Bad Spot - Example of Alias issue

[dtheriault]

In the following dump two stations incorrectly reporting a spot exactly 96 khz off from others. This is example of the alias problem I have seen on occasion. Unclear what the cause is for this effect.

Spot was 14.5kc above where skimmer band edge should be (7091).

de dx freq cq/dx snr speed time K1TTT NN0D 7105.5 CW CQ 10 dB 18 wpm 0636z 23 Mar offset 0, v4.0 N7TR NN0D 7105.5 CW CQ 21 dB 18 wpm 0636z 23 Mar no offsets, v4.0 NO1D NN0D 7009.5 CW CQ 4 dB 18 wpm 0636z 23 Mar offset 0, v4.0 W4KKN NN0D 7105.5 CW CQ 26 dB 18 wpm 0636z 23 Mar offset 0, v4.0 KM3T NN0D 7105.5 CW CQ 7 dB 19 wpm 0636z 23 Mar no report, v4.0

K5TDA NN0D 7009.5 CW CQ 4 dB 18 wpm 0628z 23 Mar offset 0, v4.0

N7TR NN0D 7105.5 CW CQ 19 dB 18 wpm 0625z 23 Mar KM3T NN0D 7105.5 CW CQ 5 dB 18 wpm 0625z 23 Mar K1TTT NN0D 7105.5 CW CQ 11 dB 18 wpm 0625z 23 Mar

Skimsvr.ini

[Skimmer] CenterFreqs48=1822750,3522750,3568250,7022750,10122750,14022750,14068250,18090750,21022750,21068250,24912750\ ,28022750,28068250,50022750,50068250,50113750,50159250 CenterFreqs96=1845500,3545500,7045500,10145500,14045500,18113500,21045500,24935500,28045500,28136500,5004550\ 0,50136500 CenterFreqs192=1891000,3591000,7091000,10191000,14091000,18159000,21091000,24981000,28091000,50091000 SegmentSel48=00000100100000000 SegmentSel96=111111000000 SegmentSel192=1111111000 CwSegments=1800000-1890000,3500000-3600000,7000000-7070000,10100000-10130000,14000000-14120000,18068000-1809\ 5000,21000000-21110000,24890000-24920000,28000000-28150000,50000000-50110000

Rate=1 FreqCalibration=1

Questions:

1. Why are we detecting station above upper band edge? Upper band edge should be 7091kc
2. What is the magic 14.5kc. Some sampling rate super sub harmonic? Ie: 28.8mhz ?
3. Is this caused by Alex's skimmer software? ie: perhaps misconfig in my CwSegments?

[n1gp]

Not sure what the magic is about the 14.5kc but there are definitely aliasing issues with the rtl dongles.

https://www.google.com/search?q=rtl+sdr+aliasing&ie=utf-8&oe=utf-8

A good read if you haven't already is:

http://www.sm5bsz.com/linuxdsp/hware/rtlsdr/rtlsdr-03.40.htm http://rtlsdrblog.rtlsdrblog.netdna-cdn.com/wp-content/uploads/2013/08/Some-Measurements-on-E4000-and-R820-Tuners.pdf

The halfband FIR filters I'm using are fairly small. I could make them tighter but that would eat up precious OC1 cpu cycles hence less rcvrs.

But just to see if it is due to improper filtering,

I believe you have all the necessary equipment, maybe you could setup a bench scenario with a couple signal generators and see if you could place a signal outside the passband and detect it using cudasdr, then compare using gqrx. gqrx doesn't filter as it displays the full spectrum from the dongle so if you see it there and cudasdr then it's the dongle.

The aliasing effects get worse as you crank up the gain as well.

The spotting offset, especially if its consistently 14.5kc off seems to be more of a filtering problem as the cutoff of the passband isn't all that sharp.

Date: Mon, 23 Mar 2015 00:16:39 -0700 From: notifications@github.com To: medusa@noreply.github.com Subject: [medusa] Bad Spot - Example of Alias issue (#11)

In the following dump two stations incorrectly reporting a spot exactly 96 khz off from others. This is example of the alias problem I have seen on occasion. Unclear what the cause is for this effect.

Spot was 14.5kc above where skimmer band edge should be (7091).

de dx freq cq/dx snr speed time

K1TTT NN0D 7105.5 CW CQ 10 dB 18 wpm 0636z 23 Mar offset 0, v4.0

N7TR NN0D 7105.5 CW CQ 21 dB 18 wpm 0636z 23 Mar no offsets, v4.0

NO1D NN0D 7009.5 CW CQ 4 dB 18 wpm 0636z 23 Mar offset 0, v4.0

W4KKN NN0D 7105.5 CW CQ 26 dB 18 wpm 0636z 23 Mar offset 0, v4.0

KM3T NN0D 7105.5 CW CQ 7 dB 19 wpm 0636z 23 Mar no report, v4.0

K5TDA NN0D 7009.5 CW CQ 4 dB 18 wpm 0628z 23 Mar offset 0, v4.0

N7TR NN0D 7105.5 CW CQ 19 dB 18 wpm 0625z 23 Mar

KM3T NN0D 7105.5 CW CQ 5 dB 18 wpm 0625z 23 Mar

K1TTT NN0D 7105.5 CW CQ 11 dB 18 wpm 0625z 23 Mar

Skimsvr.ini

[Skimmer]

CenterFreqs48=1822750,3522750,3568250,7022750,10122750,14022750,14068250,18090750,21022750,21068250,24912750\

,28022750,28068250,50022750,50068250,50113750,50159250

CenterFreqs96=1845500,3545500,7045500,10145500,14045500,18113500,21045500,24935500,28045500,28136500,5004550\

0,50136500

CenterFreqs192=1891000,3591000,7091000,10191000,14091000,18159000,21091000,24981000,28091000,50091000

SegmentSel48=00000100100000000

SegmentSel96=111111000000

SegmentSel192=1111111000

CwSegments=1800000-1890000,3500000-3600000,7000000-7070000,10100000-10130000,14000000-14120000,18068000-1809\

5000,21000000-21110000,24890000-24920000,28000000-28150000,50000000-50110000

Rate=1

FreqCalibration=1

Questions:

Why are we detecting station above upper band edge? Upper band edge should be 7091kc What is the magic 14.5kc. Some sampling rate super sub harmonic? Ie: 28.8mhz ? Is this caused by Alex's skimmer software? ie: perhaps misconfig in my CwSegments?

— Reply to this email directly or view it on GitHub.

[dtheriault]

Another example today on 10m.

K4MTP/B at 28203, was reported by me as 28011.2. This is a delta of 192hz which is 2x the 96kc bandwidth skimmer is set for. Gain was fairly high, 372. On the RX340 I see the beacon as solid S8/9 so he's pretty strong but not killing me. The Spot at 28011.2 was reported as 0dB so was right at the noise level.

Fired up another dongle on port 8 of the coupler and cannot detect him at 28011.2 even with gain all the way up.

Its up to you if you want to create a branch I could try or just send me a patch file with the changes I need to make in my workspace. I'd love to try get a better understanding of how the filter works. There used to be a FIR setting in config file and believe its no longer required so guessing you just took this code out. I don't have it set in my conf. file.

[n1gp]

I just checked in a change that filters by a factor of 3 on the incoming data from the RTL.

Give it a try. If it makes a difference I'll add switches to change the factor.

Date: Tue, 24 Mar 2015 10:42:20 -0700 From: notifications@github.com To: medusa@noreply.github.com CC: n1gp@hotmail.com Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

Another example today on 10m.

K4MTP/B at 28203, was reported by me as 28011.2. This is a delta of 192hz which is 2x the 96kc bandwidth skimmer is set for. Gain was fairly high, 372. On the RX340 I see the beacon as solid S8/9 so he's pretty strong but not killing me. The Spot at 28011.2 was reported as 0dB so was right at the noise level.

Fired up another dongle on port 8 of the coupler and cannot detect him at 28011.2 even with gain all the way up.

Its up to you if you want to create a branch I could try or just send me a patch file with the changes I need to make in my workspace. I'd love to try get a better understanding of how the filter works. There used to be a FIR setting in config file and believe its no longer required so guessing you just took this code out. I don't have it set in my conf. file.

— Reply to this email directly or view it on GitHub.

[dtheriault]

Another 192kc alias today. Using:

GNU rtl_hpsdr Version 1.8 Date Built 08:51:22 Mar 25 2015 GIT Hash 7fb8dc0

I SPOT: DX de NO1D-#: 28037.5 KA2LIM/B 3 dB 22 WPM DE 1927Z

RBN Reports: HA5PP KA2LIM 28229.4 CW BCN [LoTW] 13 dB 22 wpm 1425z 25 Mar VK3FFB KA2LIM 28229.3 CW BCN [LoTW] 6 dB 22 wpm 2318z 22 Mar

As you suggest I need to put the medusa on the RF bench and measure performance with signal generator at various amplitudes and use cuSDR to visualize it. Using single dongle, gqrx and rtl_tcp, I do not see any aliases of beacons showing up in the lower part of spectrum even with gain turned all the way up. I'm wondering if the image is possibly caused by the HamItUp Mixer and not function of the dongle at all? The PA beacon today is quite loud and no report of him in the spots. So perhaps you tightened up the filter skirt and this particular station is now gone...

[n1gp]

I'll try to make the filter as tight as I can and still get a couple rcvrs @192k

If it is a product of loose filtering I'm afraid we'll have to ditch the OC1. Maybe we can better use the Jetson and it's cuda cores to do all the filtering and decimation but with much better filtering?

I'm still curious what you may find on the bench testing the RTL's virgin signal processing.

Date: Wed, 25 Mar 2015 13:00:20 -0700 From: notifications@github.com To: medusa@noreply.github.com CC: n1gp@hotmail.com Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

Another 192kc alias today. Using:

GNU rtl_hpsdr Version 1.8 Date Built 08:51:22 Mar 25 2015 GIT Hash 7fb8dc0

I SPOT:

DX de NO1D-#: 28037.5 KA2LIM/B 3 dB 22 WPM DE 1927Z

RBN Reports:

HA5PP KA2LIM 28229.4 CW BCN [LoTW] 13 dB 22 wpm 1425z 25 Mar

VK3FFB KA2LIM 28229.3 CW BCN [LoTW] 6 dB 22 wpm 2318z 22 Mar

As you suggest I need to put the medusa on the RF bench and measure performance with signal generator at various amplitudes and use cuSDR to visualize it. Using single dongle, gqrx and rtl_tcp, I do not see any aliases of beacons showing up in the lower part of spectrum even with gain turned all the way up. I'm wondering if the image is possibly caused by the HamItUp Mixer and not function of the dongle at all? The PA beacon today is quite loud and no report of him in the spots. So perhaps you tightened up the filter skirt and this particular station is now gone...

— Reply to this email directly or view it on GitHub.

[dtheriault]

DX de NO1D-#: 28040.0 W4CLM/B 3 dB 15 WPM DE 2055Z

He's really at 28232.150 Again this is another 192kc alias example. He's fairly strong today.

[n1gp]

Can you try my old code, it had a fairly sharp filter and used a different technique, i.e. it did the full downsample in one call with filtering rather than multiple halfband stages.

I'd be curious to see the difference, that would tell me it's in the filtering algorithm.

Date: Wed, 25 Mar 2015 14:02:32 -0700 From: notifications@github.com To: medusa@noreply.github.com CC: n1gp@hotmail.com Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

DX de NO1D-#: 28040.0 W4CLM/B 3 dB 15 WPM DE 2055Z

He's really at 28232.150 Again this is another 192kc alias example. He's fairly strong today.

— Reply to this email directly or view it on GitHub.

[dtheriault]

Ok now running with version from 6463f13cdbb07530d9be8910ae7b478193d0bef7

its so easy in git. just did a

git checkout 6463f13cdbb07530d9be8910ae7b478193d0bef7

and rebuilt and installed. sweet.

Old code seemed to be picking up more: DX de NO1D-#: 28046.0 KB2SEO/B 5 dB 15 WPM DE 2106Z DX de NO1D-#: 28037.4 KA2LIM/B 3 dB 22 WPM 2107Z

Ya bench test of medusa tonight/tomorrow. About to take Dad out to Texas Roadhouse for his birthday meal. Back in couple hours...

Oh ya, with 6 dongles running 96kc, we're at 60% load. Room for more. And I still want to someday fiddle with the code once its ready and see if I can tweak it for a few more cycles. Perhaps inline segments or making sure things are cache aligned for this processor. That made a huge difference on Intels, should do same on ARM but not 100% sure.

Oh Oh ya, I figured out how to talk to the darn TI eval board this morning. Its a bit of a weird i2c write sequence but was finally able to dump out its registers/config. So I'll be programming the CDE913 chip and see if I can generate much more stable clocks than what's been running on the si5351a boards. I want this timed off Rb std so there's no excuse for drifting spots !!!

 Thanks Rick. -dt-

On Mar 25, 2015, at 02:08 PM, Rick Koch notifications@github.com wrote:

Can you try my old code, it had a fairly sharp filter and used a different technique, i.e. it did the full downsample in one call with filtering rather than multiple halfband stages.

I'd be curious to see the difference, that would tell me it's in the filtering algorithm.

Date: Wed, 25 Mar 2015 14:02:32 -0700 From: notifications@github.com To: medusa@noreply.github.com CC: n1gp@hotmail.com Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

DX de NO1D-#: 28040.0 W4CLM/B 3 dB 15 WPM DE 2055Z

He's really at 28232.150 Again this is another 192kc alias example. He's fairly strong today.

— Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

[dtheriault]

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28 2015-03-25 16:34:44Z 28011.2 K4MTP/B 25-Mar-2015 1634Z 2 dB 18 WPM <NO1D-#> 2015-03-25 19:27:45Z 28037.5 KA2LIM/B 25-Mar-2015 1927Z 3 dB 22 WPM DE <NO1D-#> 2015-03-25 20:28:33Z 28050.7 KB4W/B 25-Mar-2015 2028Z 5 dB 16 WPM DE <NO1D-#> 2015-03-25 20:39:06Z 28040.0 W4CLM/B 25-Mar-2015 2039Z 4 dB 15 WPM DE <NO1D-#> 2015-03-25 20:39:34Z 28050.7 KB4W/B 25-Mar-2015 2039Z 5 dB 16 WPM DE <NO1D-#> 2015-03-25 20:55:11Z 28040.0 W4CLM/B 25-Mar-2015 2055Z 3 dB 15 WPM DE <NO1D-#> 2015-03-25 21:05:15Z 28040.0 W4CLM/B 25-Mar-2015 2105Z 3 dB 15 WPM DE <NO1D-#> 2015-03-25 21:06:54Z 28046.0 KB2SEO/B 25-Mar-2015 2106Z 5 dB 15 WPM DE <NO1D-#> 2015-03-25 21:07:25Z 28037.4 KA2LIM/B 25-Mar-2015 2107Z 3 dB 22 WPM <NO1D-#> 2015-03-25 22:51:41Z 28012.8 KE4TWI/B 25-Mar-2015 2251Z 1 dB 14 WPM <NO1D-#> doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28 | wc -l 10 doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28 | wc -l 4 doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28 2015-03-26 14:43:20Z 28011.0 K4MTP/B 26-Mar-2015 1443Z 0 dB 18 WPM DE <NO1D-#> 2015-03-26 16:11:14Z 28010.9 K4MTP/B 26-Mar-2015 1611Z 4 dB 18 WPM DE <NO1D-#> 2015-03-26 16:46:59Z 28010.9 K4MTP/B 26-Mar-2015 1646Z 3 dB 18 WPM <NO1D-#> 2015-03-26 16:56:59Z 28010.9 K4MTP/B 26-Mar-2015 1656Z 3 dB 18 WPM DE <NO1D-#>

Granted still early... Today running old code, yesterday was running with the modified code.

[dtheriault]

Whoopsie, hit wrong button

[n1gp]

What version of old code are you running. I was referring to REAL old, like https://github.com/n1gp/librtlsdr

Date: Thu, 26 Mar 2015 10:06:59 -0700 From: notifications@github.com To: medusa@noreply.github.com CC: n1gp@hotmail.com Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28

2015-03-25 16:34:44Z 28011.2 K4MTP/B 25-Mar-2015 1634Z 2 dB 18 WPM

2015-03-25 19:27:45Z 28037.5 KA2LIM/B 25-Mar-2015 1927Z 3 dB 22 WPM DE

2015-03-25 20:28:33Z 28050.7 KB4W/B 25-Mar-2015 2028Z 5 dB 16 WPM DE

2015-03-25 20:39:06Z 28040.0 W4CLM/B 25-Mar-2015 2039Z 4 dB 15 WPM DE

2015-03-25 20:39:34Z 28050.7 KB4W/B 25-Mar-2015 2039Z 5 dB 16 WPM DE

2015-03-25 20:55:11Z 28040.0 W4CLM/B 25-Mar-2015 2055Z 3 dB 15 WPM DE

2015-03-25 21:05:15Z 28040.0 W4CLM/B 25-Mar-2015 2105Z 3 dB 15 WPM DE

2015-03-25 21:06:54Z 28046.0 KB2SEO/B 25-Mar-2015 2106Z 5 dB 15 WPM DE

2015-03-25 21:07:25Z 28037.4 KA2LIM/B 25-Mar-2015 2107Z 3 dB 22 WPM

2015-03-25 22:51:41Z 28012.8 KE4TWI/B 25-Mar-2015 2251Z 1 dB 14 WPM

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28 | wc -l

10

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28 | wc -l

4

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28

2015-03-26 14:43:20Z 28011.0 K4MTP/B 26-Mar-2015 1443Z 0 dB 18 WPM DE

2015-03-26 16:11:14Z 28010.9 K4MTP/B 26-Mar-2015 1611Z 4 dB 18 WPM DE

2015-03-26 16:46:59Z 28010.9 K4MTP/B 26-Mar-2015 1646Z 3 dB 18 WPM

2015-03-26 16:56:59Z 28010.9 K4MTP/B 26-Mar-2015 1656Z 3 dB 18 WPM DE

Granted still early... Today running old code, yesterday was running with the modified code.

— Reply to this email directly or view it on GitHub.

[dtheriault]

Whoops. Wasn't running that old code. Will tonight.

---

-dt-

On Mar 26, 2015, at 11:33, Rick Koch notifications@github.com wrote:

What version of old code are you running. I was referring to REAL old, like https://github.com/n1gp/librtlsdr

Date: Thu, 26 Mar 2015 10:06:59 -0700 From: notifications@github.com To: medusa@noreply.github.com CC: n1gp@hotmail.com Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28

2015-03-25 16:34:44Z 28011.2 K4MTP/B 25-Mar-2015 1634Z 2 dB 18 WPM

2015-03-25 19:27:45Z 28037.5 KA2LIM/B 25-Mar-2015 1927Z 3 dB 22 WPM DE

2015-03-25 20:28:33Z 28050.7 KB4W/B 25-Mar-2015 2028Z 5 dB 16 WPM DE

2015-03-25 20:39:06Z 28040.0 W4CLM/B 25-Mar-2015 2039Z 4 dB 15 WPM DE

2015-03-25 20:39:34Z 28050.7 KB4W/B 25-Mar-2015 2039Z 5 dB 16 WPM DE

2015-03-25 20:55:11Z 28040.0 W4CLM/B 25-Mar-2015 2055Z 3 dB 15 WPM DE

2015-03-25 21:05:15Z 28040.0 W4CLM/B 25-Mar-2015 2105Z 3 dB 15 WPM DE

2015-03-25 21:06:54Z 28046.0 KB2SEO/B 25-Mar-2015 2106Z 5 dB 15 WPM DE

2015-03-25 21:07:25Z 28037.4 KA2LIM/B 25-Mar-2015 2107Z 3 dB 22 WPM

2015-03-25 22:51:41Z 28012.8 KE4TWI/B 25-Mar-2015 2251Z 1 dB 14 WPM

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28 | wc -l

10

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28 | wc -l

4

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28

2015-03-26 14:43:20Z 28011.0 K4MTP/B 26-Mar-2015 1443Z 0 dB 18 WPM DE

2015-03-26 16:11:14Z 28010.9 K4MTP/B 26-Mar-2015 1611Z 4 dB 18 WPM DE

2015-03-26 16:46:59Z 28010.9 K4MTP/B 26-Mar-2015 1646Z 3 dB 18 WPM

2015-03-26 16:56:59Z 28010.9 K4MTP/B 26-Mar-2015 1656Z 3 dB 18 WPM DE

Granted still early... Today running old code, yesterday was running with the modified code.

— Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

[dtheriault]

Ok, old code now running after updating my config file to old variant.

Listening on 15m,12m,10m and will see how many 10m beacons we snarf today.  -dt-

On Mar 26, 2015, at 11:33 AM, Rick Koch notifications@github.com wrote:

What version of old code are you running. I was referring to REAL old, like https://github.com/n1gp/librtlsdr

Date: Thu, 26 Mar 2015 10:06:59 -0700 From: notifications@github.com To: medusa@noreply.github.com CC: n1gp@hotmail.com Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28

2015-03-25 16:34:44Z 28011.2 K4MTP/B 25-Mar-2015 1634Z 2 dB 18 WPM

2015-03-25 19:27:45Z 28037.5 KA2LIM/B 25-Mar-2015 1927Z 3 dB 22 WPM DE

2015-03-25 20:28:33Z 28050.7 KB4W/B 25-Mar-2015 2028Z 5 dB 16 WPM DE

2015-03-25 20:39:06Z 28040.0 W4CLM/B 25-Mar-2015 2039Z 4 dB 15 WPM DE

2015-03-25 20:39:34Z 28050.7 KB4W/B 25-Mar-2015 2039Z 5 dB 16 WPM DE

2015-03-25 20:55:11Z 28040.0 W4CLM/B 25-Mar-2015 2055Z 3 dB 15 WPM DE

2015-03-25 21:05:15Z 28040.0 W4CLM/B 25-Mar-2015 2105Z 3 dB 15 WPM DE

2015-03-25 21:06:54Z 28046.0 KB2SEO/B 25-Mar-2015 2106Z 5 dB 15 WPM DE

2015-03-25 21:07:25Z 28037.4 KA2LIM/B 25-Mar-2015 2107Z 3 dB 22 WPM

2015-03-25 22:51:41Z 28012.8 KE4TWI/B 25-Mar-2015 2251Z 1 dB 14 WPM

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28 | wc -l

10

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28 | wc -l

4

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28

2015-03-26 14:43:20Z 28011.0 K4MTP/B 26-Mar-2015 1443Z 0 dB 18 WPM DE

2015-03-26 16:11:14Z 28010.9 K4MTP/B 26-Mar-2015 1611Z 4 dB 18 WPM DE

2015-03-26 16:46:59Z 28010.9 K4MTP/B 26-Mar-2015 1646Z 3 dB 18 WPM

2015-03-26 16:56:59Z 28010.9 K4MTP/B 26-Mar-2015 1656Z 3 dB 18 WPM DE

Granted still early... Today running old code, yesterday was running with the modified code.

— Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

[n1gp]

Cool beans. I'm here slaving away at work. Trying to get our first beta release out to SQA today, oh boy, then the trouble begins...

Date: Fri, 27 Mar 2015 08:11:48 -0700 From: notifications@github.com To: medusa@noreply.github.com CC: n1gp@hotmail.com Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

Ok, old code now running after updating my config file to old variant.

Listening on 15m,12m,10m and will see how many 10m beacons we snarf today.



-dt-

On Mar 26, 2015, at 11:33 AM, Rick Koch notifications@github.com wrote:

What version of old code are you running. I was referring to REAL old, like

https://github.com/n1gp/librtlsdr

Date: Thu, 26 Mar 2015 10:06:59 -0700

From: notifications@github.com

To: medusa@noreply.github.com

CC: n1gp@hotmail.com

Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28

2015-03-25 16:34:44Z 28011.2 K4MTP/B 25-Mar-2015 1634Z 2 dB 18 WPM

2015-03-25 19:27:45Z 28037.5 KA2LIM/B 25-Mar-2015 1927Z 3 dB 22 WPM DE

2015-03-25 20:28:33Z 28050.7 KB4W/B 25-Mar-2015 2028Z 5 dB 16 WPM DE

2015-03-25 20:39:06Z 28040.0 W4CLM/B 25-Mar-2015 2039Z 4 dB 15 WPM DE

2015-03-25 20:39:34Z 28050.7 KB4W/B 25-Mar-2015 2039Z 5 dB 16 WPM DE

2015-03-25 20:55:11Z 28040.0 W4CLM/B 25-Mar-2015 2055Z 3 dB 15 WPM DE

2015-03-25 21:05:15Z 28040.0 W4CLM/B 25-Mar-2015 2105Z 3 dB 15 WPM DE

2015-03-25 21:06:54Z 28046.0 KB2SEO/B 25-Mar-2015 2106Z 5 dB 15 WPM DE

2015-03-25 21:07:25Z 28037.4 KA2LIM/B 25-Mar-2015 2107Z 3 dB 22 WPM

2015-03-25 22:51:41Z 28012.8 KE4TWI/B 25-Mar-2015 2251Z 1 dB 14 WPM

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28 | wc -l

10

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28 | wc -l

4

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28

2015-03-26 14:43:20Z 28011.0 K4MTP/B 26-Mar-2015 1443Z 0 dB 18 WPM DE

2015-03-26 16:11:14Z 28010.9 K4MTP/B 26-Mar-2015 1611Z 4 dB 18 WPM DE

2015-03-26 16:46:59Z 28010.9 K4MTP/B 26-Mar-2015 1646Z 3 dB 18 WPM

2015-03-26 16:56:59Z 28010.9 K4MTP/B 26-Mar-2015 1656Z 3 dB 18 WPM DE

Granted still early... Today running old code, yesterday was running with the modified code.

—

Reply to this email directly or view it on GitHub.

—

Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

[dtheriault]

Well first off 10m band just faded out. I was hearing that K4MTP beacon strongly, now deep in the noise... figures...

Boy, the old code was no were as sensitive as the new code. After running old code for 15m or so, no spots reported. Plenty of decoders running, but nada reports. As soon as I re-launched the new code, poof, 15m and 12m spots came poring in.

So I thought I'd try and fire up the Icom into a dummy load and see if I can pick up my own spots. For some reason I'm not seeing myself. I can hear myself through the Rx340 so some signal is getting through but skimmer not picking up my CQ.

-dt-

On Mar 26, 2015, at 11:33 AM, Rick Koch notifications@github.com wrote:

What version of old code are you running. I was referring to REAL old, like https://github.com/n1gp/librtlsdr

Date: Thu, 26 Mar 2015 10:06:59 -0700 From: notifications@github.com To: medusa@noreply.github.com CC: n1gp@hotmail.com Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28

2015-03-25 16:34:44Z 28011.2 K4MTP/B 25-Mar-2015 1634Z 2 dB 18 WPM

2015-03-25 19:27:45Z 28037.5 KA2LIM/B 25-Mar-2015 1927Z 3 dB 22 WPM DE

2015-03-25 20:28:33Z 28050.7 KB4W/B 25-Mar-2015 2028Z 5 dB 16 WPM DE

2015-03-25 20:39:06Z 28040.0 W4CLM/B 25-Mar-2015 2039Z 4 dB 15 WPM DE

2015-03-25 20:39:34Z 28050.7 KB4W/B 25-Mar-2015 2039Z 5 dB 16 WPM DE

2015-03-25 20:55:11Z 28040.0 W4CLM/B 25-Mar-2015 2055Z 3 dB 15 WPM DE

2015-03-25 21:05:15Z 28040.0 W4CLM/B 25-Mar-2015 2105Z 3 dB 15 WPM DE

2015-03-25 21:06:54Z 28046.0 KB2SEO/B 25-Mar-2015 2106Z 5 dB 15 WPM DE

2015-03-25 21:07:25Z 28037.4 KA2LIM/B 25-Mar-2015 2107Z 3 dB 22 WPM

2015-03-25 22:51:41Z 28012.8 KE4TWI/B 25-Mar-2015 2251Z 1 dB 14 WPM

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28 | wc -l

10

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28 | wc -l

4

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28

2015-03-26 14:43:20Z 28011.0 K4MTP/B 26-Mar-2015 1443Z 0 dB 18 WPM DE

2015-03-26 16:11:14Z 28010.9 K4MTP/B 26-Mar-2015 1611Z 4 dB 18 WPM DE

2015-03-26 16:46:59Z 28010.9 K4MTP/B 26-Mar-2015 1646Z 3 dB 18 WPM

2015-03-26 16:56:59Z 28010.9 K4MTP/B 26-Mar-2015 1656Z 3 dB 18 WPM DE

Granted still early... Today running old code, yesterday was running with the modified code.

— Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

[dtheriault]

Good luck with the release. Remember, no P0 or P1 bugs :-)

DX de NO1D-#: 28012.3 W0WF/B 1 dB 13 WPM DE 1922Z

Running old code indeed picked one up. Bands been in/out today.

I got the ClockBuilder software running on my very old Windows board. Bit slow, but was able to program the eval board so it generates 3 28.8 clocks that are disciplined off 10mhz std (Rb osc.) Right now I have 3 dongles off that board and 3 off the si5358. Should not see any drift now at least on those 3 dongles.

-dt-

On Mar 27, 2015, at 08:38 AM, Rick Koch notifications@github.com wrote:

Cool beans. I'm here slaving away at work. Trying to get our first beta release out to SQA today, oh boy, then the trouble begins...

Date: Fri, 27 Mar 2015 08:11:48 -0700 From: notifications@github.com To: medusa@noreply.github.com CC: n1gp@hotmail.com Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

Ok, old code now running after updating my config file to old variant.

Listening on 15m,12m,10m and will see how many 10m beacons we snarf today.



-dt-

On Mar 26, 2015, at 11:33 AM, Rick Koch notifications@github.com wrote:

What version of old code are you running. I was referring to REAL old, like

https://github.com/n1gp/librtlsdr

Date: Thu, 26 Mar 2015 10:06:59 -0700

From: notifications@github.com

To: medusa@noreply.github.com

CC: n1gp@hotmail.com

Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28

2015-03-25 16:34:44Z 28011.2 K4MTP/B 25-Mar-2015 1634Z 2 dB 18 WPM

2015-03-25 19:27:45Z 28037.5 KA2LIM/B 25-Mar-2015 1927Z 3 dB 22 WPM DE

2015-03-25 20:28:33Z 28050.7 KB4W/B 25-Mar-2015 2028Z 5 dB 16 WPM DE

2015-03-25 20:39:06Z 28040.0 W4CLM/B 25-Mar-2015 2039Z 4 dB 15 WPM DE

2015-03-25 20:39:34Z 28050.7 KB4W/B 25-Mar-2015 2039Z 5 dB 16 WPM DE

2015-03-25 20:55:11Z 28040.0 W4CLM/B 25-Mar-2015 2055Z 3 dB 15 WPM DE

2015-03-25 21:05:15Z 28040.0 W4CLM/B 25-Mar-2015 2105Z 3 dB 15 WPM DE

2015-03-25 21:06:54Z 28046.0 KB2SEO/B 25-Mar-2015 2106Z 5 dB 15 WPM DE

2015-03-25 21:07:25Z 28037.4 KA2LIM/B 25-Mar-2015 2107Z 3 dB 22 WPM

2015-03-25 22:51:41Z 28012.8 KE4TWI/B 25-Mar-2015 2251Z 1 dB 14 WPM

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28 | wc -l

10

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28 | wc -l

4

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28

2015-03-26 14:43:20Z 28011.0 K4MTP/B 26-Mar-2015 1443Z 0 dB 18 WPM DE

2015-03-26 16:11:14Z 28010.9 K4MTP/B 26-Mar-2015 1611Z 4 dB 18 WPM DE

2015-03-26 16:46:59Z 28010.9 K4MTP/B 26-Mar-2015 1646Z 3 dB 18 WPM

2015-03-26 16:56:59Z 28010.9 K4MTP/B 26-Mar-2015 1656Z 3 dB 18 WPM DE

Granted still early... Today running old code, yesterday was running with the modified code.

—

Reply to this email directly or view it on GitHub.

—

Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

[n1gp]

Worked all day today and we still didn't get a release out. SNMP issues, not me :^)

That old code had a tight filter, lots of coefficients, but timing issues.

I figure if your seeing aliases with that code then it may be more having to do with the dongle.

Actually, hmmmm, I had a version of the rtl_hpsdr where I used the liquid dsp libraries for filtering and decimation. That would be a good test. I'll try to did that up...

Date: Fri, 27 Mar 2015 12:39:44 -0700 From: notifications@github.com To: medusa@noreply.github.com CC: n1gp@hotmail.com Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

Good luck with the release. Remember, no P0 or P1 bugs :-)

DX de NO1D-#: 28012.3 W0WF/B 1 dB 13 WPM DE 1922Z

Running old code indeed picked one up. Bands been in/out today.

I got the ClockBuilder software running on my very old Windows board. Bit slow, but was able to program the eval board so it generates 3 28.8 clocks that are disciplined off 10mhz std (Rb osc.) Right now I have 3 dongles off that board and 3 off the si5358. Should not see any drift now at least on those 3 dongles.

-dt-

On Mar 27, 2015, at 08:38 AM, Rick Koch notifications@github.com wrote:

Cool beans. I'm here slaving away at work. Trying to get our first

beta release out to SQA today, oh boy, then the trouble begins...

Date: Fri, 27 Mar 2015 08:11:48 -0700

From: notifications@github.com

To: medusa@noreply.github.com

CC: n1gp@hotmail.com

Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

Ok, old code now running after updating my config file to old variant.

Listening on 15m,12m,10m and will see how many 10m beacons we snarf today.



-dt-

On Mar 26, 2015, at 11:33 AM, Rick Koch notifications@github.com wrote:

What version of old code are you running. I was referring to REAL old, like

https://github.com/n1gp/librtlsdr

Date: Thu, 26 Mar 2015 10:06:59 -0700

From: notifications@github.com

To: medusa@noreply.github.com

CC: n1gp@hotmail.com

Subject: Re: [medusa] Bad Spot - Example of Alias issue (#11)

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28

2015-03-25 16:34:44Z 28011.2 K4MTP/B 25-Mar-2015 1634Z 2 dB 18 WPM

2015-03-25 19:27:45Z 28037.5 KA2LIM/B 25-Mar-2015 1927Z 3 dB 22 WPM DE

2015-03-25 20:28:33Z 28050.7 KB4W/B 25-Mar-2015 2028Z 5 dB 16 WPM DE

2015-03-25 20:39:06Z 28040.0 W4CLM/B 25-Mar-2015 2039Z 4 dB 15 WPM DE

2015-03-25 20:39:34Z 28050.7 KB4W/B 25-Mar-2015 2039Z 5 dB 16 WPM DE

2015-03-25 20:55:11Z 28040.0 W4CLM/B 25-Mar-2015 2055Z 3 dB 15 WPM DE

2015-03-25 21:05:15Z 28040.0 W4CLM/B 25-Mar-2015 2105Z 3 dB 15 WPM DE

2015-03-25 21:06:54Z 28046.0 KB2SEO/B 25-Mar-2015 2106Z 5 dB 15 WPM DE

2015-03-25 21:07:25Z 28037.4 KA2LIM/B 25-Mar-2015 2107Z 3 dB 22 WPM

2015-03-25 22:51:41Z 28012.8 KE4TWI/B 25-Mar-2015 2251Z 1 dB 14 WPM

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-25" | grep 28 | wc -l

10

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28 | wc -l

4

doug@mugu:~/skimsvr$ grep "/B" Spots.txt | grep "03-26" | grep 28

2015-03-26 14:43:20Z 28011.0 K4MTP/B 26-Mar-2015 1443Z 0 dB 18 WPM DE

2015-03-26 16:11:14Z 28010.9 K4MTP/B 26-Mar-2015 1611Z 4 dB 18 WPM DE

2015-03-26 16:46:59Z 28010.9 K4MTP/B 26-Mar-2015 1646Z 3 dB 18 WPM

2015-03-26 16:56:59Z 28010.9 K4MTP/B 26-Mar-2015 1656Z 3 dB 18 WPM DE

Granted still early... Today running old code, yesterday was running with the modified code.

—

Reply to this email directly or view it on GitHub.

—

Reply to this email directly or view it on GitHub.

—

Reply to this email directly or view it on GitHub.

—

Reply to this email directly or view it on GitHub.

— Reply to this email directly or view it on GitHub.

[dtheriault]

Following is from master not the liquid_dsp branch. I am adding this only because it shows something I think is interesting. Seems the accuracy of the spot, even on this alias, may be influenced by the signal level. The beacon is at 28254600 as I receive off the RX340.

I've got a dongle on the signal generator now and going to try and plot out the aliasing tonight.

[dtheriault]

Been playing around tonight using rtl_tcp -a -p <PORT#> and connecting with gqrx. Usually there's some good pirate AM stations on just below 40m band. Well I was hearing a CW qso and realized its an alias from 40m. The only way I could get rid of it was reducing the gain. The station was loud, about 20 over as received on the RX340. Noticed I had to reduce gain on the dongle to a really low level for the alias to go away.

Signal levels on 40m were quite hot tonight. So lowering gain was needed or these aliases show up for sure.

This was first time I saw an alias outside of librtlhpsdr code. I mean its rtl_tcp and gqrx.

Trying to get qt5 built on my XU3 which is on the bench next to sig generator so I can build cuSDR and be able to tweak the sig gen w/o walking back and forth to the garage ! First build failed due to running out of disk space. So setup NFS mount to my linux box... its a bit slow... :-(