Forums

Antenna Amplifier Noise Figure

Started by rickman June 26, 2015
I read this post in an antenna group and I don't get how this guy is 
coming up with a negative noise figure.  Looks to me like he is 
calculating the noise figure of a resistor, not the amplifier.  Anyone 
care to explain this to me?

The part that seems bogus is this...

 > The negative NF is defined as the amplifier noise being less than the
 > increase in noise due to the amplifier gain.

I thought noise figure was NF = SNRin / SNRout

Rick


> Hello Group, > > Approximately 8 weeks ago fellow group member Steve Ratzlaff and I > had an discussion about the noise performance of amplifiers for Non > Directional Beacon work. Steve suggested that it may be a good idea > to look at the Antenna Amplifier noise floor with a simulated antenna > inductance, rather than using a resistance equal to the amplifier > input z. > > Steve's suggestion makes perfect sense when one considers that the > real and radiation resistance of loop antennas are very small in > comparison to the loops inductance. > > First I retested two active antennas; Wellbrook's ALA100 and > ALA100LN in the paddock using 20m circumference loops. Loop area 21 > sq.m. Both antenna use noiseless feedback. The ALA100 uses Bipolar > transistors, the ALA100LN uses JFETS. > > The gain comparison on MW/LW was an increase of 2-4dB in favour of > the ALA100LN. > > Unfortunately the ambient noise level was too high to see any noise > floor difference. > > Next; Bench Tests ( gain and Noise Figure NF of both ALA100 amps.) > were conducted using a Marconi 2019A Sig. Gen. and WinRadio > Excalibur. A 9dB Norton amplifier in the Antenna Interface was used > to ensure that the total gain was approx. 10dB higher than the > Excalibur NF. Thus to ensure that the higher amplifier gain would > mask the receiver NF affecting the measurements. A 20uH inductor was > used to simulate the loop’s reactance. > > The NF was measured using the gain method i.e. the excess noise above > the amplifier gain when the input is terminated with a resistor. The > negative NF is measured with a inductance connected to the amplifier > input. > > Both ALA100s have an approx. 50 Ohm resistive input z as determined > using an Array Solutions AIM 4170C Antenna Analyser. > > WinRadio Excalibur set 1kHz BW and the S Meter to RMS AVG. > > -144dBm is used as reference for the 1kHz BW > > The negative NF is defined as the amplifier noise being less than the > increase in noise due to the amplifier gain. > > Test results. > > Old ALA100 1MHz gain = 27dB (18dB with Passive Interface ) > > Noise with 50 Ohm input = �116dBm > > NF = 1.0dB > > Noise with 20uH input = �118dBm ( -127dBm with Passive > Interface ) > > NF = �1dB. > > > > New ALA100LN 1MHz gain = 28dB (19dB with Passive Interface ) > > Noise with 50 Ohm input = �116dBm > > NF = 0.0dB > > Noise with 20uH input = �124dBm ( -133dBm with Passive > Interface ) > > NF = �8dB. > > > > 3 other ALA100LNs were tested, the worst case was a -6dB NF. > > Looking at the above one can see that there is only a 1dB difference > in the gain and NF of the two Head amps with a 50 Ohm input. > > The apparent -1dB NF of the ALA100 with the 20uH input is probably > due to the fact that Inductors are noiseless compared to a 50 Ohm > resistor. > > However, what is striking, is the apparent high negative NF of the > ALA100LN. My first thoughts were when this showed up, that the gain > had dropped by 8dB because of the loop inductance and this would > show up as low antenna gain. However, the on air comparison suggested > that this wasn't the case. Also Dave Aichelman ran some comparative > tests confirming my observations. > > I do not have an explanation yet as to why the ALA100LN has an > apparent negative NF or why the noise floor drops by 8dB, but it > certainly accounts as to why some users are seeing improved > performance. The theoretical NF of the ALA100LN is close to 0dB. > > One could speculate that there is a significant mismatch of the > antenna noise from its real and radiation resistance being a small > fraction of the amplifier input z. However, I have only been able to > achieve a significant negative NF using JFETS. > > The ALA100LN uses 8 x J309 Fets. with noiseless feedback. > > The negative NF can be increased to approx. -10dB by altering the > feedback and increasing the amplifier gain. Also the negative NF > increases by a couple of dB where the loop inductance is higher e.g. > a 20m circumference thin wire loop ( 30uH ). Also the negative NF is > maintained as the frequency rises. > > The ALA100LN has a sister, “Florence” the FLG100LN with an input > matching/isolation transformer of 1152 Ohms for Terminated loop > antennas e.g. Flag and KAZ. However, the antenna termination > resistance makes the negative NF feature unrealisable. > > One may ask what is the significance of the amplifier having a > negative noise figure of -8dB? > > It basically means, that there is the potential of an 11dB s/n > improvement when compared to a conventional amplifier of the same > gain and NF of say 3dB with a similar loop size. This feature could > be used to compensate for any mismatch loss over wide bandwidths. > > Additionally the negative noise figure could compensate for usual NF > degradation when loops are run as close spaced Phase Arrays. > > I am currently exploring to see if the apparent negative NF can be > exploited with smaller loops. To this end I have extended the > bandwidth of the ALA100LN from 10MHz to 30MHz by configuring the FETs > to run in cascode with two Bipolar devices. However, there may be a > bandwidth limitation due to the fixed input impedance of the > amplifier compared to the ALA1530 with its loop/amp. impedance > tracking verses frequency.
-- Rick
On 6/26/2015 7:24 AM, rickman wrote:
> I read this post in an antenna group and I don't get how this guy is > coming up with a negative noise figure. Looks to me like he is > calculating the noise figure of a resistor, not the amplifier. Anyone > care to explain this to me? > > The part that seems bogus is this... > > > The negative NF is defined as the amplifier noise being less than the > > increase in noise due to the amplifier gain. > > I thought noise figure was NF = SNRin / SNRout > > Rick >
I once ask Ratzlaff about the FSL antenna, in his comments he said, "high Q or low Q has little relation to how the FSL will actually perform for receiving weak signals.", but then went on to say, "Of course using Litz is the only type of wire to use." When I ask him about why Litz is important if Q isn't, he got pissed off and wrote back, "When you make sarcastic replies to the info I passed along, and question what I say, then I write you off as just a tire-kicker, not interested in possibly learning something, and I have no interest in saying anything more." Seems to me he said something incorrect and didn't like having someone ask him to clarify it. I'll leave it to you to figure out which part was incorrect. I thought Q would be important, an FSL is a ferrite loaded tuned loop. With magic mixed in! ;-) If Q is not important, why use Litz? I suspect you might have already tried to ask him, and now he's mad at you too! Mikek
> >> Hello Group, >> >> Approximately 8 weeks ago fellow group member Steve Ratzlaff and I >> had an discussion about the noise performance of amplifiers for Non >> Directional Beacon work. Steve suggested that it may be a good idea >> to look at the Antenna Amplifier noise floor with a simulated antenna >> inductance, rather than using a resistance equal to the amplifier >> input z. >> >> Steve's suggestion makes perfect sense when one considers that the >> real and radiation resistance of loop antennas are very small in >> comparison to the loops inductance. >> >> First I retested two active antennas; Wellbrook's ALA100 and >> ALA100LN in the paddock using 20m circumference loops. Loop area 21 >> sq.m. Both antenna use noiseless feedback. The ALA100 uses Bipolar >> transistors, the ALA100LN uses JFETS. >> >> The gain comparison on MW/LW was an increase of 2-4dB in favour of >> the ALA100LN. >> >> Unfortunately the ambient noise level was too high to see any noise >> floor difference. >> >> Next; Bench Tests ( gain and Noise Figure NF of both ALA100 amps.) >> were conducted using a Marconi 2019A Sig. Gen. and WinRadio >> Excalibur. A 9dB Norton amplifier in the Antenna Interface was used >> to ensure that the total gain was approx. 10dB higher than the >> Excalibur NF. Thus to ensure that the higher amplifier gain would >> mask the receiver NF affecting the measurements. A 20uH inductor was >> used to simulate the loop’s reactance. >> >> The NF was measured using the gain method i.e. the excess noise above >> the amplifier gain when the input is terminated with a resistor. The >> negative NF is measured with a inductance connected to the amplifier >> input. >> >> Both ALA100s have an approx. 50 Ohm resistive input z as determined >> using an Array Solutions AIM 4170C Antenna Analyser. >> >> WinRadio Excalibur set 1kHz BW and the S Meter to RMS AVG. >> >> -144dBm is used as reference for the 1kHz BW >> >> The negative NF is defined as the amplifier noise being less than the >> increase in noise due to the amplifier gain. >> >> Test results. >> >> Old ALA100 1MHz gain = 27dB (18dB with Passive Interface ) >> >> Noise with 50 Ohm input = �116dBm >> >> NF = 1.0dB >> >> Noise with 20uH input = �118dBm ( -127dBm with Passive >> Interface ) >> >> NF = �1dB. > >> >> >> New ALA100LN 1MHz gain = 28dB (19dB with Passive Interface ) >> >> Noise with 50 Ohm input = �116dBm >> >> NF = 0.0dB >> >> Noise with 20uH input = �124dBm ( -133dBm with Passive >> Interface ) >> >> NF = �8dB. > >> >> >> 3 other ALA100LNs were tested, the worst case was a -6dB NF. >> >> Looking at the above one can see that there is only a 1dB difference >> in the gain and NF of the two Head amps with a 50 Ohm input. >> >> The apparent -1dB NF of the ALA100 with the 20uH input is probably >> due to the fact that Inductors are noiseless compared to a 50 Ohm >> resistor. >> >> However, what is striking, is the apparent high negative NF of the >> ALA100LN. My first thoughts were when this showed up, that the gain >> had dropped by 8dB because of the loop inductance and this would >> show up as low antenna gain. However, the on air comparison suggested >> that this wasn't the case. Also Dave Aichelman ran some comparative >> tests confirming my observations. >> >> I do not have an explanation yet as to why the ALA100LN has an >> apparent negative NF or why the noise floor drops by 8dB, but it >> certainly accounts as to why some users are seeing improved >> performance. The theoretical NF of the ALA100LN is close to 0dB. >> >> One could speculate that there is a significant mismatch of the >> antenna noise from its real and radiation resistance being a small >> fraction of the amplifier input z. However, I have only been able to >> achieve a significant negative NF using JFETS. >> >> The ALA100LN uses 8 x J309 Fets. with noiseless feedback. >> >> The negative NF can be increased to approx. -10dB by altering the >> feedback and increasing the amplifier gain. Also the negative NF >> increases by a couple of dB where the loop inductance is higher e.g. >> a 20m circumference thin wire loop ( 30uH ). Also the negative NF is >> maintained as the frequency rises. >> >> The ALA100LN has a sister, “Florence” the FLG100LN with an input >> matching/isolation transformer of 1152 Ohms for Terminated loop >> antennas e.g. Flag and KAZ. However, the antenna termination >> resistance makes the negative NF feature unrealisable. >> >> One may ask what is the significance of the amplifier having a >> negative noise figure of -8dB? >> >> It basically means, that there is the potential of an 11dB s/n >> improvement when compared to a conventional amplifier of the same >> gain and NF of say 3dB with a similar loop size. This feature could >> be used to compensate for any mismatch loss over wide bandwidths. >> >> Additionally the negative noise figure could compensate for usual NF >> degradation when loops are run as close spaced Phase Arrays. >> >> I am currently exploring to see if the apparent negative NF can be >> exploited with smaller loops. To this end I have extended the >> bandwidth of the ALA100LN from 10MHz to 30MHz by configuring the FETs >> to run in cascode with two Bipolar devices. However, there may be a >> bandwidth limitation due to the fixed input impedance of the >> amplifier compared to the ALA1530 with its loop/amp. impedance >> tracking verses frequency. > >
--- This email has been checked for viruses by Avast antivirus software. http://www.avast.com
On Fri, 26 Jun 2015 08:24:25 -0400, rickman <gnuarm@gmail.com> wrote:

>I read this post in an antenna group and I don't get how this guy is >coming up with a negative noise figure. Looks to me like he is >calculating the noise figure of a resistor, not the amplifier. Anyone >care to explain this to me? > >The part that seems bogus is this... > > > The negative NF is defined as the amplifier noise being less than the > > increase in noise due to the amplifier gain. > >I thought noise figure was NF = SNRin / SNRout > >Rick > > >> Hello Group, >> >> Approximately 8 weeks ago fellow group member Steve Ratzlaff and I >> had an discussion about the noise performance of amplifiers for Non >> Directional Beacon work. Steve suggested that it may be a good idea >> to look at the Antenna Amplifier noise floor with a simulated antenna >> inductance, rather than using a resistance equal to the amplifier >> input z. >> >> Steve's suggestion makes perfect sense when one considers that the >> real and radiation resistance of loop antennas are very small in >> comparison to the loops inductance. >> >> First I retested two active antennas; Wellbrook's ALA100 and >> ALA100LN in the paddock using 20m circumference loops. Loop area 21 >> sq.m. Both antenna use noiseless feedback. The ALA100 uses Bipolar >> transistors, the ALA100LN uses JFETS. >> >> The gain comparison on MW/LW was an increase of 2-4dB in favour of >> the ALA100LN. >> >> Unfortunately the ambient noise level was too high to see any noise >> floor difference. >> >> Next; Bench Tests ( gain and Noise Figure NF of both ALA100 amps.) >> were conducted using a Marconi 2019A Sig. Gen. and WinRadio >> Excalibur. A 9dB Norton amplifier in the Antenna Interface was used >> to ensure that the total gain was approx. 10dB higher than the >> Excalibur NF. Thus to ensure that the higher amplifier gain would >> mask the receiver NF affecting the measurements. A 20uH inductor was >> used to simulate the loop&rsquo;s reactance. >> >> The NF was measured using the gain method i.e. the excess noise above >> the amplifier gain when the input is terminated with a resistor. The >> negative NF is measured with a inductance connected to the amplifier >> input. >> >> Both ALA100s have an approx. 50 Ohm resistive input z as determined >> using an Array Solutions AIM 4170C Antenna Analyser. >> >> WinRadio Excalibur set 1kHz BW and the S Meter to RMS AVG. >> >> -144dBm is used as reference for the 1kHz BW
That is 174 dBm/Hz, indicating about 300 K noise temperature (room temperature). UHF people would call that 3 dB reference level and it is not so hard to go down to 100 K (1 dB NF) in a good preamplifier. If the noise level is below 3 dB, someone would call it a negative NF:-). Anyway, at VLF/LF the band noise is so huge, that I do not understand what a low NF would help, unless the antenna is extremely small and lossy, such a ferrite bar with -60 to -80 dB antenna efficiency and hence similar gain. A several meter in diameter loop will have a half decent efficiency, so extremely low amplifier NF is seldom justified. Of course, putting the amplifier at the antenna will help keeping interference from entering the downlead.
On Fri, 26 Jun 2015 08:24:25 -0400, rickman wrote:

> I read this post in an antenna group and I don't get how this guy is > coming up with a negative noise figure. Looks to me like he is > calculating the noise figure of a resistor, not the amplifier. Anyone > care to explain this to me? > > The part that seems bogus is this... > > > The negative NF is defined as the amplifier noise being less than the > > increase in noise due to the amplifier gain. > > I thought noise figure was NF = SNRin / SNRout
Yes, it is. NF = 0dB is about as low as you can go without magic. -- Tim Wescott Wescott Design Services http://www.wescottdesign.com
On 6/26/2015 12:38 PM, amdx wrote:
> On 6/26/2015 7:24 AM, rickman wrote: >> I read this post in an antenna group and I don't get how this guy is >> coming up with a negative noise figure. Looks to me like he is >> calculating the noise figure of a resistor, not the amplifier. Anyone >> care to explain this to me? >> >> The part that seems bogus is this... >> >> > The negative NF is defined as the amplifier noise being less than the >> > increase in noise due to the amplifier gain. >> >> I thought noise figure was NF = SNRin / SNRout >> >> Rick >> > I once ask Ratzlaff about the FSL antenna, in his comments he said, > "high Q or low Q has little relation to how the FSL will actually > perform for receiving weak signals.", but then went on to say, "Of > course using Litz is the only type of wire to use." > When I ask him about why Litz is important if Q isn't, he got pissed > off and wrote back, "When you make sarcastic replies to the info I > passed along, and question what I say, then I write you off as just a > tire-kicker, not interested in possibly learning something, and I have > no interest in saying anything more." > Seems to me he said something incorrect and didn't like having someone > ask him to clarify it. I'll leave it to you to figure out which part was > incorrect. > I thought Q would be important, an FSL is a ferrite loaded tuned loop. > With magic mixed in! ;-) > If Q is not important, why use Litz? > I suspect you might have already tried to ask him, and now he's mad at > you too! > Mikek
I suspect he said he didn't like your sarcastic replies because you made sarcastic replies much more than he didn't want to discuss anything he said. -- Rick
amdx wrote:
> On 6/26/2015 7:24 AM, rickman wrote: >> I read this post in an antenna group and I don't get how this guy is >> coming up with a negative noise figure. Looks to me like he is >> calculating the noise figure of a resistor, not the amplifier. Anyone >> care to explain this to me? >> >> The part that seems bogus is this... >> >> > The negative NF is defined as the amplifier noise being less than the >> > increase in noise due to the amplifier gain. >> >> I thought noise figure was NF = SNRin / SNRout >> >> Rick >> > I once ask Ratzlaff about the FSL antenna, in his comments he said, > "high Q or low Q has little relation to how the FSL will actually > perform for receiving weak signals.", but then went on to say, "Of > course using Litz is the only type of wire to use." > When I ask him about why Litz is important if Q isn't, he got pissed off > and wrote back, "When you make sarcastic replies to the info I passed > along, and question what I say, then I write you off as just a > tire-kicker, not interested in possibly learning something, and I have > no interest in saying anything more." > Seems to me he said something incorrect and didn't like having someone > ask him to clarify it. I'll leave it to you to figure out which part was > incorrect. > I thought Q would be important, an FSL is a ferrite loaded tuned loop. > With magic mixed in! ;-) > If Q is not important, why use Litz? > I suspect you might have already tried to ask him, and now he's mad at > you too! > Mikek > > >
** SNIPPED for brevity ** Even at DC frequencies (eg: broadcast FM or even AM) Litz wire gives an edge for higher Q; that is a proven fact. BUT.... If you do not give a rat about Q or gain losses or noise figure (again, at DC frequencies), then bag Litz and pay the price on all of them....
On 26/06/2015 13:24, rickman wrote:
> I read this post in an antenna group and I don't get how this guy is > coming up with a negative noise figure. Looks to me like he is > calculating the noise figure of a resistor, not the amplifier. Anyone > care to explain this to me? > > The part that seems bogus is this... > > > The negative NF is defined as the amplifier noise being less than the > > increase in noise due to the amplifier gain. > > I thought noise figure was NF = SNRin / SNRout > > Rick >
Both definitions are correct and mean the same thing; a negative NF, when expressed in dB, would be when the SNRout is less than the SNRin. However, the big but is that an negative NF is not possible. It only appears to be the case due to the fact that the OP is not comparing like with like, the test method used is only valid if the system impedance remains the same. You cannot compare oranges with lemons. Jeff
In message <mmllkh$bgl$1@speranza.aioe.org>, Jeff <jeff@ukra.com> writes
>On 26/06/2015 13:24, rickman wrote: >> I read this post in an antenna group and I don't get how this guy is >> coming up with a negative noise figure. Looks to me like he is >> calculating the noise figure of a resistor, not the amplifier. Anyone >> care to explain this to me? >> >> The part that seems bogus is this... >> >> > The negative NF is defined as the amplifier noise being less than the >> > increase in noise due to the amplifier gain. >> >> I thought noise figure was NF = SNRin / SNRout >> >> Rick >> > >Both definitions are correct and mean the same thing; a negative NF, >when expressed in dB, would be when the SNRout is less than the SNRin. >However, the big but is that an negative NF is not possible. > >It only appears to be the case due to the fact that the OP is not >comparing like with like, the test method used is only valid if the >system impedance remains the same. You cannot compare oranges with >lemons. >
Is that formula correct? If the input SNR is poor, an amplifier with a high NF has very impact on the output SNR. Also, are the units ratios, or are they in dB?
>
-- Ian
On Fri, 26 Jun 2015 23:04:43 -0700, Robert Baer
<robertbaer@localnet.com> wrote:

>amdx wrote: >> On 6/26/2015 7:24 AM, rickman wrote: >>> I read this post in an antenna group and I don't get how this guy is >>> coming up with a negative noise figure. Looks to me like he is >>> calculating the noise figure of a resistor, not the amplifier. Anyone >>> care to explain this to me? >>> >>> The part that seems bogus is this... >>> >>> > The negative NF is defined as the amplifier noise being less than the >>> > increase in noise due to the amplifier gain. >>> >>> I thought noise figure was NF = SNRin / SNRout >>> >>> Rick >>> >> I once ask Ratzlaff about the FSL antenna, in his comments he said, >> "high Q or low Q has little relation to how the FSL will actually >> perform for receiving weak signals.", but then went on to say, "Of >> course using Litz is the only type of wire to use." >> When I ask him about why Litz is important if Q isn't, he got pissed off >> and wrote back, "When you make sarcastic replies to the info I passed >> along, and question what I say, then I write you off as just a >> tire-kicker, not interested in possibly learning something, and I have >> no interest in saying anything more." >> Seems to me he said something incorrect and didn't like having someone >> ask him to clarify it. I'll leave it to you to figure out which part was >> incorrect. >> I thought Q would be important, an FSL is a ferrite loaded tuned loop. >> With magic mixed in! ;-) >> If Q is not important, why use Litz? >> I suspect you might have already tried to ask him, and now he's mad at >> you too! >> Mikek >> >> >> >** SNIPPED for brevity ** > Even at DC frequencies (eg: broadcast FM or even AM) Litz wire gives >an edge for higher Q; that is a proven fact. > BUT.... > If you do not give a rat about Q or gain losses or noise figure >(again, at DC frequencies), then bag Litz and pay the price on all of >them....
Litz is useful at VLF/LF/MF due to the high inductance and hence large number of turns required. On a small coil former, the requirement is to use the available copper cross section as effectively as possible. At high HF and VHF, fitting a few turns into a coil former is required, so using 0.5 mm solid thick wire to contain the RF within the skin dept is not a problem. For a single turn loop antenna, there are no such size restrictions. A 1-2 m loop antenna made of 10-20 mm copper tubing works quite well as a transmitting antenna at 3.5 MHz with perhaps 1 % efficiency (-20 dB gain). I once made a loop antenna into a hoola hoop using a flat cable using "off by one" connection to create a multiturn loop. The problem was that the self capacitance between turns made it tunable only at VLF frequencies.
On 6/27/2015 1:04 AM, Robert Baer wrote:
> amdx wrote: >> On 6/26/2015 7:24 AM, rickman wrote: >>> I read this post in an antenna group and I don't get how this guy is >>> coming up with a negative noise figure. Looks to me like he is >>> calculating the noise figure of a resistor, not the amplifier. Anyone >>> care to explain this to me? >>> >>> The part that seems bogus is this... >>> >>> > The negative NF is defined as the amplifier noise being less than the >>> > increase in noise due to the amplifier gain. >>> >>> I thought noise figure was NF = SNRin / SNRout >>> >>> Rick >>> >> I once ask Ratzlaff about the FSL antenna, in his comments he said, >> "high Q or low Q has little relation to how the FSL will actually >> perform for receiving weak signals.", but then went on to say, "Of >> course using Litz is the only type of wire to use." >> When I ask him about why Litz is important if Q isn't, he got pissed off >> and wrote back, "When you make sarcastic replies to the info I passed >> along, and question what I say, then I write you off as just a >> tire-kicker, not interested in possibly learning something, and I have >> no interest in saying anything more." >> Seems to me he said something incorrect and didn't like having someone >> ask him to clarify it. I'll leave it to you to figure out which part was >> incorrect. >> I thought Q would be important, an FSL is a ferrite loaded tuned loop. >> With magic mixed in! ;-) >> If Q is not important, why use Litz? >> I suspect you might have already tried to ask him, and now he's mad at >> you too! >> Mikek >> >> >> > ** SNIPPED for brevity ** > Even at DC frequencies (eg: broadcast FM or even AM) Litz wire gives > an edge for higher Q; that is a proven fact. > BUT.... > If you do not give a rat about Q or gain losses or noise figure > (again, at DC frequencies), then bag Litz and pay the price on all of > them....
Yes, his comment made no sense to me. So I ask about it, that was a mistake! Mikek