Elecraft › [K3]

APF

Classic

List

Threaded

30 messages Options

Barry N1EU

Re: APF measurements: FW 4.17 better results by 2 dB

4.16 APF has the same response curve as all the other versions.

BArry N1EU

On Thu, Jan 27, 2011 at 8:16 PM, Barry N1EU <[hidden email]> wrote:

> I'll check the APF in 4.16 after work and see if it's any different.
>
> Barry N1EU
>
>
> On Thu, Jan 27, 2011 at 7:58 PM, Steve Ellington [via Elecraft] <
> [hidden email]<ml-node%[hidden email]>
> > wrote:
>
>> FW pre-beta 4.16 release was actually the first to have APF, not 4.17.
>> Here's an interesting comment made shortly APF was finally released to the
>>
>> public.
>> http://elecraft.365791.n2.nabble.com/K3-APF-td5735159.html#a5736718<http://elecraft.365791.n2.nabble.com/K3-APF-td5735159.html?by-user=t#a5736718>
>>
>>
>>
>> Steve
>> N4LQ
>> ----- Original Message -----
>> From: "Arie Kleingeld PA3A" <[hidden email]<http://user/SendEmail.jtp?type=node&node=5967599&i=0>>
>>
>> To: <[hidden email]<http://user/SendEmail.jtp?type=node&node=5967599&i=1>>
>>
>> Sent: Thursday, January 27, 2011 6:44 AM
>> Subject: [Elecraft] APF measurements: FW 4.17 better results by 2 dB
>>
>>
>> > Hello all,
>> >
>> > Did some measurements on the different frimware versions. This time not
>> > a BW-plot but I measured S+N/N ratio's.
>> >
>> >
>> > I have no lab equipment but I have a K3 with dBV measurements, XG2 and
>> > the stepped attenuator from elecraft.
>> > I measured the difference of the K3 dBV-reading in signal off (=N) and
>> > signal on (=S+N)
>> > AGC off, mode CW, 400Hz BW and 400Hz Roofing.
>> > RX gain of the K3 has been calibrated with XG2 and the K3 utility.
>> >
>> >
>> > This is what I found.
>> >
>> > _FW 4.25 and 4.21 (both give the same measurement values):_
>> > At MDS level (stepped att on 20dB to reach this level)
>> > normal 400 Hz BW: S+N/N = 3dB (as expected, hi)
>> > 400 Hz BW + dual PB: S+N/N = 10 dB
>> >
>> > Now signal level 6dB under MDS (26dB att instead of 20dB):
>> > normal 400 Hz BW: S+N/N = 0.5 dB
>> > 400 Hz BW + dual PB: S+N/N = 4 dB
>> >
>> >
>> > _FW 4.17_
>> > At MDS level (stepped att on 20dB)
>> > normal 400 Hz BW: S+N/N = 3dB (as expected)
>> > 400 Hz BW + dual PB: S+N/N = 12 dB (THIS is 2 dB BETTER)
>> >
>> > Now signal level 6dB under MDS:
>> > normal 400 Hz BW: S+N/N = 0.5 dB
>> > 400 Hz BW + dual PB: S+N/N = 6 dB (THIS is 2dB BETTER)
>> >
>> >
>> > There are probably people that can measure this far more better than I
>> > did with better equipment.
>> > But this is what I can come up with here.
>> >
>> > 73,
>> > Arie PA3A
>> >
>> >
>> >
>> > Op 27-1-2011 0:52, Barry N1EU schreef:
>> >> Forget it - full stop!
>> >>
>> >> I just ran a spectral analysis of the current firmware versus the very
>> >> first
>> >> APF implementation (mcu 4.17 dsp 2.65) from 11/3/10 AND THE PLOTS ARE
>> >> ABSOLUTELY CONGRUOUS! The APF implementation hasn't changed a bit (or
>> a
>> >> byte).
>> >>
>> >> Have a look: http://n1eu.com/k3apf.gif
>> >>
>> >> magenta and green plots are taken with width = 200hz
>> >> aqua and yellow plots are taken with width = 100hz
>> >>
>> >> Barry N1EU
>> > ______________________________________________________________
>> > Elecraft mailing list
>> > Home: http://mailman.qth.net/mailman/listinfo/elecraft
>> > Help: http://mailman.qth.net/mmfaq.htm
>> > Post: mailto:[hidden email]<http://user/SendEmail.jtp?type=node&node=5967599&i=2>
>> >
>> > 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]<http://user/SendEmail.jtp?type=node&node=5967599&i=3>
>>
>> This list hosted by: http://www.qsl.net
>> Please help support this email list: http://www.qsl.net/donate.html
>>
>>
>> ------------------------------
>> If you reply to this email, your message will be added to the discussion
>> below:
>> http://elecraft.365791.n2.nabble.com/APF-tp5963894p5967599.html
>> To unsubscribe from APF, click here<
>>
>>
>
>

______________________________________________________________
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: APF measurements: FW 4.17 better results by 2 dB

Barry N1EU wrote

4.16 APF has the same response curve as all the other versions.

Q.E.D. If it has an identical response curve it has identical S/N performance. Anything else is either differences in noise conditions or human imagination.

73, Bill

Don Wilhelm-4

Re: APF measurements: FW 4.17 better results by 2 dB

But Bill,

I had a sage manager many years ago who often stated "Perception is 99%
of reality". Being of a scientific mind, I had my doubts at the time
thinking "people will be convinced by the facts of science".
In later years, I have mellowed and now agree with that manager.
Perception IS 99% of reality!

73,
Don W3FPR

On 1/27/2011 7:28 PM, Bill W4ZV wrote:

>
> Barry N1EU wrote:
>> 4.16 APF has the same response curve as all the other versions.
>>
> Q.E.D. If it has an identical response curve it has identical S/N
> performance. Anything else is either differences in noise conditions or
> human imagination.
>
> 73, Bill
>
>

Bill W4ZV

Re: APF measurements: FW 4.17 better results by 2 dB

Don Wilhelm-4 wrote

I had a sage manager many years ago who often stated "Perception is 99%
of reality". Being of a scientific mind, I had my doubts at the time
thinking "people will be convinced by the facts of science".
In later years, I have mellowed and now agree with that manager.
Perception IS 99% of reality!

OK let's follow your logic. Let's assume Elecraft now wants to duplicate the "magical" 4.16 APF performance. The firmware code is identical according to the author and the frequency response curve is identical. What do you suggest they do? Sprinkle pixie dust?

73, Bill

Don Wilhelm-4

Re: APF measurements: FW 4.17 better results by 2 dB

Bill,

I am not suggesting Elecraft do anything other than the explanations
that have already transpired.
Yet even with such dialog and tests that say "Q. E. D.", there will be
many who remain unconvinced, and will want to stick with the old version
"because it is better". They will not be convinced until they have
received countering information with their own ears.

There may be other factors involved -- although there were no
intentional changes in the DSP firmware to change the noise response of
the K3, that is one possible factor that comes to mind. Another factor
is operator induced -- it is known that the 50 Hz filter has some
ringing which will change minute by minute with band noise. Add APF on
top of that, and the perception from one moment to another can be all
"over the map".

I believe the test data, and look seriously for other factors that may
influence the perception of "better or worse" when confronted with the
facts that things (like APF) are equal. I believe Wayne has explained
that there are many variables (QSB, band noise) that can influence the
final APF response, and to my mind, that is sufficient. But there will
be those who will remain unconvinced and will continue to use 4.16
because the APF is "better". There is no cure for perceptions - it is
just something to be accepted.

Pixie dust may help, measurements will not.

73,
Don W3FPR

On 1/27/2011 7:55 PM, Bill W4ZV wrote:

>
> Don Wilhelm-4 wrote:
>> I had a sage manager many years ago who often stated "Perception is 99%
>> of reality". Being of a scientific mind, I had my doubts at the time
>> thinking "people will be convinced by the facts of science".
>> In later years, I have mellowed and now agree with that manager.
>> Perception IS 99% of reality!
>>
> OK let's follow your logic. Let's assume Elecraft now wants to duplicate
> the "magical" 4.16 APF performance. The firmware code is identical
> according to the author and the frequency response curve is identical. What
> do you suggest they do? Sprinkle pixie dust?
>
> 73, Bill

Arie Kleingeld PA3A-2

APF measurements: FW 4.17 vs FW4.25 revisited

In reply to this post by Arie Kleingeld PA3A-2

All,

Measured the two FW's 4.17 and 4.25 with APF on S+N/N ratio at MDS
again, and again.
Question was: if the filter plots are exactly the same, where does the
measured (marginal) difference of 2 dB come from?

Found no difference this time. Seems that the XG2 drifted a bit in
yesterdays measurements which, I think, caused the 2dB difference
(despite tuning the K3 on the right freq before measurement started).

Positive thing is that I could measure the improvement of S+N/N ratio
with the APF switched on. Even with the 200Hz roofer and 50Hz DSP BW it
adds a few dB. So for me it is a good tool for digging holes in the
noise looking for weak signals.

73,
Arie PA3A

______________________________________________________________
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: APF measurements: FW 4.17 vs FW4.25 revisited

PA3A wrote:

> Measured the two FW's 4.17 and 4.25 with APF on S+N/N ratio at MDS
again, and again.
Question was: if the filter plots are exactly the same, where does the
measured (marginal) difference of 2 dB come from?

> Found no difference this time. Seems that the XG2 drifted a bit in
yesterdays measurements which, I think, caused the 2dB difference
(despite tuning the K3 on the right freq before measurement started).

Measurements don't lie when done properly. I should have thought about drift since peaking APF is so critical.

I believe the lesson to be learned is to not blame the APF if it doesn't work like our memories think it did previously. Look elsewhere for other reasons...such as:

1. The character of the noise (atmospheric vs galactic).
2. Operator error (gain and/or bandwidth settings, tuning errors, etc).

Human nature is to blame anything but ourselves.
http://www.igopogo.com/we_have_met.htm

> Positive thing is that I could measure the improvement of S+N/N ratio
with the APF switched on. Even with the 200Hz roofer and 50Hz DSP BW it
adds a few dB. So for me it is a good tool for digging holes in the
noise looking for weak signals.

Yes it definitely helps. Conditions on 160 were excellent last night so hopefully they will hold for the contest. RV9CX, UA9MA and 4L/UT5EO were all worked with decent signals. I also heard UN5J in deep Central Asia (near the BY border) for the first time since 1997!

73, Bill

wayne burdick

APF re-check: Rev. 4.17 vs. 4.25 (no difference found)

Administrator

In reply to this post by Arie Kleingeld PA3A-2

Lyle (KK7P), our DSP engineer--a very meticulous guy--just did a very
careful re-test of the two versions of code referred to in this
thread. He found no difference between 4.17 and 4.25.

I see that Arie (PA3A) also confirmed no difference when he repeated
his own test. (Thanks for doing this, Arie.)

73,
Wayne
N6KR

On Jan 28, 2011, at 2:51 AM, Arie Kleingeld PA3A wrote:

> All,
>
> Measured the two FW's 4.17 and 4.25 with APF on S+N/N ratio at MDS
> again, and again.
> Question was: if the filter plots are exactly the same, where does the
> measured (marginal) difference of 2 dB come from?
>
> Found no difference this time. Seems that the XG2 drifted a bit in
> yesterdays measurements which, I think, caused the 2dB difference
> (despite tuning the K3 on the right freq before measurement started).
>
> Positive thing is that I could measure the improvement of S+N/N ratio
> with the APF switched on. Even with the 200Hz roofer and 50Hz DSP BW
> it
> adds a few dB. So for me it is a good tool for digging holes in the
> noise looking for weak signals.
>
> 73,
> Arie PA3A
>
>
>
>
> ______________________________________________________________
> 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

KD8NNU

Re: APF re-check: Rev. 4.17 vs. 4.25 (no difference found)

I am not smart enough to comment regarding the real stuff going on
behind the sceines in the code.

However, since the new update the NR and NB pretty much take care of my
Plasma TV noise.

So there is something that changed in the combination and I say thanks
for this improvement.

Don
KD8NNU

On Fri, Jan 28, 2011 at 11:57 AM, Wayne Burdick wrote:

> Lyle (KK7P), our DSP engineer--a very meticulous guy--just did a very
> careful re-test of the two versions of code referred to in this
> thread. He found no difference between 4.17 and 4.25.
>
> I see that Arie (PA3A) also confirmed no difference when he repeated
> his own test. (Thanks for doing this, Arie.)
>
> 73,
> Wayne
> N6KR
>
>
> On Jan 28, 2011, at 2:51 AM, Arie Kleingeld PA3A wrote:
>
>> All,
>>
>> Measured the two FW's 4.17 and 4.25 with APF on S+N/N ratio at MDS
>> again, and again.
>> Question was: if the filter plots are exactly the same, where does
>> the
>> measured (marginal) difference of 2 dB come from?
>>
>> Found no difference this time. Seems that the XG2 drifted a bit in
>> yesterdays measurements which, I think, caused the 2dB difference
>> (despite tuning the K3 on the right freq before measurement started).
>>
>> Positive thing is that I could measure the improvement of S+N/N ratio
>> with the APF switched on. Even with the 200Hz roofer and 50Hz DSP BW
>> it
>> adds a few dB. So for me it is a good tool for digging holes in the
>> noise looking for weak signals.
>>
>> 73,
>> Arie PA3A
>>
>>
>>
>>
>> ______________________________________________________________
>> 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

Eric Swartz - WA6HHQ

Re: APF measurements: FW 4.17 better results by 2 dB

Administrator

In reply to this post by Bill W4ZV

Bill, the K3s dBv and AF Mv meters are true RMS. We added that feature for automated MDS testing in K3 production.

73,
Eric

www.elecraft.com
_..._

On Jan 27, 2011, at 11:17 AM, Bill W4ZV <[hidden email]> wrote:

>
>
> Arie Kleingeld PA3A-2 wrote:
>>
>> I have no lab equipment but I have a K3 with dBV measurements, XG2 and
>> the stepped attenuator from elecraft.
>>
>
> Arie I believe you may have some measurement error. I'm sure you know that
> noise can only be measured correctly with a True RMS responding
> meter...which is not what is in the K3 dBV meter. This is the only way I
> can reconcile any difference based on N1EU's measurements:
>
>>