EMRFD Message Archive 12237

Message Date From Subject
12237 2016-01-26 00:44:50 Ashhar Farhan electret biasing
we use the electret mics all the time with varying results. finally i got it all sorted out.
watch this :

- f
12248 2016-01-31 05:49:14 Roelof Bakker Noise figure measurement
Hello all,

I have build a 1 to 4 distribution amplifier to be used between an
active antenna and up to four receivers. The amplifier is a 11 dB
gain balanced Norton Amplifier with two BFG591's, followed by a
Mini-Circuit 4-port splitter.

To evaluate the noise figure of this amplifier, I have used a Rigol
DSA-815-TG spectrum analyser. With the measurement kit installed,
one can measure the channel power. The channel (noise) power is
displayed in dBm and also in dBm/Hz.

With the analyser's pre-amplifier on the noise floor is about -158
dBm/Hz. The noise figure of the balanced Norton amplifier is < 3 dB
or < -171 dBm/Hz. With 11 dB gain the level is < -160 dBm/Hz at the
output of the amplifier. So, it can not be measured directly.

I have build a "measurement amplifier" consisting of two 15 dB gain
stages in series of the 2N3904 amplifier from page 2.20, fig 2.57
from the book. Using the Rigol, the noise figure was found to be 7.7
dB.

The noise figure of the Norton amplifier could be found by cascading
it with the measurement amplifier. Knowing the noise figure of the
cascade the noise figure of the first amplifier can be calculated.
It is 1.6 dB.

The amplifier / power splitter has a bandwidth of 50 kHz - 80 MHz.
The noise figure was measured at 5 MHz in a 2 MHz channel.

This is a simple and straight forward way of measuring the noise
figure. A calibrated noise source is not needed.

I wonder how good the accuracy of the Rigol is for this kind of
measurements. I have searched the internet, but nothing turned up.

So, I wonder if anyone on this list has used it for this purpose and
can say something about the accuracy.

Thank you in advance.

73,
Roelof Bakker, pa0rdt
12250 2016-01-31 09:12:47 Thomas S. Knutsen Re: Noise figure measurement
Hello Roelof. 

Unfortunatley there are a large error in your reasoning, but to get there, we need to extend on the noise figure term a bit with noise temprature. Observe that for the following, all tempratures will be in Kelvin. 

All noise figures are measured by the principle of how a given receiver (amplifier) behave with regard to a temprature change over 0K (absolute zero).  For a given noise factor (Fn, linear not dB) we can find the equivalent input temprature of the receiver by Te = (Fn-1)*T0 where T0 is the ambivalent temprature, and Te the receiver temprature that produces the same input noise as the receiver. 
So for a 7.7dB amplifier (Fn = 5.9) at room temprature (290K): Te = (5.9 -1)*290 = 1421K. This is the temprature as you would need to heat a resistor to in order to get the same amount of noise. 

From this, we get that the noise factor Fn = (Te/T0) +1 or in the more familiar noise figure NF = 10*log((Te/T0) +1). The problem is then reduced to solving for a number. The problem with the approach you have taken, is that Te = T0, and that makes the uncertainty go to infinity. The whole uncertainty calculation would be too large to take in this e-mail. See the link at the bottom of the e-mail. 

The Y factor usualy used in the measurement of noise figure can be used by the same method where Y = Thot / Tcold. 
a couple 50 ohm resistors, one in boiling water, the other in melting ice to get a temprature differential. The nice thing about this is that the accuracy is only depending on the accuracy  in temprature measurement. The bad thing is that if the receiver is not at Tcold, you will need to do some more math to get the true noise factor.  

The nice thing about having the dBm/Hz function on the spectrum analyzer is that it should be compensated for true noise bandwith. This means we can determine Thot or the ENR of a homebrew noise source, with the same accuracy as the amplitude readout. The only thing is that we would need to know the NF of the analyzer. The simple way around this, is to add a pre-amplifier with known noise figure and gain. Getting a pre-amp NF measured at a VHF meeting or other OM's shack is usualy not too hard.  
Then the ENR can be determined by: ENR = (Ton - Toff) / T0.  You would need to convert from dB into temperature and back, and compensate for cascaded Noise figure. This requires some measurements, and math, but its not too hard to do. 
Make sure that the reading you get off the spectrum analyzer is 10dB or more above the noise floor of the analyzer to avoid problems with the linearity-in-dB of the log amplifiers, altough the digital DSP in the Rigol and other modern analyzers are quite good at this linearity compared to the older analyzers.

Noise figure measurements are difficult to do accurate. While the temperature measurements that it all boils down to is not difficult to do, there are both cascades of noise figure inaccuracy and there are the temprature difference between the devices that changes the accuracy. When doing measurements near 0 dB, you fast approach the accuracy of the noise source, and usualy cryogenic or celestial noise are used as the Tcold reference.  Its a fast spiral where you end up with liquid nitrogen, and hotter than sun semiconductor sources.  

The best reference on NF measurement I know is this app note by HP: http://cp.literature.agilent.com/litweb/pdf/5952-3706E.pdf 

73 es GL de Thomas LA3PNA.


2016-01-31 14:49 GMT+01:00 'Roelof Bakker' roelof@ndb.demon.nl [emrfd] <emrfd@yahoogroups.com>:
 

Hello all,

I have build a 1 to 4 distribution amplifier to be used between an
active antenna and up to four receivers. The amplifier is a 11 dB
gain balanced Norton Amplifier with two BFG591's, followed by a
Mini-Circuit 4-port splitter.

To evaluate the noise figure of this amplifier, I have used a Rigol
DSA-815-TG spectrum analyser. With the measurement kit installed,
one can measure the channel power. The channel (noise) power is
displayed in dBm and also in dBm/Hz.

With the analyser's pre-amplifier on the noise floor is about -158
dBm/Hz. The noise figure of the balanced Norton amplifier is < 3 dB
or < -171 dBm/Hz. With 11 dB gain the level is < -160 dBm/Hz at the
output of the amplifier. So, it can not be measured directly.

I have build a "measurement amplifier" consisting of two 15 dB gain
stages in series of the 2N3904 amplifier from page 2.20, fig 2.57
from the book. Using the Rigol, the noise figure was found to be 7.7
dB.

The noise figure of the Norton amplifier could be found by cascading
it with the measurement amplifier. Knowing the noise figure of the
cascade the noise figure of the first amplifier can be calculated.
It is 1.6 dB.

The amplifier / power splitter has a bandwidth of 50 kHz - 80 MHz.
The noise figure was measured at 5 MHz in a 2 MHz channel.

This is a simple and straight forward way of measuring the noise
figure. A calibrated noise source is not needed.

I wonder how good the accuracy of the Rigol is for this kind of
measurements. I have searched the internet, but nothing turned up.

So, I wonder if anyone on this list has used it for this purpose and
can say something about the accuracy.

Thank you in advance.

73,
Roelof Bakker, pa0rdt

12251 2016-01-31 13:35:00 kerrypwr Re: electret biasing
G'day Roelof.

I have used my Rigol '815 to measure noise figures using the method described in HP AN-150-9 which seems to be the method you used.

http://www.hpmemoryproject.org/an/pdf/an_150-9.pdf

The results were believable; this is an extract from my notes on the experiments;


The method used was the “direct” method (HP AN 150-9) using a signal generator and a spectrum analyser; it is not particularly accurate but it is simple to do and provides a useful comparison.

 

Noise figures at 10 MHz for the three devices in the test amplifier were measured as follows;

 

2N3866; 6.6 dB;

 

2N5109; 7.2 dB and

 

2N3053; 6.8 dB.

 

Given the limitations of the test method, no important conclusions can be drawn from these figures except to say that the noise figure of all three is in an acceptable range for a post-mixer amplifier.



I used a home-made amplifier in the chain; it has two MMICs, over 40dB of gain and a NF of about 3dB.


Kerry VK2TIL.







12252 2016-01-31 14:36:56 Glen Torr Re: electret biasing
Hi Kerry,

Thanks for the reference to the HP Ap. Note.

Cheers,

Glen VK1FB
12253 2016-01-31 18:46:51 kerrypwr Re: electret biasing
Glad to help, Glen.

It's been a while since I played-around with NF measurements; just looking over my files and remembered that I had this excellent paper;

esl.eng.ohio-state.edu/~rstheory/iip/nfmeas.pdf


It largely follows the HP app note but is slightly different if you study it.

I also wrote some notes on the method, mainly so that I could remember how it's done.  :)

PM me if you want it; I am good on QRZ.

Kerry VK2TIL.
12254 2016-02-01 00:52:08 Ephemeral Re: Noise figure measurement

Thermionic noise diodes are readily available cheaply on Ebay. They have the advantage their calibration simply depends on the electron charge and anode current. You will need an HT and LT power supply. Google for details.



---- 'Roelof Bakker' roelof@ndb.demon.nl [emrfd] wrote ----

 

Hello all,

I have build a 1 to 4 distribution amplifier to be used between an
active antenna and up to four receivers. The amplifier is a 11 dB
gain balanced Norton Amplifier with two BFG591's, followed by a
Mini-Circuit 4-port splitter.

To evaluate the noise figure of this amplifier, I have used a Rigol
DSA-815-TG spectrum analyser. With the measurement kit installed,
one can measure the channel power. The channel (noise) power is
displayed in dBm and also in dBm/Hz.

With the analyser's pre-amplifier on the noise floor is about -158
dBm/Hz. The noise figure of the balanced Norton amplifier is < 3 dB
or < -171 dBm/Hz. With 11 dB gain the level is < -160 dBm/Hz at the
output of the amplifier. So, it can not be measured directly.

I have build a "measurement amplifier" consisting of two 15 dB gain
stages in series of the 2N3904 amplifier from page 2.20, fig 2.57
from the book. Using the Rigol, the noise figure was found to be 7.7
dB.

The noise figure of the Norton amplifier could be found by cascading
it with the measurement amplifier. Knowing the noise figure of the
cascade the noise figure of the first amplifier can be calculated.
It is 1.6 dB.

The amplifier / power splitter has a bandwidth of 50 kHz - 80 MHz.
The noise figure was measured at 5 MHz in a 2 MHz channel.

This is a simple and straight forward way of measuring the noise
figure. A calibrated noise source is not needed.

I wonder how good the accuracy of the Rigol is for this kind of
measurements. I have searched the internet, but nothing turned up.

So, I wonder if anyone on this list has used it for this purpose and
can say something about the accuracy.

Thank you in advance.

73,
Roelof Bakker, pa0rdt

12255 2016-02-01 08:01:54 Roelof Bakker Re: Noise figure measurement
Hello Thomas,

Thank you for your thorough advice and the link to the HP
application note. Interesting reading for sure.

I only like to have a coarse indication of the NF of amplifiers at
VLF / LF/ MW / HF, nothing overly fancy!

73,
Roelof, pa0rdt
12256 2016-02-01 15:46:23 Rod KM6SN need crystal filter info
Hi All,

I have a few crystal filters marked

CF Networks, Inc.
Phoenix, Arizona
Quartz Filter Network
MODEL NO. 10.9-2.8
6830-82

and I need input/output impedance information for matching purposes.

I would appreciate any information on these specific filters.

Thanks

Rod KM6SN
12257 2016-02-01 16:56:25 rodwall1234 Re: need crystal filter info

Hi Rod,

If you are not able to find your filter information. You might know someone who has a VNA. And then use rfsim99 as in this video.
https://m.youtube.com/watch?v=qnfd-jrSxnA

Regards,

Roderick Wall, vl3yc.

12258 2016-02-02 04:23:18 John Lawson Re: need crystal filter info
Ron, see if the company is still in business. That is what I did when I picked up some Network Sciences Crystal Filters from a Surplus Store in Phoenix while on vacation many years ago. The Engineer who I talked with over the phone was more that happy to provide the needed information. John K5IRK 

Sent from my iPad

12261 2016-02-02 12:49:22 Sandeep Lohia Re: electret biasing
All my ECM mic, thousands of them works even if I swap two pins with
each other,

never understood why it gives more gain? if - ECM body pin connected
to that of + output if IP phones, eg Grand steam, Yelaink, Avaya etc.

connecting + to + also works, connecting - to plus also works, don't know why ?

J-fets also works on wrong polarity?
but never experienced same with home brewed BJT amp, eg BITX.