Soundcard based Noise Figure Alignment aid


Andy G4JNT
 

If you've read December's RadCom (Design Notes) you'll have seen my
outline for a soundcard replacement for the G4COM box and request for
someone to write some code to do it.

Even before opening my copy of the magazine (having been away in
Birmingham for the RSGB EGM) Peter G3PLX had written some suitable
software and emailed it to me for testing.    Got back home, built up
a suitable driver, tested, and after a bit of tweaking it all works
FB.

Wrote it up,  and this plus the software can be found at
http://www.g4jnt.com/NoiseMeter.zip

The new design cantake in audio up to 22kHz, so by using both
sidebands from a direct downconverter you have 44kHz of noise
bandwidth to smooth out the reading better than an SSB filter
bandwidth allows.    Also, with a 16 bit soundcard means it can work
over 80dB or so of dynamic range

Enjoy

Andy
www.g4jnt.com


mike.willis@...
 

This is all very well but it still needs a receiver. How about using something like the Funcube Dongle and use the bias output as a switch for the noise source? This could then be used to optimise pre-amps for all bands up to 23cms. It needs software and very little hardware.

 

Mike


--
Scanned by iCritical.



Brian Flynn GM8BJF
 

I thought that was a really neat idea. I just happen to have an HP 346B noise head and no meter. I must try it out when I get some time.

73s
Brian, GM8BJF.

--- In ukmicrowaves@yahoogroups.com, <mike.willis@...> wrote:

This is all very well but it still needs a receiver. How about using something like the Funcube Dongle and use the bias output as a switch for the noise source? This could then be used to optimise pre-amps for all bands up to 23cms. It needs software and very little hardware.

Mike

--
Scanned by iCritical.


Ian White
 

Andy Talbot wrote:
If you've read December's RadCom (Design Notes) you'll have seen my
outline for a soundcard replacement for the G4COM box and request for
someone to write some code to do it.

Even before opening my copy of the magazine (having been away in
Birmingham for the RSGB EGM) Peter G3PLX had written some suitable
software and emailed it to me for testing.    Got back home, built up
a suitable driver, tested, and after a bit of tweaking it all works
FB.

Wrote it up,  and this plus the software can be found at
http://www.g4jnt.com/NoiseMeter.zip

The new design cantake in audio up to 22kHz, so by using both
sidebands from a direct downconverter you have 44kHz of noise
bandwidth to smooth out the reading better than an SSB filter
bandwidth allows.    Also, with a 16 bit soundcard means it can work
over 80dB or so of dynamic range
A very similar design with comprehensive software was published by Dave
Roberts G8KBB in Radcom January 2007. Details of the downconverter and
software are downloadable from
<http://g8kbb.roberts-family-home.co.uk/html/downloads.html> along with
some very detailed assessments of errors and performance.

Dave's software allows both manual control and automatic switching of
the noise source (and if needed, the DUT as well) through either a COM,
LPT or USB port. The automatic switching cycle includes a settling time
to allow the noise source and DUT to stabilise before the measurement
begins, and a delay after the end of each measuring period before the
system changes state. When the settling time is set correctly, this
should also take care of timing uncertainties in the PC ports.

If a 28V switched supply is provided for the noise head, that interface
needs to conform to industry standards: the instrument provides a
switched voltage source and the current regulator is in the noise head.
(I'm not sure what happens when two series-connected current sources are
switched on and off, but it probably isn't good.)

G(M)4PMK's RATS design in Radcom July-August 1995 includes a suitable
hardware interface, an LM317LZ voltage regulator with a MOSFET switch.



--

73 from Ian GM3SEK
http://www.ifwtech.co.uk/g3sek


Andy G4JNT
 

One thing RSGB need to do as part of their revamp  is to place past copies of RadCom on the website - and not just the previous year's.
 
They can't, surely,  expect us to pay for stuff we've already read once and 99/100 times have no further interest in.  
Needless to say I chucked out all magazines before 2008 :-)  Who wants paper copies of stuff now.
Hopefully my Rad Club library cupboard will go back that far
 
Tried to find that design on the website you list, but he only shows circuit diagrams in the native .SCH form - which is most definitely not readable by my EasyPC which what .SCH and .PCB files load into here.   
 
Why authors cannot give circuit diagrams on web sites as .GIF or whatever I just don't know.  They're just asking not to be found or looked at.
 
As for the two current sources in series by the way:  I did that deliberately. When feeding a separate I-source,  the PNP transistor one which is set to a higher current (20mA) just saturates and delivers an on-off switched 28V or whatever.  Teh main reason for doing it like that is, since teh PNP is needed anyway as a switch, all it wanted was another resistor and a zener to give a nice current protected output.
 
My MaCom noise diode does not have its own integral current source, but does have the value of 12.3mA engraved on it which is what I set the separate LM317 source to deliver.    Homebrew ones can probably work just with the PNP source adjusted appropriately.
 
Andy


 

On 25 November 2011 21:49, Ian White GM3SEK <gm3sek@...> wrote:
 

A very similar design with comprehensive software was published by Dave
Roberts G8KBB in Radcom January 2007. Details of the downconverter and
software are downloadable from
<http://g8kbb.roberts-family-home.co.uk/html/downloads.html> along with
some very detailed assessments of errors and performance.

Dave's software allows both manual control and automatic switching of
the noise source (and if needed, the DUT as well) through either a COM,
LPT or USB port. The automatic switching cycle includes a settling time
to allow the noise source and DUT to stabilise before the measurement
begins, and a delay after the end of each measuring period before the
system changes state. When the settling time is set correctly, this
should also take care of timing uncertainties in the PC ports.

If a 28V switched supply is provided for the noise head, that interface
needs to conform to industry standards: the instrument provides a
switched voltage source and the current regulator is in the noise head.
(I'm not sure what happens when two series-connected current sources are
switched on and off, but it probably isn't good.)

G(M)4PMK's RATS design in Radcom July-August 1995 includes a suitable
hardware interface, an LM317LZ voltage regulator with a MOSFET switch.

--

73 from Ian GM3SEK
http://www.ifwtech.co.uk/g3sek



dave powis <g4hup@...>
 

Andy,

Under the 'Noise Meter Hardware' row of Dave Download section there is a folder which contains the .pdf's of the schematics, and also the .sch files - which were generated in Eagle - Freeware version of Eagle downloadable from cadsoft.usa or cadsoft.de.  There are some other files in another dirctory listed as Gerber files - but these are not generated from Eagle, and I think relate to the USB interface, rather than the noise meter RF bits.

73,
Dave


From: Andy Talbot
To: ukmicrowaves@...
Sent: Saturday, 26 November 2011, 11:10
Subject: Re: [ukmicrowaves] Soundcard based Noise Figure Alignment aid

 
One thing RSGB need to do as part of their revamp  is to place past copies of RadCom on the website - and not just the previous year's.
 
They can't, surely,  expect us to pay for stuff we've already read once and 99/100 times have no further interest in.  
Needless to say I chucked out all magazines before 2008 :-)  Who wants paper copies of stuff now.
Hopefully my Rad Club library cupboard will go back that far
 
Tried to find that design on the website you list, but he only shows circuit diagrams in the native .SCH form - which is most definitely not readable by my EasyPC which what .SCH and .PCB files load into here.   
 
Why authors cannot give circuit diagrams on web sites as .GIF or whatever I just don't know.  They're just asking not to be found or looked at.
 
As for the two current sources in series by the way:  I did that deliberately. When feeding a separate I-source,  the PNP transistor one which is set to a higher current (20mA) just saturates and delivers an on-off switched 28V or whatever.  Teh main reason for doing it like that is, since teh PNP is needed anyway as a switch, all it wanted was another resistor and a zener to give a nice current protected output.
 
My MaCom noise diode does not have its own integral current source, but does have the value of 12.3mA engraved on it which is what I set the separate LM317 source to deliver.    Homebrew ones can probably work just with the PNP source adjusted appropriately.
 
Andy


 
On 25 November 2011 21:49, Ian White GM3SEK <gm3sek@...> wrote:
 
A very similar design with comprehensive software was published by Dave
Roberts G8KBB in Radcom January 2007. Details of the downconverter and
software are downloadable from
<http://g8kbb.roberts-family-home.co.uk/html/downloads.html> along with
some very detailed assessments of errors and performance.

Dave's software allows both manual control and automatic switching of
the noise source (and if needed, the DUT as well) through either a COM,
LPT or USB port. The automatic switching cycle includes a settling time
to allow the noise source and DUT to stabilise before the measurement
begins, and a delay after the end of each measuring period before the
system changes state. When the settling time is set correctly, this
should also take care of timing uncertainties in the PC ports.

If a 28V switched supply is provided for the noise head, that interface
needs to conform to industry standards: the instrument provides a
switched voltage source and the current regulator is in the noise head.
(I'm not sure what happens when two series-connected current sources are
switched on and off, but it probably isn't good.)

G(M)4PMK's RATS design in Radcom July-August 1995 includes a suitable
hardware interface, an LM317LZ voltage regulator with a MOSFET switch.

--

73 from Ian GM3SEK
http://www.ifwtech.co.uk/g3sek




Ian White
 

Andy Talbot wrote:


One thing RSGB need to do as part of their revamp  is to place past
copies of RadCom on the website - and not just the previous year's.
 
Strongly agree - the technical articles for several years back used to
be downloadable but those have disappeared.

They can't, surely,  expect us to pay for stuff we've already read once
and 99/100 times have no further interest in.  
Needless to say I chucked out all magazines before 2008 :-)  Who
wants paper copies of stuff now.
Hopefully my Rad Club library cupboard will go back that far
 
If all else fails, I will scan you a copy - you've already paid for it.


Tried to find that design on the website you list, but he only shows
circuit diagrams in the native .SCH form - which is most definitely not
readable by my EasyPC which what .SCH and .PCB files load into
here.   
 
Why authors cannot give circuit diagrams on web sites as .GIF or
whatever I just don't know.  They're just asking not to be found or
looked at.
 
They are there in pdf.

A small correction: G8KBB's 'USB' option is specific to the high
performance Cypress chip that Dave was favouring a few years ago, but
perhaps he may have become reconciled to FTDI by now :-)


As for the two current sources in series by the way:  I did that
deliberately. When feeding a separate I-source,  the PNP transistor one
which is set to a higher current (20mA) just saturates and delivers an
on-off switched 28V or whatever.  Teh main reason for doing it like that 
is, since teh PNP is needed anyway as a switch, all it wanted was
another resistor and a zener to give a nice current protected output.
 
My MaCom noise diode does not have its own integral current source,
but does have the value of 12.3mA engraved on it which is what I set 
the separate LM317 source to deliver.    Homebrew ones can probably
work just with the PNP source adjusted appropriately.
 
Overcurrent protection in the instrument side of the interface is a good
idea, but it isn't good to promote an interface that *appears* to be
standard ("28V") but actually isn't, in ways that other people might not
be aware of.



--

73 from Ian GM3SEK
http://www.ifwtech.co.uk/g3sek