Date   

Re: Maximum input level usb audio

DAVID IW5BNL
 

Hi Giovanni

Thank you!

David

Il Ven 14 Ago 2020, 18:28 Giovanni Franza <gfranza@...> ha scritto:
Hi David,
when I built a 4 phases example I fed I/Q in L/R audio using stock audio block in Gnuradio.
Only thing to remember is to launch periodically the CAT command allowing audio as I/Q.
No idea of the levels, sorry :-(

Regards,

Giovanni - HB9EIK


Re: Eladit.com site down?

Neil Smith G4DBN
 

Thanks Andy, I must have missed an announcement about the site change.  It looks like the shop section is still being worked on.

Neil G4DBN

The old site can still be found at the Wayback Machine


On 14/08/2020 22:15, Andreas Rehberg wrote:
It's elad.it

Andy, DF4WC
Am 14.08.20, 21:16 schrieb "MM6IXE via groups.io" <sota@...>:
It’s been like that for a while

for me

Iain


Re: Eladit.com site down?

Andreas Rehberg
 

It's elad.it

Andy, DF4WC
Am 14.08.20, 21:16 schrieb "MM6IXE via groups.io" <sota@...>:

It’s been like that for a while

for me

Iain


Re: Eladit.com site down?

 

It’s been like that for a while

for me

Iain


Eladit.com site down?

Neil Smith G4DBN
 

I can't reach sdr.eladit.com or any of the variants other than shop.eladit.com, have I missed a maintenance notice, or is there a fault?

Neil
http://g4dbn.uk


USB Audio I/Q Mode - what is it?

Andy G4JNT
 

What IS this IQ mode that can be set by the CAT IQ command m?
Is it to the internal soundcard, and if so, what advantage does if offer over normal single audio channel ?  
What bandwidth is available, is it plus/minus the normal Tx filter ?
If so I assume it goes down to true DC, unlike a real soiundcard that typically has a low cut off at a few tens of Hz



On Fri, 14 Aug 2020 at 17:28, Giovanni Franza <gfranza@...> wrote:
Hi David,
when I built a 4 phases example I fed I/Q in L/R audio using stock audio block in Gnuradio.
Only thing to remember is to launch periodically the CAT command allowing audio as I/Q.
No idea of the levels, sorry :-(

Regards,

Giovanni - HB9EIK


Re: Maximum input level usb audio

Giovanni Franza
 

Hi David,
when I built a 4 phases example I fed I/Q in L/R audio using stock audio block in Gnuradio.
Only thing to remember is to launch periodically the CAT command allowing audio as I/Q.
No idea of the levels, sorry :-(

Regards,

Giovanni - HB9EIK


Maximum input level usb audio

DAVID IW5BNL
 

Hi All,

I do some math supposing that ssb modulator is based on hilbert transformer. 
With little information available and my calculation may be possible that maximum input level is 0.4-0.5 in order to keep normalized modulator output to -1/+1.
I'm going to try IQ input in the next days..I assume that I (or Q) is right input of tx USB port and Q(or I) is left input of tx USB port.
Giovanni HB9EIK or I2PHD do you have some more informations (that can be shared..of course) on this topic.

Regards

David
IW5BNL 


Thank you

Francesco Di Giovanni
 

Thank you for accept my request to join the group.

I am waiting for the availability of the S3.

73s
Francesco Di Giovanni - IN3XZP


Re: FDM-DUO feature suggestion #elad

Giovanni Franza
 

To all people interested,
I've made a first attempt to compile in win32 using mingw.
As aspected portmidi is there and it seems to exists a port of pthreads.
But no hope with termio (for now) so, at the moment, there are some work to be done to port the program under Win10.
Regards,

Giovanni - HB9EIK


Re: FDM-DUO feature suggestion #elad

Giovanni Franza
 

Hi All,
attached you can find the last version of the program.
Using -i switch you can "program" the interface:
move a command then input the field requested, using the keyboard and following the instrucions.
Remember that "keys" send 2 commands: one at key down other at key up.
Hope this helps,

Giovanni - HB9EIK


Re: FDM-DUO feature suggestion #elad

Giovanni Franza
 

Hi All,
attached is a tarball with a working version also with jog "query".
When the operator is turning jog after a certain period of inactivity the program send a query CAT command to know the value stored in the DUO.
As for the previous version the tar, unrolled, create a folder. Go on this folder and do a "make" to generate the program.
To use the program launch the program as, for example:
./play -d 1 -m 3 -n /dev/ttyUSB0 -s 115200
where:
-d 1 ask for debug
-m 3 ask for midi device 3
-n /dev/ttyUSB0 indicate the device connected with DUO CAT
-s 115200 set the serial CAT speed (must match the one on the DUO menu #70)
to obtain help use
./play -h
to obtain a list of midi devices use
to avoid debug omit -d 1 from the parameters.
The file init.txt sets the command used: the one in the folder is for DJControlCompact and set NoisBlanker, NoiseLimiter, Att, Low pass filter, TX/RX and frequency.
If you have other midi devices use the debug to know bank and keys values sent and modify init.txt accordingly.
Beta tester are wellcomed!

Giovanni - HB9EIK


Re: Maximum usb input driving signal elad duo

DAVID IW5BNL
 

Hi All

I agree with Andy 88db of dynamic are still extremely good.
I suspect that this reduced dynamics depend how the internal modulator is implemented.
I'm doing some math..I will show my calculations in the next days. 
I think that is also available a IQ input for transmitting,infact the duo cat documentation show a CAT command to turn on IQ input for transmitting.
I suspect that sw2 software use this input for transmitting and,probably,full resolution is available here. I note that in transmitting part of sw2 software IQ graph are show with axis limit -1/+1..
I will investigate more.....

Regards

David
IW5BNL 

Il Lun 10 Ago 2020, 17:58 Andy G4JNT <andy.g4jnt@...> ha scritto:
It's not a lot of dynamic range loss really.
If the maximum value for the soundcard input is +/- 13000  out of 32768, that's still 14 bits worth of dynamic range (actually 14.6 bits if you can get your head around fractional bit sizes)

That still allows for 88dB of dynamic range - and we're talking about transmit here.   Which is still rather better than you'll ever be interested in.




On Mon, 10 Aug 2020 at 16:33, Chris Bartram <chris@...> wrote:

Thanks Andy,

That was my original understanding, so many thanks for the clarification. I think I went into brain fart mode - probably as the result of excessive plumbing and tiling!

It's interesting that Elad seem to have thrown-away quite a lot of potential TX dynamic range. However, without full design data, I'm certainly not going to comment, except to say that I applaud the detective work which you and David are doing to understand it.

73

Chris G4DGU


On 10/08/2020 14:41, Andy G4JNT wrote:
NO NO NO  Chris

The 0dBm output is linear up to 0dBm output in analogue / RF terms
What we're talking about is the digital drive word to the internal soundcard in order to get that signal level.
It looks as if only values in the range up to perhaps +/-13000 are valid.   So there is something like 7-8 dB of effective gain from  DDS level to 0dBm point.

Don't forget, the 0dBm comes direct from the DDS so it's a hard upper limit.   Set by logic rather than compression or voltage clipping or anything.


It's odd, though as David reports, that matches WSJT-X drive spot-on.    I doubt very much it was designed that way.  



On Mon, 10 Aug 2020 at 14:28, Chris Bartram <chris@...> wrote:

Hello Andy,

I'm just starting to think about connecting the 0dBm output of my FDM-DUO to my transverters. I'm a little unclear about the results you have reported - it's probably just too hot! I think you are saying that the output is linear up to about -8dBm from the 0dBm port on 144MHz. Is that correct?

73

Chris G4DGU

On 10/08/2020 12:43, Andy G4JNT wrote:
Sri Folks
Looks like a BIG correction is needed.
I've just done a quick test using the pre-recorded .WAV files and it's as you suggested.   Full output from the 0dBm port is achieved with, in my case, about 7 to 8dB back-off from full scale.

Setting the DUO as default soundcard, controlling PTT via the CAT interface and playing the .WAV files using the Windows inbuilt playwav facility I looked at the 0dBm output on a scope.   It needed a file generated at -8dB (0.4 voltage)  of full scale to see a noticeable decrease in amplitude on the scope.   Which is higher than you saw, at I think 0.1, but still lower than I expected.

With the desktop PC, large screen monitor, oscilloscope and stuff on, it was far too hot in the upstairs shack in this weather to spend more than a few minutes doing this so I'll investigate fuller, later.   I can use the .WAV files as calibration, then use TCUBE written by I2PHD o generate a continuously valuable tone in amplitude or frequency.   That will really map-out the ampltude response.

I did generate some two-tone .WAV files, and TCUBE can generate a two tone waveform, so that will be the way forward.    I'm pretty certain TCUBE gives full-scale digits at it's maximum setting, but at least I can now properly calibrate it.




On Mon, 10 Aug 2020 at 06:13, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

As you make some measurement in the next days I suggest to generate a wave file with two tones so you can measure the IM.
I'm also reading the data sheet of duo integrated sound card.
The chip is capable 16 or 24 bits resolutions..
If Franco or someone from Elad can just comment this result and providing us scale Factor (If any...) will very useful.

Regards

David
IW5BNL 


Il Dom 9 Ago 2020, 18:55 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi Andy

Very good idea.
I'm working with Linux.
I suspect that in the gnuradio /Linux pulse audio chain is some scale Factor that I don't know.
I will investigate more.

David
IW5BNL


Il Dom 9 Ago 2020, 18:49 Andy G4JNT <andy.g4jnt@...> ha scritto:
I generate .WAV files from first principles.   They use a pure 16 bit integer
Then replay them via teh soundcard of choice.

Unfortunately the wav.play function works with the default soundcard so for this test I will need to temporarily make the Duo my default soundcard !   So allteh Windows bings and chimes will appear on it.

I have plenty of third party software that generates tones and itest signals, but no proof that full scale is actually filling the 16 bit integer in any of them.    
I could write my own, but it's easier to generate .Wav, and I have a whole suite of software to generate all sorts of signals as a wav.

One USB soundcard I use a lot is the PCM2900 USB audio device.   That has one wonderful property - the reference for the A/D appears on a pin of the chip.   So it is possible, by measuring this DC level, to relate the exact voltage that will appear for any 16 bit integer sent, and the same for the sound input.  




On Sun, 9 Aug 2020 at 17:41, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

How you generate the audio USB signal.


David
IW5BNL 

Il Dom 9 Ago 2020, 18:34 Andy G4JNT <andy.g4jnt@...> ha scritto:
I'll do some tests tomorrow.  But I'm pretty certain that when tested before, a peak audio of the full 16 bit word was required to get maximum out of the 0dBm port.  I do remember thinking "how well designed"



On Sun, 9 Aug 2020 at 17:30, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all

I do some meeasuraments.
As I'm on country home I have only a power meter and the capability to measure the tx signal using the monitoring rx of duo.
I generate a two tone signal with gu radio and send it to duo USB.
By monitoring the internal rx I can measure the spectrum of duo rx by a another gnuradio script that capture the 192khz sample rate rx out (available on rx usbvport).
Everithink is perfectly linear until a USB audio signal of 0.1 pk normalized signal.
Increasing over 0.1 we start to see intermodulation.
0.1 is exactly the point where the output power is saturated.
For this measurement the audio input of duo was adjusted to 100% in pulse audio  control.
Of course reducing the pulse audio below 100% the level of saturation increase.
So the asumption that -1/1 is the limit isn't correct as depend on pulse audio level adjustment.
The good news is that I now know the maximum USB audio level before start to intermodulated.
The modulator inside duo working extremely well in the correct range of driving signal linear is perfect.

Ciao

David
IW5BNL 





Il Sab 8 Ago 2020, 22:04 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi all,
Thank for information and answer.
I'm driving a small amp on hf band (60w) so duo power is adjusted on 3 watts out.
I have to make some more accurate measurement but,look like,that changing USB driving from 0.5/-0.5 to 1/-1 peak power don't change.May be issue is on the saturation of my external amplifier.Next days I will made some more accurate meeasuraments.
I encourage elad crew to produce more documentation on it.

David
IW5BNL 

Il Sab 8 Ago 2020, 19:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
Ditto, and even if I want to, I have a perfectly good HF PA, -3dBm in gives 200 Watts out, all boxed up with switched low pass filters and absolutely bomb-proof with mismatched loads (that's been tested, intentionally, several times)



On Sat, 8 Aug 2020 at 18:38, Neil Smith G4DBN <neil@...> wrote:

Good point, I would have bought that version in a heartbeat.  The amp and TX filters are almost zero use to me.  But then I have weird minority interests and don't use the HF bands much at all these days.

Neil G4DBN

On 08/08/2020 18:28, Andy G4JNT wrote:
A full scale signed 16 bit integer, sine wave, (RMS = 23170) generates 0dBm out of the low power port.   Can't speak for teh power amp - never used it.

In fact, what isn't there a lower cost version of thd FDM withuut the 5W amp and all harmonic filters.   Just a 0dBm port on transmit



On Sat, 8 Aug 2020 at 18:20, Neil Smith G4DBN <neil@...> wrote:

Hi David, I found it worked OK with normalised +-1 peak amplitude when using the 0dBm output, but I didn't try it when driving the PA stage.  I was only using FM, so I don't know what the IMD performance was like at higher levels.

It would be interesting to know where you start to see any sign of non-linearity between the level in gnuradio and the actual power output via the PA.  Also, what happens when the power is set lower than MAX.  Does full scale on the USB device input still correspond to full output, or does the USB device level start to flatten out sooner?

Easy enough to test with simple tones (or pairs of tones) I guess.

Neil G4DBN

On 08/08/2020 17:43, DAVID IW5BNL wrote:
Thank you!

David
IW5BNL 



Il Sab 8 Ago 2020, 18:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
If that translates to a full scale 16 bit integer drive to the soundcard then I guess so



On Sat, 8 Aug 2020 at 17:38, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

Thank you for your prompt answer. 
Normalize -1/1 as gnu radio audio output is normalize.
Is it correct?

David




Il Sab 8 Ago 2020, 18:32 Andy G4JNT <andy.g4jnt@...> ha scritto:
Full scale -32768 / 32767  or   0x8000 / 0x7FFF   peak to peak is full power
Nice and simple, easy to remember



On Sat, 8 Aug 2020 at 17:29, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all,

I'm writing a ssb compressor using gnuradio.It work well but I don't really know maximum driving signal of duo USB internal sound card to produce full output power.
I proceed adjusting the input USB audio level little bit lower that the level where the output peak power is saturated ( for saturated output power I mean the maximum level on USB audio that even increasing the audio output power don't increase). Apparently look like a sort of alc is present or simply increasing the level the modulator is saturate. 
Is it somewhere a document and/or information showing correlation between input USB audio and output power (on ssb mode..)

Thank you

David 
IW5BNL

-- 
Neil
http://g4dbn.uk

--
Chris Bartram G4DGU

--
Chris Bartram G4DGU


RFRScanner replaces all RadioFrequency programs.

 


Re: Maximum usb input driving signal elad duo

Andy G4JNT
 

It's not a lot of dynamic range loss really.
If the maximum value for the soundcard input is +/- 13000  out of 32768, that's still 14 bits worth of dynamic range (actually 14.6 bits if you can get your head around fractional bit sizes)

That still allows for 88dB of dynamic range - and we're talking about transmit here.   Which is still rather better than you'll ever be interested in.




On Mon, 10 Aug 2020 at 16:33, Chris Bartram <chris@...> wrote:

Thanks Andy,

That was my original understanding, so many thanks for the clarification. I think I went into brain fart mode - probably as the result of excessive plumbing and tiling!

It's interesting that Elad seem to have thrown-away quite a lot of potential TX dynamic range. However, without full design data, I'm certainly not going to comment, except to say that I applaud the detective work which you and David are doing to understand it.

73

Chris G4DGU


On 10/08/2020 14:41, Andy G4JNT wrote:
NO NO NO  Chris

The 0dBm output is linear up to 0dBm output in analogue / RF terms
What we're talking about is the digital drive word to the internal soundcard in order to get that signal level.
It looks as if only values in the range up to perhaps +/-13000 are valid.   So there is something like 7-8 dB of effective gain from  DDS level to 0dBm point.

Don't forget, the 0dBm comes direct from the DDS so it's a hard upper limit.   Set by logic rather than compression or voltage clipping or anything.


It's odd, though as David reports, that matches WSJT-X drive spot-on.    I doubt very much it was designed that way.  



On Mon, 10 Aug 2020 at 14:28, Chris Bartram <chris@...> wrote:

Hello Andy,

I'm just starting to think about connecting the 0dBm output of my FDM-DUO to my transverters. I'm a little unclear about the results you have reported - it's probably just too hot! I think you are saying that the output is linear up to about -8dBm from the 0dBm port on 144MHz. Is that correct?

73

Chris G4DGU

On 10/08/2020 12:43, Andy G4JNT wrote:
Sri Folks
Looks like a BIG correction is needed.
I've just done a quick test using the pre-recorded .WAV files and it's as you suggested.   Full output from the 0dBm port is achieved with, in my case, about 7 to 8dB back-off from full scale.

Setting the DUO as default soundcard, controlling PTT via the CAT interface and playing the .WAV files using the Windows inbuilt playwav facility I looked at the 0dBm output on a scope.   It needed a file generated at -8dB (0.4 voltage)  of full scale to see a noticeable decrease in amplitude on the scope.   Which is higher than you saw, at I think 0.1, but still lower than I expected.

With the desktop PC, large screen monitor, oscilloscope and stuff on, it was far too hot in the upstairs shack in this weather to spend more than a few minutes doing this so I'll investigate fuller, later.   I can use the .WAV files as calibration, then use TCUBE written by I2PHD o generate a continuously valuable tone in amplitude or frequency.   That will really map-out the ampltude response.

I did generate some two-tone .WAV files, and TCUBE can generate a two tone waveform, so that will be the way forward.    I'm pretty certain TCUBE gives full-scale digits at it's maximum setting, but at least I can now properly calibrate it.




On Mon, 10 Aug 2020 at 06:13, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

As you make some measurement in the next days I suggest to generate a wave file with two tones so you can measure the IM.
I'm also reading the data sheet of duo integrated sound card.
The chip is capable 16 or 24 bits resolutions..
If Franco or someone from Elad can just comment this result and providing us scale Factor (If any...) will very useful.

Regards

David
IW5BNL 


Il Dom 9 Ago 2020, 18:55 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi Andy

Very good idea.
I'm working with Linux.
I suspect that in the gnuradio /Linux pulse audio chain is some scale Factor that I don't know.
I will investigate more.

David
IW5BNL


Il Dom 9 Ago 2020, 18:49 Andy G4JNT <andy.g4jnt@...> ha scritto:
I generate .WAV files from first principles.   They use a pure 16 bit integer
Then replay them via teh soundcard of choice.

Unfortunately the wav.play function works with the default soundcard so for this test I will need to temporarily make the Duo my default soundcard !   So allteh Windows bings and chimes will appear on it.

I have plenty of third party software that generates tones and itest signals, but no proof that full scale is actually filling the 16 bit integer in any of them.    
I could write my own, but it's easier to generate .Wav, and I have a whole suite of software to generate all sorts of signals as a wav.

One USB soundcard I use a lot is the PCM2900 USB audio device.   That has one wonderful property - the reference for the A/D appears on a pin of the chip.   So it is possible, by measuring this DC level, to relate the exact voltage that will appear for any 16 bit integer sent, and the same for the sound input.  




On Sun, 9 Aug 2020 at 17:41, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

How you generate the audio USB signal.


David
IW5BNL 

Il Dom 9 Ago 2020, 18:34 Andy G4JNT <andy.g4jnt@...> ha scritto:
I'll do some tests tomorrow.  But I'm pretty certain that when tested before, a peak audio of the full 16 bit word was required to get maximum out of the 0dBm port.  I do remember thinking "how well designed"



On Sun, 9 Aug 2020 at 17:30, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all

I do some meeasuraments.
As I'm on country home I have only a power meter and the capability to measure the tx signal using the monitoring rx of duo.
I generate a two tone signal with gu radio and send it to duo USB.
By monitoring the internal rx I can measure the spectrum of duo rx by a another gnuradio script that capture the 192khz sample rate rx out (available on rx usbvport).
Everithink is perfectly linear until a USB audio signal of 0.1 pk normalized signal.
Increasing over 0.1 we start to see intermodulation.
0.1 is exactly the point where the output power is saturated.
For this measurement the audio input of duo was adjusted to 100% in pulse audio  control.
Of course reducing the pulse audio below 100% the level of saturation increase.
So the asumption that -1/1 is the limit isn't correct as depend on pulse audio level adjustment.
The good news is that I now know the maximum USB audio level before start to intermodulated.
The modulator inside duo working extremely well in the correct range of driving signal linear is perfect.

Ciao

David
IW5BNL 





Il Sab 8 Ago 2020, 22:04 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi all,
Thank for information and answer.
I'm driving a small amp on hf band (60w) so duo power is adjusted on 3 watts out.
I have to make some more accurate measurement but,look like,that changing USB driving from 0.5/-0.5 to 1/-1 peak power don't change.May be issue is on the saturation of my external amplifier.Next days I will made some more accurate meeasuraments.
I encourage elad crew to produce more documentation on it.

David
IW5BNL 

Il Sab 8 Ago 2020, 19:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
Ditto, and even if I want to, I have a perfectly good HF PA, -3dBm in gives 200 Watts out, all boxed up with switched low pass filters and absolutely bomb-proof with mismatched loads (that's been tested, intentionally, several times)



On Sat, 8 Aug 2020 at 18:38, Neil Smith G4DBN <neil@...> wrote:

Good point, I would have bought that version in a heartbeat.  The amp and TX filters are almost zero use to me.  But then I have weird minority interests and don't use the HF bands much at all these days.

Neil G4DBN

On 08/08/2020 18:28, Andy G4JNT wrote:
A full scale signed 16 bit integer, sine wave, (RMS = 23170) generates 0dBm out of the low power port.   Can't speak for teh power amp - never used it.

In fact, what isn't there a lower cost version of thd FDM withuut the 5W amp and all harmonic filters.   Just a 0dBm port on transmit



On Sat, 8 Aug 2020 at 18:20, Neil Smith G4DBN <neil@...> wrote:

Hi David, I found it worked OK with normalised +-1 peak amplitude when using the 0dBm output, but I didn't try it when driving the PA stage.  I was only using FM, so I don't know what the IMD performance was like at higher levels.

It would be interesting to know where you start to see any sign of non-linearity between the level in gnuradio and the actual power output via the PA.  Also, what happens when the power is set lower than MAX.  Does full scale on the USB device input still correspond to full output, or does the USB device level start to flatten out sooner?

Easy enough to test with simple tones (or pairs of tones) I guess.

Neil G4DBN

On 08/08/2020 17:43, DAVID IW5BNL wrote:
Thank you!

David
IW5BNL 



Il Sab 8 Ago 2020, 18:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
If that translates to a full scale 16 bit integer drive to the soundcard then I guess so



On Sat, 8 Aug 2020 at 17:38, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

Thank you for your prompt answer. 
Normalize -1/1 as gnu radio audio output is normalize.
Is it correct?

David




Il Sab 8 Ago 2020, 18:32 Andy G4JNT <andy.g4jnt@...> ha scritto:
Full scale -32768 / 32767  or   0x8000 / 0x7FFF   peak to peak is full power
Nice and simple, easy to remember



On Sat, 8 Aug 2020 at 17:29, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all,

I'm writing a ssb compressor using gnuradio.It work well but I don't really know maximum driving signal of duo USB internal sound card to produce full output power.
I proceed adjusting the input USB audio level little bit lower that the level where the output peak power is saturated ( for saturated output power I mean the maximum level on USB audio that even increasing the audio output power don't increase). Apparently look like a sort of alc is present or simply increasing the level the modulator is saturate. 
Is it somewhere a document and/or information showing correlation between input USB audio and output power (on ssb mode..)

Thank you

David 
IW5BNL

-- 
Neil
http://g4dbn.uk

--
Chris Bartram G4DGU

--
Chris Bartram G4DGU


Re: Maximum usb input driving signal elad duo

Chris Bartram
 

Thanks Andy,

That was my original understanding, so many thanks for the clarification. I think I went into brain fart mode - probably as the result of excessive plumbing and tiling!

It's interesting that Elad seem to have thrown-away quite a lot of potential TX dynamic range. However, without full design data, I'm certainly not going to comment, except to say that I applaud the detective work which you and David are doing to understand it.

73

Chris G4DGU


On 10/08/2020 14:41, Andy G4JNT wrote:
NO NO NO  Chris

The 0dBm output is linear up to 0dBm output in analogue / RF terms
What we're talking about is the digital drive word to the internal soundcard in order to get that signal level.
It looks as if only values in the range up to perhaps +/-13000 are valid.   So there is something like 7-8 dB of effective gain from  DDS level to 0dBm point.

Don't forget, the 0dBm comes direct from the DDS so it's a hard upper limit.   Set by logic rather than compression or voltage clipping or anything.


It's odd, though as David reports, that matches WSJT-X drive spot-on.    I doubt very much it was designed that way.  



On Mon, 10 Aug 2020 at 14:28, Chris Bartram <chris@...> wrote:

Hello Andy,

I'm just starting to think about connecting the 0dBm output of my FDM-DUO to my transverters. I'm a little unclear about the results you have reported - it's probably just too hot! I think you are saying that the output is linear up to about -8dBm from the 0dBm port on 144MHz. Is that correct?

73

Chris G4DGU

On 10/08/2020 12:43, Andy G4JNT wrote:
Sri Folks
Looks like a BIG correction is needed.
I've just done a quick test using the pre-recorded .WAV files and it's as you suggested.   Full output from the 0dBm port is achieved with, in my case, about 7 to 8dB back-off from full scale.

Setting the DUO as default soundcard, controlling PTT via the CAT interface and playing the .WAV files using the Windows inbuilt playwav facility I looked at the 0dBm output on a scope.   It needed a file generated at -8dB (0.4 voltage)  of full scale to see a noticeable decrease in amplitude on the scope.   Which is higher than you saw, at I think 0.1, but still lower than I expected.

With the desktop PC, large screen monitor, oscilloscope and stuff on, it was far too hot in the upstairs shack in this weather to spend more than a few minutes doing this so I'll investigate fuller, later.   I can use the .WAV files as calibration, then use TCUBE written by I2PHD o generate a continuously valuable tone in amplitude or frequency.   That will really map-out the ampltude response.

I did generate some two-tone .WAV files, and TCUBE can generate a two tone waveform, so that will be the way forward.    I'm pretty certain TCUBE gives full-scale digits at it's maximum setting, but at least I can now properly calibrate it.




On Mon, 10 Aug 2020 at 06:13, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

As you make some measurement in the next days I suggest to generate a wave file with two tones so you can measure the IM.
I'm also reading the data sheet of duo integrated sound card.
The chip is capable 16 or 24 bits resolutions..
If Franco or someone from Elad can just comment this result and providing us scale Factor (If any...) will very useful.

Regards

David
IW5BNL 


Il Dom 9 Ago 2020, 18:55 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi Andy

Very good idea.
I'm working with Linux.
I suspect that in the gnuradio /Linux pulse audio chain is some scale Factor that I don't know.
I will investigate more.

David
IW5BNL


Il Dom 9 Ago 2020, 18:49 Andy G4JNT <andy.g4jnt@...> ha scritto:
I generate .WAV files from first principles.   They use a pure 16 bit integer
Then replay them via teh soundcard of choice.

Unfortunately the wav.play function works with the default soundcard so for this test I will need to temporarily make the Duo my default soundcard !   So allteh Windows bings and chimes will appear on it.

I have plenty of third party software that generates tones and itest signals, but no proof that full scale is actually filling the 16 bit integer in any of them.    
I could write my own, but it's easier to generate .Wav, and I have a whole suite of software to generate all sorts of signals as a wav.

One USB soundcard I use a lot is the PCM2900 USB audio device.   That has one wonderful property - the reference for the A/D appears on a pin of the chip.   So it is possible, by measuring this DC level, to relate the exact voltage that will appear for any 16 bit integer sent, and the same for the sound input.  




On Sun, 9 Aug 2020 at 17:41, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

How you generate the audio USB signal.


David
IW5BNL 

Il Dom 9 Ago 2020, 18:34 Andy G4JNT <andy.g4jnt@...> ha scritto:
I'll do some tests tomorrow.  But I'm pretty certain that when tested before, a peak audio of the full 16 bit word was required to get maximum out of the 0dBm port.  I do remember thinking "how well designed"



On Sun, 9 Aug 2020 at 17:30, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all

I do some meeasuraments.
As I'm on country home I have only a power meter and the capability to measure the tx signal using the monitoring rx of duo.
I generate a two tone signal with gu radio and send it to duo USB.
By monitoring the internal rx I can measure the spectrum of duo rx by a another gnuradio script that capture the 192khz sample rate rx out (available on rx usbvport).
Everithink is perfectly linear until a USB audio signal of 0.1 pk normalized signal.
Increasing over 0.1 we start to see intermodulation.
0.1 is exactly the point where the output power is saturated.
For this measurement the audio input of duo was adjusted to 100% in pulse audio  control.
Of course reducing the pulse audio below 100% the level of saturation increase.
So the asumption that -1/1 is the limit isn't correct as depend on pulse audio level adjustment.
The good news is that I now know the maximum USB audio level before start to intermodulated.
The modulator inside duo working extremely well in the correct range of driving signal linear is perfect.

Ciao

David
IW5BNL 





Il Sab 8 Ago 2020, 22:04 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi all,
Thank for information and answer.
I'm driving a small amp on hf band (60w) so duo power is adjusted on 3 watts out.
I have to make some more accurate measurement but,look like,that changing USB driving from 0.5/-0.5 to 1/-1 peak power don't change.May be issue is on the saturation of my external amplifier.Next days I will made some more accurate meeasuraments.
I encourage elad crew to produce more documentation on it.

David
IW5BNL 

Il Sab 8 Ago 2020, 19:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
Ditto, and even if I want to, I have a perfectly good HF PA, -3dBm in gives 200 Watts out, all boxed up with switched low pass filters and absolutely bomb-proof with mismatched loads (that's been tested, intentionally, several times)



On Sat, 8 Aug 2020 at 18:38, Neil Smith G4DBN <neil@...> wrote:

Good point, I would have bought that version in a heartbeat.  The amp and TX filters are almost zero use to me.  But then I have weird minority interests and don't use the HF bands much at all these days.

Neil G4DBN

On 08/08/2020 18:28, Andy G4JNT wrote:
A full scale signed 16 bit integer, sine wave, (RMS = 23170) generates 0dBm out of the low power port.   Can't speak for teh power amp - never used it.

In fact, what isn't there a lower cost version of thd FDM withuut the 5W amp and all harmonic filters.   Just a 0dBm port on transmit



On Sat, 8 Aug 2020 at 18:20, Neil Smith G4DBN <neil@...> wrote:

Hi David, I found it worked OK with normalised +-1 peak amplitude when using the 0dBm output, but I didn't try it when driving the PA stage.  I was only using FM, so I don't know what the IMD performance was like at higher levels.

It would be interesting to know where you start to see any sign of non-linearity between the level in gnuradio and the actual power output via the PA.  Also, what happens when the power is set lower than MAX.  Does full scale on the USB device input still correspond to full output, or does the USB device level start to flatten out sooner?

Easy enough to test with simple tones (or pairs of tones) I guess.

Neil G4DBN

On 08/08/2020 17:43, DAVID IW5BNL wrote:
Thank you!

David
IW5BNL 



Il Sab 8 Ago 2020, 18:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
If that translates to a full scale 16 bit integer drive to the soundcard then I guess so



On Sat, 8 Aug 2020 at 17:38, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

Thank you for your prompt answer. 
Normalize -1/1 as gnu radio audio output is normalize.
Is it correct?

David




Il Sab 8 Ago 2020, 18:32 Andy G4JNT <andy.g4jnt@...> ha scritto:
Full scale -32768 / 32767  or   0x8000 / 0x7FFF   peak to peak is full power
Nice and simple, easy to remember



On Sat, 8 Aug 2020 at 17:29, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all,

I'm writing a ssb compressor using gnuradio.It work well but I don't really know maximum driving signal of duo USB internal sound card to produce full output power.
I proceed adjusting the input USB audio level little bit lower that the level where the output peak power is saturated ( for saturated output power I mean the maximum level on USB audio that even increasing the audio output power don't increase). Apparently look like a sort of alc is present or simply increasing the level the modulator is saturate. 
Is it somewhere a document and/or information showing correlation between input USB audio and output power (on ssb mode..)

Thank you

David 
IW5BNL

-- 
Neil
http://g4dbn.uk

--
Chris Bartram G4DGU

--
Chris Bartram G4DGU


Re: Maximum usb input driving signal elad duo

DAVID IW5BNL
 

Hi Andy

I think we need some feedback from elad team or some comment on this result.
Maybe also i2phd or Giovanni
HB9EIK can help 

Regards

David
IW5BNL 


Il Lun 10 Ago 2020, 15:41 Andy G4JNT <andy.g4jnt@...> ha scritto:
NO NO NO  Chris

The 0dBm output is linear up to 0dBm output in analogue / RF terms
What we're talking about is the digital drive word to the internal soundcard in order to get that signal level.
It looks as if only values in the range up to perhaps +/-13000 are valid.   So there is something like 7-8 dB of effective gain from  DDS level to 0dBm point.

Don't forget, the 0dBm comes direct from the DDS so it's a hard upper limit.   Set by logic rather than compression or voltage clipping or anything.


It's odd, though as David reports, that matches WSJT-X drive spot-on.    I doubt very much it was designed that way.  



On Mon, 10 Aug 2020 at 14:28, Chris Bartram <chris@...> wrote:

Hello Andy,

I'm just starting to think about connecting the 0dBm output of my FDM-DUO to my transverters. I'm a little unclear about the results you have reported - it's probably just too hot! I think you are saying that the output is linear up to about -8dBm from the 0dBm port on 144MHz. Is that correct?

73

Chris G4DGU

On 10/08/2020 12:43, Andy G4JNT wrote:
Sri Folks
Looks like a BIG correction is needed.
I've just done a quick test using the pre-recorded .WAV files and it's as you suggested.   Full output from the 0dBm port is achieved with, in my case, about 7 to 8dB back-off from full scale.

Setting the DUO as default soundcard, controlling PTT via the CAT interface and playing the .WAV files using the Windows inbuilt playwav facility I looked at the 0dBm output on a scope.   It needed a file generated at -8dB (0.4 voltage)  of full scale to see a noticeable decrease in amplitude on the scope.   Which is higher than you saw, at I think 0.1, but still lower than I expected.

With the desktop PC, large screen monitor, oscilloscope and stuff on, it was far too hot in the upstairs shack in this weather to spend more than a few minutes doing this so I'll investigate fuller, later.   I can use the .WAV files as calibration, then use TCUBE written by I2PHD o generate a continuously valuable tone in amplitude or frequency.   That will really map-out the ampltude response.

I did generate some two-tone .WAV files, and TCUBE can generate a two tone waveform, so that will be the way forward.    I'm pretty certain TCUBE gives full-scale digits at it's maximum setting, but at least I can now properly calibrate it.




On Mon, 10 Aug 2020 at 06:13, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

As you make some measurement in the next days I suggest to generate a wave file with two tones so you can measure the IM.
I'm also reading the data sheet of duo integrated sound card.
The chip is capable 16 or 24 bits resolutions..
If Franco or someone from Elad can just comment this result and providing us scale Factor (If any...) will very useful.

Regards

David
IW5BNL 


Il Dom 9 Ago 2020, 18:55 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi Andy

Very good idea.
I'm working with Linux.
I suspect that in the gnuradio /Linux pulse audio chain is some scale Factor that I don't know.
I will investigate more.

David
IW5BNL


Il Dom 9 Ago 2020, 18:49 Andy G4JNT <andy.g4jnt@...> ha scritto:
I generate .WAV files from first principles.   They use a pure 16 bit integer
Then replay them via teh soundcard of choice.

Unfortunately the wav.play function works with the default soundcard so for this test I will need to temporarily make the Duo my default soundcard !   So allteh Windows bings and chimes will appear on it.

I have plenty of third party software that generates tones and itest signals, but no proof that full scale is actually filling the 16 bit integer in any of them.    
I could write my own, but it's easier to generate .Wav, and I have a whole suite of software to generate all sorts of signals as a wav.

One USB soundcard I use a lot is the PCM2900 USB audio device.   That has one wonderful property - the reference for the A/D appears on a pin of the chip.   So it is possible, by measuring this DC level, to relate the exact voltage that will appear for any 16 bit integer sent, and the same for the sound input.  




On Sun, 9 Aug 2020 at 17:41, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

How you generate the audio USB signal.


David
IW5BNL 

Il Dom 9 Ago 2020, 18:34 Andy G4JNT <andy.g4jnt@...> ha scritto:
I'll do some tests tomorrow.  But I'm pretty certain that when tested before, a peak audio of the full 16 bit word was required to get maximum out of the 0dBm port.  I do remember thinking "how well designed"



On Sun, 9 Aug 2020 at 17:30, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all

I do some meeasuraments.
As I'm on country home I have only a power meter and the capability to measure the tx signal using the monitoring rx of duo.
I generate a two tone signal with gu radio and send it to duo USB.
By monitoring the internal rx I can measure the spectrum of duo rx by a another gnuradio script that capture the 192khz sample rate rx out (available on rx usbvport).
Everithink is perfectly linear until a USB audio signal of 0.1 pk normalized signal.
Increasing over 0.1 we start to see intermodulation.
0.1 is exactly the point where the output power is saturated.
For this measurement the audio input of duo was adjusted to 100% in pulse audio  control.
Of course reducing the pulse audio below 100% the level of saturation increase.
So the asumption that -1/1 is the limit isn't correct as depend on pulse audio level adjustment.
The good news is that I now know the maximum USB audio level before start to intermodulated.
The modulator inside duo working extremely well in the correct range of driving signal linear is perfect.

Ciao

David
IW5BNL 





Il Sab 8 Ago 2020, 22:04 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi all,
Thank for information and answer.
I'm driving a small amp on hf band (60w) so duo power is adjusted on 3 watts out.
I have to make some more accurate measurement but,look like,that changing USB driving from 0.5/-0.5 to 1/-1 peak power don't change.May be issue is on the saturation of my external amplifier.Next days I will made some more accurate meeasuraments.
I encourage elad crew to produce more documentation on it.

David
IW5BNL 

Il Sab 8 Ago 2020, 19:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
Ditto, and even if I want to, I have a perfectly good HF PA, -3dBm in gives 200 Watts out, all boxed up with switched low pass filters and absolutely bomb-proof with mismatched loads (that's been tested, intentionally, several times)



On Sat, 8 Aug 2020 at 18:38, Neil Smith G4DBN <neil@...> wrote:

Good point, I would have bought that version in a heartbeat.  The amp and TX filters are almost zero use to me.  But then I have weird minority interests and don't use the HF bands much at all these days.

Neil G4DBN

On 08/08/2020 18:28, Andy G4JNT wrote:
A full scale signed 16 bit integer, sine wave, (RMS = 23170) generates 0dBm out of the low power port.   Can't speak for teh power amp - never used it.

In fact, what isn't there a lower cost version of thd FDM withuut the 5W amp and all harmonic filters.   Just a 0dBm port on transmit



On Sat, 8 Aug 2020 at 18:20, Neil Smith G4DBN <neil@...> wrote:

Hi David, I found it worked OK with normalised +-1 peak amplitude when using the 0dBm output, but I didn't try it when driving the PA stage.  I was only using FM, so I don't know what the IMD performance was like at higher levels.

It would be interesting to know where you start to see any sign of non-linearity between the level in gnuradio and the actual power output via the PA.  Also, what happens when the power is set lower than MAX.  Does full scale on the USB device input still correspond to full output, or does the USB device level start to flatten out sooner?

Easy enough to test with simple tones (or pairs of tones) I guess.

Neil G4DBN

On 08/08/2020 17:43, DAVID IW5BNL wrote:
Thank you!

David
IW5BNL 



Il Sab 8 Ago 2020, 18:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
If that translates to a full scale 16 bit integer drive to the soundcard then I guess so



On Sat, 8 Aug 2020 at 17:38, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

Thank you for your prompt answer. 
Normalize -1/1 as gnu radio audio output is normalize.
Is it correct?

David




Il Sab 8 Ago 2020, 18:32 Andy G4JNT <andy.g4jnt@...> ha scritto:
Full scale -32768 / 32767  or   0x8000 / 0x7FFF   peak to peak is full power
Nice and simple, easy to remember



On Sat, 8 Aug 2020 at 17:29, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all,

I'm writing a ssb compressor using gnuradio.It work well but I don't really know maximum driving signal of duo USB internal sound card to produce full output power.
I proceed adjusting the input USB audio level little bit lower that the level where the output peak power is saturated ( for saturated output power I mean the maximum level on USB audio that even increasing the audio output power don't increase). Apparently look like a sort of alc is present or simply increasing the level the modulator is saturate. 
Is it somewhere a document and/or information showing correlation between input USB audio and output power (on ssb mode..)

Thank you

David 
IW5BNL

-- 
Neil
http://g4dbn.uk

--
Chris Bartram G4DGU


Re: Maximum usb input driving signal elad duo

Andy G4JNT
 

NO NO NO  Chris

The 0dBm output is linear up to 0dBm output in analogue / RF terms
What we're talking about is the digital drive word to the internal soundcard in order to get that signal level.
It looks as if only values in the range up to perhaps +/-13000 are valid.   So there is something like 7-8 dB of effective gain from  DDS level to 0dBm point.

Don't forget, the 0dBm comes direct from the DDS so it's a hard upper limit.   Set by logic rather than compression or voltage clipping or anything.


It's odd, though as David reports, that matches WSJT-X drive spot-on.    I doubt very much it was designed that way.  



On Mon, 10 Aug 2020 at 14:28, Chris Bartram <chris@...> wrote:

Hello Andy,

I'm just starting to think about connecting the 0dBm output of my FDM-DUO to my transverters. I'm a little unclear about the results you have reported - it's probably just too hot! I think you are saying that the output is linear up to about -8dBm from the 0dBm port on 144MHz. Is that correct?

73

Chris G4DGU

On 10/08/2020 12:43, Andy G4JNT wrote:
Sri Folks
Looks like a BIG correction is needed.
I've just done a quick test using the pre-recorded .WAV files and it's as you suggested.   Full output from the 0dBm port is achieved with, in my case, about 7 to 8dB back-off from full scale.

Setting the DUO as default soundcard, controlling PTT via the CAT interface and playing the .WAV files using the Windows inbuilt playwav facility I looked at the 0dBm output on a scope.   It needed a file generated at -8dB (0.4 voltage)  of full scale to see a noticeable decrease in amplitude on the scope.   Which is higher than you saw, at I think 0.1, but still lower than I expected.

With the desktop PC, large screen monitor, oscilloscope and stuff on, it was far too hot in the upstairs shack in this weather to spend more than a few minutes doing this so I'll investigate fuller, later.   I can use the .WAV files as calibration, then use TCUBE written by I2PHD o generate a continuously valuable tone in amplitude or frequency.   That will really map-out the ampltude response.

I did generate some two-tone .WAV files, and TCUBE can generate a two tone waveform, so that will be the way forward.    I'm pretty certain TCUBE gives full-scale digits at it's maximum setting, but at least I can now properly calibrate it.




On Mon, 10 Aug 2020 at 06:13, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

As you make some measurement in the next days I suggest to generate a wave file with two tones so you can measure the IM.
I'm also reading the data sheet of duo integrated sound card.
The chip is capable 16 or 24 bits resolutions..
If Franco or someone from Elad can just comment this result and providing us scale Factor (If any...) will very useful.

Regards

David
IW5BNL 


Il Dom 9 Ago 2020, 18:55 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi Andy

Very good idea.
I'm working with Linux.
I suspect that in the gnuradio /Linux pulse audio chain is some scale Factor that I don't know.
I will investigate more.

David
IW5BNL


Il Dom 9 Ago 2020, 18:49 Andy G4JNT <andy.g4jnt@...> ha scritto:
I generate .WAV files from first principles.   They use a pure 16 bit integer
Then replay them via teh soundcard of choice.

Unfortunately the wav.play function works with the default soundcard so for this test I will need to temporarily make the Duo my default soundcard !   So allteh Windows bings and chimes will appear on it.

I have plenty of third party software that generates tones and itest signals, but no proof that full scale is actually filling the 16 bit integer in any of them.    
I could write my own, but it's easier to generate .Wav, and I have a whole suite of software to generate all sorts of signals as a wav.

One USB soundcard I use a lot is the PCM2900 USB audio device.   That has one wonderful property - the reference for the A/D appears on a pin of the chip.   So it is possible, by measuring this DC level, to relate the exact voltage that will appear for any 16 bit integer sent, and the same for the sound input.  




On Sun, 9 Aug 2020 at 17:41, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

How you generate the audio USB signal.


David
IW5BNL 

Il Dom 9 Ago 2020, 18:34 Andy G4JNT <andy.g4jnt@...> ha scritto:
I'll do some tests tomorrow.  But I'm pretty certain that when tested before, a peak audio of the full 16 bit word was required to get maximum out of the 0dBm port.  I do remember thinking "how well designed"



On Sun, 9 Aug 2020 at 17:30, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all

I do some meeasuraments.
As I'm on country home I have only a power meter and the capability to measure the tx signal using the monitoring rx of duo.
I generate a two tone signal with gu radio and send it to duo USB.
By monitoring the internal rx I can measure the spectrum of duo rx by a another gnuradio script that capture the 192khz sample rate rx out (available on rx usbvport).
Everithink is perfectly linear until a USB audio signal of 0.1 pk normalized signal.
Increasing over 0.1 we start to see intermodulation.
0.1 is exactly the point where the output power is saturated.
For this measurement the audio input of duo was adjusted to 100% in pulse audio  control.
Of course reducing the pulse audio below 100% the level of saturation increase.
So the asumption that -1/1 is the limit isn't correct as depend on pulse audio level adjustment.
The good news is that I now know the maximum USB audio level before start to intermodulated.
The modulator inside duo working extremely well in the correct range of driving signal linear is perfect.

Ciao

David
IW5BNL 





Il Sab 8 Ago 2020, 22:04 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi all,
Thank for information and answer.
I'm driving a small amp on hf band (60w) so duo power is adjusted on 3 watts out.
I have to make some more accurate measurement but,look like,that changing USB driving from 0.5/-0.5 to 1/-1 peak power don't change.May be issue is on the saturation of my external amplifier.Next days I will made some more accurate meeasuraments.
I encourage elad crew to produce more documentation on it.

David
IW5BNL 

Il Sab 8 Ago 2020, 19:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
Ditto, and even if I want to, I have a perfectly good HF PA, -3dBm in gives 200 Watts out, all boxed up with switched low pass filters and absolutely bomb-proof with mismatched loads (that's been tested, intentionally, several times)



On Sat, 8 Aug 2020 at 18:38, Neil Smith G4DBN <neil@...> wrote:

Good point, I would have bought that version in a heartbeat.  The amp and TX filters are almost zero use to me.  But then I have weird minority interests and don't use the HF bands much at all these days.

Neil G4DBN

On 08/08/2020 18:28, Andy G4JNT wrote:
A full scale signed 16 bit integer, sine wave, (RMS = 23170) generates 0dBm out of the low power port.   Can't speak for teh power amp - never used it.

In fact, what isn't there a lower cost version of thd FDM withuut the 5W amp and all harmonic filters.   Just a 0dBm port on transmit



On Sat, 8 Aug 2020 at 18:20, Neil Smith G4DBN <neil@...> wrote:

Hi David, I found it worked OK with normalised +-1 peak amplitude when using the 0dBm output, but I didn't try it when driving the PA stage.  I was only using FM, so I don't know what the IMD performance was like at higher levels.

It would be interesting to know where you start to see any sign of non-linearity between the level in gnuradio and the actual power output via the PA.  Also, what happens when the power is set lower than MAX.  Does full scale on the USB device input still correspond to full output, or does the USB device level start to flatten out sooner?

Easy enough to test with simple tones (or pairs of tones) I guess.

Neil G4DBN

On 08/08/2020 17:43, DAVID IW5BNL wrote:
Thank you!

David
IW5BNL 



Il Sab 8 Ago 2020, 18:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
If that translates to a full scale 16 bit integer drive to the soundcard then I guess so



On Sat, 8 Aug 2020 at 17:38, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

Thank you for your prompt answer. 
Normalize -1/1 as gnu radio audio output is normalize.
Is it correct?

David




Il Sab 8 Ago 2020, 18:32 Andy G4JNT <andy.g4jnt@...> ha scritto:
Full scale -32768 / 32767  or   0x8000 / 0x7FFF   peak to peak is full power
Nice and simple, easy to remember



On Sat, 8 Aug 2020 at 17:29, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all,

I'm writing a ssb compressor using gnuradio.It work well but I don't really know maximum driving signal of duo USB internal sound card to produce full output power.
I proceed adjusting the input USB audio level little bit lower that the level where the output peak power is saturated ( for saturated output power I mean the maximum level on USB audio that even increasing the audio output power don't increase). Apparently look like a sort of alc is present or simply increasing the level the modulator is saturate. 
Is it somewhere a document and/or information showing correlation between input USB audio and output power (on ssb mode..)

Thank you

David 
IW5BNL

-- 
Neil
http://g4dbn.uk

--
Chris Bartram G4DGU


Re: Maximum usb input driving signal elad duo

Chris Bartram
 

Hello Andy,

I'm just starting to think about connecting the 0dBm output of my FDM-DUO to my transverters. I'm a little unclear about the results you have reported - it's probably just too hot! I think you are saying that the output is linear up to about -8dBm from the 0dBm port on 144MHz. Is that correct?

73

Chris G4DGU

On 10/08/2020 12:43, Andy G4JNT wrote:
Sri Folks
Looks like a BIG correction is needed.
I've just done a quick test using the pre-recorded .WAV files and it's as you suggested.   Full output from the 0dBm port is achieved with, in my case, about 7 to 8dB back-off from full scale.

Setting the DUO as default soundcard, controlling PTT via the CAT interface and playing the .WAV files using the Windows inbuilt playwav facility I looked at the 0dBm output on a scope.   It needed a file generated at -8dB (0.4 voltage)  of full scale to see a noticeable decrease in amplitude on the scope.   Which is higher than you saw, at I think 0.1, but still lower than I expected.

With the desktop PC, large screen monitor, oscilloscope and stuff on, it was far too hot in the upstairs shack in this weather to spend more than a few minutes doing this so I'll investigate fuller, later.   I can use the .WAV files as calibration, then use TCUBE written by I2PHD o generate a continuously valuable tone in amplitude or frequency.   That will really map-out the ampltude response.

I did generate some two-tone .WAV files, and TCUBE can generate a two tone waveform, so that will be the way forward.    I'm pretty certain TCUBE gives full-scale digits at it's maximum setting, but at least I can now properly calibrate it.




On Mon, 10 Aug 2020 at 06:13, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

As you make some measurement in the next days I suggest to generate a wave file with two tones so you can measure the IM.
I'm also reading the data sheet of duo integrated sound card.
The chip is capable 16 or 24 bits resolutions..
If Franco or someone from Elad can just comment this result and providing us scale Factor (If any...) will very useful.

Regards

David
IW5BNL 


Il Dom 9 Ago 2020, 18:55 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi Andy

Very good idea.
I'm working with Linux.
I suspect that in the gnuradio /Linux pulse audio chain is some scale Factor that I don't know.
I will investigate more.

David
IW5BNL


Il Dom 9 Ago 2020, 18:49 Andy G4JNT <andy.g4jnt@...> ha scritto:
I generate .WAV files from first principles.   They use a pure 16 bit integer
Then replay them via teh soundcard of choice.

Unfortunately the wav.play function works with the default soundcard so for this test I will need to temporarily make the Duo my default soundcard !   So allteh Windows bings and chimes will appear on it.

I have plenty of third party software that generates tones and itest signals, but no proof that full scale is actually filling the 16 bit integer in any of them.    
I could write my own, but it's easier to generate .Wav, and I have a whole suite of software to generate all sorts of signals as a wav.

One USB soundcard I use a lot is the PCM2900 USB audio device.   That has one wonderful property - the reference for the A/D appears on a pin of the chip.   So it is possible, by measuring this DC level, to relate the exact voltage that will appear for any 16 bit integer sent, and the same for the sound input.  




On Sun, 9 Aug 2020 at 17:41, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

How you generate the audio USB signal.


David
IW5BNL 

Il Dom 9 Ago 2020, 18:34 Andy G4JNT <andy.g4jnt@...> ha scritto:
I'll do some tests tomorrow.  But I'm pretty certain that when tested before, a peak audio of the full 16 bit word was required to get maximum out of the 0dBm port.  I do remember thinking "how well designed"



On Sun, 9 Aug 2020 at 17:30, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all

I do some meeasuraments.
As I'm on country home I have only a power meter and the capability to measure the tx signal using the monitoring rx of duo.
I generate a two tone signal with gu radio and send it to duo USB.
By monitoring the internal rx I can measure the spectrum of duo rx by a another gnuradio script that capture the 192khz sample rate rx out (available on rx usbvport).
Everithink is perfectly linear until a USB audio signal of 0.1 pk normalized signal.
Increasing over 0.1 we start to see intermodulation.
0.1 is exactly the point where the output power is saturated.
For this measurement the audio input of duo was adjusted to 100% in pulse audio  control.
Of course reducing the pulse audio below 100% the level of saturation increase.
So the asumption that -1/1 is the limit isn't correct as depend on pulse audio level adjustment.
The good news is that I now know the maximum USB audio level before start to intermodulated.
The modulator inside duo working extremely well in the correct range of driving signal linear is perfect.

Ciao

David
IW5BNL 





Il Sab 8 Ago 2020, 22:04 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi all,
Thank for information and answer.
I'm driving a small amp on hf band (60w) so duo power is adjusted on 3 watts out.
I have to make some more accurate measurement but,look like,that changing USB driving from 0.5/-0.5 to 1/-1 peak power don't change.May be issue is on the saturation of my external amplifier.Next days I will made some more accurate meeasuraments.
I encourage elad crew to produce more documentation on it.

David
IW5BNL 

Il Sab 8 Ago 2020, 19:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
Ditto, and even if I want to, I have a perfectly good HF PA, -3dBm in gives 200 Watts out, all boxed up with switched low pass filters and absolutely bomb-proof with mismatched loads (that's been tested, intentionally, several times)



On Sat, 8 Aug 2020 at 18:38, Neil Smith G4DBN <neil@...> wrote:

Good point, I would have bought that version in a heartbeat.  The amp and TX filters are almost zero use to me.  But then I have weird minority interests and don't use the HF bands much at all these days.

Neil G4DBN

On 08/08/2020 18:28, Andy G4JNT wrote:
A full scale signed 16 bit integer, sine wave, (RMS = 23170) generates 0dBm out of the low power port.   Can't speak for teh power amp - never used it.

In fact, what isn't there a lower cost version of thd FDM withuut the 5W amp and all harmonic filters.   Just a 0dBm port on transmit



On Sat, 8 Aug 2020 at 18:20, Neil Smith G4DBN <neil@...> wrote:

Hi David, I found it worked OK with normalised +-1 peak amplitude when using the 0dBm output, but I didn't try it when driving the PA stage.  I was only using FM, so I don't know what the IMD performance was like at higher levels.

It would be interesting to know where you start to see any sign of non-linearity between the level in gnuradio and the actual power output via the PA.  Also, what happens when the power is set lower than MAX.  Does full scale on the USB device input still correspond to full output, or does the USB device level start to flatten out sooner?

Easy enough to test with simple tones (or pairs of tones) I guess.

Neil G4DBN

On 08/08/2020 17:43, DAVID IW5BNL wrote:
Thank you!

David
IW5BNL 



Il Sab 8 Ago 2020, 18:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
If that translates to a full scale 16 bit integer drive to the soundcard then I guess so



On Sat, 8 Aug 2020 at 17:38, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

Thank you for your prompt answer. 
Normalize -1/1 as gnu radio audio output is normalize.
Is it correct?

David




Il Sab 8 Ago 2020, 18:32 Andy G4JNT <andy.g4jnt@...> ha scritto:
Full scale -32768 / 32767  or   0x8000 / 0x7FFF   peak to peak is full power
Nice and simple, easy to remember



On Sat, 8 Aug 2020 at 17:29, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all,

I'm writing a ssb compressor using gnuradio.It work well but I don't really know maximum driving signal of duo USB internal sound card to produce full output power.
I proceed adjusting the input USB audio level little bit lower that the level where the output peak power is saturated ( for saturated output power I mean the maximum level on USB audio that even increasing the audio output power don't increase). Apparently look like a sort of alc is present or simply increasing the level the modulator is saturate. 
Is it somewhere a document and/or information showing correlation between input USB audio and output power (on ssb mode..)

Thank you

David 
IW5BNL

-- 
Neil
http://g4dbn.uk

--
Chris Bartram G4DGU


Re: Maximum usb input driving signal elad duo

DAVID IW5BNL
 

Hi Andy,

I have exactly same result.
This morning I generate a full scale 1500khz tone and the point of output power saturation is achieved adjusting audio level at-7.8 db.
This means that the wait adjustment suggested by elad application note aren't correct.
The right setting for wsjt-x software is:
1) full power setting on wsjt-x software 
2) -7.8db setting in elad duo input.

With this setting the system is perfectly linear and the output RF spectrum match perfectly the spectrum generated by wsjt-x software.

David
IW5BNL 

Il Lun 10 Ago 2020, 13:43 Andy G4JNT <andy.g4jnt@...> ha scritto:
Sri Folks
Looks like a BIG correction is needed.
I've just done a quick test using the pre-recorded .WAV files and it's as you suggested.   Full output from the 0dBm port is achieved with, in my case, about 7 to 8dB back-off from full scale.

Setting the DUO as default soundcard, controlling PTT via the CAT interface and playing the .WAV files using the Windows inbuilt playwav facility I looked at the 0dBm output on a scope.   It needed a file generated at -8dB (0.4 voltage)  of full scale to see a noticeable decrease in amplitude on the scope.   Which is higher than you saw, at I think 0.1, but still lower than I expected.

With the desktop PC, large screen monitor, oscilloscope and stuff on, it was far too hot in the upstairs shack in this weather to spend more than a few minutes doing this so I'll investigate fuller, later.   I can use the .WAV files as calibration, then use TCUBE written by I2PHD o generate a continuously valuable tone in amplitude or frequency.   That will really map-out the ampltude response.

I did generate some two-tone .WAV files, and TCUBE can generate a two tone waveform, so that will be the way forward.    I'm pretty certain TCUBE gives full-scale digits at it's maximum setting, but at least I can now properly calibrate it.




On Mon, 10 Aug 2020 at 06:13, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

As you make some measurement in the next days I suggest to generate a wave file with two tones so you can measure the IM.
I'm also reading the data sheet of duo integrated sound card.
The chip is capable 16 or 24 bits resolutions..
If Franco or someone from Elad can just comment this result and providing us scale Factor (If any...) will very useful.

Regards

David
IW5BNL 


Il Dom 9 Ago 2020, 18:55 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi Andy

Very good idea.
I'm working with Linux.
I suspect that in the gnuradio /Linux pulse audio chain is some scale Factor that I don't know.
I will investigate more.

David
IW5BNL


Il Dom 9 Ago 2020, 18:49 Andy G4JNT <andy.g4jnt@...> ha scritto:
I generate .WAV files from first principles.   They use a pure 16 bit integer
Then replay them via teh soundcard of choice.

Unfortunately the wav.play function works with the default soundcard so for this test I will need to temporarily make the Duo my default soundcard !   So allteh Windows bings and chimes will appear on it.

I have plenty of third party software that generates tones and itest signals, but no proof that full scale is actually filling the 16 bit integer in any of them.    
I could write my own, but it's easier to generate .Wav, and I have a whole suite of software to generate all sorts of signals as a wav.

One USB soundcard I use a lot is the PCM2900 USB audio device.   That has one wonderful property - the reference for the A/D appears on a pin of the chip.   So it is possible, by measuring this DC level, to relate the exact voltage that will appear for any 16 bit integer sent, and the same for the sound input.  




On Sun, 9 Aug 2020 at 17:41, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

How you generate the audio USB signal.


David
IW5BNL 

Il Dom 9 Ago 2020, 18:34 Andy G4JNT <andy.g4jnt@...> ha scritto:
I'll do some tests tomorrow.  But I'm pretty certain that when tested before, a peak audio of the full 16 bit word was required to get maximum out of the 0dBm port.  I do remember thinking "how well designed"



On Sun, 9 Aug 2020 at 17:30, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all

I do some meeasuraments.
As I'm on country home I have only a power meter and the capability to measure the tx signal using the monitoring rx of duo.
I generate a two tone signal with gu radio and send it to duo USB.
By monitoring the internal rx I can measure the spectrum of duo rx by a another gnuradio script that capture the 192khz sample rate rx out (available on rx usbvport).
Everithink is perfectly linear until a USB audio signal of 0.1 pk normalized signal.
Increasing over 0.1 we start to see intermodulation.
0.1 is exactly the point where the output power is saturated.
For this measurement the audio input of duo was adjusted to 100% in pulse audio  control.
Of course reducing the pulse audio below 100% the level of saturation increase.
So the asumption that -1/1 is the limit isn't correct as depend on pulse audio level adjustment.
The good news is that I now know the maximum USB audio level before start to intermodulated.
The modulator inside duo working extremely well in the correct range of driving signal linear is perfect.

Ciao

David
IW5BNL 





Il Sab 8 Ago 2020, 22:04 DAVID IW5BNL via groups.io <martini.david.p1=gmail.com@groups.io> ha scritto:
Hi all,
Thank for information and answer.
I'm driving a small amp on hf band (60w) so duo power is adjusted on 3 watts out.
I have to make some more accurate measurement but,look like,that changing USB driving from 0.5/-0.5 to 1/-1 peak power don't change.May be issue is on the saturation of my external amplifier.Next days I will made some more accurate meeasuraments.
I encourage elad crew to produce more documentation on it.

David
IW5BNL 

Il Sab 8 Ago 2020, 19:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
Ditto, and even if I want to, I have a perfectly good HF PA, -3dBm in gives 200 Watts out, all boxed up with switched low pass filters and absolutely bomb-proof with mismatched loads (that's been tested, intentionally, several times)



On Sat, 8 Aug 2020 at 18:38, Neil Smith G4DBN <neil@...> wrote:

Good point, I would have bought that version in a heartbeat.  The amp and TX filters are almost zero use to me.  But then I have weird minority interests and don't use the HF bands much at all these days.

Neil G4DBN

On 08/08/2020 18:28, Andy G4JNT wrote:
A full scale signed 16 bit integer, sine wave, (RMS = 23170) generates 0dBm out of the low power port.   Can't speak for teh power amp - never used it.

In fact, what isn't there a lower cost version of thd FDM withuut the 5W amp and all harmonic filters.   Just a 0dBm port on transmit



On Sat, 8 Aug 2020 at 18:20, Neil Smith G4DBN <neil@...> wrote:

Hi David, I found it worked OK with normalised +-1 peak amplitude when using the 0dBm output, but I didn't try it when driving the PA stage.  I was only using FM, so I don't know what the IMD performance was like at higher levels.

It would be interesting to know where you start to see any sign of non-linearity between the level in gnuradio and the actual power output via the PA.  Also, what happens when the power is set lower than MAX.  Does full scale on the USB device input still correspond to full output, or does the USB device level start to flatten out sooner?

Easy enough to test with simple tones (or pairs of tones) I guess.

Neil G4DBN

On 08/08/2020 17:43, DAVID IW5BNL wrote:
Thank you!

David
IW5BNL 



Il Sab 8 Ago 2020, 18:40 Andy G4JNT <andy.g4jnt@...> ha scritto:
If that translates to a full scale 16 bit integer drive to the soundcard then I guess so



On Sat, 8 Aug 2020 at 17:38, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi Andy

Thank you for your prompt answer. 
Normalize -1/1 as gnu radio audio output is normalize.
Is it correct?

David




Il Sab 8 Ago 2020, 18:32 Andy G4JNT <andy.g4jnt@...> ha scritto:
Full scale -32768 / 32767  or   0x8000 / 0x7FFF   peak to peak is full power
Nice and simple, easy to remember



On Sat, 8 Aug 2020 at 17:29, DAVID IW5BNL <martini.david.p1@...> wrote:
Hi all,

I'm writing a ssb compressor using gnuradio.It work well but I don't really know maximum driving signal of duo USB internal sound card to produce full output power.
I proceed adjusting the input USB audio level little bit lower that the level where the output peak power is saturated ( for saturated output power I mean the maximum level on USB audio that even increasing the audio output power don't increase). Apparently look like a sort of alc is present or simply increasing the level the modulator is saturate. 
Is it somewhere a document and/or information showing correlation between input USB audio and output power (on ssb mode..)

Thank you

David 
IW5BNL

-- 
Neil
http://g4dbn.uk

1081 - 1100 of 18877