 9 signal 2 Khz away or is this just a theoretical exercise?
Would not the transmitted Phase Noise, IMD, and perhaps key clicks be
the limiting factor or am I wrong about this?

Is there is a point of no return where increased close in dynamic range
no long matters because modern transmitters simply can not or do not
transmit a signal that clean. If so, then has the K3 and the Perseus
reached that point of no return and the question of which one has a
better close in dynamic range is more of a theoretical then a practical
exercise.

I would think that with very strong signals (like 80 db over 9), the
more important number from a practical point of view would be at wider
spacings, as I would want to know if the receiver would hold up when I
move away from the phase noise and IMD of the transmitted signal. I have
a neighbor ham that I receive a very strong signal from. It would be
impossible to operate 2 Khz from him regardless of receiver performance,
but I do operate on CW while he is on SSB on the same band using the K3
and I do not hear a thing. In my case, wide spacing performance is
important to me, and is a concern that I have when considering
purchasing a Windows based SDR. When I see wide spaced numbers of
117-123, compared to the 140db plus of the K3, I feel that the wider
spaced number are the more critical numbers to look at, or am I wrong
about that?

73,

John, KD8K

_______________________________________________
Elecraft mailing list
Post to: [hidden email]
You must be a subscriber to post to the list.
Subscriber Info (Addr. Change, sub, unsub etc.):
 http://mailman.qth.net/mailman/listinfo/elecraft   

Help: http://mailman.qth.net/subscribers.htm
Elecraft web page: http://www.elecraft.com